AU4556501A - Phytomics: a genomic-based approach to herbal compositions - Google Patents

Description

WO 01/66803 PCT/USO1/07608 PHYTOMICS: A GENOMIC-BASED APPROACH TO HERBAL COMPOSITIONS FIELD OF THE INVENTION This invention relates to herbal compositions. More specifically, this invention 5 provides tools and methodologies for improving the selection, testing, quality control and manufacture of herbal compositions, and to help guide the development of new herbal compositions and identify novel uses of existing herbal compositions. BACKGROUND OF THE INVENTION 10 All publications and patent applications herein are incorporated by reference to the same extent as if each individual publication or patent application was specifically and individually indicated to be ihicorporated by reference. Herbal medicine has been in use for centuries by people of Asia and Europe. In the United States (US), herbs have become commercially valuable in the dietary supplement 15 industry as well as in holistic Medicine. Approximately one third of the US population has tried some form of alternative medicine at least once (Eisenberg et al., 1993, N. Engl. J. Med. 328:246-252). Botanicals, including hei-bs, have also become a focal point for the identification of new active agents to treat diseases. Active compounds, derived from plant extracts, are of 20 continuing interest to the yhemaceutical industry. For example, taxol is an antineoplastic drug obtained from the bafk of the western yew tree. It is estimated that approximately 50 percent of the thousands of drugs commonly used and prescribed today are either derived from a plant source or cofitaih chemical'imitatirns of a plant compound (Mindell, E.R., 1992, Earl Mindell's Herb Bible, Fireside Book). 25 Currently, a number of medicinal formulations, food supplements, dietary supplements and the likeoialin herbal components or extracts from herbs. Herbal medicines have been iisedNfomt16r acting various diseases of humans and animals in many different countries for a very long period of time (see, e.g., I.A. Ross, 1999, Medicinal Plants of the World, Chemical cohstituents Traditional and Modern Medicinal Uses, WO 01/66803 PCT/USO1/07608 -2 Humana Press; D. Molony, 1998, The American Association of Oriental Medicine's Complete Guide to Chinese Herbal Medicine, Berkley Books; Kessler et al., 1996, The Doctor's Complete Guide to Healing Medicines, Berkley Health/Reference Books); Mindell, supra). Herbal Medicines. There are many branches of herbal medicine around the world, 5 such as Ayurveda, Unani, Sida and Traditional Chinese Medicine (TCM). While modern Western medicine typically consists of administering a single chemical entity capable of intervening, specific biocheinical pathways, each formula of TCM typically contains hundreds of chemical entities from several herbs which are designed to interact with multiple targets in the body in a coordinated manner. Although empirical practice contributed in a significant 10 way to the herbal composition and prescription of these ancient herbal medicines, they are also supported, to a varying degree, bya' set of theories which all are distinct from that of modern Western medicine in ternns .o'f'atdniy, pharmacology, pathology, diagnosis treatment, etc. Among the different herbad'iedidiie fields, TCM has developed a more complete set of theories over several'centuies which have been well documented and practiced by local 15 physicians caring for a huge population (>1.3 billion people) in greater China and in East Asia including Korea and Japen. Western medicine generally uses purified -compounds, either natural or synthetic, mostly directed tow ards a single physiological target. However, the compositions used in TCM are usually composed of multiple herbs and compounds which are aimed at multiple 20 targets in the body based on unique and holistic concepts. TCM mainly used processed crude natural products, with various combiiations and formulations, to treat different conformations resulting in fewer side effects: The great potential of TCM has yet to be realized for the majority of the world'spe6pl' The herbs in a typical TCM prescription are assigned roles as the principal herb and the 25 secondary herbs, including assistant, adjuvant and guiding herbs. The principal herb produces the leading effects in treating the cause or the main symptom of a disease. An assistant herb helps to strengthen the effect of the principal herb and produces leading effects in the treatment of the accompanying ymptffis. There are three types of adjuvant herbs: 1) those that enhance the therapeutic effects of theprincipal and assistant herbs or treat tertiary symptoms, 2) those 30 that reduce or eliminate the' txicity and ofher side effects of the principal and the assistant herbs and 3) those which actbni complementary target tissues not specifically affected by the principal herb. A guiding herb directs the effect of other herbs to the affected site and/or coordinates and mediates the'ffects'of the other herbs in the prescription or formulation. In WO 01/66803 PCT/USO1/07608 contrast to most of the herbal medicines or supplements that consist of one or more parts of a single plant, the intended effects of TCM, are directed at multiple tissues. For example, a well-known TCM recipe, "Ephedra Decoction" used for treating asthma is composed of ephedra, cinnamon twig, bitter apricot kernel and licorice. Ephedra, is the 5 principal herb, which expels, cold, induces diaphoresis and facilitates the flow of the Lung Qi to relieve asthma, the main symptom. Cinnamon twig, as the assistant herb, enhances ephedra's induction of diaphoresis and warms the Channels to ensure the flow of Yang Qi for reducing headache and pantalgia. Bitter apricot kernel, as the adjuvant herb, facilitates the adverse flow of the Lung Qi and strengthens the asthma relief by ephedra. Licorice as the 10 guiding herb moderates the effects ofboth ephedra and cinnamon to ensure ahomeostasis of the vital Qi. While each of the four herbs clearly exhibits its respective activity, they complement as well as suoplement each other when they are combined. In practice, the principal herb can be prescribed with one or more secondary herbs, depending on the symptoms at a patient's presentation (Prescriptions of Traditional Chinese Medicine, Chapter 15 One, pp 10-16, E. Zhang, editor in Chief, Publishing House, Shanghai University of Traditional Chinese Medicine, 1998). The main theories'of TCM that guide the treatment of sickness with herbal medicine and other means, such as acupuncture, are 1) the theory of Yin and Yang, 2) the theory of Five Elements, 3) the theory of Yiscera and Bowels, 4) the theory of Qi, Blood and Body Fluid, and 20 5) the theory of Channels and Collaterals. In TCM, the first'imp6rtant aspect of making the proper diagnosis is to ascertain whether the disease is Yin otang. For exam]l6, those patients who have a fever, are thirsty, constipated or have a rapid pulse condition are of Yang character. Those individuals who have an aversion to cold, are not thirsty, and diarrhea and a slow pulse condition are of Yin 25 character. The property, flavor pand function of herbs can also be classified according to Ying and Yang theory. For dxan'ile, herb f cold and cool nature belong to Ying, while herbs which are warm and hot innature belong to Yang. Herbs with sour, bitter and salty flavor belong to Ying, while h~rbs with pungent, sweet and bland flavor belong to Yang. Herbs with astringent and subsiding function belong to Yin, while herbs with dispersing, ascending and 30 floating function belong to Yang. In TCM, the principles of treatment are based on the predominance or weakness of Yin and Yang. Herbs are prescribed according to their property of Ying and Yang and their function for restoring the imbalance of the Ying and Yang. In so doing, the benefit of treatnient is achieved.

WO 01/66803 PCT/US01/07608 -4 According to the theory ofFive Elements there are five basic substances that constitute the material world (i.e., wood, fire, earth, metal and water). In TCM, this theory has been used to explain the physiology and pathology of the human body and to guide clinical diagnosis and treatment. Herbal physicians, have applied the laws of generation, restriction, subjugation and 5 reverse restriction of the five elements to workout many effective and specific treatment regimens, such as reinforcig earth to generate metal (strengthening the function of the spleen to benefit the lung), replenishing water to nourish wood (nourishing the essence of the kidney to benefit the liver), supporting earth to restrict the wood (supplementing the function of the spleen to treat the hyperactivity of the liver), and strengthening water to control fire 10 (replenishing the essence of the kidney to treat hyperactivity of the heart). Specifically, the property of some herbs is assigned to each of the five Elements for the purposes of guiding the prescription of a TCM recipe. In TCM, the internal organs of the human body are divided into three groups: five Viscera (the Heart, the Live'fthe Spleen, the Lung and the Kidney), Six Bowels (the Gall 15 Bladder, the Stomach, the large Intestine, the Small Intestine, the Urinary Bladder, and the Triple Warmer), the Extradrdinary Organs (the Brain, the Medulla, the Bone, the Blood Vessel, the Gall Bladder, and:the Uterus). In TCM, the Viscera or the Bowel are not only anatomic units, but are also. concepts of physiology and pathology about interactions between different organs. For exaiilpfe;' the heart also refers to some of the mental functions and 20 influence functions of blood, hair, tongue and skin. Ying-Yang and the Five Elements influence the interactions am6iig thee Viscera, Bowels and Organs. The complexity of interplay of the theories is used*t6'ekplain the pathology of diseases to which herbs are prescribed, as discussed below. The prescription of heibal medicine in TCM starts with the diagnosis, which consists of 25. four main items: interrogatil, inspection, auscultation and olfaction, pulse taking and palpation. During the interrogation phase, much information is gathered, including the characteristics of the mainiyinpton . For instance, if the main symptom is characterized by dull pain of epigastria regi:n,9v hich maybe relieved by warming and pressing, this suggests the insufficiency of the Splebn-Yang. Soreness and weakness of the loins and knees, 30 intolerance of coldness with cold extremities manifests a weakness of the Kidney-Yang. During inspection, observations- are made for vitality, skin color and the general appearance and the condition of the tongue For 'example, a pale complexion corresponds internally to the Lung of autumn, whose Qi Is dry. This may occur when Yang Qi is lacking and the circulation WO 01/66803 PCT/USO1/07608 -5 of Qi and blood is impeded, or when the coldness in the channels and collaterals causes them to contract. In TCM, it is from Qi, blood and body fluid that come energy needed by the Viscera and Bowels, Channels and Collaterals, tissues and other organs for carrying-out their 5 physiological functions; and on which the formation and metabolism of Qi, blood and body fluid depend. Prescriptions of TCM consider the herbal effects on Qi and blood for treatments. TCM holds that Channels, Collaterals and their subsidiary parts are distributed over the entire body. It is through them that herbs exert influence on pathological targets and achieve the improvement of sickness. For example, ephedra acts on the Channels of the Lung and 10 Urinary Bladder so as to induce sweat for relieving asthma and promoting diuresis. As noted above, clinical applications of acupuncture are also guided by the theory of Channels and Collaterals. In summary, while he nature or property of each herb in TCM may be assigned as Yin or Yang, and to one of the Five elements, they act through Channels and Collaterals and are 15 mediated via Qi, Blood aid Fluid Ito yield therapeutic effects on targets, such as Viscera and Bowels. Pathogenic factors may be disguised as decoy through the very same systems of Channels and Collaterals to adversely affect the functions of Viscera and Bowels and thus cause sickness. From the foregoing discussion, it is clear that the TCM terminology is as much of a 20 philosophical concept as-an anatomical one. For example, the Heart represents a host of tissues, organs or systems in the body that contribute to a function described in TCM. Thus, the concept of the Heart rfiuires a multiple dimension data set to describe each concept of TCM. Once this is accomplished, a molecular holistic medicine can be developed. U.S. Regulatory Prdeess. :In the US, dietary supplements (such as botanical products, 25 vitamins and minerals, amino acids and tissue extracts) are regulated under the Dietary Supplement Health and Eidatio5 Act of 1994 (the DSHE Act). This Act removed the ingredients of dietary supplements from regulation as food additives under the Federal Food, Drug, and Cosmetic Act. In addition, the DSHE Act requires that The Food and Drug Administration (FDA) bear the burden of proof that a marketed dietary supplement presents a 30 serious or unreasonable risk under the conditions of use on the label or as commonly consumed. Thus, there are- citently no federal regulations that establish specific criteria for purity, identification and manufacturing procedhrs for dietary supplements. In addition, few WO 01/66803 PCT/USO1/07608 -6 published papers on herbal quality have resulted from the establishment of the Office of Alternative Medicine by Coingress in 1992 (Angell et al., 1998, N. Engl. J. Med. 339:839-841). At the present time, the FDA must approve each one of the chemical entities in a drug composition or cocktail, and then clinical trials must be undertaken so as to obtain separate 5 FDA approval for marketing the drug. This process is extremely tedious and costly. A molecular holistic medicine may require a less arduous evaluation since the previous use of a particular herbal composition as a botanical drug permits clinical trials with multiple chemicals at the outset (i.e., clinical trials using the herbal composition or specific components of the herbal composition). Recently, the FDA has approved the testing of some herbal medicines in 10 clinical trials as botanical drugs (FDA Guidance on Botanical Drugs, April, 1997). While these events represent a positive development for health care in general, it also raises important issues regarding the formulation, manufacturing and quality control of herbal medicines and dietary supplements, iicludingthe traditional Chinese medicines. The multitude of relevarit biological responses induced by the multiple chemicals in 15 herbs are not currently available and will be increasingly important to support marketing approval by the FDA. Herbal-based industries are coming under increasing pressure to upgrade their current practices (see, e.g., Angell etal.,, supra). The need to apply scientific testing to the preparation and administration of herbal iedioines and food supplements has been highlighted by several 20 recent reports of toxicity iesulting from ingesting herb-based formulations. For example, one patient who took anherbal-based dietary supplement experienced digitalis toxicity (Slifinan et al., 1998, N. Engl. J. Med. 339:806-811). It was subsequently determined that the herb ingredient labeled as plantain in the supplement was actually contaminated with Digitalis lanata, an herb known to contain atleast'60 cardiac glycosides. In another instance, an herbal 25 preparation was found to be the cause of chronic lead intoxication in a patient (Beigel et al., 1998, N. Engl. J. Med. 339:827-830). This is not-a completely unexpected occurrence since contamination of traditional Asian herbal remedies by lead and other heavy metals is well documented (Woolf et at., 1994, Ann. Intern. Med. 121:729-735). Characterization of Botaliials. It is well known that the genetic identity (e.g., 30 genera, species, cultivar, Vality, 5 lone), age of herbal growth, harvest time, the specific plant part utilized, processing nefhod,geographical origin, soil type, weather patterns, type and rate of fertilizer, and other growth 'factois have a great impact on the particular chemical composition of any particular herb "harvested" from any particular area.

WO 01/66803 PCT/USO1/07608 Increasing numbers of various types of tests have been instituted to assure the consistent quality of herbs uiied ih'medicine and as dietary supplements; including inspections at the macro- and microscopic 1eiels as well as a variety of chemical analyses. Recently, high performance liquid chromatography (HPLC) profile of marker molecules in an herbal extract 5 has become one reference standard. However, there are problems with this approach, including that some of the bioactive molecules may not adsorb UV or the visible lights for HPLC detection, and the amount of a chemical is not necessarily proportional to its biological potency. For these reasons, herbal manufacturers resort to a practice of mixing raw herbs from different sources to minimize chemical variations. 10 Mass spectrometry (MS) is an analytical method for determining the relative masses and relative abundances of components of a beam of ionized molecules or molecular fragments produced from a sample iii a high vacuum. MS, unlike HPLC, is not optical density dependent. In practice it is iised in conjunction with HPLC or capillary electrophoresis (CE): the HPLC separates the chemicals and the MS then can be used to identify what they are. 15 Commercial systems are available which integrate MS and HPLC for biological uses. Mass spectrometry is limited -to samples that'are gaseous or volatile at low pressure, or that can be so rendered by derivatization. These steps are no longer adequate. Recent publications report a greater variation in the quality of herbs by specific 'suppliers, and the difficulty of providing biological equivalence 20 of herbal extracts.. Furthermore, the correlation between safety and efficacy and chemicals in an herb is not well defined in, iost cases. Recently, in response to complaints from consumer groups and regulatory agencies (Federal Register, February 6, 1997, Volume 62, No. 25, Docket No. 96M-0417, cG in Manufacturing, Packing or Holding Dietary Supplements, Proposed Rules), some herbaf manufacturers have begun to implement Good Manufacturing 25 Practice (GMP) which requires stiringent controls at all levels. Chemical and spctroscoyic methods have been used to characterize the components of herbal medicines and food supplements. Forexample, three new hederagenin-based acetylated saponins were isolated from the fruits of Gliricidia sepium using these two methods (Kojima et al., 1998, Phytochemistry48'5):885-888). The botanical sources of Chinese herbal drugs in a 30 number of commercial samples were inferred by comparing the contents of some characteristic constituents which were ainilyzed with high-performance chromatography (HPLC) or capillary electrophoresis (CE) (Shuenn-Jyi Sheu, 1997, Journal of Food and Drug Analysis 5(4):285 294). For example, the ratio of ephedrine/pseudoephedrine was used as a marker to WO 01/66803 PCT/USO1/07608 -8 differentiate Ephedra intermedia from other species; total alkaloid contents were used to distinguish between species of Phellodendron; and the contents of ginsenosides were used to differentiate between species of Panax. However, these methods do not provide a direct measurement of the effect of the various herbs on the molecular, physiological or 5 morphological responses following human treatment with the herbs. Using gas chromatography-mass spectrometry and atomic-absorption methods, the California Department of Health Sciences, Food and Drug Branch, recently tested Asian medicines obtained from herbal stores for contaminants (R. J. Ko, 1998, N. Engl. J. Med. 339:847). Of the 260 products they tested, at least 83 (32 percent) contained undeclared 10 pharmaceuticals or heavy metals, and 23 had more than one adulterant. Using high performance liquid chromatography, gas chromatography, and mass spectrometry, a commercially available combination of eight herbs (PC-SPES), was found to contain estrogenic organic compounds (DiPaola et al., 1998, N. Engl. J. Med. 339:785-791). The researchers concluded that PC-SPES has potent estrogenic activity and that prostate cancer 15 patients that took PC-SPES could confound the results of standard therapies and may experience clinically significant adverse effects. Gas chromatography data was also collected for different samples of the traditional Chinese medicine 'wei ling xian' and correlated to the antiinflammatory activity of the sarnples (Wei et al., Study of chemical pattern recognition as applied to quality assessment of the traditional Chinese medicine "wei ling xian," Yao Hsueh 20 Pao 26(10): 772-772 (1991)). This study did not provide relevant HBR Array data, such as time course, dose dependent response, control samples to substantiate the differential power of the biomarkers, nor it utilize a reiterative type. of data construction process to establish a comprehensive database for ghara6terizing effects of the herbal composition. Changes in protein levels have also been used to characterize the effects of herbal 25 compositions or specific components of herbs. ,For example, the production of granulocyte colony-stimulating factor (G-CSF) from peripheral blood mononuclear cells was found to vary depending on which specific Chinese herb was added to the culture (Yamashiki et al., 1992, J. Clin. Lab. Immunol. 37(2):83290). Expression of interleukin-1 alpha receptors was markedly up regulated in cultured humfrepiderinal keratinocytes treated with Sho-saiko-to, the most 30 commonly.used herbal medicine in Japan (Matsumoto et al., 1997, Jpn. J. Pharmacol. 73(4):333-336). The expressioinof F ganma 11/111 'receptors and complement receptor 3 of macrophages were increased by treatment with Toki-shakuyakusan (TSS) (J. C. Cyong, 1997, Nippon Yakurigaku Zasshi 1 0(Suppl. 1):87-92). Tetrandrine, an alkaloid isolated from a WO 01/66803 PCT/US01/07608 -9 natural Chinese herbal medicine, inhibited signal-induced NF-kappa B activation in rat alveolar macrophages (Clden et al., 1997, 3iochem. Biophys. Res. Commun. 231(1):99-102). The herbs Sairei-to, alismatis rhizoma(Japanese name "Takusha") and hoelen (Japanese name "Bukuryou") inhibited the sypthesis and expression of endothelin-1 in rats with anti-glomerular 5 basement membrane nephritis (Hattori et al., 1997, Nippon Jinzo Gakkai Shi 39(2):121-128). The increase or decrease in mRNA levels has also been used as an indicator of the effect of various herbs and herbal components. Intraperitoneal injection of Qingyangshen (QYS), a traditional Chinese medicine with antiepileptic properties, and diphenylhydantoin sodium reduced alpha- and beta-tublin mRNAs and hippocampal c-fos mRNA induction 10 during kainic acid-induced chronic seizures in rats (Guo et al., 1993, J. Tradit. Chin. Med. 13(4):281-286; Guo et al, 1995, J. Tradit. Chin. Med. 15(4):292-296; Guo et al., 1996, J Tradit. Chin. Med. 16(1):48 51) Treatment of cultured human umbilical vein endothelial cells (IUVECs) with the sapohin astragaloside IV, a component purified from Astragalus membranaceus, decreased plasminogen activator inhibitor type I (PAI-1) specific mRNA 15 expression and increased tissue-type plasminogen activator (t-PA) specific mRNA (Zhang et al., 1997, J. Vasc. Res. 34(4):273-280). One component isolated from the root of Panax ginseng was found to be a potent inducer of interleukin-8 (IL-8) production by human monocytes and by the hurhan nionocytic cell line THP-1, with this induction being accompanied by increased IL-8 iRNA expression (Sonoda et al., 1998, Immunopharmacology 20 38:287-294). Recent advances in nucleic acid microarray technology enable massive parallel mining of information on gene exptession. This process has been used to study cell cycles, biochemical pathways, genome-wide expression in yeast, cell growth, cellular differentiation, cellular responses to a single chemical.compound, and genetic diseases, including the onset 25 and progression of the diseases (M. Schena et al., 1998, TIBTECH 16:301). Because cells respond to the micro-envitotient changes by changing the expression level of specific genes, the identities of genes expressed in a cell determine what the cell is derived of and what biochemical and regulatory systems are involved, among other things (Brown et al. 1999, Nature genet.; 21 (1) supplement:33). Thus, cellular gene expression profiles portray the 30 origin, the present'diffe-entiation of the cell, and the cellular responses to external stimulants. No researchers to date, if any, have attempted to apply these new technologies to study the molecular effects of whole herbal treatments and supplements.

WO 01/66803 PCT/USO1/07608 -10 Some researchers have attempted to characterize the effects of the major active constituents isolated from selected herbs. For example, treatment of HUVECs with notoginsenoside R1 (NR1), purified from Panax notoginseng, resulted in a dose- and time dependent increase in TPA syIthesis (Zhang et al., 1994, Arteriosclerosis and Thromobosis 5 14(7):1040-1046). Treatment with NRI did not change urokinase-type plasminogen activator and PAI- 1 antigen synthesis, nor did it effect the deposition of PAI-1 in the extracellular matrix. TPA mRNA increased as much as twofold when HUVECs were treated with NR1, whereas expression of PAI-1-specific mRNA was not significantly affected by NRl. Since most studies on P. notoginseng have involved its mixture with other herbs, the researchers 10 noted that it was difficult to assess how their results relate to the situation in vivo when is used therapeutically in humans ( ., at 1045, second column, first paragraph). In addition, since the researchers only studied oie iajor componentof the herb, it is not possible to ascertain the molecular effect of the whole hdrb or the interactions among components of the herb from this study. 15 Dobashi et al. (1995, Nuroscience Letters 197:235-238) studied the effect of two of the main components of saiko agents, a Chinese herbal drug used to treat nephrotic syndrome, bronchial asthma and chronic rheumatoid arthritis. Administration of SS-d increased plasma adrenocorticotropin (ACTH) levels, proopiomelanocortin mRNA levels in the anterior pituitary and the CRF mRNA level in the rat hypothalamus in a dose dependent manner. In 20 contrast, treatment with SS-'tailed to affect the levels of these molecular markers. While this study indicates that administration of SS-d may have an important role in saiko agents-induced CRF release and CRF gene expression in rat hypothalamus, it fails to address the molecular effect of the herbal medication as a whole. Kojima et al. (1998; il. Pharm. Bull. 4:426-428) describe the utilization of 25 differential display of mRNA to isolate and identify genes transcriptionally regulated in mouse liver by sho-saiko-to, an herbal medicine used for treating various inflammatory diseases in Japan. These researchers limited their study to the use of mRNA differential display techniques in investigating the molecuar mechanisms of herbal medicine. It also failed to address effects in multiple'drgans of treated animals and did not provide any guidance for 30 quality control, new use, and standardization of effects. In addition, the study failed to analyze the individual components of th& herb and compare the individual results with the results obtained using the whole herbal mixture.

WO 01/66803 PCT/USO1/07608 -11 Ma Ji et al. (1998; Chinese Medical Journal 111(1): 17-23) investigated the therapeutic effect of the herb Astragali Mnembranaceus on sodium and water retention in rats experiencing aortocaval fistula-caused experimental congestive heart failure. Chronic heart failure rats with and without Astraglia treatment were compared for changes in various morphological 5 characteristics (e.g., body weight, serum sodium concentration); physiological characteristics (e.g., mean arterial pressure, heart rate, hematocrit and plasma osmolality); mRNA expression levels (e.g., hypothalamic arginine vasopressin (AVP), AVP Via receptor, renal AVP V 2 receptor, aquaporin-2 (AXP2)) and protein excretion (e.g., plasma atrial monophosphate peptide (ANP) and urinary cyclic guanidino monophosphate (cGMP)). The researchers found 10 that treatment with Astraglia- improved cardiac and renal functions, partially corrected abnormal mRNA expressions of theAVP system and AQP2, and improved the renal reaction to ANP. This study did not addies using the collected data to guide the development of new formulations or for elueidating the synergistic or other interactions among various herbs in a formula, or validate the differential power of the effects for quality control purposes. 15 As shown by the aboiiereview of relevant scientific articles, molecular-based technology has not been used to explore and validate cellular and molecular responses in biological systems that are treated'or challenged with multiple chemicals at the same time, such as herbal medicines and TCM. Furthermore, these recent advances have not been integrated with other technologies and methods to produce a process for the systematic 20 exploration of biologicaleffects of herbal medicines and TCM. SUMMARY OF THE INVENTION This invention provides the tools and methodologies for creating, maintaining, improving and utilizing Hebal PioR6sponse Arrays (HBR Arrays), wherein the HBR Arrays constitute data sets associated with particular' herbal compositions. The HBR Arrays of the 25 present invention may include information on'the plant-related parameters of the herbal constituents, marker infornition collected following the exposure of a biosystem to the herbal composition, and biological response information collected following the exposure of a biosystem to the herbal coinposition.. The present invention rovides the tools and methodologies necessary for establishing 30 standardized HBR Arrays for particular herbal compositions, wherein the standardized HBR Arrays are used as benchmarks by which to evaluate batches of similar or different herbal compositions. The present invention further provides the tools and methodologies necessary to update and maintain the staridadized.IHBR Arrays. Particular embodiments of the present WO 01/66803 PCT/USO1/07608 -12 invention involve iterative processes whereby data for additional batches of the herbal composition, additional plant-related data, additional marker information, and/or additional BioResponse information is periodically added to the standardized HBR Arrays. Thus, the present invention provides the tools and methodologies for creating, maintaining, updating and 5 using HBR Arrays on an ongoing basis. The present invention provides the tools and methodologies necessary to guide the standardization of herbalcompositions; to determine which specific components of herbal compositions are responsible for particular biological activities; to predict the biological activities of herbal compositions; for the development of improved herbal therapeutics; for 10 adjusting or modifying an herbal composition; for measuring the relatedness of different herbal compositions; for identifying specific molecules in the batch herbal composition which retain the desired biological activity; for determining which herbal components of a known herbal composition can be eliininated from the known herbal composition while maintaining or improving the desired biblogical activity of the known herbal composition; for identifying new 15 uses and previously unknowii biological activities for the batch herbal composition; and for using the predicted biol6gi6i activity of the batch herbal composition to aid in the design of therapeutics which include heibal components and synthetic chemical drugs, including the design of therapeutics using the methods of conibinatorial chemistry. More specifically, the present invention provides methods of establishing standardized 20 Herbal BioResponse ArrayIs BR Arrays) for herbal compositions, wherein the methods comprise:, a) selecting a characterized heibal.composition; b) exposing a biosystm to a btch of the characterized herbal composition and collecting data on two or more markers wherein one of the markers is a change in gene expression 25 determined through theuidof a nucleic acid microarray, produced by the steps comprising: i) producing Ia celfbanliiig system; ii) profiling tfie n gee xyression pattern of cells from the cell banking system before andafter exposure to the herbal composition; iii) selecting as markers'those genes whose expression levels are changed by 30 exposure to tli herbal composition; c) storing the marker data ofstep b) as a standardized HBR array. The present invention further provides such methods which further comprise exposing a biosystern to one or morebatdies of the herbal "composition, collecting the data on one or more WO 01/66803 PCT/USO1/07608 -13 BioResponses, and adding the collected BioResponse data to the standardized HBR Array for that herbal composition. The present invention provides methods of evaluating herbal compositions, wherein the methods comprise exposing a biosystem to a batch of the herbal composition and collecting 5 data on two or more markers; and comparing the collected marker data with a standardized HBR Array for the same o a substantially same herbal composition as that of the batch herbal compositions. The present invention provides a system for predicting the biological activity of an herbal composition comprising: 10 1). a biosystem comprising one or more different types of cells, tissues, organs or in vitro assays; 2). a batch herbal composition; 3). 'two or more molecular markers; 4). a means forexposing the biosystem to the batch herbal composition and 15 measuring the differential responses of the molecular markers; 5). a computerprocessor, including memory, for analyzing and storing the differential response measurements of the molecular markers so as to create an Herbal BioResponse Array (HBR Array) data set for the batch herbal composition; 6). a compute Trocessor; including memory, for comparing the HBR Array of the 20 batch herbal composition to 61ne or more previously-stored HBR Arrays so as to predict the biological activity of the batch herbal composition, wherein the biological activities of the herbal compositions used to generate the one or more previously-stored HBR Arrays are known. BRIEF DESCRIPTION OF THE DRAWINGS 25 Figure 1. .Figure 1 provides a schematic of the basic method steps for constructing a Standardized Herbal BiolesposeAriay (HBR Array) for any selected herbal composition. The figure is shown in its most basic form for ease of understanding. As discussed herein, each of the pathways of thedschematic can be done iteratively. Furthermore, any information contained in one box can be dsed to guide decisions regarding gathering information for any 30 other box. In this Way, numerous fdedback loops are also possible throughout the scheme. Figure 2. Figure 2.

t prdvides a schematic of the basic method steps for constructing a an Herbal BioResponse Array (HBR Array) for any batch herbal composition and for comparing WO 01/66803 PCT/USO1/07608 -14 this batch J{RB Array to a selected subset of information from the Standardized HBR Array. The figure is shown in its'most basic form for ease of understanding. As discussed herein, each of the pathways of the schematic can be done iteratively. Furthermore, any information contained in one box can be used to guide decisions regarding gathering information for any 5 other box. In this way, numerous feedback loops are possible throughout the scheme. Figure 3. Figure 3 provides a schematic of the basic method steps for establishing and using a major data set. The figure-is shown in its most basic form for ease of understanding. As discussed herein, each of the pathways of the schematic can be done iteratively. 10 Furthermore, any information contained in one box can be used to guide decisions regarding gathering information for another box. In this way, numerous feedback loops are possible throughout the scheme. Figure 4. Western blot for various herbal compositions. 15 A. No herbal composition. B. Huang Qing Tang A (HQT A) (0.2 mg/ml). C. HQT A (4 mg/ml). D. HQT B (0.2 mg/ml). E. HQT B (4 mg/ml). 20 F. Scute (0.2 mg/ml). G. Scute (4 mg/ml). Figure 5. HPLC for Paeonie lactiflorapallus. 25 Figure 6. HPLC for Ziziphifructus. Figure 7. Figure"7 provides a schematic- for establishing a bio-response data set for an herbal composition. The data set is based on differentially expressed gene induced by the herbal medicine for more than three different concentrations in a mammalian cell culture. 30 Figure 8. Figure 8 provides a schematic for establishing a characteristic expression profile database or HBR Array for an herbal medicine or a complex herbal preparation.

WO 01/66803 PCT/USO1/07608 -15 Figure 9. Figure 9 provides a schematic for identifying an unknown herbal composition. The' expression profiles induced by the unknown herbal medicine are aligned with the expression profile database and statistical method is employed to score the possible identities of herbal medicines archived in the database. 5 Figure 10. Figure 10 provides a schematic for extracting signature genes for an herbal composition or a complex herbal preparation. Figuire 11. Figure 11 provides a'schematic for extracting signature genes for individual chemical constituents in an herbal medicine or a complex herbal preparation. 10 Figure 12. Clustered, display of gene expression data from cells treated with three types of single-element herbal ektracts (Cordyceps Sinensis Mycelium(CSM), ST024, ST1 17) with high and low' concentrations (indicated with H and L, respectively). (A) Cluster analysis w"as performed by the program "Cluster" (Eisen et al., 1999) with 492 selected genes (see text). 15 (B) Enlarged image of genes up-regulated by ST1 17 treatment but down-regulated by other herbal extract treatments. The clone ID and putative gene name are indicated. (C) The clustering' agorithm separated CSM, ST024 and ST 117 into 3 distinct clusters. The distance between each cluster as displayed by the hierarchical dendrogram can be viewed as the difference betweenithe expregsion profiles of the three herbal extracts treated cells. 20 Figure 13. (A) Pseudocolm encoded display of clustering results as calculated based on the selected 492 genes. The bx es in (A) indicate the positions of the three clusters of genes described above. 25 (B) Enlarged'image of genes down-regulated by the CSM but up-regulated by the others. (C) Genes up-regulated by all kinds of herbal treatments. (D) Genes down-regulated by CSM and up-regulated by the others. The IMAGE clone ID and putative- gene name are indicated. 30 Figiue 14. Clustered display of expression data from 2 batches of multi-element herbal preparations of the Huang Chin Tang (PHY906-303503 (#11) and PHY906-284003 (#12)) WO 01/66803 PCT/USO1/07608 -16 treated cells with Ingh and low concentrations (indicated with H and L, respectively). The data were averaged based on three repeated experiments on three different dates. Cluster analysis was performed based on the selected 500 genes (see text). (B) The clustering algorithm separated #11-L, #11H and (#12-H and #12-L) into 3 distinct clusters. Distance between 5 clusters or resemblance coefficient is indicated by the hierarchical clustering dendrogram. Figure 15. Enlarged image of (A) averaged and (B) individual gene expression levels measured by three independent experiments. Box1 encloses genes that were down regulated in #1 1-L treated cells but up regulated in others, Box2 encloses the genes that were up regulated 10 by all the herbal treatments. Box3 enclosed the genes that showed no response by #11 -L treatment but were down regulated by the others. Box4 encloses the genes highly down regulated by low concentration herbal treatments but show mild response at high concentration herbal treatments. The clone ID and putative gene name are indicated beside each gene. 15 Figure 16. Classification of gene expression profiles in the cells treated by herbal medicines. Hierarchical clustering of (A) the data sets normalized with the expression data of the untreated control cells and (B) data sets standardized to have zero-mean and unit-variance. (C) The result of a non-hierarchical clustering by the self-organizing maps algorithm. 20 Figure 17. Candidate class predictors for the classification of herbal medicines based on the gene expression profiles induced by the medicines. 50 class predictors with their expression profiles for discriminating #11 and #12 herbal preparations are shown in this figure. The IMAGE clone ID and putative-gene name are indicated beside each gene. 25 Figure 18. The gene expression profiles induced by a batch of a complex herbal preparation of five different 6ncentrations. A 6x4 clustering of expression profiles is shown in (A), and the details of theg6ne expression profiles for the selected clusters are shown in (B). Figure 19. Figure 19 illustrates how the expression profiles in Figure 18 are categorized into three different groups for subsequent hamming distance calculation. 30 Figure 20. Figure 20 shows the analysis results of gene expression profiles induced by five batches of a complex herbal preparation. The numbers in the table are hamming distance. The smaller the distance,-the more similar are 'the expression profiles.

WO 01/66803 PCT/USO1/07608 -17 Figure 21. Shown in (A) is a table of integrated peak intensities of 4 chemical constituents in HPLC analyses of five batches of a complex herbal preparation. Two additional parameters, BG+B and BG/B are introduced td the table and a 6 parameter radial plot is shown 5 in (B) to illustrate that one batch is more similar to a second batch #18 than to the other batches by the'HPLC analysis. Figure 22. A display of the signature genes induced by a complex herbal preparation, the Huang Chin Tang, in Jurkat T cells. 10 Figure 23. Figure 23 illustrates the principle of identifying signature genes induced by individual chemical constituents in a mix of herbal medicines. The signature genes are those whose expression levels c6rfelate with the amount of chemical constituents in the herbal medicine and that the correlation.coefficient is larger than 0.99 or smaller than -0.99. (A) shows that the R value between the, gene and Glycyrrhizin was 0.998, and (B) shows that the 15 gene whose. expression levels increase with the decrease of Wogonin has an R value of -0.997. Figure 24. The signature genes induced by the chemical constituent Albiflorin in a complex herbal preparations,: Huang Chin Tang, in Jurkat T cell. (A) show the genes that were positively correlated with Albiflorin, and (B) shows the genes that were negatively correlated with Albiflorin. 20 Figure25. Correlation of gene expression profiles to a control group. (A) is the gene expression profile of a control group, and (B) is the gene expression profile of a sample group. (C) shows the number of genes with a differential expression ratio having greater than 2-fold increase with concentration of herbal treatment. Figure 26. Clusters of expression profiles clustered by a non-hierarchical analysis 25 program, wherein the program is based on a self-organizing map (SOM) principle. The X-axis represents the herbal conctnition from low to high and the Y-axis is the gene-expression ratio. Figure 27. Figure 27 shows the induced and repressed genes commonly found in two batches of Huang Chin Tang.

WO 01/66803 PCT/USO1/07608 -18 Figure 28. SOM'clnstering results for two batches of Huang Chin Tang. (A) shows the SOM clustering results for theexpression profiles of two batches of Huang Chin Tang. (B) shows that ten genes have similarly responded to the two batches, and (C) shows how the weighing factor decreases as cluster I and cluster j become more different. 5 Figure 29. Calculation of S score between pairs of herbal preparations in cluster analysis. (A) is a tabulation of the scores, and (B) is demonstrates how 5 batches of similar herbal preparations are related. DETAILED DESCRIPTION OF THE INVENTION Unless defined otherwise, all technical and scientific terms used herein have the same 10 meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. Although any methods and materials similar or equivalent to those described herein can be used in the practice o resniiig of the present invention, the preferred methods and materials are described. Overview of the Invention 15 As set forth above, the present invention is directed to tools and methods useful for predicting the biological response of an herbal composition. More particularly, this invention provides methods of creating Herbal BioResponse Array (HBR Array) databases as well as methods for using such databases to improve the design of effective herbal-based therapeutics. The goal of the present in 6nti6n is thd overall design, creation, improvement and use of HBR 20 Arrays for the preparation, testing and administration of herbal compositions, and guide development of new herbal positionss and novel uses of existing herbal compositions. Phytomics. As ustl herein, depending on the context in which it is used, "phytomics" refers to using bioinfonnatics and statistical approaches to address the qualitative and quantitative aspects of the components of herbal compositions or to the actual data bases 25 which are developed for addressing such aspects. Herbal BioResponseArray As used herein, an HBR Array constitutes a data set of two or more observations'or nieasdtments associated with an herbal composition. The HBR Array may include quaifatiVe arid quantitative data on the plants in the composition (plant related data), marker information obtained after exposure of a biosystem to the herbal 30 composition including a dose dependent study, and a database of BioResponse data obtained after exposure of a biosystem to the herbal composition. The data in any particular HBR Array can be statistically analyzed ifn either 2- or 3-dimensional space.

WO 01/66803 PCT/USO1/07608 -19 HBR Arrays may be designated is batch HBR Arrays and standardized HBR Arrays. Batch HBR Arrays are arrays of data associated with specific batches of an herbal composition. Standardized HBR Arrays are arrays of data associated with a standardized herbal composition. 5 Major Data Set. As used herein, the term "major data set" refers to the data set which acts as the baseline set of data by which various other sets of data are compared or otherwise analyzed for the same or different herbal compositions. Generally, the major data set is created using biotechnological techniques to ascertain some genetic or protein aspect of the herbal compositions. Thus, the major data set will usually, but not always, be based on a genomic or 10 proteomic set of data. For example, nucleic acid microarray results could be the major data set which is used to compare to other, dependent or minor data sets. Minor or Dependent ata Set. As used herein, the "minor data set" or "dependent data set" refers to one or more data sets which are used for comparing to the major data set. Generally, but not always; the minor data set Will consist of information on an herbal 15 composition which are collected by more traditional methods. For example, the minor, or dependent, data set may consist of a collection of plant-related data obtained by more conventional means. Examlpes of plant-related data include, but are not limited to, the genus/species of the herb(s) in the herbal composition, the particular plant parts of the herb(s) in the composition and the geographic location where the herb(s) were located. Another 20 example of a minor data set might consist of a set of biological responses of a cell, tissue, organ or organism after treatment with. one or more different amounts of the herbal composition. Examples of siclbiological data or a whole organism may include, but are not limited to, cell toxicity stdies, enzyme treatment studies, growth rates, weight gain or loss, changes in motor skills an'd changes in mental abilities. 25 Herb. Technically speaking an herb is a small, non-woody (i.e., fleshy stemmed), annual or perennial seed-bearing plant in which all the aerial parts die back at the end of each growing season. Herbs are valued for their medicinal, savory or aromatic qualities. As the word is more generally used and as the word is used herein, an "herb" refers to any plant or plant part which has a food supplement, medicinal, drug, therapeutic or life-enhancing use. 30 Thus, as used herein, an herbis not limited to the botanical definition of an herb but rather to any botanical, plant or plant pArt used for such purposes, including any plant or plant part of any plant species or subspecids of the Metaphyta kingdom, including herbs, shrubs, subshrubs, and trees. Plant parts used iiherbaltomyositions include, but are not limited to, seeds, leaves, WO 01/66803 PCT/USO1/07608 -20 stems, twigs, branches, budei flowers, bulbs, corms, tubers, rhizomes, runners, roots, fruits, cones, berries, cambium and bark. Herbal Composition. As used herein, an "herbal composition" refers to any composition which includes herbs, herbal plants or herbal plant parts. Thus, as used herein, an 5 herbal composition is any herbal preparation, including herbal food supplements, herbal medicines, herbal drugs and medical foods. Examples of herbal compositions include, but are not limited to, the following components: a whole plant or a plant part of a single plant species; whole plants or plant parts of multiple plant species; multiple components derived from a single plant species; multiple components derived from multiple plant species; or any 10 combination ofthese various components. For a thorough review of various herbal compositions, see, for example, Kee.Chang Huang, The Pharmacology of Chinese Herbs, CRC Press (1993), herein incorjorad in its, entirety. Representative examples of various herbal compositions are provided in'the following paragraphs. Herbal compositions which include the bark of the willow tree have been used to treat 15 fever since the mid-eighteenith century in England. The active ingredient in willow bark is a bitter glycoside called salicii, which on hydrolysis yields glucose and salicylic alcohol. Aspirin (acetylsalicylic acid)' and aspirin-like drugs (e.g., ibuprofen), all of which are often called nonsteroidal antiinflammaiory drugs (NSAIDs), are frequently used to treat pain, fever, and inflammation. Meadowsweet'is another hrb- that contains salicylates. Treatment of 20 arthritic and arthritic-like Smptoms'with willow bark or meadowsweet requires the consumption of large quantities of herbal teas made from these plants. The entire Populus species (i.e., poplar trees and shrubs) also contains salicylate precursors and poplar-buds have been used in antiinflammatory, antipyretic and analgesic medications. U.S. Patents have been.is946d for herbal compositions used for the treatment of various 25 diseases and other health4r1ated problems afflicting humans and animals. For example, U.S. Patent No. 5,417,979 discloses composition comprising a mixture of herbs, including species of Stephania and Glycyrrhiza as well as their extracts, which is used as an appetite stimulant and for the treatment of pain. -Herbal compositions which include Glycyrrhiza uralensis have been found useful for treating eczema; psoriasis, pruritis and inflammatory reactions of the 30 skin (U.S. Patent No. 5,4664562). U.S. Patent No. 5,595,743 discloses various herbal compositions which include licorice extract (Glycyrrhiza) and siegesbeckia, sophora, stemona and tetrandra herbs used for-the treatment of various mammalian diseases, including inflammation and rheumatoid.arthritis. Ocular inflammation can be treated with a WO 01/66803 PCT/USO1/07608 -21 pharmaceutical composition containing the plant alkaloid tetrandrine (U.S. Patent No. 5,627,195). U.S. Patent No. 5,683,697 discloses a pharmaceutical composition having anti inflammatory, anti-fever, expectorant or anti-tussive action, wherein the composition includes 5 plant parts from the, species Melia, Angepica, Dendrobium, Impatiens, Citrus, Loranthus, Celosia, Cynanchum and Glehnia. An herbal composition which includes extracts of the roots, rhizomes, and/or vegetation ofAlphinia, Smilax, Tinospora, Tribulus, Withania and Zingiber has been found to reduce or alleviate the symptoms associated with rheumatoid arthritis, osteoarthritis, reactive arthritis and for reducing the production of proinflammatory cytokines 10 (U.S. Patent No. 5,683,698).. Herbal compositions are available in many forms, including capsules, tablets, or coated tablets; pellets; extracts or finctures; powders; fresh or dried plants or plant parts; prepared teas; juices; creams and ointments; essential oils;.or, as combinations of any of these forms. Herbal medicines are adniiriitered by any one of various methods, including orally, rectally, 15 parenterally, enterally, transdermally, intravenously, via feeding tubes, and topically. Herbal compositions encompassed by the present invention include herbal compositions which also contain nonL.herbal components. Examples of such non-herbal components include, but are not linfited to, whole insects and insect parts, worms, animal or insect feces, natural or petroleum ils, carbonate of ammonia, salt of tartar, liquor, water, 20 glycerin, steroids, pharmaceuticals, vitamins, nutrient extracts, whey, salts, and gelatin. For oral administration, the herbal compositions disclosed may take the form of, for example, tablets or capsules prepared by conventional means in admixture with generally acceptable excipients such as binding agents (e.g., pregelatinised maize starch, polyvinylpyrrolidone or hydrbxypropyl methylcellulose); fillers (e.g., lactose, microcrystalline 25 cellulose or-calcium phospiiate); lubricants (e.g., magnesium stearate, talc or silica); disintegrants (e.g., potato starch or sodium starch glycolate); or wetting agents (e.g., sodium lauryl sulphate); glidants artificial and natural flavors and sweeteners; artificial or natural colors and dyes; and solubilisers. The herbal compositions may be additionally formulated to release the active agents in atime-i Iease manner as is known in the art and as discussed in 30 U.S. Patent Nos. 4,690,825 and5,055,300. The tablets may be coated by methods well known in the art. Liquid preparations for oral administration may take the form of, for example, solutions, syrups, suspensions or slurries (such, as the liquid nutritional supplements described WO 01/66803 PCT/USO1/07608 -22 in Mulchandani et al., 1992 U.S..Patent No. 5,108,767), or they may be presented as a dry product for reconstitution with water or other suitable vehicles before use. Liquid preparations of folic acid, and other vitamins and minerals may come in the form of a liquid nutritional supplement specifically designed for ESRD patients. Such liquid preparations may be 5 prepared by conventional means with pharmaceutically acceptable additives such as suspending agents (e.g., sorbitol syrup, methyl cellulose or hydrogenated edible fats); emulsifying agents (e.g., jecithin or acacia); non-aqueous vehicles (e.g., almond oil, oily esters or ethyl alcohol);.preservatives (e.g., methyl or propyl p-hydroxybenzoates or sorbic acid); and artificial or natural colors and/or sweeteners. 10 For topical administration, herbal components may be combined in admixture with at least one other ingredient constituting an acceptable carrier, diluent or excipient in order to provide a composition, gIch as 'acreati, gel, solid, paste, salve, powder, lotion, liquid, aerosol treatment, or the like, which is niost suitable for topical application. Sterile distilled water alone and simple cream, ointment and gel bases may be employed as carriers of the herbal 15 components. Preservatives and buffers may also be added. The formulation may be applied to a sterile dressing, biodegradable, absorbable patches or dressings for topical application, or to slow release implant systems with a high initial release decaying to slow release. For a more complete -oveiview and discussion of herbal-based compositions see Earl Mindell, Earl Mindell's HerbBible. Simon & Schuster (1992); Culpeper's Complete Herbal, 20 W. Foulsham & Co., Ltd. (originally published in the mid 1600's); and, Rodale's Illustrated Encyclopedia of Herbs!,Rdle Press (1987). Standardized Herbal Composition. As used herein, a "standardized herbal composition" or a "charadterized herbial composition" refers to a particular herbal composition which is chosen as the staidard-'ierbal composition for evaluating batch herbal compositions 25 which have the same, similar or different components as the components of the standardized herbal composition. Sometinies herein also.referred to as the "master herbal composition." Standardized herbal cornposiiins: aie generally herbal compositions which have been well characterized and which deoinnstrae the desired biological responses in a particular biosystem. Standardizedheibal compositions Ere usually standardized by chemical tests well 30 known to one skilled in the'ari and are properly stored for long term usage and reference. The standardized herbal composition is used to establish a standardized HBR Array based on observations and me'asui-ei6fis for the plants (i.e., plant-related data), markers and BioResponses so -as to characterine the herbal composition.

WO 01/66803 PCT/USO1/07608 -23 Batch Herbal Composition. As used herein, a "batch herbal composition" refers to any test herbal composition which is used to establish a HBR Array based on observations and measurements for the plants and markers so as to characterize the herbal composition. Sometimes herein also referred to as a "test" or "batch" herbal composition. Observations and 5 measurements of BioResponses may or may not be included. The herbal compositions used to establish the standardized herbal composition may also be referred to as "batch herbal compositions" until designated as "standardized herbal compositions." Batch. As used herein, a "batch" refers to a particular quantity of an herbal composition which can be identified as to some particular attribute so as to distinguish it from 10 any other particular quantity of that same herbal composition. For example, one batch of an herbal composition may differ from another batch of that same herbal composition in that one of the batches was harvesteda a different time or in a different geographical location than the other batch. Other differences that distinguish particular batches may include, but are not limited to, the following .1) the particular plant part used (e.g., the root of an herb was used in 15 one batch while the leaves of that same herb were used in a different batch); 2) the post-harvest treatment of the individual, herbs'or herbal composition (e.g., one batch may be processed with distilled water while a different batch may be processed with Hydrogen Chloride to simulate the acidity of the human stomach); and, 3) the relative proportions of the individual herbs in an herbal composition (e.g., one batch may have equal parts by weight or volume of three 20 different herbs while another batch has proportionally more of one herb than the other two). Biosystem. As used herein, a "biosystem" refers to any biological entity for which biological responses may be observed or measured. Thus, a biosystem includes, but is not limited to, any cell,'tissu ofigan, whole organism or in vitro assay. Biological Activity. As used herein, the "biological activity" of an herb refers to the 25 specific biological effect peculiar to ag herbal composition on a given biosystem. Plant-Related Data. As used herein, "Plant-related data" refers to the data collected on the herbal composition including, but not limited to, data about the plants, their growing conditions and the handling of the b giants during and after harvesting. The plant-related data also includes the relative proportions of the components in an herbal compositions, wherein the 30 components may be different plant parts, different plant species, other non-plant ingredients (e.g., insect parts, chemical'drigs) or any 'combinations of these variables. Plant-related data which madybe gathered for an herbal composition includes, but is not limited to, the following: 1) theplant species (and, if available, the specific plant variety, WO 01/66803 PCT/USO1/07608 -24 cultivar, clone, line, etc.) and specific plant parts used in the composition; 2) the geographic origin of the herbs, including the longitude/latitude and elevation; 3) the growth conditions of the herbs, including fertilizer types and amounts, amounts and times of rainfall and irrigation, average microEinsteins received per day, pesticide usage, including herbicides, insecticides, 5 miticides and fungicides, and tillage methods; 4) methods and conditions used for processing the herbs, including age/maturity of the herbs, soaking times, drying times, extraction methods and grinding methods; and 5) storing methods and conditions for the herbal components and the final herbal composition. Additionally, the standardized herbal composition may be analyzed chemically. 10 Chemical characterization may be accomplished by any chemical analysis method generally known by one skilled in the art. Examples of applicable chemical analyses include, but are not limited to, HPLC, TLC,- cherhical fingerprinting, mass spectrophotometer analyses and gas chromatography. Cell Banking Systeim. As used herein, a "cell banking system" includes a Master Cell 15 Bank (MCB) and a Working Cell Bank (WCB) of cells. The use of a cell banking system minimizes cell variability for herbal medicine testing, and is used for all types of cells in nucleic acid microarraf studies. Bioinformatics. As used herein, "bioinformatics" refers to the use and organization of information of biological interest. Bioinformatics covers, among other things, the following: 20 (1) data acquisition and analysis; (2) database development; (3) integration and links; and (4) further analysis of theresulting database. Nearly all bioinformatics resources were developed as public domain fteeware until the early 1990s, and much is still available free over the Internet. Some companieshaedevelobed proprietary databases or analytical software. Genomic or Genomics. As used herein, the term "genomics" refers to the study of 25 genes and their function. Genomics emphasizes the integration of basic and applied research in comparative gene mapping,;molecular cloning, large-scale restriction mapping, and DNA sequencing and computational analysis. 'Genetic information is extracted using fundamental techniques, such as DNA sequencing, protein sequencing and PCR. Gene function is deteriniied (1) by analyzing the effects of DNA mutations in genes on 30 normal development and health of the cell, tissue, organ or organism; (2) by analyzing a variety of signals encoded intheDNA sequence; and (3) by studying the proteins produced by a gene or system of related genes.

WO 01/66803 PCT/USO1/07608 -25 Proteomic or Proteoniics. As used herein, the term "proteomics", also called "proteome research" or "phenomire", refers to the quantitative protein expression pattern of a genome under defined conditions. As used generally, proteomics refers to methods of high throughput, automated analysis using protein biochemistry. 5 Conducting prote ome 'research in addition to genome research is necessary for a number of reasons. First, the level of gene expression does not necessarily represent the amount of active protein in a cell. Also, the gene sequence does not describe post-tranIsational modifications which are essential for the function and activity of a protein. In addition, the genome itself does not describe the dynamic cell processes which alter the protein level either 10 up or down. Proteome programs seek to characterize all the proteins in a cell, identifying at least part of their amino acid s6egaence of an isolated protein. In general, the proteins are first separated using 2D gels or HPLC and then the peptides or proteins are sequenced using high throughput mass spectromety. Using a 'computer, the output of the mass spectrometry can be 15 analyzed so as to link a gone and the particular protein for which it codes. This overall process is sometimes referred to as "functional genomics". A number of commercial ventures now offer proteomic services (ea., Pharmaceutical ProteomicsTM, The ProteinChipTM System from Ciphergen Biosystem; PerSepive Biosystems). For general information on proteome research, see, for example, J.S. Fruton, 1999, 20 Proteins, Enzymes, Genes: The Interplay of Chemistiry and Biology, Yale Univ. Pr.; Wilkins et al., 1997, Proteome Research: New Frontiers in Functional Genomics (Principles and Practice), Springer Verlag; A.J. Link1, 199, 2-D Proteome Analysis Protocals (Methods in Molecular Bioloay, -112, Humana Pr. Kamp et al., 1999, Proteome and Protein Analysis, Springer Verlag. 25 Signal Transduction As used herein, "signal transduction", also known as cellular signal transduction, refers to the athways through which cells receive external signals and transmit, amplify and diro thei initerially. Signaling pathways require intercommunicating chains of proteins that transmit the signal in a stepwise fashion. Protein kinases often participate in this cascade of reactions, since many signal transductions involve receiving an 30 extracellular chemical signal, which triggers the phosphorylation of cytoplasmic proteins to amplify the signal. Post-translational Modification. As used herein, "post-translational modification" is a blanket term used to cox'erthe alterations that-happen to a protein after it has been WO 01/66803 PCT/USO1/07608 -26 synthesized as a primary polypeptide. Such post-translational modifications include, but are not limited to, glycosylation, removal of the N-terminal methionine (or N-fonnyl methionine), signal peptide removal, aceIylation, formylation, amino acid modifications, internal cleavage of peptide chains to release smaller proteins or peptides, phosphorylation, and modification of 5 methionine. Array or Microarray. As used herein, an "array" or "microarray" refers to a grid system which has each position or probe cell occupied by a defined nucleic acid fragment. The arrays themselves are sometimes referred to as "chips", "biochips", "DNA chips" or "gene chips". High-density nucleic acid microarrays often have thousands of probe cells in a variety 10 of grid styles. Once the array is fabricated, DNA or protein molecules derived from a biosystem are added and some form of chemistry occurs between the DNA or protein molecules and the array to give some recognition pattern that is particular to that array and biosystem. Autoradiography of radiolabeled batches is a traditional detection strategy, but other options 15 are available, including fluorescence, colorimetry, and electronic signal transduction. Markers. As used herein, the term "markers" refers to any biological-based measurement or obseraiibh for a particular herbal composition that is characteristic of a particular biosystem which is being exposed to a particular batch of an herbal composition. The term "marker" encompasses both qualitative and qualitative measurements and 20 observations of a biosystem. The marker database constitutes a data set which characterizes gene expression patterns in response to -herbal therapies, wherein the patterns show which genes are turned on, off, ip, or down in response to specific herbal compositions. Thus, "markers" refers to any'bibloicallybased measurement or observation whose up- and down or temporal regulations; or qualitative or quantitative changes of expression levels in a 25 biosystem are used to characterize differential biological responses of a biosystem to an herbal composition. The particular batch'bf an herbal composition to which the biosystem is exposed may be an unknown herbal coniposition, a known herbal composition, or a standardized herbal composition. Examples 6fmarkers useful in accomplishing the present invention include, but 30 are not limited to, moleculainatkers, cytogenetic markers, biochemical markers or macromolecular markers. Macroinolecular markers include, but are not limited to, enzymes, polypeptides, peptides, sugars,, antibodies, DNA, RNA, proteins (both translational proteins and post-translational proteins),:nuleic !acids, polysaccharides.

WO 01/66803 PCT/USO1/07608 -27 Any marker that satisfies the definition of "marker" herein is appropriate for conducting the present invention. The term "markers" includes related, alternative terms, such as "biornarker" or "genetic marker" or "gene marker." There may be one or more primary markers along with secondary markers, or a hierarchy of markers for achieving the purposes of 5 increasing the discriminating power of a HBR array. Thus, selected molecular markers may be combined with various other molecular, cytogenetic, biochemical or macromolecular markers to enable an even more accurate, extended HBR Array. A molecular marker comprises one or more microscopic molecules from one or more classes of molecular compounds, such as DNA, RNA, cDNA, nucleic acid fragments, proteins, 10 protein fragments, lipids, fatty acids, carbohydrates, and glycoproteins. The establishment, generation and use 6f applicable molecular markers are well known to one skilled in the art. Examples of particularly useful technologies for the characterization of molecular markers include differential display, reverse transcriptase polymerase chain reactions (RT-PCR), large-sd ale sequencing of expressed sequence tags (ESTs), serial analysis 15 of gene expression'(SAGE), Western immunoblot or 2D, 3D study of proteins, and microarray technology. One skilled in the artof molecular marker technology is familiar with the methods and uses of such technology (06e, e.g:,' Bernard R. Glick and Jack J. Pasternak, Molecular Biotechnology, Principles and 'Applications of Recombinant DNA, Second Edition, ASM Press (1998); Mathew R. Walker and Ralph Rapley, Route Maps in Gene Technology, Blackwell 20 Science (1997); Roe et al., DNA-Isolation' and Sequencing, John Wiley & Sons (1996)James D. Watson et al., Recombinant DNA Second Edition, Scientific American Books (1992)). DNA, RNA and protein isolation and sequencing methods are well known to those skilled in the 'att. Exampl6s'of such well known techniques can be found in Molecular Cloning: A Laboratory Manual 2nd Edition, Sambrook et al., Cold Spring Harbor, N.Y. 25 (1989); Hanspeter Saluz afid J. P ost, A Laboratory Guide to Genomic Sequencing: The Direct Sequencing of Nativ'e Unloned 'NA (Biomethods Vol 1), Birkhauser (1988); and B. Roe et al., DNA Isolation and Sequencing, Wiley (1996). Examples of conventional molecular biology techniques include, bit are not limited to, in vitro ligation, restriction endonuclease digestion, PCR, cellular'transformation, hybridization, electrophoresis, DNA sequencing, cell 30 culture, and the like. Specific 'kits and tools available commercially for use in the present invention include, but are not limited lo, those useful for RNA isolation, PCR cDNA library construction, retroviral express n libraries, vectors, gene expression analyses, protein antibody purification; cyotowdiity assays, protein expression and purification, and high- WO 01/66803 PCT/USO1/07608 -28 throughput plasmid purification (see, e.g., CLONTECHniques product catalog, XIII(3), 1-32 (1998) or www.clontech.com; AtlasTM cDNA Expression Assays product catalog (1998); SIGMA@ product catalog (1997)). For discussions, methodologies and applications of oligonucleotide arrays, microarrays, 5 DNA chips or biochips, see, for example, U.S. Patent Numbers 5,445,934, 5,605,662, 5,631,134, 5,736,257, 5,741,644, 5,744,305, 5,795,714; Schena et al., Parallel human genome analysis: Microarray-based expression monitoring of 1000 genes, Proc. Natl. Acad. Sci. USA 93, 10614-10619 (1996); DeRisi etal., Exploring the Metabolic and Genetic Control of Gene Expression on a Genomic Scale, Science 278, 680-686 (1997); Wodicka, et al., Genome-wide 10 Expression Monitoring in Saccharomyces cerevisiae, Nature Biotechnology 15, 1359-1367 (1997); Pardee, Complete Genome Expression Monitoring: The Human Race, Nature Biotechnologv 15, 1343-134 (1997); Schafer et al., DNA Variation and the Future of Human Genetics, Nature Biotechnology 16, 3,3-39 (1998); DeRisi et al., Use of a cDNA Microarray to Analyze Gene Expression Patterns in Human Cancer, Nature Genetics 14, 457-460 (1996); 15 Heller et al., Discovery and Analysis of Inflammatory Disease-Related Genes Using cDNA Microarrays, Pr 6 c. Natl.rAcad.' Sci. USA 94, 2150-2155 (1997); Marshall et al., DNA Chips: An Array of Possibilities, Naturb Biotechnology 16, 27-31 (1998); Schena et al., Microarrays: Biotechnology's Discovery Platfoi- for Functional Genomics, Tibtech 16, 301-306 (1998); Ramsay, DNA Chips: State-bf-theart, Nature Biotechnology 16, 40-44 (1998); Chee et al., 20 Accessing Genetic Information with High-Density DNA Arrays, Science 274, 610-614 (1996); and Chen et al., Profiling ExprossionPatterns and Isolating Differentially Expressed Genes by cDNA Microarray System with Colorimetry Detection, Genomics 50, 1-12 (1998); P. Andrew Outinen et al., Characterizatibon of the stress-inducing effects of homocysteine, Biochem. J. 332, 213-221 (1998); and Gelbert et al., Will genetics really revolutionize the drug discovery 25 process, Curr Opin Biotechnol 8(6), 669-674'(1997). Other, more specific, references applicable to the instant invention include, but are not limited to, those addressing the expression technologies, such as ESTs (see, e.g., Michael R. Fannon, Gene expression-in normal and disease states - identification of therapeutic targets, TIBTECH 14, 294-298 (1996)); the generation of protein profiles (see, e.g., Robinson et al., A 30 Tyrosine Kinase Profile of Prostate Carcinoma, Proc. Natl. Acad. Sci. USA 93, 5958-5962 (1996)); chemical and'spectrosc6pic methods for identifying components of herbal compositions (Kojima et al., Siponins from Gliricidia sepium, Phytochemistry 48(5), 885-888 (1998)); the determination of fAictioenal antigens (see, e.g., Aris Persidis, Functional WO 01/66803 PCT/USO1/07608 -29 antigenics, Nature Biotechnology 16, 305-307 (1998)); HPLCs (see, e.g., Milton T. W. Hearn (Editor), HPLC of Proteins dptids. and Polynucleotides: Contemporary Topics and Applications (Analytical Techniques in Clinical Chemistry and Laboratory Manual), VCH Pub. (1991); electrophoresis (see, e.g., Westermeier et al., Electrophoresis in Practice: A Guide 5 to Methods and Applicaios of DNA aind Protein Separations, John Wiley & Sons (1997)); and cross-reactivity marker assays (see, e.g., Irving Millman et al., Woodchuck Hepatitis Virus: Experimental Infection and Natural Occurrence, Hepatology 4(5):817-823 (1984)). The use of structural genomics for solving the structures of all the proteins encoded for in completed genonies, wherein the methodology includes high-throughput direct structure 10 determinations and computational methods, is discussed by Terry Gaasterland, Structural genomics: Bioinformatics in the driver's seat, Nature Biotechnology 16, 625-627. For bioinformatics methodologies, see, for example, Andreas Baxevanis (Editor), Bioinformatics: A Practical Guide to the Analysis of Genes and Proteins, John Wiley & Sons (1998) and Luke Alphey, DNA Sequencing: From Experimental Methods to Bioinformatics (Introduction to 15 Biotechniques Series), 'Sprifnger Verlag (1997). Cytogenetic parameters include, but are not limited to, karyotype analyses (e.g., relative chromosome lengths, centromere positions, presence or absence of secondary constrictions), ideograms: (i.g. a diagraminatic representation of the karyotype of an organism), the behavior of chromosomes during mitosis and meiosis, chromosome staining and banding 20 patterns, DNA-protein interactions (also known as nuclease protection assays), neutron scattering studies, rolling circles (A.M. Diegelman and E.T. Kool, Nucleic Acids Res 26(13):3235-3241 (1998); Backert et alt, Mol. Cell. Biol. 16(11):6285-6294 (1996); Skaliter et al., J. Viol. 70(2)1132-1136 (1996); A. Fire and S.Q. Xu, Proc. Natl. Acad. Sci. USA 92(10):4641-4645 (1995)), and auoradiography of whole nuclei following incubation with 25 radiolabelled ribonucleotides. Biochemical Parameters include, but are not limited to, specific pathway analyses, such as signal transduction, proteiit'synthesis and transport, RNA transcription, cholesterol synthesis and degradation, glucogenesis and glycolysis. Fingerprinting. As used herein, the term "fingerprinting" as used herein refers to the 30 means of making a characteristic.profile of a substance, particularly an herb, in order to identify it. The term "fingerprint" as used herein refers to the display of the result of the particular means employed, for the fingerprinting.

WO 01/66803 PCT/USO1/07608 -30 Examples of tie vafious types of fingerprinting means include, but are not limited to, DNA fingerprinting, protein inigeiprinting, chemical fingerprinting and footprinting. DNA fingerprinting, or profiling, refers to a way of making a unique pattern from the DNA of particular biological source (e.g., a particular plant, plant species, genus of plant, plant 5 part or plant tissue). The NXfingerprint, or profile, can be used to distinguish that particular biological source from a different biological source. The patterns obtained by analyzing a batch using microarrays, 61lig6nucletide arrays, DNA chips or biochips are also referred to as "fingerprints". Protein fingerprinting refers to generating a pattern of proteins in a cell, tissue, organ or 10 organism, such as a plant,, which provides a completely characteristic "fingerprint" of that cell, tissue, organ or organism at that time. Chemical fingerprinting refers to the analysis of the low molecular weight chemicals in a cell and the resulting pattern used to identify a cell, tissue, organ or organism, such as a plant. The analysis is usually done using Gas Chromatography (GC), HPLC or mass spectrometry. 15 Footprinting refers to a imethod of finding how two molecules stick together. In the case of DNA, a protein is bbund to a labeled piece of DNA, and then the DNA is broken down, by enzymes or by chemical attack. This process produces a "ladder" of fragments of all sizes. Where the DNA is protected 1y the bound protein it is degraded less, and so the "ladder" appears fainter. Footprinting is a common technique for homing in on where the proteins that 20 regulate gene activity actdallybind to the DNA. The means, or methods, used to accomplish each type of fingerprinting are described in detail elsewhere herein. BioResponses. As sad herein, a "BioResponse" refers to any observation or measurement of a biological response of a biosystem following exposure to an herbal 25 composition. Sometimes herein also referred to as a "biological effect." A BioResponse is a qualitative or quantitative data pint for the biological activity of a particular herbal composition. BioRespons data includes'both dosage and temporal information, wherein such information is well known'to one skilled in the art of measuring responses of biosystems to various treatments. Thus, BioR sponse data includes information on the specific biological 30 response of a specific biosystein to a specific dosage of herbal composition administered in a particular manner for a'specificpe iod of time. BioResponses inclnie, blut are not limited to, physiological responses, morphological responses, cognitive responses, motivational responses, autonomic responses and post- WO 01/66803 PCT/USO1/07608 -31 translational modifications, such as signal transduction measurements. Many herbal compositions demonstrate more than one BioResponse (see, e.g., Kee Chang Huang, The Pharmacology of Chinese Hetbs, CRC Press (1993)). Some particular BioResponses may be included in more than one of the delineated groups or have aspects or components of the 5 response that encompassimore than one group. BioResponses applicable to the instant invention are well known to one skilled in the art. The following references are representative of the state of art in-the field: Kee Chang Huang, The Pharmacology of Chinese Herbs, CRC Press (1993); Earl Mindell, Earl Mindell's Herb Bible, Simon & Schuster (1992); Goodman & Gilman's The Pharmacological Basis of Therapeutics, Ninth Edition, Joel G. Hardman, et. al. 10 (eds.), McGraw Hill, Health Professions Division (1996); P. J. Bentley, Elements of pharmacology A prim6r ondag action, Cambridge University Press (1981); P. T. Marshall and G. M. Hughes, Physiology of mammals and other vertebrates, Second Edition, Cambridge University Press (1980); 6pot of the Committee on Infectious Diseases, American Academy of Pediatrics (1991); Knut Schmidt-Nielsen, Animal Physiology: Adaptation and Environment, 15 5th Edition, Cambridge University Press (1997); Elain N. Marieb, Human Anatomy & Physiology, Addison-WessleyPb. Co. (1997); William F. Ganong, Review of Medical Physiology (18th Ed), Appletbi & Lange (1997); Arthur C. Guyton and John E. Hall, Textbook of Medical Physiology, W. B. Saunders Co. (1995). A "physiological response" refers to any characteristic related to the physiology, or 20 functioning, of a biosysten Physiological responses on a cellular, tissue or organ level include, but are not limited to, temperature, blood flow rate, pulse rate, oxygen concentration, bioelectric potential; pH value, cholesterol levels, infection state (e.g., viral, bacterial) and ion flux. Physiological respon se on a whole organism basis include gastrointestinal functioning (e.g., ulcers, upset stomach' indigestion, heartburn), reproductive tract functioning (e.g., 25 physiologically-based in otinc& uterine cramping, menstrual cramps), excretory functions (e.g., urinary tract problems, kidiey ailments, diarrhea, constipation), blood circulation (e.g., hypertension, heart disordens) oxygen consumption, skeletal health (e.g., osteoporosis), condition of the cartilage and connective tissues (e.g., joint pain and inflammation), locomotion, eyesight (e.g.;imyopia, blindness), muscle tone (e.g. wasting syndrome, muscle 30 strains), presence or absence'of pain, epidermal and dennal health (e.g., skin irritation, itching, skin wounds), functioning of the endocrine system, cardiac functioning, nervous coordination, head-related health (e.g., hadaches, dizziness), age (e.g., life span, longevity) and respiration (e.g., congestion, respiratoiy aihiients).

WO 01/66803 PCT/USO1/07608 -32 A "morphological response" refers to any characteristic related to the morphology, or the form and structure, of a biosystem following exposure to an herbal composition. Morphological responses, regardless of the type of biosystem, include, but are not limited to, size, weight, height, widtholor, degree of inflammation, general appearance (e.g., 5 opaqueness, transparency,-paleriess), degree of wetness or dryness, presence or absence of cancerous growths, and the presence or lack of parasites or pests (e.g., mice, lice, fleas). Morphological responses of" awhole organism basis include, but are not limited to, the amount and location of hair growth (eg., hirsutism, baldness), presence or absence of wrinkles, type and degree of nail and skin growth, degree of blot clotting, presence or absence of sores or 10 wounds, and presence or absence of hemorrhoids. A "cognitive respbuse" refers to any characteristic related to the cognitive,'or mental state, of a biosystem following exposure to an herbal composition. Cognitive responses include, but are not limited t6, 'erceiving, recognizing, conceiving, judging, memory, reasoning and imagining. 15 A "motivational response" refers to any' characteristic related to the motivation, or induces action, of'a biosysteni following exposure to an herbal composition. Motivational responses include, but are not limited-to, emotion (e.g., cheerfulness), desire, learned drive, particular physiological needs (e.g, appetite sexual drive) or similar impulses that act as incitements to action (e.g., stamina, sex drive). 20 An "autonomic response"' efers to any characteristic related to autonomic responses of a biosystem following exposure to n herbal composition. Autonomic responses are related to the autonomic nervous system of the biosystem. Examples of autonomic responses include, but art not limited to, invol Intary-functioning (e.g., nervousness, panic attacks), or physiological needs (e.g., respiration, cardiac rhythm, hormone release, immune responses, 25 insomnia, narcolepsy).: BioResponses of cels, tissues, organs and whole organisms treated with various herbal compositions or herbal componients -are well known in the herbal arts. For example, the herbal compositions Sairei-to (TJ- i4), alismatis rhizoma (Japanese name 'Takusha') and hoelen (Japanese name 'Bukuryu)'wrdeach found to inhibit the synthesis and expression of 30 endothelin-1 in rafs (Hattori et al., Sairei-to may inhibit the synthesis of endothelin-1 in nephritic glomeruli,Nippon Jinzo Gakkai Shi 39(2), 121-128 (1997)). Interleukin (IL)-1 alpha production was significantly promoted by treatment of cultured human epidermal keratinocytes with the herbal medicine Sho-saiko-to (Matsumoto et al., Enhancement of interleukin-1 alpha WO 01/66803 PCT/USO1/07608 -33 mediated autocrine growth of dultured human keratinocytes by sho-saiko-to, Jpn J. Pharmacol 73(4), 333-336 (1997). Adding Sho-saiko-to to a culture of peripheral blood mononuclear cells obtained from healthy volunteers resulted in a dose-dependent increase in the production of granulocyte colony-stimulating-factor (G-CSF) (Yamashiki et al., Herbal medicine "sho 5 saiko-to" induces in vitro'grafiulocyte colony-stimulating factor production on peripheral blood mononuclear cells, J Clintab Immunol 37(2), 83-90 (1992)). These researchers concluded that the administration of Sho-saiko-to may be useful for the treatment of chronic liver disease, malignant diseases and acute infectious diseases where G-CSF is efficacious. Plasminogen activator inhibitor type 1 (PAI-1)-specific mRNA expression decreased and tissue-type 10 plasminogen activator (t-PA)-specific mRNA increased after treatment of human umbilical vein endothelial cells (HUVECs) with the saponin astragaloside IV (AS-IV) purified from the Chinese herb Astragalus ineibrainaceus'(Zhang et al., Regulation of the fibrinolytic potential of cultured human umbilicWi ein ndothelial cells: astragaloside IV down regulates plasminogen activator iihibitor-l and up regulates tissue-type plasminogen activator 15 expression, J Vase Res 34(4), 273-280(1997)). One component out of four components isolated from the roots of Panax ginseng was found to be a potent inducer of IL-8 production by human monocytes and THP.l cells, and this induction was accompanied by increased IL-8 mRNA expression (Sonoda e al., Stimulation of interleukin-8 production by acidic polysaccharides from the root of panax ginseng, Immunopharmacology 38(3), 287-294 20 (1998)). By flow cytometric -analysis, the expression of Fc gamma 11/111 receptors and complement receptor 3 (CR3) on macrophages were found to be increased by treatment with the Kampo-herbal medicine Toki-shakuyakusan (TSS) (Cyong, New BRM from kampo-herbal medicine, Nippon Yakurigak Zaishi 110 Suppi 1, 87P-92P (1997)). Using computer image analysis, Chen metal. (m aly si for intercellular adhesion molecule-1 expression in 25 MRI/lpr mice: effects of Chinese herb medicine, Chung Hua I Hsueh Tsa Chih 75(4), 204-206 (1995)) found that the distribution intensity of intercellular adhesion molecule-1 (ICAM-1), immunoglobulins and C3were significantly decreased in MRL/lpr mice after treatment with the Chinese herb stragalin. Western blot analysis showed that tetradrine, isolated from a natural Chinese herbal medicine, inliibited'signal-induced NF-kappa B activation in rat 30 alveolar macrophages (Chei et al., Tetrandrine inhibits signal-induced NF-kappa B activation in rat alveolar macrophages, Biocliem BiophysiRes Commun 231(1), 99-102 (1997)). Algorithn.' As usedherein, an "algorithm" refers to a step-by-step problem-solving procedure, especially an established, recursive computational procedure with a finite number WO 01/66803 PCT/USO1/07608 -34 of steps. Appropriate algorithms for two- and three-dimensional analyses of the plant-related, marker and-BioResponse data sets are well known to one skilled in the computational arts. Such algorithms are useful in constructing the Herbal BioResponse Arrays of the present invention. For general information on algorithiis, see, for example, Jerrod H. Zar, 5 Biostatistical Analysis, second edition, Prentide Hall (1984); Robert A. Schowengerdt, Techniques for image prodessing and classification in remote sensing, Academic Press (1983); Steven Gold et al., New Algorithms for 2D and 3D Point Matching: Pose Estimation and Correspondence, Pattern Recognition, 31(8):1019-1031 (1998); Berc Rustem, Algorithms for Nonlinear Programming an Mluultiple-Objective Decisions, Wiley-Interscience Series in 10 Systems and Optimization, John Wiley & Sons (1998); Jeffrey H. Kingston, Algorithms and Data Structures: Design, Correctness, Analysis, International Computer Science Series, Addison-Wesley Pub. Co.,(197); Steven S. Skiena, The Algorithm Design Manual, Springer Verlag (1997); and Marcel F.Neuts, Algorithm Probability: A Collection of Problems (Stochastic Modeling), Chapman & Hall (1995). For information more specific to the 15 application of algorithins to genetic-based data, see, for example, Dan Gusfield, Algorithms on Strings, Trees, and Sequences: Computer Science and Computational Biology, Cambridge University Press (1997); Melanie Mitchell, An Introduction to Genetic Algorithms (Complex Adaptive Systems), MIT Press (1996); David E. Goldberg, Genetic Algorithms in Search, Optimization and Machine earning, Addison-Wessley Pub. Co. (1989); Zbigniew 20 Michalewicz, Genetic Algorithms + Data Structures = Evolution Programs, Springer Verlag (1996); Andre G. Uitterlindei 'and Jan Vijg, Two-Dimensional DNA Typing: A Parallel Approach to Genome Analysis, Ellis Horwood Series in Molecular Biology, Ellis Horwood Ltd. (1994); and Pierre Baldi and Soren Brunak, Bioinformatics: The Machine Learning Approach (Adaptive Computatin and Machine Learning), MIT Press (1998). 25 Combinatorial Cheiistry. As used herein, "combinatorial chemistry" refers to the numerous technologies usea tocdreat6 hundreds or thousands of chemical compounds, wherein each of the chemical compounds differ for one or more features, such as their shape, charge, and/or hydrophobic characteristics C6mbinatorial chemistry can be utilized to generate compounds which are cheii!al v anations of herbs or herbal components. Such compounds 30 can be evaluated using the methods of the present invention. Basic combinatorial. chemistry concepts are well known to one of ordinary skill in the chemical arts and can also be foiind in Nicholas K. Terrett, Combinatorial Chemistry (Oxford Chemistry, Masters), Oxford Univ. Press (1998); Anthony W. Czarnik and Sheila Hobbs WO 01/66803 PCT/USO1/07608 .35 Dewitt (Editors), A Practical Guide to Corbinatorial Chemisf, Amer. Chemical Society (1997); Stephen R. Wilson (Editor) and Anthony W. Czarnik (Contributor), Combinatorial Chemistry: Synthesis and Application, John Wiley & Sons (1997); Eric M. Gordon and James F. Kerwin (Editors), Combinatorial Chemistry and Molecular Diversity in Drug Discovery, 5 Wiley-Liss (1998); Shmuel Cabilly (Editor), Combinatorial Peptide Library Protocols (Methods in Molecular Biology), Human Press (1997); John P. Devlin, High Throughput Screening, Marcel Dekker (1998); Larry Gold and Joseph Alper, Keeping pace with genomics through combinatorial chemistry, Nature Biotechnology 15, 297 (1997); Aris Persidis, Combinatorial chemistry, Nature Biotechnology 16, 691-693 (1.998). 10 EXAMPLES Example 1. Establishing a Standardized HBR Array for Selected Herbal Compositions. The basic scheme for establishing a Standardized HBR Array is provided in Figure 1. Definitions of each component of the schematic are provided above. 15 Following selection ofan herbal Composition of interest, data is collected for various traits associated with theherblal composition, including, but not limited to plant-related characteristics and marker and Biokesponse infonnation. Plant-related data includes, but is not limited to, the plant species, specific plant parts, geographic origin of the plants in the herbal composition, the growth conditions of the plants, 20 the processing methods usedtolprepare the herbal components, storage methods and conditions, and various chemical analyses of the herbal composition. Marker information includes qualitative and quantitative data for markers collected after exposure of a biosystem to the herbal compost. Applicable makers include, but are not limited to, molecular markers, cytogenetic markers biochemical markers and macromolecular markers. BioResponse 25 information includes qualitativ& and quantitative data for biological responses collected after exposure of a biosystem. toth6 herbal composition. Each type of data (eg. .chemical, marker, BioResponse) can be obtained using one or more assays on the samb, similar, substantially similar, or different batches of the herbal composition of interest. Such di ffeirnt assays can be conducted at the same or different times. 30 In addition, data can be colleted for the same or different markers at the same or different times. Similarly, BioResponse data can be collected for the same or different biological responses at the same or different times. Thus, collection of the data for the HBR Array is either collected at one tim or collected on an on-going basis. Where a biosystem is exposed WO 01/66803 PCT/USO1/07608 -36 to an herbal composition so as to collect data, information is recorded on the administered dosages of the herbal composition as Well as treatment times. BioResponse data may also consist of post-translational modifications, such as measurements of signal transduction. After collection of tivo or more types of data (e.g., data for two or more markers and a 5 BioResponse; data for plant-related traits and data for a BioResponse), the data is analyzed using algorithms so as to create 2- and/or 3-dimensional Herbal BioResponse Arrays. Various statistical parameters may be calculated for the HBR Array and may become part of the HBR Array data set. These statistical parameters may include, but are not limited to, means, standard deviations, correlation or match (or mismatch) matrices, ratios, regression 10 coefficients, and transformed values (e.g., arcsin percentage transformations of the raw data). Thus, the HBR Array may consist of the raw data as well as certain calculations, distributions, graphical presentations and other data manipulations associated with the raw data. Particular examples of such infonnafion include, but are not limited to, digital images, scatter graphs, cluster analyses and large scale gene. expression profiles for marker data. 15 The-total accumulated data and resultant analyses constitute a standardized HBR Array for the particular herbal composition used to establish the HBR Array data set. Due to the iterative nature of the processed to establish and maintain an HBR Array for an herbal composition, such arrays can be viewed as either static at any one point in time or dynamic over time. 20 The resulting analyses canl identify subsets of the standardized HBR Arrays which are correlated (positively or negatively) or associated (i.e., showing a general trend) with one or more specific biological' activities of any particular herbal composition. Example 2. Establishing a Batch HBR Array for Batch Herbal Compositions. The basic scheme for establishing a HIBR Array for a batch of an herbal composition is 25 provided in Figure 2. Definitions of each component of the schematic are provided above. The procedure for establishing such an array is the same as that set forth immediately above for the standardized HBR Array. Generally, the amount, of data collected for a batch HBR Array will be less than that collected to establish a standardized HBR Array. However, data collected for a batch herbal 30 composition may be added to an established HBR Array or used to establish a new standardized HBR Array.*' Generally, the only data collected for a batch herbal composition is that data which has been found to be highly-correlfed or associated with the desired biological activities of the WO 01/66803 PCT/USO1/07608 -37 herbal composition being tested. Tor example, if it has been determined that a particular subset of plant-related and marker data is highly correlated to a desired biological activity of a particular herbal compositiogn(based on the standardized HBR Array data and analyses discussed above), it is only necessary to test the batch herbal composition for that subset of 5 traits in order to determine whether or not the batch has the desired biological activity. By comparing the data obtained-for that subset of traits obtained from the batch (i.e., the batch HBR Array) with the -standardized HBR Array.for that particular herbal composition, one skilled in the art can determine whether or not that particular batch has the desired biological activity. 10 Example 3. Establishing and Using a Major Data Set. The basic scheme for establishing and using a major data set for an herbal composition is provided in Figure 3.. Defiftions of each component of the schematic are provided above. The first step, is the establishment of a major data set for a selected herbal composition or batch herbal composition This is accomplished by exposing a biosystem to the herbal 15 composition and collecting the resultaiit marker information which will constitute the major data set. In most, but not instances, the major data set will consist of genomics and/or proteomics data in'the form of an array, such as an array obtained with a DNA biochip. Next, the majot data set is analyzed to see if differential expression/results have been obtained for the tested herbal composition. Differential expression/results are necessary in 20 order to generate meaningful algorithms in the next step. Examples of such differential expression/results include, but are not limited to, indications that certain genes are up- or down-regulated in respoisieto exposure to the herbal composition or that the levels of certain proteins have been increased 6r decreased in response to the exposure. If no meaningful orusefal differential expression/results are obtained, then it is 25 necessary to repeat the exposure and marker collection step. If it is believed that experimental error lead to the lack of aadeq'uate result the first time then the exposure/data collection step can be repeated with all of the variables the same as the first time (e.g., same biosystem, same marker set, same experineital protocol, etc.). However, it may be necessary to vary the biosystem sampling (e.g., type of cell utilized, stage of cell growth), use a different marker set 30 and/or change the experimental protocol in order to get differential expression/result.. Example 4. Using HBR Array Information. The HBR Array ifiirfiation discussed herein can be used for many different purposes including, but not limited to, the following: 1) evaluating the components of an herbal WO 01/66803 PCT/USO1/07608 -38 composition; 2) predicting the BioResponse of an herbal composition; 3) determining which marker information is most'highly correlated with a particular BioResponse of an herbal composition; 3) determining what data set of information (i.e., plant-related data, marker data, and BioResponse data) is most correlated with a particular BioResponse of an herbal compost; 5 4) determining which type of biosy stem is best for evaluating the biological activity of an herbal composition; 5) adjusting or changing the components of a herbal composition so that the HBR Array of that herbal composition corresponds to a standardized HBR Array for the same or substantially the same herbal composition; 6) adjusting or changing the components of an herbal composition so that the herbal composition will have the desired biological activity; 10 7) measuring the relatedness of different herbal compositions; 8) creating and updating standardized HBR Arrays; 9) identifying specific components (e.g., plant parts, proteins, molecules) which retain the desired biological activity of an herbal composition; 10) determining which compoinntiof afrherbal composition can be eliminated while maintaining or improving the desired biological activity of the herbal composition; 11) identifying one or 15 more previously unknown biological activities for an herbal composition; 12) aiding in the design of therapeutics which"incltide herbal and non-herbal components, such as chemically synthesized drugs or pharmaceuticals and 13) utilizing the HBR Array information to complement combinatorial cheimistlry methods of designing therapeutics. Each of these embodiments of the present invention can be accomplished by one skilled in the applicable art 20 using the methods and tools provided herein. Example 5. Quality Control. The HBR Array technology of the present invention is used to correlate or to determine a substantial equivalence ifaspecific batch of an herbal composition (single herb or multiple herbs of a formula) to 'a tandardized, or master, batch of a same or substantial similar herbal 25 composition. The HBR Ariays utilized in this process include the 'acceptable range of quantitative variation for 'eachof the biological effects (i.e., BioResponse), and possibly a global score composed of weighed values assigned to each of the biological effects, which may consist of markers from multiple biochemical pathways of a biosystem. "Data mining" refers to aprocess used to determine or select which subset of biological 30 effects is the minimum numiber of biological effects required in any specific HBR Array. The information for data mining results from exposing a biosystem (e.g., a cell line) in a dose dependent manner to a standaidized'herbal composition to establish a standardized HBR Array. This standardized BR Array can then be compared to various HBR Arrays WO 01/66803 PCT/USO1/07608 -39 established for test herbal conpositions. These test herbal compositions include, but are not limited to, different batches prepared at different dates; different batches prepared from raw herbs collected at different times; and different batches prepared from raw herbs collected at different locations. 5 Example'6. Improving an Herbal Composition or Identifying New Uses for an Herbal Composition. HBR Arrays are generated by exposing biosystems to either extracts from individual herbs of a formula, or to extracts from the whole formula, and examining the biological effects of the extracts. The observed biological effects can be from multiple biochemical pathways of 10 a biosystem and/or from multiple tissues of an animal, wherein various markers are evaluated for their corresponding qualitative and/or quantitative changes. The resulting HBR Arrays can be compared to novel HBR Arrays or to similar HBR Arrays from different herbal compositions or herbal conipositions prepared by different processes. This procedure is useful for selecting a given set of biological effects and the minimum number of markers required to 15 predict that a given batch herbal composition has the given set of biological effects. In order to construct HBR' Acrays, one skilled in the art utilizes various data mining tools including, but are not limited to. statistical analyses, artificial intelligence, and database research on neural work. The statistical niethods of choice include, but are not limited to, basic exploratory data analysis (-DA), graphic EDA (such as bushing) and multivariate exploratory 20 techniques (e.g., cluster analysis, discriminating factor analyses, stepwise linear on non-linear regression, classification, tree) (see,"e.g., STATISTICATM, software packages from StatSoft, Tulsa, OK 74104; Tel: 918-749-1119; Fax: 918-749-2217; www.statsoft.com). Data mining tools sreiused to explore large amounts of HBR Array data in search of constructing an HBR Array and consistent pattern within, between or among various HBR 25 Arrays. The procedure consists of exploration, construction of an HBR array, and validation. This procedure is typicallyrepeated iteratively until a robust HBR Array, or standardized HBR Array, is identified. Example 7. Establishing a Standardized HBR Array for Ginseng Recipes. For the purposes o his exaniple, standard ginseng is chosen to be Panax Ginseng C.A. 30 Meyer G115 gr6wn either in Manchuria or in Korea. The climate for growth is between -10 to +10 0 C with an annual rainfall'f 50-100 cm'(see Huang in The Pharmacology of Chinese Herbs, (1993) pp 21-45, CRC Prss, Boca Raton, FL, fully incorporated by reference). Ginseng batches will first be 6haratterized by geographic origin, species, plant part (e.g., rhizome, root, WO 01/66803 PCT/USO1/07608 -40 leaf skin, seed, bud and flower); growth conditions, processing methods and storage conditions both before and after processing. Verification of chemical content for these batches will be performed by qualitative HPLC analysis for determination of ginsenoside saponins (e.g., Ro, Ral, Ra2, Rbl, Rb2, Rb3, Rc, Rgl, Rg2, Rd, Re, Rf, Rhl, Rh2, NG-R2 and Z-R1), including 5 TLC qualitative analysis for lipophilic constituents (see, Elkin et al., Chung Kuo Yao Li Hsueh Pao (1993) 14: 97-100 and Yoshikawa et al., Yakugaku Zasshi (1993) 113: 460-467). The saponin content of different herbs should be between 2.1 and 20.6% (by weight) depending on the species (see Table 1). These data will then be stored, preferably in the memory of a computer processor, for further manipulation. 10 Table 1. Saponin Content of Different Ginseng Herbs.* Species Total saponins (% by weight) Panax ginseng C.A. Meyer 2.1-4.4% Panax quiquefolius 4.9% Panax notoginseng and Panaxjaponica 13.6-20.6% Panaxjaponica var. majoi 9.34% *from Huang in The Pharmacolg of Chinese Herbs, (1993) page 29, CRC Press, Boca Raton, FL. Expression biomarkers for standard ginseng (i.e., G1 15) include the following: IL-8, IL-2, GM-CSF, NfKB, ICAM-1, interferon gamma, choline acetyl transferase, trk A, nerve 15 growth factor (Kim et al.,,Planta Med (1998) 64: 110-115; Sonoda et al., Immunopharmacology (1998) 38: 287-294;.Baum et al., Eur J Appl Physiol (1997) 76: 165 169; Iwangawa et a., Free. Radic Biol Med (1998) 24: 1256-1268; Rhind et al., Eur J Appl Physiol (1996) 74: 348-360). Alternatively, for a broader batch size, the 400,000 oligonucleotide group/1.6 cm 2 chip of Affymetrix can be used (U.S. Pat. No.5,556,752). The 20 expression biomarkers foi standard ginseng will be prepared by nucleic acid microarray technology using either photolithography, mechanical microspotting or ink jet application (see Schena et al., TIBTECH (1998),16: 301-306). Selected sets of cells will be contacted with standard ginseng for varying periods of time, under varying conditions to generate multiple microarray sets. The micioariay set'swill then be analyzed by hybridization-based expression 25 monitoring of biochemical extracts via deduction of steady state mRNA levels from fluorescence intensity at each position on the microarrays (Schena et a., Science (1995) 270: 467-470; Schena et al., Proc Natl Acad Sci USA (1996) 93: 10614-10619; Lockhart et al. Nat WO 01/66803 PCT/USO1/07608 -41 Biotechnol (1996) 14: 1675-1686; DeRisi et al., Nat Genet (1996) 14: 457-460; Heller et al., Proc Natl Acad Sci USA (1997) 94: 2150-2155). The array data sets are then input into algorithms to generate statistical expression biomarker values for standard ginseng. Biochemical biomarkers for standard ginseng include quantitative analysis for increases in 5 cycloheximide sensitive [ H]-leucine incorporation proportional to protein synthesis and [ 3

H]

thymidine incorporation ,reflective of mitosis. (.see Yamamoto et al., Arzneimittelforschung (1977) 27: 1169-1173). For biochemical biomarkers, bone marrow cells will be contacted with standard ginseng for varying time periods under varying conditions in the presence of [ 3

H]

thymidine (for DNA synthesis) or in the presence and absence of cycloheximide and [ 3

H]

10 leucine (for protein synthesis) to perform multiple quantitative analysis of biochemical biomarkers (i.e., BBM sets). The BBM sets are then input into algorithms to generate statistical biochemical biomarker vahies. foi standard ginseng. Statistical data will then be stored, preferably in the memory of a computer processor, for further manipulation. Biological response of a biosystem (i.e., BioResponses) will be determined using cells 15 and whole animals. -For cells, ginseng batches will be exposed to specific cell types, including, but not limited to, fibroblasts, macrophages, monocytes, PMNL, LAK cells, B16-F1O melanoma cells, THP-I cell and hippocampalheurons at a concentration of 0.5 mg/ml to 100 mg/ml. For animal treatments,0.5-100 mg/kg of ginseng herbal extract will be administered orally, by intraperitoneal injection or subcutaneous injection. 20 To determine a biological response of a biosystem to standardized ginseng, human ovarian cancer cells willbe inoculated into nude mice, which results in the formation of palpable tumors. After tumor, formation the mice will be treated by co-administration of cis diamminecichloroplatinum and 'stidaid ginseng. Mice will be examined for tumor growth inhibition, increase in survival time and lowered adverse side-effects on hematocrit values and 25 body weight (Nakata et'al., Jpn J Caneer Res (1998) 89:733-740). The assay will be repeated using various concentrations of standard ginseng to generate measures of central tendency, dispersion and variability for each Variable. The data collected will then'be subjected to multidimensional analysis to generate multivariant normal distribution sets as a means of determining a baseline correlation between 30 biological activity and standard ginseng (see Zar, J. H., in Biostatistical Analysis, 2nd ed. (1984), pp 328-360, Preritic'2Hall, Englewood Cliffs, NJ). A second independent determination of a biological response of a biosystem to standard ginseng will be the effect of standard ginseng onphysicalperforinance during exercise. Rats will be treated for 4 days with WO 01/66803 PCT/USO1/07608 -42 standard ginseng at various ofncentiations (between 0.5-100 mg/kg/day) and animals will be tested for increased plasma free fatty acid level and maintenance of glucose level during exercise at approximately70% VO2max (see Wang et aL, Planta Med (1998) 64130-133). The data generated will be collectdd and then subjected to multidimensional analysis to 5 generate multivariant normal distribution sets as a means of.determining a baseline correlation between biological activity and standard ginseng (see Zar, J. H., in Biostatistical Analysis. 2nd ed. (1984), pp 328-360, Prentice Hall, Englewood Cliffs, NJ, Herein, fully incorporated by reference). The distribution sets for each BioResponse are then put into algorithms to generate statistical values for standard ginseng. Statistical data will then be stored, preferably in a 10 memory of a computer processor, for further manipulation. Each of these steps (i.e., chemical analysis, generation of biomarker information and determination of responseg:of 'a biosystem) is reiterated to generate a large database of statistical values. These values are compiled and input into an algorithm to generate 2- and 3 dimensional Herbal Respons Arrays (HBR Array) for standardized ginseng. Through 15 reiteration, the resulting arrays (i.e',Standardized Arrays) display the highest correlation between composition (iiiluding' growth conditions), biomarker information and biological response for standardized gindgii . By determining two or more known associated variables for composition and bionaikr information values via display on an HBR Array for a test batch, the values for biological response variables can be predicted for the test batch by 20 comparing test valuesagainst Standardized HBR Array values for standardized ginseng. The resulting prediction will b6 used to evaluate the quality of a given ginseng batch without necessitating the use-of ai'observecbiological-response of a biosystem (see Example 2). Example 8. Evaluation of a Selected Herbal Composition of Ginseng Using a Subset of Variables Correlated with a Specific Biological Response. 25 To evaluate the quality of 'a test batch herbal composition, data is first collected concerning the plant-relatedPrameteis for the herbs in the selected herbal composition (e.g., plant species, plant parts, geog aphic origin, growth conditions, processing methods and storage conditions). The selected herbal composition is then manipulated such that chemical analysis can be performed determine the chemical content of the herb (see Elkin et aL., 30 Chumg Kuo Yao Li Hsueh Pao (1993)14: 9'-100 and Yoshikawa et aL., Yakugaku Zasshi (1993) 113: 460-467). Preiously dbtainbd ginseng data has demonstrated a strong correlation between oxygen consumption during aerobic exercise performance and the presence of a WO 01/66803 PCT/USO1/07608 -43 subset of saponin components, especially Rgl and Rbl (Wang et al., Planta Med (1998) 64: 130-133). The test batch is then exposed to test cells including, but not limited to, fibroblasts, macrophages, monocytes, PMNL, LAK cells, B16-Fl0 melanoma cells, TBP-1 cells and 5 hippocampal neurons at a concentration of 0.5 mg/ml to 100 mg/ml to determine expression biomarker values. .mRNA is isolated from exposed cells which is subsequently manipulated to serve as a substrate for hybridization-based expression monitoring of biochemical extracts using microarrays comprising IL-8, IL-2 and Interferon gamma cDNA (Schena et al., Science (1995) 270: 467-470; Schena et at., Proc Natl Acad Sci USA (1996) 93: 10614-10619; 10 Lockhart et al., Nat Biotechnol (1996) 14: 1675-1680; DeRisi et al., Nat Genet (1996) 14: 457 460; Heller et al., Proc NatlAcad Sci USA (1997) 94: 2150-2155). Previously obtained ginseng data has demonstrate a'strong correlation between oxygen consumption during aerobic exercise performance and the induction of the expression biomarkers IL-8, IL-2 and Interferon gamma in test celi (Venkatraman et al.; Med Sci Sports Exerc (1997) 29: 333-344 15 and Wang et al., Planta Med (1998) 64: 130-133). For biochemical biomarkers, rat bone marrow cells will then be exposed to the test batch and assayed for [ 3 H]-thymidine incorporation reflective ofmitosis. Previously obtained ginseng data has demonstrated that Rbl and Rgl show a strong correlation with DNA synthesis in rat bone marrow cells (Yamamoto et al., Arzneirnittelforschung (1978) 28: 2238-2241). 20 After reiterative aiialysis, data from each assay will be input into an algorithm to generate a test HBR ari-aytf6r the-selected herbal composition based on the enumerated plant related data, including chemical analyses, and data concerning the subset of biomarkers. The quality of a test batch will bedefermined by comparing test HBR and standard ginseng Standardized HBR Array variables directed toward analysis of the above observations and 25 subsets, wherein the demonstration 6f the induction of IL-2, IL-8 and INF gamma mRNA in vitro and an increase in [ 3 H]thymidine incorporation in rat bone marrow cells (including data collected on growth conditions, origin, and verification of the saponins Rgl and Rbl) is predictive of an equivalent BioResponse effect of the test batch on oxygen consumption as that exhibited by standard ginseng. Based on this procedure it can be determined whether or not 30 the test batch is of a similar r different quality than that of the standard for the given biological response or biological response of interest.

WO 01/66803 PCT/USO1/07608 -'44 Example 9. Establishing a Standardized HBR Array for Huang Ling (HL) Recipes. For the purposes of this example, standard huang ling (HL) is chosen to be Coptis chinesis France, from southwest Asia, wherein growth conditions are well known to one 5 skilled in the art (see Huang in The Pharmacology of Chinese Herbs, (1993), pp 69 and 287 288, CRC Press, Boca Raton, FL). Dried rhizomes of Coptis chinesis France will be verified for chemical content by quantitative chemical analysis for determination arsenic, berberine, caeruleic acid, columbamine, copsine, coptine, coptiside-I, coptiside-II, coptisine, coreximine, epiberberine, ferulic acid, greenlandicine, isocoptisine, lumicaerulic acid, magnoflorine, 10 oxybererine, thalifendine, umbellatine, urbenine, worenine, palmatine, j atrorrhizine and colubamine (see also Zhu M., Chung Yao Tung Pao (1984) 9: 63-64). Content of the alkaloid berberine of different herbs should be-between 7-9% (by weight). These data will be stored, preferably in the memory of a computer processor, for further manipulation. Expression biomaikers fo standard HL include the following: NfKB; bel-2 analog, Al; 15 zinc finger protein, A20; IL-2 receptor; cell cycle probes; c-Ki-ras2; growth regulators probes and glucocorticoid receptor dependent apoptosis probes (see Chi et al., Life Sci (1994) 54: 2099-2107; Yang et al., Nannyn Schmiedebergs Arch Pharmacol (1996) 354: 102-108; Miura et al., Biochem Pharmacol (1997) 53; Chang K.S., J Formos Med Assoc (1991) 90: 10-14). Alternatively, for a broader batch size, the 400,000 oligonucleotide group/1.6 cm 2 chip of 20 Affymetrix can be used (U.S. Pt. No.5,556,752). The expression biomarkers for standard HL will be prepared by microarray technology as described in Example 1, including analysis and statistical data generation..Biochemical biomarkers for standard HL include increase in glucocorticoid receptor and inhibition of alpha-fetoprotein secretion in HL exposed HepG2 cells (see Chi et al., Life Sdi(1994) 54: 2099-2107). BBM sets are generated and analyzed as 25 described in Example 1 Statisticl data will then be stored, preferably in the memory of a computer processor, for fithr manipulation. Biological response abidsystem will be determined using cells and whole animals. Batches of the selected herbal composition will be exposed to specific cell types, including but not limited to, human HepG2 iepatoma cells, human embryonal carcinoma cells and 30 thymocytes at concentrationsfroin0.l-100mg/ml. For animal treatments 0.lmg-2g/kg of coptic herbal composition (i.e., HL) will be administered orally, by intraperitoneal injection or subcutaneous injection. To determine a biological response of a biosystem to standardized HL, human embryonal carcinomimclone, NT2/D1 is exposed to various concentrations of standard WO 01/66803 PCT/USO1/07608 -45 HL and cells will be examined for differentiation into cells with neuronal-like cell morphology (Chang K.S., J Formos Med Assoc (1991) 90: 10-14). The assay will be repeated to generate measures and analysis will be performed as described for ginseng in Example 1. A second independent determination of a biological response of a biosystem to standard HL will be the 5 effect of standard HL on dia rea due to enterotoxigenic Escherichia coli (ETEC). Patients with active diarrhea due to ETEC will be treated with various concentrations of HL (e.g., 2g/kg) and stool volumes will be determined (see, e.g., Rabbani G.H., Dan Med Bull (1996) 43: 173-185). The assay will be repeated to generate measures and analysis will be performed as described for ginseng in -Example 1. The distribution sets for each biological system are 10 then put into algorithms to generate statistical values for standard HL. Statistical data will then be stored, preferably in the memory of a computer processor, for further manipulation. Lastly, as in Exaimple 1, the Steps are reiterated to generate HBR arrays for standardized HL, wherein the resulting HBR arrays will then be used to predict biological activity and evaluate batch qutIity. Using'this method, a Standardized HBR Array can be 15 generated and updated periodicaliy. Example 10. Evaluation of a Selected Herbal Composition of Huang Ling Using a Subset of Variables Correlated with a Specific Biological Response. To evaluate the quality of a selected test batch of an herbal composition of Huang Ling, data is first collected concerning the plant-related characteristics (e.g., plant species, 20 plant parts, geographic origin, growth conditions, processing methods and storage conditions). The herbal composition isthen manipulated such that chemical analysis can be performed to determine the chemical conteiit of the composition (see also Zhu M., Chung Yao Tung Pao (1984) 9: 63-64). Previously obtained HL data has demonstrated terminal differentiation of human 25 embryonal carcinoma clones into, neuronal-like cells is strongly correlated with the presence of berberine (see Chang K.S., J Formos Med Assoc (1991) 90: 10-14). The test batch is then exposed to test cells including human -embryonal carcinoma clone, NT2/D 1 at a concentration starting at a non-toxic 6otientration (determination of which is within the skill of the ordinary artisan). mRNA is isolated from exposed cells which is subsequently manipulated to serve as 30 substrate for hybridization based expression monitoring of biochemical extracts using microarrays comprising 11-2 receptor and NficB; (see Chi et al., Life Sci (1994) 54: 2099 2107; Yang et al., Naunyn Schmiedebergs Arch Pharmacol (1996) 354: 102-108; Miura et al., Biochem Pharmacol (1997)53; Chang K.S., J Formos Med Assoc (1991) 90: 10-14; U.S. Pat.

WO 01/66803 PCT/USO1/07608 -46 No.5,556,752), and which can be used to determine down regulation of e-Ki-ras2 gene expression in said cells. r eviousliy obtained HL data has demonstrated terminal differentiation of human embryo oial carcinoma clones into neuronal-like cells is strongly correlated with induction of nitogen probes and down regulation of c-Ki-ras2 gene expression 5 (see Chang K.S., J Formos Med Assoc (1991) 90: 10-14). For biochemical markers, HepG2 cells are exposed to the test composition and cells are assayed for increase in glucodorticoid receptor and inhibition of alpha-fetoprotein secretion (see Chi et al., Life Sci (1994) 54: 2099-2107). Previously obtained HL data has demonstrated that inhibition of glucocorticoid induced apoptosis is strongly correlated with berberine-type 10 alkaloids (see Miura et al.; Biochem Pharmacol (1997) 53: 1315-1322). After reiterative analysis, data from each assay will be input into an algorithm to generate a test HBR array based on the enumerated observational data, chemical data and data concerning the subset of biomarkers. The quality of a test batch will be determined by comparing test HBR and standard HL 15 HBR Array variables directed toward analysis of the above observations and subsets, wherein the demonstration of the induction of IL-2 receptor and NfiB, the down regulation of c-Ki ras2 gene expression, an increase in glucocorticoid receptor and inhibition of alpha-fetoprotein secretion for HepG2 cells (to including data collected on growth conditions, origin, and verification of berberine alkaloid) is predictive of an equivalent BioResponse effect of the test 20 batch on terminal differentiation of human embryonal carcinoma clones into neuronal-like cells and inhibition of dexamethasone induced apoptosis as that exhibited by standard HL. Based on this procedure it cani be determined whether or not the test batch is of a similar or different quality than that of the standard. Example 11. Evaluation of Xiao Chai Hu Tang (sho-saiko-to) Using Two 25 Bioassays. To evaluate the qualityof three sources of Xiao Chai Hu Tang, two bioassays were used: 1) cell growth inhibitioniaiid 2) hepatitis B virus secretion from infected cells. The Xiao Chai IHu Tang composition is made from a mixture of 6-7 herbal plants (Radix Bupeuri, Rhizoma Pinelliae, Rhiiomd Zingiberis, Radix Scutellariae, Fructus Ziziphi, Codonopsis 30 Pilosula, Radix Ginseng and Radix Glycyrrhizae, see Table 2 for relative amounts, by weight).

WO 01/66803 PCT/US01/07608 -47 Table 2. Composition of Xiao Chai Hu Tang. Source Plant Species Radix Rhizoma Rhizoia Radix Fructus Codonopsis Radix Radix bupleuri pinelliae zingiberis scutellariae ziziphi pilosula ginseng glycyrrhizae Relative Amount by Weight Singapore 1 1 0.375 0.375 0.375 --- 0.375 0.375 Korea 1 0717 0.571 0.492 --- 0.429 -- 0.288 Taiwan 1 0.25 0.375 0.375 0.25 ---- 0.375 0.375 The three "recipes" Brigiite in either Singapore, Korea or Taiwan. Batches were evaluated for toxicity and f6 the ability to inhibit hepatitis B virus as detected by DNA 5 quantitation or detection of hepatitis B surface antigen (HbsAg) (see Dong et al., Proc Natl Acad Sci USA (1991) 88: ,8495-8499). Briefly, one gram of preparation was added with 10 ml of water. The mixture was boiled for 30 minutes. 'The supernatant was collected after centrifugation and filtered through a 0.22 ptm filter. Two cell tyes were used: a) 2.2.15 cells which secrete hepatitis B virons 10 (kindly provided by Professor . Ace; see Ace et al. Proc Natl Acad Sci USA (1987) 84: 1005 1009) and b) HepG2 cells (ATCC cat # HB-8065). One to fifty dilutions were used for each assay. The cell growth inhibition assay was performed for 72 hours. All other procedures were performed as described by Dnget al., Proc Natl-Acad Sci USA (1991) 88: 8495-8499. The results of the assays using the three batches is displayed in Table 3. Based on these data, 15 the Taiwan source would be selected as a standard herbal composition because of its low toxicity combined with its effedfiveness in reducing secretion HbsAG (which is proportional to viral release) by more thanaiaif.

WO 01/66803 PCT/USO1/07608 -48 Table 3. Bioassay of Xiao.Chai Hu Tang (Sho-saiko-to). Source Cell Growth Inhibition (%) Hepatitis B Virus (secreted) % Inhibition HepG2 2.2.15 DNA HbsAG Singapore 73 100 65 38 Korea 13 60 20 42 Taiwan 0 42 0 47 The data presented in Tables 2 and 3 for the Taiwan herbal composition constitute the 5 initial data for the standardized HBR Array for this herbal composition. Therefore, this data set would initially include the source of the herbal composition, the plant species and relative amounts of each the herbal composition, and two BioResponses (i.e., cell growth inhibition and hepatitis B virus secretion frin infected cells). Using the procedures set forth in the schematic of Figure 1 and in Examples 1 and 3, 10 additional data can be collected on plant-related data, markers and BioResponses for the standard herbal composition. This additional data is added to the initial standardized HBR Array to generate an expanded standardized HBR Array. Appropriate analyses of the resulting database can be conducted as set forth in the detailed description and the examples in order to ascertain the subset of Viriables which is most highly correlated or associated with the 15 BioResponse of interest. Batch HBR Arrays may be determined using the methods depicted in Figure 2 and in the-procedures of Examples 2 and 4. The resultant batch HBR Array can be compared to the standardized HBR Array so as to predict the BioResponss of the batch herbal compositions. Example 12. Herbal Preparation' 20 The standardized protocol for the herbal extract preparation was as follows: The ingredients of herbal raw materials With proper ratios were placed in a jacketed reactor and extracted with water at an elevated constant temperature with mixing. The solid was separated from the liquid with a 120-mesh screen. The resultant filtrate was collected and then concentrated by evaporating the water under reduced pressure. The concentrated liquor was 25 spray dried at elevated temperature-to yield granulated powder. This bulk substance was then formulated into the desireddosage form.

WO 01/66803 PCT/USO1/07608 -49 Example 13. Evaluation of Huang Qing Tang Huang Qing Tang (HQT) is an ancient Chinese botanical formula composed of four distinct herbs: Scutellariae (scute) I Glyyrrhizae (licorice), Paeonie lactiflora pallus (white peony root), and Fructus ziziph6 (date).' (Table 4). This herbal formula has been long used in Asia to 5 treat a variety of gastrointestinal ailments since 300 AD. Table 4. Herbal Ingredients of TCM Formula HQT Scientific Name Common Name Traditional Use Scutellariae Radix Scute Baical Used to reduce capillary permeability: to reduce Skullcap root inflammation: to treat enteritis and dysentery: increase the secretion of bile to treat jaundice: to relieve muscle spasms to treat coughing: to expel parasites. Glycyrrhizae Licorice iRot Used to-moisten the lungs and stop coughs: to relax spasm Radix and stop pain: to moderate the action of herbs; to reduce (Gancao) fire and release toxins. Fructus Ziziphi Date Has diuretic and strengthening effects. Paeonie lactiflora White Peony Used to suppress and soothe pain; to soothe ligaments and pallus radix Root ,purify the blood. Biological and Enzyme Assays 10 Table .5. Batch Properties (HQT) Property Batch'A Batch BC Origin Taiwan; Sun-Ten Taiwan, Sun-Ten Taiwan, Sun-Ten Preparation method Standard Standard Boiled 30 min. Plant part Root . Rdot Briefly, one gram of each batch of Huang Qing Tang (HQT) was added with 10 ml of water (1 mg/ml). The mitre was treated as outlined in Table 5. The supernatant was collected after centrifugation and filtered through a 0.22 pm filter. Two cell types were used to 15 test for biological effects of each batch of HQT: a) Jurkat T cells (ATCC cat #TIB-152) and b) HepG2 cells (ATCC cat # HB'8065). One to fifty dilutions were used for each assay. Frozen cells (10 7 /ml) were quickly thawed-in a water bath at 37 C. The cells were then diluted in 10 ml of pre-warmed media (see Life Technologies, Inc., Catalogue and Reference Guide, 1998 1999, Cell Culture section) followed by centrifugation at 1500 rpm for 5 min. The supernatant WO 01/66803 PCT/USO1/07608 -50 was then discarded and the cells were cultured in 100 ml media at 37 "C, 5% CO 2 . After 2 days, the cells were counted (approximately 8 x 10 5 /ml, total 100 ml). Batches were also evaluated for the ability to inhibit hepatitis B virus as detected by DNA quantitation (see Dong et al., Proc Natl Acad Sci USA (1991) 88: 8495-8499). Briefly, 5 one gram of preparation wag'addedf with 16'ml of water. The mixture was boiled for 30 minutes. The supernatant was collected after centrifugation and filtered through a 0.22 pLm filter. HepG2.2.15 cells which secrete hepatitis B virons (kindly provided by Professor G. Ace; see Ace et al. Proc Natl Acad Sci USA (1987) 84: 1005-1009) were used in this assay. One to fifty dilutions were used for each assay. The cell growth inhibition assay was 10 performed for 72 hours. All other procedures were performed as described by Dong et al., Proc Natl Acad Sci USA (1991) 88: 8495-8499. p-glucuronidase was assayed as HQT is known for its anti-diarrhea properties. Different HQT extracts were added to triplicate wells of a 96-well plate which contained 0.1mM phenolphthalein glucuronidate, 70 mM Tris-HC1 (pH 6.8) and 0.8 ng of dialyzed P 15 glucuronidase (from E.Coli, purchased from SigmaTM) to a final volume of 80 gl. After 2 hr incubation. at 37 0 C, the reactions were terminated with 200 pl of stopping solution which contained 0.2 M Glycine and 0.2 M NaCl (pH 10.4), and the OD was monitored with a kinetic microplate reader at 540 nm. The results of the assays using the three batches are displayed in Table 6. Based on 20 these data, HQT sources A and B have relatively low toxicities combined with higher inhibitory activity relative to batch HQT C (i.e., approximately 5 fold greater toxicity toward HepG2 cells and 3.3 fold less inhibitory activity against p-glucuronidase than either HQT A or B, see Table 6). 25 30 WO 01/66803 PCT/USO1/07608 -51 Table 6. Biolo2ical Assay of Three Preparations of HOT* E. Coli HepG2 Jurkat HBV$ @-Glucuronidase DNA HQTA 0.6 1.50 0.76 None HQT B . 0.7 1.6 0.81 ND HQT C 2.2 0.32 ND ND *Values $, % 6f bontr61 represent IC 50 values. ND, not determined. Evaluation of HQT Effects on Protein Expression 5 HepG2 cells (1 x 10 6) were seeded in 25 cm 2 flasks in 3.0 ml of RPMI-1640 medium (see Life Technologies, ik' Catalogie and-Reference Guide, 1998-1999, Cell Culture section) 24 hr before the drug addition. The- cells were treated with or without herbal medicine, where the former is added at two final concentrations of 0.2 mg/ml or 4 mg/ml, respectively, and 10 incubated at 37*C for 24 hours: The medium was removed and the cells were washed twice with cold PBS. The cells were harvested into 1 ml of PBS and centrifuged at 10,000 rpm for 2 minutes, extracted on ice witli a buffer containing 50 mM Tris-Cl (pH 7.5), 0.2 mM PMSF and 10% glycerol, followed by three freeze-thaw cycles. Potassium chloride was added to the cell lysate at a final concentration bf 0.15 M prior to centrifugation. The protein concentration was 15 determined and the cell extract was electrophoresed according to the method of Laemmli (Nature (1970) 227:680-685). Western blots were performed by standard techniques known in the art, see for example Sainbrook, et al (1989). The antibodies used were directed to the following proteins Topo ;Stit (20707); Cyclin B1; MAPK (Ab2) and Nm 23 HI. Figure 4 d6monstiff ih{f the higher concentrations of HQT A or HQT B differentially 20 effects the expression of cycling BIpolypeptide. HPLC Analysis The herbal batches'Were analyzed by HPLC with a Beckman ODS UltrasphereTM column (5 micron particles, 4.6 mm X 25 cm) and detected with an UV spectrophotometer (Perkin Elmer). The wavereil ths for- UV detection were monitored at 280 nm and 340 nm. The 25 mobile phase was pumped f 1 'ml/min'and consisted of Solvent A: H20 and Solvent B: 20% WO 01/66803 PCT/US01/07608 -52 MeOH with the following gradient: 1) the solvent was 100% solvent A for the first 5 minutes; 2) the solvent composition as changed to 10% solvent A / 90% solvent B for the next 10 minutes; and 3) the solvent was changed to 10% solvent A / 90% solvent B for the next 40 minutes. This was followediby the addition of 100 % solvent A for 5 minutes. The HPLC 5 markers are baicalin and baicalein. Mass Spectrometry The herbal extractwas analyzed by Mariner T M ESI-TOF Mass Spectrometry (MS) from PE Biosystems. Control tracings were generated using baicalein and baicalin, two known active ingredients in HQT. 10 HQT samples in water and acid treated batches were been analyzed by HPLC and Mass Spectrometry. While waer' treafed HQT batches A and B had distinct HPLC and MS tracings, acid treated batches gave almost identical patterns (data not shown). Algorithm The data collected fdrm part of the multidimensional analysis used to generate 15 multivariant normal distrib ution sets as a means of determining a baseline correlation between biological activity and standard HQT chemical (HPLC and Mass Spec), and origin/growth characteristics. Example 14. Individual components A. Licorice. 20 Evaluation of G1v vrhizae Radix (licorice) Licorice is useful for moistening the lungs and reducing coughs, helps to relax spasm and pain. The properties of the licorice batches used in this example are presented in Table 7. Table 7. Batch Properties (licorice) Property Bateh A Batch B Batch C Batch D Plant Name Glycyrrhizae".' Glycyrrhizae Glycyrrhizae Glycyrrhizae Radix Radix Radix Radix Origin Inner Mongolia Inner Mongolia U.S., Kin Man U.S., Kin Man Herb Center Herb Center Preparation Standard Standard Boiled 30 min. Warm H 2 0, 30 method mnm. Plant part -Root, Root 25 Biological' and Enzyme Assays, WO 01/66803 PCT/USO1/07608 -53 To assay the quality of herbal sources, each herbal extract supernatant was assayed and the analysis was repeated three times. For a given sample to be assayed, 1 gram of herbal powder was dissolved in 10 ml of 80' C deionized water (neutral pH) in a polypropylene tube. The tube was then incubated as outlined in Table 7, then centrifuged to obtain the supernatant. 5 Batches of licorice were tested against either HepG2 cells (ATCC cat # HB-8065) or Jurkat T cells (ATCC cat #TIB-152) or both. Cells were cultured for 24 hours as described above. Batches were also evaluated for the ability to inhibit hepatitis B virus as detected by DNA quantitation (see Dong et al., Proc Natl Acad Sci USA (1991) 88: 8495-8499). Briefly, one gram of preparation was added with 10 ml of water. The mixture was boiled for 30 10 minutes. The supernatant was collected after centrifugation and filtered through a 0.22 pim filter.2.2.15 cells which secrete hepatitis B virons (kindly provided by Professor G. Ace; see Ace et al. Proc Natl Acad Sci USA (1987) 84: 1005-1009) were used in this assay. One to fifty dilutions were used for ea6fassaj. The cell growth inhibition assay was performed for 72 hours. All other procedures wAereperformed as described by Dong et al., Proc Natl Acad Sci 15 USA (1991) 88: 8495-8499t Again, p-glucuronidase was assayed. Different'licorice extracts were added to triplicate wells of a 96-well plate whicli contained 0. 1mM phenolphthalein glucuronidate, 70 mM Tris HCl (pH 6.8) and 0.8 ng of dialyzed beta-glucuronidase (from E.Coli, purchased from Sigma) to a final volume of 80 pl and'assayed as above. 20 The results of the assays using the two batches is displayed in Table 8. Based on these data, licorice batch A was much more toxic to Jurkat cells than batches B (approximately 9 fold) and a more effective inhibitor of p-glucuronidase (see Table 8). Table 8. Biologi al.Assay of Four Preparations of Licorice* E Coli HepG2 Jurkat HBV$ p--Gucuronidase DNA Licorice A 1.1 1.07 0.41 None Licorice B ND ND 3.6 ND Licorice C 2.1 ND ND ND Licorice D ND ND - >2.0 53.8 *Values represent ICso o/ of Control values. ND, not determined.

WO 01/66803 PCT/USO1/07608 -54 Expression Assay In order to assay gene expression, Jurkat T cells were treated with herbal extract as follows: Jurkat cells (10 7 /ml) were quickly thawed in a water bath at 37 'C. The cells were then diluted in 10 ml of pre-warmed media (see Life Technologies, Inc., Catalogue and 5 Reference Guide, 1998-1999, Cell Culture section) followed by centrifugation at 1500 rpm for 5 min. The supernatant was then discarded and the cells were cultured in 100 ml media at 37 *C, 5% CO 2 . After 2 days ,the cells were counted (approximately 8 x 10 5 /ml, total 100 ml). The herbal extract solution was prepared as outlined above (e.g., 2 g of an herbal powder to obtain 20 ml of sterile solution (0.1 g/ml). The cells were divided into 3 flasks at a 10 density of 2.5 x 10 5 /ml, 100 mlleach flask. Assays were carried out with control (no extract), and 10 ml of extract at 10 mg/ml, and 1 mg/ml. Again, toxicity results were used to determine the "high" and "low", concentrations for any given extract. After extract addition, cell cultures were incubated for 24 houis'unde# conditions as outlined above. The cells were counted and subsequently collected in 50 mI centrifuge tubes. The resulting cell pellet was treated with an 15 RNA isolation means to extract rnRNA (see, for example, Sambrook et al., 1989 at pages 7.3 7.39). Microarray Microarray printing was barred out as follows: Human gene blones wer obtained from the IMAGE Consortium libraries through its 20 distributors and comprise gnes frodm various tissues. Most clones have been partially sequenced and the sequeis- Werei available as expressed sequence tags in the dbEST database of GenBank. Clones were cultuked and amplified using commercially available primers prior to application on nylon meibranes (Chen et al., Genomics (1998) 51:313-324). Approximately 10 ng of each amplified aetas applied on a positively charged nylon membrane using a PC 25 (personal computer) controle arraing system. The arraying system allows high density spotting and is capable of depositing 31,000 spots on a piece of nylon membrane measuring 18 by 27 mm using a 24-pin arraying tool. cDNA probe and Mmbrane Hybridization Two microgram of each mRNA sample (mRNA was isolated as outlined above) was 30 labeled with biotin and/oi dfgoxigenin using random primed reverse transcription. The labeled samples were treated'with alkali and thie resulting labeled nucleic acids were precipitated prior to use in hybridization. 1Vembrai hybridization and washing were carried out using the labeled probes as disclosed inChon et al. (1998). To detect the spots on the membrane in dual WO 01/66803 PCT/USO1/07608 -55 color mode (i.e., both biotin and digoxigenin), p-galactosidase-conjugated streptavidin (Strept Gal) and alkaline phosphatae. conjugated digoxigenin antibody (anti-Dig-AP) were employed. After color development, iinage digitization using an imaging means was employed (e.g., a flatbed scanner or digital.camera). Quantitative measurements were determined by computer 5 analysis which uses a program that measures the integrated density of the primary color components of each spot, performs regression analysis of the integrated density data and locates statistical outliers as differentially expressed genes. Gene expression data for samples 1, 2. and licorice (ST1 17) Extract 1, 2 and 6 corresponding to extract of Cordyceps sinensis, Poria cocos (ST 10 027) and licorice, respectively, were assayed by the following method: Batches were evaluated for toxicity using Jurkat T cells. The extractswbre prepared as outlined in Example 6. The cells were divided into 24 well culture -plates by addiig 1 ml of Jurkat cells at a density of 5 x 10 5 /ml. Assays were carried out with control (no extract), and 5 concentrations of extracts as described (see Table 15 9). The high and low concentrations for the cell culture assays were varied between 10 mg/ml and 0.05 mg/ml (i.e., mg dry 'weight of herbal extract per ml) depending on the toxicity of the extract to cells. For certain samples the toxicities at 10 mg/ml were such that "high" and "low" concentrations were adjusted'downward, nevertheless, at least one order of magnitude between extremes was maintained.' For example, for licorice (STI 17) the "high" was 0.5 mg/ml and the 20 "low" was 0.05 mg/ml (see.Table :9). After extract addition, cell cultures were incubated for 24 hours under conditions as. outlined in Example 6. The cells were counted and the resulting data tabulated to demonstrate extract toxicity. The resulting data is shown in Table 9. 25 30 35 WO 01/66803 PCT/USO1/07608 -56 Table 9. Survival Cell Number at Different Concentration of Herbal Extract Solution experiment concentration no.x10'mi . 10 mg/mil 5 mg/ml 2 mg/ml' 1 mg/nml 0.5 mg/m No drug High conc Low conc. (mg/m1) (mg/mi) 1 Cordyceps sinensis 8.4 11.9 11.5 9.2 9.0 12.4 10 1 mycelium 2 ST027 4.2 8.7 10 7.5 10 10 10 1 3 STO-44 - 5.9 8.4 9.9 9.4 5 0.5 4 ST051 - 1.7 5.4 0.5 0.05 5 ST093 - 1.9 3.8 4.4 0.5 0.05 6 ST117 - . i 6 3.6 4.6 5.8 0.5 0.05 7 ST123 3.4 6.4 8 9.3 7.8 5 0.5 8 ST128 3.5 7.7 7.9 7.7 8.3 5 0.5 9 ST134 2.9 6.1 11.2 9.6 9.8 5 0.5 10 ST237 - 2.5 6.6 8.7 1 0.1 Note: original cell number is 5x10/ml and the number to 1l0x10/ml after 24h incubation. "-" describes all dead cells. 5 Protocol: 1. Add 1 ml of 5x10 5 /ml Jurkat cells into 24 well culture plates. 2. Prepare 12 kinds of herbalextract solutions and sterilize. 3. Test 5 concentration per sample. 10 mg/ml, 5 mg/ml, 2 mg/ml, 1 mg/ml, 0.5 mg/ml 4. Culture the cells for 24 h iii 37C with 5% CO 2 incubator. 10 In each analysis, 144X96 genes (i.e., 13,824 genes) were analyzed (data not shown) and about 100 genes showed significant differences in comparison with that of control (Table 10). Some of the gen6s wereup-regulated and others were down-regulated. The magnitude of the difference with the control sometimes varied depending on the relative amount of the 15 herbal composition to.which the particular cells were exposed. Numbers under C1 (control treatment) and H or L (herbs) represent intensities of mRNA expressed after subtraction of background (Table 10). The gene designation is encoded in Array AD, which can be traced to a specific GenBank clone." The level of expression was determined by H or L divided by C. Only a fraction of 13,824 genes in each herb treated samples showed significant changes, 20 namely, up, down or unchanged (see' Table 10).

WO 01/66803 PCT/USO1/07608 -57 V V V V V V V V VVV V V V V V V V V v mz CDC 12 ~ 14 > > 0. > IT = C 00 > 0 o,- C o P El 0 1+ 0 CDQ- CDS CLJ\0O~ > CD C w -41.. C) w 0O w ,c %,0 UA t4. ~ 0 ~ 0 ') ~ ~ --. ,- t'D w 1O ti cn . .

WO 01/66803 PCT/USO1/07608 -58 a? t -4o~ 44. CWo w \1 L/0 00 (A0 O = " - 7 ui j W) C' 00 -.4 \0 W 0)0 4:, -.. (A- (- 0 00 0 , ON \10-0o 0-a0000JI' '00~ b~ bN ,:. (J~~ ~ p'. CS 41. . L ~ ~ ~ tJ 0CJ 00-P o o ~Lo w t , - -4 ON C) -P kA -* 4. Z = , t0 0 - 0 C)~ O.~ >) - 0 sJ - - ~ 0 L3~ C tJ , awU -- J W W--1 0 's\ 0 c t WO 01/66803 PCT/USO1/07608 -59 AA AA AA AAAA AAAA AA A AA A AA A V V V V V V VV V VV vV VV V vVvV VV VVV V Z m > 01-. > 0D co 0 ~c~ o + >~ E3 = > CD~ 0 t7' H 01 n R.I -- 0 jw0 - 00-l -- 4 00 0 00 00 00 k \1 -- C -C ~-00 UJc (A "o0-4 0 0 c00 C a w00 "0 0C -j -. 4 w - -4 ul~'J \0 0 A-0 O\4. 4w4 A-3 0 = p " " " j-"N , " \C-b N10-14:- u WO 01/66803 PCT/USO1/07608 -59/1 cl ~00 .)C\ 0 "oC 0 COk w ~ 0 4 Q'o 4 ' (-A (0) ,a w \o o0 4tCJ CD W- ~ ~0 \ 0 -. o 00 N)0 Q 0 00 006 -4. N 0O(.A w -. t 00 00OCN 0(A 00 0000 001 0 -. J " " \O = w ON10 \1 1 " - k)-4C)-j w )C 00 c0 Q A A-o 0u 0 j z ' -p. 0 00 C-' uj0 " .A A00-P C - j C) 00 0\"" 00ON aNOk 00 00 0U00 -- l ON C) -~0 0 4 -4- C 0 tAC jC CD CD00 0J -Ok1=)0C) *-k Q , -000iol 00 - o 0o A -- \c 0 . 020 (=(-,IC O W00 0 ~ - ) = 00 0= 0 00 0* = W-j t. 00 , -- " -P- 0 -l00 ,\(A 1, " - 0~ N U 0(0 - 0- Q0 C)00 0-0 ) )C 000 0j il00 LA -0 0 C -I\oW0 C 0 00 0 P" C N (-AO 00-W -- "\C -0 O0C) w-1 \0 N r-(=0 0 0 -a0-j-- -0 0 - WO 01/66803 PCT/USO1/07608 -5 9/2 A A A AA AA A AA AA AA A AA A AA A AAA t~ IiN tiOO. . . (. w w w 1J -Pl4 I' -- UJj -4 ONa,-4" P )W O > Eg 0 cr H >~ > 0 CD > ~ H n -. l >r , >~ 0 0C -N w .4. C O0 1,04 \1 0 0 0 - -4 :41 -4 :1 \0- -41u 00)0 u 0U -. .A 00 -. j- .46 u \c ,' - ~ 0 *N 110 0-4,Lu CN -4, C C> ON \10 00 (-A -N 110 100000 002 00 w Cl C'N a) a) O \c -,a00 - ~t 00 LA t0 b" O \- - N 0 0 CN Lit --A \C --J Lit C :. - 0040 CN W J1 t-J w C. - p 6.6C \ j i.a jzi(A o6 p 00 ',0II** ',C* 4E:I *I ICI --. \, WO 01/66803 PCT/USO1/07608 -59/3 "A Ap(A~o C 00 000 ALA: Oo ~ ccN0 NJA-(A Ow-o A"DCo 0=100 =0- 0 A00 0A 0 0 0 0 00 0 , , -0 4:4. \ C ',C) 0 ~ O - -0-OC " - A -- ] -.4 w CO-4 "C) 0 00- ~ (A -QC -C4j - -]00N4 00 00 0 0 " 0 0 040 -~L 0c-0 O WA 0 IQ (AOQ4a o ~ ~ (11 -J -4 -oJ ( ~O OC O\ -O C', CDN4 ~ U U 0 ' ONr- 10 Q - ~0 -= - 4 0 ~ W-A- \ 0) -0~- - 0Q t~ 0 J 000t4 :-- -. 4 (Aj - . j :1.-. :0.:" 0 0-O-4 "O J O- (A',= w a)-- 0 0 000 0k (.-- - A- - W 0 \, C)t4 C) C> C 0 0\0\0 0-.10 - 0000 000 0 4U .= " - = I -)t(A A r, 0. ~O (.A 4 ( C' (AOQ wOl ) 10 110 0A 10 UIJ~O( A (A ( 4 - A 0 Lt00 0N 0, 00 0000 ~ (A (AO% -- 4--A WCJ41 ( Q 'C CC \, 0 w a\ - OO O' 00\c 0 04-C WO 01/66803 PCT/USO1/07608 -59/4 A A AA AA A AA A A A AA AA AA A AA A A A A - - - W - ? - A - - - - 4- - - -W- 41, Vv vvv VV VV VV VV VV VV VV VV v v vvv c~S*n Dz -D -t Hc rL t"~ 0D > o> > CD0 qq a* H CD o ~wZ3 -- A0Q-4 tQtJ N 0 0 0 C,< -~~~-J -, " - ~~oo-~~o -j ~ 0 ~ - CjJ O 00 tJJ 00 0-Pk) (-A C0 - 0 S oo-06. o oo oi-iloa bc- 11 1 I& WO 01/66803 PCT/USO1/07608 -5 9/5 0000~~00 LA W I j 00 0 =o-- -4.0 -coj c-, o c 00coo 0-, 4Oo. o -(A j ( Ok) Ck k JL J Otoo o10 l -0~~0 0 0 - - < 00 w~LJk V, ) o LAula NJ00 -- (JJ-A J- )Q Ck 0 0 -- j 0 0o .4 wk 0 0 j 0-0u 00 - o~O k " o(- = A00 - ONO k k OON 0j O 0 = CS 0 "AJO W 00 W W (jr 4,\ 00o WO 01/66803 PCT/USO1/07608 -59/6 o1 -P 0N ON ON ON 0 04 04 0o 0a 0 0 0 0 0 0 0o UIUJ (o0 0C 00\ 0 l - - P - - - L -A -- 0A ' ~ ~ ~ Po. C4~o H 0 0D -CD0 4 , -4 0 'P... 7C', 00 j ww C Uj -, 0 t-jp, A o o ] 0 0n LA W t W w t C) 0 !J 3000 4j ~0 J ~ Ull c ~ oo ~ oe e U - -4t C 1 Un ~ ~ ~ 00 00L j bN c b00:1 i 04J0 (-04,W0 W N-0 000t Ak-, WO 01/66803 PCT/USO1/07608 -59/7 00 CD 't. C iQ~o~ 0;- o6 -- j - " -j o -0 0U~4 tA 00 WO W tJ)\O - -- CN -A W t./A w ) w-- ON tQ Q w -- tA LA~~ * t.bJ (A bN Uj Lo LA bN b 0( ~O- o ~ o 00 WO 01/66803 PCT/USO1/07608 -60 In this manner, we are able to correlate specific gene expression with the exposure of a cell to no, low (L) or high (H) amounts of an herbal composition. Many of the genes identified in this way code for proteins important in known metabolic or biochemical pathways. Many of 5 these proteins have direct and indirect effects on certain physiological, morphological and psychological parameters. Thus, this method permits the association of a particular genetic fingerprint of an herbal composition with its array biological effects. Such associations can be used to profile or characterize an herbal composition for the purposes of Quality Control and Quality Assurance and evaluating pharmacological or toxicological properties. The role of 10 primary and secondary herbs in an herbal formula can also be assessed by this approach. HPLC Analysis The herbal batches were analyzed by HPLC with a Beckman ODS UltrasphereTM column (5 micron particles, 4.6 mm X 25 cm) and detected with an UV spectrophotometer (Perkin Elmer). The wavelengths for UV detection were monitored at 280 nm and 340 nm. The 15 mobile phase was pumped at '1 ml/min and consisted of Solvent A: H 2 0 and Solvent B: 20% MeOH with the following gradient: 1) the solvent was 100% solvent A for the first 5 minutes; 2) the solvent composition was changed to 10% solvent A / 90% solvent B for the next 10 minutes; and 3) the solvent Was changed to 10% solvent A / 90% solvent B for the next 40 minutes. This was followed by the addition of 100 % solvent A for 5 minutes. The HPLC 20 marker is glycyrrhizin. Algorithm The data collected form part of the multidimensional analysis used to generate multivariant normal distribution sets as a means of determining a baseline correlation between biological activity and standard licorice molecular, chemical (HPLC and Mass Spec), and 25 origin/growth characteristics. B. Scute Evaluation of Radix Scitellariae (Scute) Scute has been found to b useful in reducing capillary permeability and inflammation. It can also be used treat entefitis and dysentery, increases the secretion of bile to treat jaundice; 30 to relieve muscle spasms; to treat coughing and to expel parasites. The properties of the scute batches used in this example are presented in Table 11.

WO 01/66803 PCT/USO1/07608 -61 Table 11. Batch Properties (Scute) Property Batch A Batch B Batch C Batch D Plant Name Scutellariae Scutellariae radix Scutellariae Scutellariae radix radix radix Origin Sanxi Province. U.S., Kin Man Herb U.S., Kin Man U.S., Boston Center Herb Center Preparation Standard Boiled, 30 min Warm H 2 0, 30 Boiled , 2 hours method min. Plant part Root Biological and Enzyme Assays Briefly, one gram of each preparation of scute extract was added with 10 ml of water (1 5 mg/ml). The mixture was treated as Outlined in Table 11. The supernatant was collected after centrifugation and filtered through a 0.22 im filter. Batches of scute were tested against either HepG2 cells (ATCC cat # HB-8065) or Jurkat T cells (ATCC cat #TIB-152) or both. One to fifty dilutions were used for each assay. Cells were cultured for 24 hours as described above. Batches were also evaluated for the ability to inhibit hepatitis B virus as detected by 10 DNA quantitation (see Dong et al.; Proc Natl Acad Sci USA (1991) 88: 8495-8499). Briefly, one gram of preparation was added with 10 ml of water. The mixture was treated as outlined in Table 11. The supernatant was collected after centrifugation and filtered through a 0.22 ptm filter.2.2.15 cells which secrete hepatitis B virons (kindly provided by Professor G. Ace; see Ace et al. Proc Natl Acad S'ciUSA(1987) 84: 1005-1009) were used in this assay. One to fifty 15 dilutions were used for each issay 'The cell growth inhibition assay was performed for 72 hours. All other procedures weie performed as described by Dong et al., Proc Natl Acad Sci USA (1991) 88: 8495-8499. For p-glucuronidase, different scute extracts were added to triplicate wells of a 96-well plate which contained 0.imM phenolphthalein glucuronidate, 70 mM Tris-HC1 (pH 6.8) and 20 0.8 ng of dialyzed p-glucuronidase (from E. Coli, purchased from Sigma) to a final volume of 80 pl. After 2 hr incubation at 37 0 C, the reactions were terminated with 200 j1 of stopping solution which contained 0.2 M Glycine and 0.2 M NaCl (pH 10.4), and the OD was monitored with a kinetic microplate reader at 540 nm.

WO 01/66803 PCT/USO1/07608 -62 The results of the assays using the three batches is displayed in Table 12. Table 12. Biological Assay of Four Preparations of Scute* t Coli HepG2 Jurkat HBV$ DNA p-Glucuronidase Scute A 1.5 0.33 0.45 None Scute B 1.8 ND ND ND Scute C 0.3 ND ND ND Scute D ND 0.65 ND 27.5 *Values represent $, % of Control

IC

50 values. ND, not determined. Evaluation of Scute Effects dir Pr6tein Expression HepG2 cells (1 x 109) were seeded in 25 cm 2 flasks in 3.0 ml of RPMI-1640 medium 5 (see Life Technologies, cataldgue and Reference Guide, 1998-1999, Cell Culture section) 24 hr before the extract addition. The cells were treated with or without herbal medicine, where the former is added at.two final concentrations of 0.2 mg/ml or 4 mg/ml, respectively, and incubated at 3 7*C for 24 hours:. The medium was removed and the cells were washed twice with cold PBS. The cells were harvested into 1 ml of PBS and centrifuged at 10,000 rpm 10 for 2 minutes, extracted on ice with a buffer containing 50 mM Tris-Cl (pH 7.5), 0.2 mM PMSF and 10% glycerol, 'followed by three freeze-thaw cycles. Potassium chloride was added to the cell lysate at a final concentration of 0.15 M prior to centrifugation. The protein concentration was deternfiinedunid c' cell extract was electrophoresed according to the method of Laemmli U.K. (Nature (1970) 227:680-685). Western blots were performed by standard 15 techniques know in the art,'see for example Sambrook, et al (1989). The antibodies used were directed to the following proteins: Topo I; Stat (20707); Cyclin B1; MAPK (Ab2) and Nm 23 H1. Figure 4 demonstrates that scute batches A and B do not differentially affect the expression of the polypeptides resolved on Western blots. 20 HPLC Analysis The herbal batches' wer analyzed by HPLC with a Beckman ODS Ultrasphere TM column (5 micron particles, 4.6mm X 25 cm) and detected with an UV spectrophotometer (Perkin Elmer). The wavelengths for UV detection were monitored at 280 nm and 340 nm. The mobile phase 'was pumped al l il/niin and consisted of Solvent A: H20 and Solvent B: 20% WO 01/66803 PCT/US01/07608 -63 MeOH with the following gradien: 1)'the solvent was 100% solvent A for the first 5 minutes; 2) the solvent composition was changed to 10% solvent A / 90% solvent B for the next 10 minutes; and 3) the solvent was changed to 10% solvent A / 90% solvent B for the next 40 minutes. This was followed by the addition of 100 % solvent A for 5 minutes. The HPLC 5 markers are baicalin and baicalein. Scute batches in witer and acid treated samples were analyzed by HPLC. Water and acid treated batches were virtually indistinguishable. Algorithm The data collected form part of the multidimensional analysis used to generate 10 multivariant normal distribution sets as a means of determining a baseline correlation between biological activity and standard scute chemical (HPLC), and origin/growth characteristics. C. White Peony Root Evaluation of Paeonie ladflor'a pallus radix (Peony) Peony is used to suppress and soothe pain. It is also known to soothe ligaments and 15 purify the blood. The properties of the peony batches used in this example are presented in Table 13. Table13. Batch Properties (Peony) Property Batch A Batch B Plant Name Paeonie Jactiflora palus Paeonie lactiflora pallus Origin Anwey Province U.S., Boston Preparation method Standard Boiled 2 hours. Plant part Root Root Biological and Enzyme Assays 20 Briefly, one gram of each preparation of scute extract was added with 10 ml of water (1 mg/ml). The mixture was treated as outlined. in Table 13. The supernatant was collected after centrifugation and filtered- through a 0.22 pm filter. Batches of peony were tested against either HepG2 cells (ATCC cat # B-8065) or Jurkat T cells (ATCC cat #TIB-152) or both. One to fifty dilutions were used for each assay. Cells were cultured for 24 hours as described above. 25 Batches were also evaluated for the ability to inhibit hepatitis B virus as detected by . DNA quantitation (see Dong-etal,-Proc Natl Acad Sci USA (1991) 88: 8495-8499).

WO 01/66803 PCT/USO1/07608 -64 Briefly, one gram of preparation was added with 10 ml of water. The mixture was treated as outlined in Table 13. The supernatant was collected after centrifugation and filtered through a 0.22 pn filter. 2.2.15 cells'which secrete hepatitis B virons (kindly provided by Professor G. Ace; see Ace et al. Proc Natl Acad Sci USA (1987) 84: 1005-1009) were used in 5 this assay. One to fifty dilutions were used for each assay. The cell growth inhibition assay was performed for 72 hours. All other procedures were performed as described by Dong et al., Proc Natl Acad Sci USA (1991) 88: 8495-8499. Different peony extracts were added to triplicate wells of a 96-well plate which contained 0.1mM phenolphihalein glucuronidate, 70 mM Tris-HCl (pH 6.8) and 0.8 ng of 10 dialyzed beta-glucuronidase (from E. coli, purchased from Sigma) to a final volume of 80 pl. After 2 hr incubation at 37C, the reactions were terminated with 200 pl of stopping solution which contained 0.2 4 Glycine and 0.2 M NaCl (pH 10.4), and the OD was monitored with a kinetic microplate reader at 540 nm. Results are shown in Table 14. Table 14. Biological Assay of Two Preparations of Peony* E. Coli HepG2 Jurkat HBV$ P-Glucuronidase Peony A 2.8 >1.5 1.1 None Peony B >2.5 ND ND ND *Values represent IC 50 $, % of Control values. ND, not determined. 15 HPLC Analysis The herbal batches were analyzed by HPLC with a Beckman ODS Ultrasphere column (5 micron particles, 4.6-mm X 25 cn) and detected with an UV spectrophotometer (Perkin Elmer). The wavelengths for UV detection were monitored at 280 nm and 340 nm. The mobile phase was pumped at 1 ml/hin and consisted of Solvent A: H20 and Solvent B: 20% MeOH 20 with the following gradient::1) the solvent was 100% solvent A for the first 5 minutes; 2) the solvent composition was changed to 10% solvent A / 90% solvent B for the next 10 minutes; and 3) the solvent was changed to 10% solvent A / 90% solvent B for the next 40 minutes. This was followed by the addition of 100 % solvent A for 5 minutes. HPLC marker is paeoniflorin. 25 Peony batches ere analyzed by HPLC as shown in Figure 5. Algorithm WO 01/66803 PCT/USO1/07608 -65 The data collected form part- of the multidimensional analysis used to generate multivariant normal distribution sets'as a means of determining a baseline correlation between biological activity and standard peony chemical (HPLC), and origin/growth characteristics. D. Date 5 Evaluation of Ziziphi Fructus (Date) Date has been used for diuretic properties and strengthening effects. The properties of the date batches used in this example are presented in Table 15. 10 Table 15. Batch Properties (Date) Property Batch A Batch B Batch C Plant Name Ziziphi Jrub'tus Ziziphi Fructus Ziziphi Fructus Origin Hebei Piovin e. U.S., Kin Man Herb U.S., Kin Man Herb Center Center Preparation method Standard Boiled, 30 min Warm H 2 0, 30 min. Plant part Fruit Biological and Enzyme Assays Briefly, one gram of each batch of scute extract was added with 10 ml of water (1 mg/ml). The mixture was treated as'outlined in Table 15. The supernatant was collected after 15 centrifugation and filtered through a 0.22 ptm filter. Batches of date were tested against either HepG2 cells (ATCC cat # HB-8065) orJurkat T cells (ATCC cat #TIB-152) or both. One to fifty dilutions were used for-each assay. Cells were cultured for 24 hours as described above. Batches were also evaluated for the ability to inhibit hepatitis B virus as detected by DNA quantitation (see Dong ital., Proc'Natl Acad Sci USA (1991) 88: 8495-8499). Briefly, 20 one gram of preparation was added with 10 ml of water. The mixture was treated as outlined in Table 15. The supernatant was collected after centrifugation and filtered through a 0.22 pim filter. HepG2.2.15 cells which secrete hepatitis B virons (kindly provided by Professor G. Ace; see Ace et al. Proc Natl Acad Sci USA (1987) 84: 1005-1009) were used in this assay. One to fifty dilutions were use for each assay. The cell growth inhibition assay was WO 01/66803 PCT/USO1/07608 -66 performed for 72 hours. All other procedures were performed as described by Dong et al., Proc Natl Acad Sci USA (1991) 88: 8495-8499. Different peony extracts were added to triplicate wells of a 96-well plate which contained 0.1mM phenolphthalein glucuronidate, 70 mM Tris-HCI (pH 6.8) and 0.8 ng of 5 dialyzed beta-glucuronidase (from E. Coli, purchased from Sigma) to a final volume of 80 p1. After 2 hr incubation -at 37 0 , the -reactions were terminated with 200 pl of stopping solution which contained 0.2 M Glycine and 0.2 M NaCl (pH 10.4), and the OD was monitored with a kinetic microplate reader at,540 nm. Results are shown in Table 16. Table 16. Biological Assay of Three Preparations of Date* E Coli HepG2 Jurkat HBV$ DNA p-Glucuronidase Date A 1.2 1.5 5.1 None Date B ND >2.0 ND 52.3 Date C 2.5 ND ND ND *Values represent , % of Control

IC

50 values. ND, not determined. 10 HPLC Analysis The herbal batches-were analyzed by HPLC with a Beckman ODS Ultrasphere column (5 micron particles, 4.6 mm X 25 cm) and detected with an UV spectrophotometer (Perkin Elmer). The wavelengths for UV detection were monitored at 280 nm and 340 nm. The mobile phase was pumped at 1 ml/min and consist of Solvent A: H 2 0 and Solvent B: 20% MeOH with 15 the following gradient: 1) the solvent was 100% solvent A for the first 5 minutes; 2) the solvent composition was chaged to 10% solvent A / 90% solvent B for the next 10 minutes; and 3) the solvent was changed to 10% solvent A / 90% solvent B for the next 40 minutes. This was followed by the iaddii6n of 100 % solvent A for 5 minutes. HPLC markers for date are chelidonic acid and cAMP. 20 Date batches samples were analyzed by HPLC as shown in Figure 6. Algorithm The data collected form part of the multidimensional analysis used to generate multivariant normal distribution sets as a means of determining a baseline correlation between biological activity and standard peoy chemical (HPLC), and origin/growth characteristics.

WO 01/66803 PCT/USO1/07608 -67 Example 15. Characterizing Herbal Medicines by Nucleic Acid Microarray Analysis. Introduction. The rapid development of nucleic acid microarray technology has led to an explosion 5 of gene expression data (Lander, 1999, Duggan et al., 1999). Four characteristics of the gene expression account for the great value of using nucleic acid microarrays to study the gene expression profiles. (i) Nucleic acid microarray makes it easier to measure the transcripts of thousands of genes at once. (ii) Close association between the function of a gene product and its expression pattern makes gene function predictable. (iii) Cells respond to the micro 10 environmental changes by changing the expression level of specific genes. (iv) The sets of genes expressed in a cell determine what the cell is derived of, what biochemical and regulatory systems are involved, and so on (Brown and Botstein, 1999). By using a microarray system, the above features can be studied in an ensemble manner. The expression of any, desired number of genes can be detected using the nucleic acid 15 microarray technology Ior example, up to about 20;000 genes may be placed on a single array. We have developed a incleic acid microarray with colorimetric detection system (Microarray/CD) (Chen et al., 1998). Gene expression profiles of different cell lines were studied using microarray filter membranes (2.7 cm x 1.8 cm) with about 10,000 cDNA representing approximately 10,000 distinct human transcripts. The sensitivity and detection 20 limits of the microarray/CD system have been characterized and are comparable to the system with radioactive detection ortliie, ytem with laser induced fluorescence detection (Bertucci et al., 1999). As previously ddscribed, cellular gene expression profiles portray the origin, the present differentiation of the cell, and the cellular responses to external stimulants. In other 25 words, the gene expression profiles reveal the state of the cell and microarray is a perfect tool for the rendering purpose. Inithe present studies, we apply the microarray/CD system to characterize cellular responses to external stimulants, in this case, the Chinese herbal medicines. Conversely, we also based on the stimulated gene expression profiles to classify different herbal medicines, 30 Figure 7 is a flowchart depicting a general method that may be used for establishing an expression response data set for cells treated with an herbal composition. The method comprises the steps of: WO 01/66803 PCT/US01/07608 -68 (a) Determine the IC 5 concentration oftan herbal composition by incubating various concentrations of the herbal medicine in mammalian cell cultures and identify the concentration that leaves 50% of survival cells after a predetermined time. (b) Incubate the maminmalian cell cultures with herbal extracts of various fractions of 5 IC 5 0 concentrations. (c) Harvest and count the cultured cells after a predetermined culture time. (d) Immediately lysd the cells after they are removed from the incubator and extract mRNA from cell lysate. (e) Label the mRNA by reverse transcription reaction to turn mRNA into labeled cDNA. 10 (f) Mix the labeled cI)NA with control cDNA of plant origin and perform hybridization to a microarray of mammalian gene probes. (g) Measure expressidniilevel of genes by analyzing digitized images of the microarray hybridization resulfs. (h) Perform data pre-pkocessing to select data for statistical analysis. 15 (i) Acquired 'expression data generated by microarray experiments of an herbal composition with arious concentrations. (j) Data pre-processing to select the genes with statistical significance in cells treated with different concentrations of the herbal medicine. (k) Categorize expression profiles into clusters by statistical methods such as the self 20 organizing-map algorithm. (1) Deduce the characteristic expression profiles for the herbal medicine based on the expression profile clusters. Figure 8 is a flowchart demionstrating how data sets of expression data for various batches of the herbal composition are integrated to make an expression profile database for the particular 25 herbal composition. The expression profile database then becomes part of the HBR Array. HBR Arrays containing expression profiles may also be used to identify an unknown herbal composition. Figure 9 is a flowchart depicting a general method for identifying an unknown herbal composition, the method comprising the steps of: (a) Construct an HBR Array containing characteristic expression profiles for an herbal 30 medicine or a collection of expression profiles of various herbal medicines by the aforementioned ste p. (b) Obtain the characteristic expression profile data set of the unknown herbal composition.

WO 01/66803 PCT/USO1/07608 -69 (c) Compare the HBl Aray containing the characteristic expression profile induced by the said unknown herbal composition with a standardized HBR Array containing expression data by algorithms such as the Hamming distance algorithm. (d) Score possible alignments to identify the most probable herbal composition whose 5 characteristic expression profiles are archived in the said HBR Array. Scoring possible alignments of HBR Arrays containing expression profiles may be performed using hierarchical cluster analysis of the Hamming distance matrix. Use of hierarchical cluster analysis for the Hamming distance matrix is well known in the art. The gene expression profiles may also be incorporated into the standardized HBR Array. 10 As has been already discussed, the standardized HBR Array containing such gene expression profiles induced by an herbal composition can be used for studying the pharmacological mechanisms of the herbal composition, for discovering new application of the herbal composition, and for designig optimized formulation of a complex herbal preparation. As can be seen from the flowchart-of Figure 10, the method may be generally outlined as comprising 15 the steps of: (a) Construct a data set containing the characteristic gene expression profiles for an herbal composition. (b) Score each gene by the consistency of its expression profiles in the data set using known statistical parameters, such as the coefficient of variation. 20 (c) Based on the statisti6l scoring, gene expression profiles for an herbal medicine are selected to be incorporated into the standardized HBR Array. HBR Arrays containing gene depression profiles may also be used to identify signature gene expression profiles induced by individual chemical constituents in an herbal composition consisting of complex chemical constituents, as outlined in the flowchart of Figure 11. The 25 method comprises the steps of: (a) Construct a HBR Array containing characteristic gene expression profiles for an herbal composition by the aforementioned steps. (b) Determine the com position of chemical constituents in an herbal medicine by high performance liquid chromatography (HIPLC) or liquid chromatography mass 30 spectrometry (LC-1ASS). (c) Repeat the step (b) forvarious batches of herbal medicine preparations. (d) Score the correlatioticofficients between the expression levels of each gene with the amount of individual chemical constituent in an herbal preparation.

WO 01/66803 PCT/USO1/07608 -70 (e) The signature gene expression profiles for individual chemical constituent are selected with a Pearson coirefation coefficient exceeding 0.99 or smaller than -0.99. Any herbal composition can then be characterized through the use of gene expression profiles generated through the use of nucleic acid microarrays. Moreover, one can choose any 5 number of genes that are differentially expressed to be included in the data set represented the gene expression profiles. Forexample, one may choose about 10 genes, about 100 genes, about 500 genes, about 1000 genes, about 1500 genes, abut 2000 genes, about 2500 genes or more, or any number in between. The prescription of the Chinese herbal medicine Scute and Licorice combination 10 (Huang Chin Tang) stops diarrhea, relieves spasms and clears fever. The ingredients of Huang Chin Tang are Scute, Peony Licorice and Jujube. This recipe has been used for more than 1000 years but the chemical and biomedical studies on the prescription have not been carried out until recent decades. In this study we used the nucleic acid microarray technology to study the gene expression profiles'of herbal medicines treated cells. Our aims are to demonstrate the 15 feasibility of using the niicrarray/CD system for classification of different herbal compositions or different'preparations and to find the predictor genes (marker genes) for the Huang Chin Tang prescription. The long-term goals are to find the correlation of the biochemical ingredients1in each herbal composition with the gene expression profiles of various treated cells and to decipher the molecular pharmacological mechanisms of the 20 Chinese herbal medicines i rarational fashion. Materials and Methods. 1. Development of acell banking system. Purpose: Microarray system is a sensitive detection method to monitor gene expression 25 patterns of cefl6J Itsi necessary to build a Cell Banking System with a Master Cell Bank (MC8):and a Working Cell Bank (WCB) to minimize cell variability for herbal medicine testing. Scope: The Cell Baik System is used for all types of cells in microarray studies. Apparatus: CO 2 Air-Jacketed Incubator (NUAIRETM DH autoflow) 30 Centrifuge (IUBOTA 2100) Freezing vial (Corning Costar, Cat. #430659) Tissue culture flask 750 ml (Falcon, Cat. #3045) Tissue culture dish 150x25 mm (Falcon, Cat. #3025) WO 01/66803 PCT/USO1/07608 -71 Cell: Jurkat T cell from Dr. Alexandra Ho Reagents: RPMI Medium 1640 (GIBCO BRL, Cat. #31800-014) Dimethyl Suiphoxide (DMSO) (Sigma, Cat. #D-2650) Fetal Bovine Serum (HyClone, Cat. #SH30070.03, Lot#AGL7258) 5 2-mercaptoethanol (GIBCO BRL, Cat. #21985-023, 5x10- 2 M) Media: I. Culture medium: 90% RPMI + 10% fetal bovine serum + 2 mercaptoethanol (5x1 05 M) II. Freezing medium: 90% RPMI 1640 + 10% DMSO Procedure: 10 A. Master Cell Bank 1. Follow the standard sterile procedure of cell culture. 2. Seed Jurkat T cells in *ultuie medium in flask at 37'C with 5% CO 2 incubator. 3. After incubation for two days, count the cell number and spread the cells to two flasks. Note. Cell density is kept about 5x1 04- 2x1 06 per ml. 15 4. Culture and count the cell number until the number reaches 2x1 07. 5. Collect the cells in 50 ml centiifuge tubes and spin at 1300 rpm (300xg) for 5 min. 6.- Discard the superntant and re-suspend the cell pellet with chilled freezing media. The cell number in each vial is about 1xl06 per ml. 7. Slowly freeze the cells by the following temperature profile: -20"C for 2 hr, 20 -80'C for 24 hr and then place the cells in liquid N 2 storage. Store a total of 20 frozen vials in MCB. B. Working Cell Bank 1. Retrieve one vial of cell from MCB in liquid N 2 tank and quickly thaw at 3 7"C water bath. 25 2. Transfer the cells into,10 ml of warm culture medium. 3. Spin down the cells at 1300 rpm (300xg) for 5 min. Discard the supernatant. Culture the cells with 20 ml medium in a flask. 4. Sub-culture the cells to'2 flasks. 5. Seed 5x10 7 cells with 500 ml culture medium in each flask with stirring for a total of 2 30 flasks. Culture the eels for 2 days. 6. Culture until the ceiflensity reaches lx10 6 /ml and a total volume of 1 L. 7. Prepare freezing media by adding 100 ml of fetal bovine serum and 10 ml of DMSO.

WO 01/66803 PCT/USO1/07608 -72 8. Centrifuge, discard the supernatant and re-suspend the cells to 110 ml of freezing medium. 9. Dispense 1 ml to every freezing vial (10 million cells per vial) for a total of 100 vials. Slowly freeze the cell by the temperature profile described above. 5 2. Determination of growth inhibition concentration of herbal extract in cell cultures. Purpose: Most drugs are toxic to cells. This experiment is designed to examine the toxicity of herbal extracts in Jurkat T cells and to determine the growth 10 inhibition concentration of herbal extracts that keeps the cells alive. Scope: This assaycan-be used in all kinds of herbal extracts to examine the toxicity. Apparatus: CO 2 'Air-Jacketed Incubator (NUAJRE DH autoflow) Counting cham'ube (Hemacytometer, Reichert, USA) Microscop e (Zeiss, Akiovert 100) 15 Cell: Jurkat T cell Reagents: RPMI Medium 1640 (GIBCO BRL, Cat. #31800-014) Fetal Bovine Sxum (HClone, Cat. # H30070.03, Lot #AGL7258) 2-mercaptoethanol (G hBCO BRL, Cat. #21985-023, 5x1 0 2 M) Culture media: 90%RPMI + 10% fetal bovine serum + 2-mercaptoethanol 20 (5x10 5 M) Disosabl cspiler syringe filters (0.2 m, Coming, Cat. #21052-25) Herbal extracts: 1M. Cordyceps Sinensis Mycliuor t 2. ST 024: 25 3. ST 044:. 4. MT 051: 5. ST 093: 6. ST 117: 7. ST 123: 30 8. ST 128: 9. ST 134: 10. ST 237: 11. P DY906-30350: C6mplex mix composed of 4,6,7, 10 WO 01/66803 PCT/USO1/07608 -73 12. PHY906-284003: Complex mix composed of 4, 6, 7, 10 Procedure: A. Herbal Extract Preparation 1. Dissolve 1 gram of herbal powder in 10 ml of 80 *C deionized water (neutral pH) in a 5 polypropylene tube. 2. Incubate the tube at 80 'C water bath for 30 minutes with gentle shaking then centrifuge at 4000 rpm (1500xg) for 5 min to obtain the supernatant. 3. Centrifuge at 11000 rpm (14000xg) for 10 min to collect the supernatant. 4. Using the disposable sterile syringe filter to filter the supernatant. 10 B. Cell Survival Test 1. Culture Jurkat T cells as described above. 2. Dispense 1 ml of 5x10mil cells per well to 24-well culture plates. 3. Prepare 12 kinds of herbal extract solutions. The extract solutions must be freshly prepared and used immediately. 15 4. Add 100, 50, 20, 10, 5p41 of each herbal extract solution into the 24 well culture plates to get the five different concentrations: 10, 5, 2, 1, 0.5 mg/ml. 5. Culture the cells for*24 li at 370C in an incubator filled with 5% CO 2 . 6. Count the number of cells per well. Mix 10 p1 of cell solution with 10 pl of Trypan blue dye and load into cell counting chamber. 20 7. Count the four najor square areas to calculate the cell number. (number of cells in 4 areas)/4 x 104 x dilution factor =number of cells per ml 3. Profiling gene expression patterns of Jurkat T cells treated with herbal extracts. Purpose: Profile the gene expression patterns of Jurkat T cells treated with herbal 25 extracts. A igh-density nucleic acid microarray with colorimetric detection system is used. Apparatus: Heat block (Boekel, Model 110002) Spectrophotonieter (Beckman, DV640) Centrifuge.(kIBOTA 1910) 30 Water bath (SLM AMINCO, Model 800) Hybridization incubator (YIH DER OH-800) Heat sealer(TISH-300, TEW) Reagents: RNAzolM, B (Tel-Test, Cat. #CS-104) WO 01/66803 PCT/USO1/07608 -74 Oligotex mRNA Midi Kit (Qiagen, Hilden, Germany) Hybridizatioii Bags (GIBCO BRL, Cat. #18278-010) EasiSeal (Hybaid,'O at.' #HBOSSSEZ1E) Glass slides (Matsunami, S2214, Japan) 5 Aerosol Resistant Tips (ART'tip) (molecular BIO-Products, Cat. #2139) Random hexamer primer (GIBCO BRL, Cat. #48190-011) Reverse transcriptase and 5x buffer (GIBCO BRL, Cat. #18064-014) RNase inhibitor (GIBCO BRL, Cat. #10777-019) Biotin-16-dUTP (Boehringer Mannheim, Cat. #1093070) 10 Dig-i 1-dUTP (Boehringer Mannheim, Cat. #1558706,) Blocking powder for hybridization (Boehringer Mannheim, Cat. #1096176) Bovine serum albumin (Sigma, Cat. #A2153) 20X SSC (Aresco, Cat.#0918S-2-20XPTM5L) SDS (Meick, Cat.#113760) 15 Dextran Sulfate (Sigma, Cat. #D6001) Streptavidinp-galactosidase (GIBCO BRL, Cat. #19536-010) Anti-digoxigeiiin-AP Fab fragments (Boehringer Mannheim, Cat. #1093274,) X-gal (GIBCO BRL, C'at. #15520-018) Maleic acid (Sigma, Cat. #M1 125) 20 N-lauroylsarcosin (Sigma, Cat. #L5777) Fast red TR/AS-MX substrate kit (PIERCE, Cat. #34034) Polyethylene glycol (Sigma, Cat.#P2139) Reagent Preparation: 1x hybridization buffer (4X SSC, 0.1% N-lauroylsarcosine, 0.02% SDS, 1% BM 25 blocking reagent),'-, 20x SSC 16 m 1% N-lauroylsarcosine 8 ml 10% SDS 160 pl BM blocking powder 0.8 g 30 H 2 0 51'ml total 80 ml WO 01/66803 PCT/USO1/07608 -75 Heat to 65'C to dissolve the powder then store at -20'C. 50% PEG-8000 (Polyethylene Glycol) PEG-8000 10g 5 H20 up to 20 ml Heat to 65'C to dissolve then autoclave. Aliquot and store at -20'C. 1Ox TBS (100 mM Tris, 1.5M-NaCl, pH 7.4) Tris base '12.1 g 10 NaC1 87.6 g

H

2 0 up to 1000 ml 120 mM X-gal X-gal- 100 mg 15 DMF 2 ml Store at -20*C. X-gal Substrate Buffer (1mM MgCl 2 , 3 mM K 3 Fe(CN) 6 , 3 mM K 4 Fe(CN) 6 in 1X TBS buffer) 500 ml 20 1OX TBS/pH7.4 50 ml Potassium Ferrocyanide 633.5 mg Potassium Ferricyanide 493.9 mg MgCl 2 101.6 mg 25 Filter and store at -20 C. BM Blocking Dilution Buffer/pH7.5 (0.1 M maleic acid, 0.15 M NaCl) 1M Maleic Acid 100 ml 5M NaCl 30 ml 30 Solid NaOH 7.5 g H20 up to 1000 ml 10% Blocking Reagent 100 ml Blocking Powder - 10 g WO 01/66803 PCT/USO1/07608 -76 Blocking Dilution Buffer (no tween 20) 100 ml Heat to 70*C then autoclave. Store at 4*C. 20% Dextran Sulfate 5 Dextran Sulfate '2g

H

2 0 upto 8 ml Autoclave then store at -20 0 C. DEPC-treated water 800 ml 10 Diethyl Pyrocarbonate (store at 4'C) 400 pl

H

2 0 800 ml Put in 37'C shaking water bafi for 4 hrs and then in 37'C warm room for overnight. Autoclave the solution for 45 minutes o'r 25 "minutes each for two times. Procedure: 15 A. Preparation of Jurkai'T Cells 1. Thaw one vial of Jurkat T cells. Transfer to 10 ml growth medium. The number of vials to thaw depends on the number of test to perform. In general, one vial of cells is needed for 2 herbal extract tests. 2. Re-suspend the cells in 50 ml of culture medium. 20 3. Incubate the cells for one day. Add 150 ml of culture medium and divide into two flasks with 100 ml each. 4. Culture the cells for 3 days. 5. Change the medium ind distribute 4 flasks with 100 ml of medium each. 6. Culture the c ells for 2 days. 25 7. Count the cell number. Collect the cells and spin down. Re-suspend the cell pellet with culture medium to 5x10 5 cells per ml and 100 ml per flask. 8. Culture the cells for 3,hrs before adding the herbal extract. B. Herbal extract treatment, 1. Prepare the herbal extract (see herbal extract preparation). 30 2. According to the growth inhibition concentration of herbal extract determined by the cell survival experiments, calculate the'50% growth inhibition concentration of each herbal extract.

WO 01/66803 PCT/USO1/07608 -77 3. Define the 50% growth inhibition concentration as H. Treat the cell with serial dilution of herbal extracts with the following concentrations: H, H/2.5, H/5, H/10, H/20. 4. Culture the cell for 24 hrs. 5. Collect cells and count the number of cells. Centrifuge to obtain cell pellet. 5 6. Wash the cell pellet with 1xPBS once. 7. Discard the supernatant. The cell pellet is ready for total RNA isolation. C. Isolation of total RNA 1. Add 1 ml of RNAzolTM B per 10 7 cells. Homogenize the cell pellet but do not vortex. 2. Add 0.1 ml chloroform per ml of homogenate, cover the samples tightly, shake 10 vigorously for 1 min(do not vortex). Place on ice for 15 min. 3. Centrifuge at 12000 rpm (13500xg) at 4 0 C for 15 min. 4. After centrifugation, the homo1genate develops two phases: a lower blue phenol chloroform phase and a colorless upper aqueous phase. DNA and proteins are in the interphase and the organic phase. Transfer the aqueous phase to a new tube, add an 15 equal volume of isopropanol and store the samples at -80'C. Note. The range of isopropanol addition isd'rom 0.7 to 1 volume of the aqueous phase solution. 5. Keep the samples atO -*0C until use. Let the sample completely thaw before centrifuging and mix 2 to 3 times by inverting the tube. Centrifuge samples for 15 min at 13000 rpm (15000fixg). 20 6. Remove the supernatant and wash the RNA pellet once with 1 ml of 75% ethanol. Centrifuge for 3 mii at'13090 rpm (15000xg) and at 4 0 C. 7. Discard the-supernatant. Dry the pellet under vacuum for 1 min. Note. Do not let the RNA pellet dry comp lately. It will greatly decrease its solubility. 8. Dissolve the RNA pellet Yfn5i100 pl of diethylpyrocarbonate (DEPC) - treated water 25 by pipetting. Note. If he pellet is hard to dissolve, incubating the pellet for 10 - 15 min at 60'C may help. 9. Measure absorbance at 260 nm (A 2 60 ) and 280nm (A 280 ) with a spectrophotometer. Concentration analysis. OD 26 0'x 40 ng/pl x dilution factor = total RNA (ng/ 1). D. Isolation of p olv-A + 'iiRNA from total RNA 30 1. Determine the amount of starting RNA and the appropriate volume of Buffer OBB and Oligotex Suspension solution to be added in the RNA solution according to the Table 17.

WO 01/66803 PCT/USO1/07608 -78 Table 17. Buffer amounts for Oligotex mRNA Spin-Column Protocol. Total RNA Add RNase-free Buffer OBB Oligotex Prep size water to: (pL) (pl) Suspension (p1) 20 pg 100 100 6 Mini 0.25 mg 250- 250 15 Mini 0.25-0.5 mg 500 500 30 Midi 0.5-0.75 mg 500 500 45 Midi 0.75-1 mg 500 500 55 Midi The following procedures are based on using 500 pg total RNA as an example. 2. Add 500 p1 of 2x Binding Buffer and 30 pl Oligotex Suspension to the total RNA 5 sample. Mix the contents thoroughly by flipping the tube. 3. Incubate the sample for 10 min at 70'C. 4. Incubate for 20 min at room. temperature. 5. Centrifuge for 2 min at maximum speed (14000 to 18000xg) and aspirate the supernatant. 10 6. Re-suspend the pellet in 400 pl of Wash Buffer OW2 and transfer onto spin column and centrifuge the spin column for 1 min. 7. Wash with 400 pl of 0W2 and centrifuge as above. 8. Add 20 pl of preheated (70'C) Elution Buffer onto the column and re-suspend the resin. Close the microcentrifuge tube. 15 9. Put the spin column with 1.5 ml microcentrifuge tube at 70'C for 3 min. 10. Centrifuge the colunn at maximum speed for 2 min at room temperature. 11. Elute again. (repeat step 8 to'10 to get better yield) E. cDNA Labeling 1. Mix 2pg mRNA, 11 of control plants mRNA for single color label (Hat22: 1xiO 9 , 20 Rbcl: 5x1O8, Ga4: 1x108, Rca: 5x10 7 , Asal: 1X107, Atps: 5x106 molecule/[L1), 6 pl of 50 mM random hexamer and DEPC-H 2 0 to 28.88 pl final volume. For dual-color mode, use 2 tg of mRNA each in Biotin or Dig labeling and individual addition of control plants mRNA: 1. biotin labeling: Hat22: 1x10 8 , Rbcl: 5x10 7 , Ga4: 2x107, Rca: 1x107, Asal: 1x10 7 Atps:. 1x107, Hat4: 1x10 7 /pl. 2. Dig labeling: Hat22: 1x10 7 , Rbcl: 25 1x10 7 , Ga4: 1x107 Rea: 1x.107, Asal: 1x10 8 , Atps: 5x10 7 , Hat4: 2x10 7 /pl. 2. Denature for 10 min at 70 0 C, then chill quickly in ice for 5 min. 3. Add 10 p1 of 5x first strand buffer, 5 pl of 0.1 M DTT, 1 pl of 25 mM dATP, dCTP, dGTP mixture, 1 pl of 2 mM dTTP , 2 pl of 1 mM biotin-16-dUTP, or Dig- 11-dUTP (1 WO 01/66803 PCT/USO1/07608 -79 mM), 0.63 pl of 40 U/pl RNAsin and 1.5 pl of Superscript II (reverse transcriptase, GIBCO BRL)(200U/pl). 4. Mix well and incubate for 10 min at 25'C, then for 90 min at 42'C. 5. Stop the reaction for 5 min at 94 0 C. 5 6. Add 5.5 pl of 3 M NaOH for 30 min at 50'C. 7. Add 5.5 pl of 3 M CH 3 COOH for 30 min at 50*C. 8. Precipitate the labeled cDNA by adding 34 pl of water, 50 p.

1 of 7.5 M ammonia acetate, 10 pg of linear polyacrylamide as carrier and 380 pl of absolute alcohol. 9. Incubate the sample for 30 min at -80'C. Centrifuge at 13000 rpm for 15 min. 10 10. Wash the pellet with 1 ml of 70% ethanol and centrifuge at 13000rpm for 5 min. 11. Dissolve the pellet in 36 p.1 of autoclaved H 2 0. For dual color, combine two labeled cDNA together. F. Array Hvbridization 1. The filter membrane carrying the 9600 EST PCR products is pre-hybridized in 5 ml of 15 lx hybridization buffer (4X SSC, 0.1% N-lauroylsarcosine, 0.02% SDS, 1% BM blocking reagent (Boehringer Mannheim)), and 50 pg/ml salmon sperm DNA (GIBCO BRL) at 63'C for 1.5 hours. Note. You can prepare 80 ml of 1x hybridization buffer and store it at -2OC.' Thaw the buffer at 60'C before use. 2. Stick one side of adhesive EasiSeal* square to a clean glass slide and place the pre 20 hybridized membran in the center of the square with the spots facing up. 3. Mix the probe with 2 pl of poly-d(A)10 (10 [ig/pl) and 2 pl of human Cot-1 DNA (10 pg/pl) (GIBCO BRL) and 40 pl of 2x hybridization buffer to 80 pl final volume. 4. Denature the probe mixture at 95'C for 5 min and then cool on ice. 5. Seal the filter membrane with the probe solution in the hybridization bag. 25 6. Incubate at 95'C for 5 min and then at 63'C for 12-16 h (overnight). 7. Wash the filter membrane twice with 5 ml of 2x SSC, 0.1% SDS for 5 min at room temperature. 8. Wash three times for 15 min each with 5 ml of 0.lx SSC, 0.1% SDS at 63 0 C. 9. Block the filter membrane with 5 ml of 1% BM blocking reagent containing 2% 30 dextran sulfate at room temperature for 1 h. 10. Incubate with 5 ml nixtuie containing 700x diluted Streptavidin-p-galactosidase (1.38U/ml, enzyme activity)(GIBCO BRL), 10000x diluted anti-Digoxigenin-alkaline WO 01/66803 PCT/USO1/07608 -80 phosphatase (0.075U/ml, enzyme activity)(Boehringer Mannheim), 4% polyethylene glycol 8000 (Sigma), and 0.3% BSA in 1x TBS buffer at room temperature for 2 hours. Note. This formula is for dual-color mode. For single color mode, anti-Dig-AP is not needed and the incubation time can be reduced to 1 hour. 5 11. Wash with 1x TBS buffer three times for 5 minutes each. 12. Freshly prepare X-gal substrate solution (1.2 mM X-gal, 1mM MgCl 2 , 3 mM

K

3 Fe(CN) 6 , 3 mM K 4 Fe(CN) 6 in 1x TBS buffer) by mixing 50 pl of 120 mM X-Gal and 5 ml of X-Gal substrate buffer. Immerse the filter membrane in the X-gal substrate solution for 45 min at 37 C with gentle shaking. 10 13. WashwithlxTBS. 14. Dual color development: stain the membrane with 5 ml of Fast red TR/naphthol AS MX substrate (Pierce, Rockford, IL) at room temperature for 30 minutes with gentle shaking. 15. Wash with deionized water. Stop the reaction with lx PBS containing 20 mM EDTA 15 for 20 min. 16. Air dry the filter menibrane'. Results. 1. Determine the growth inhibition concentrations of herbal extracts in cell cultures. 20 Each herbal extract has different cellular toxicity, thus it is necessary to determine the growth inhibition concentration of every herbal medicine before treating the cells. Five serial dilutions of herbal extrac1'(lO 5, 2, 1, 0.5 mg/ml) were added to 5x10 5 /ml cultured cells and incubated for 24 hours in an incubator at 37*C with 5% Co 2 . The numbers of survival cells at different concentrations of herbal extracts are shown in Table. 18. 25 30 35 WO 01/66803 PCT/USO1/07608 -81 Table 18. Survival Cell Number at Different Concentrations of Herbal Extracts experimental concentration no.x105/ml 10. 5 2 11 0.5 No High conc. Low conc. mg/ml mg/ml mg/ml mg/ml mg/ml drug (mg/m1) (mg/m1) 1 Cordyceps sinensis 8.4 11.9 11.5 9.2 9.0 12.4 10 1 mycelium 2 ST024 4.2 8.7 10 7.5 10 10 10 1 3 ST044 - .5.9 8.4 9.9 9.4 5 0.5 4 ST051 - 1.7 5.4 0.5 0.05 5 ST093 - _- 1.9 3.8 4.4 0.5 0.05 6 ST117 - 1.6 3.6 4.6 5.8 0.5 0.05 7 ST123 3.4 6.4 8 9.3 7.8 5 0.5 8 ST128 3.5 7.7 7.9 7.7 8.3 5 0.5 9 ST134 2.9 6.1 11.2 9.6 9.8 5 0.5 10 ST237 - - 2.5 6.6 8.7 1 0.1 11 PHY906-303503 - - 4.5 4.9 8.2 1 0.1 12 PHY906-284003 - - 3.8 5.7 7.6 1 0.1 Note: The original cell number was 5x107/ml. The number increased to 10x10"/ml after 24h incubation.."-" indicates alldead cells. 5 The cell number without herbal extract addition doubled after 24 hours incubation. On the other hand, the number of survival cells varies with different herbal medicine treatments. We chose the 50% growth inhibition concentration (IC 5 o) as the high concentration and one tenth of it as the low concentration. In order to maintain consistence of the Jurkat T cell line, a 10 cell banking system was established. In the cell bank, a total of 100 vials of cells (10 million cells per vial) were frozen in a.-1 50'C freezer. 2. Molecular classification of herbal medicines by nucleic acid microarray analysis. Analysis of 3 single-element herbal medicines. Three single-element herbal medicines, 15 Cordyceps Siniensis Mycelium (CSM), ST024, and ST1 17 were used to treat the cell cultures as described in the methods section. Gene expression measurements were performed by using microarrays of 13824 cDNAfragments each representing a distinct human transcript. For the data analyses, gene spots of high data quality were selected. The selection was based on signal to background ratio greater than -. 5 or the coefficient of variation (CV) of spot area smaller 20 than 10%. All the data sets were normalized with the control cells, which received no herbal treatments. The spot intensity a's rounded up to 10 for those intensities that were less than 10. Based on the selection criteria; a total of 492 genes with differential expression ratio greater WO 01/66803 PCT/USO1/07608 -82 than 1.5 were selected forelisier analysis. These data pre-processing procedures were performed by the prograiu !'DatExfract" and "Ratio2" developed in-house. These 492 genes were cliiter analyzed by the average-linkage method. The distance between genes is used as'tie linear correlation coefficient or resemblance coefficient. The 5 cluster analysis programs,'Cluster and TreeView, were based on hierarchical clustering method and were written by Dr. Michael Eisen of Stanford University (Eisen, 1999, Eisen et al., 1998). The results are shown in Figure 12. From Figure 12C, one can clearly identify that three different herbal medicines of high and low concentrations are each clustered together. For instance, CSM-L is closer to CSM-H and less similar to ST024 or ST 117 in the clustering tree. 10 A different clustering algorithm, the self-organizing map algorithm, which is based on non hierarchical method, yields the same results (data not shown). From the clustering results shown in Figures 12A & 13A,ee&ral features are noted. (1) 4 genes were up regulated by ST1 17 treatment but down regulated by other herbal treatments (Figure 12B). (2) 34 genes were down regulated by CSMtreatment but up regulated by others (Figure 13B). (3) 2 genes 15 were up regulated by all the three herbal treatments, one is Malic enzyme 2 and the other one is an anonymous gene (clone 10: 328351) (Figure 13C). (4) 12 genes were highly induced by the high concentration treatment and less induced by the low concentration treatment in all the three herbal medicines (Fig. 13D). Analysis of 2 preparations'of multi-elerfient herbal medicines. Two batches of Huang 20 Chin Tang, PI-iY906-30350'3(#1 1) & PHY906-284003 (#12), each with low and high concentrations were used to treat the cell cultures in three independent experiments. The gene expression profiles were acquird with microarrays of 9600 non-redundant cDNA elements. After the data pre-prqocessing procedures as described above about 5000 genes were selected for the subsequent data analysis. There were 3 repeats for each herbal treatment. For data 25 analysis, we use a modified method based on the one reported by Slonim et al. (Slonim, 1999). The following algorithm is designed to search for the candidate marker genes that have high differential expression ratios but low deviation in the three repeats. We designate a P(i) value to account for the gene i With the aforementioned features. 30 P(i) square root of'(X(pr-p)/ (ac+m) p: Mean expression levels.in three repeat experiments for herbal treated cells (pm) or untreated control ceilsp WO 01/66803 PCT/USO1/07608 -83 a : Standard deviationf he expression levels in three repeat experiments for herbal treated cells (am) or untreated control cells (ac). We calculated the P(i) value for each gene and selected 500 genes with the highest 5 scores as candidate genes for Cluster analysis (Figure 14). The values of each gene were averaged over the 3 repeats. As shown in Figure 14B, two different concentrations of #12 are clustered together (12-H &1l2-L). The higher concentration of the #11 preparation is closer to the #12 preparation cluster than the lower concentration of #11 preparation. However, all these clusters have similar resemblaiice coefficient (distance between clusters) compared with the 10 tree shown in Figure 12. These results suggest that the gene expression profiles of #11 and #12 preparations of Huang Chin Tang are similar. The results are justified based on the fact that these two preparations are based on the same herbal medicine mix. Several features are noted in the expression profiles illustrated in Figures 14A & 15. The averaged gene expression levels are shown in Figure 15A The Box1 encloses genes that 15 were down regulated in #11-L treated cells but up-regulated in others. These genes include 2 tRNA synthetase (isoleudire adnd methion), RNA polymerase II polypeptide B (Clone ID 42020), KIAA0212 gene (Clone ID 310497, containing ATP/GTP-binding site motif A), and KIAA0577 (Clone'ID 29263;,ATP-dependent RNA helicase). It is interesting to note that 3 out of the 6 genes were involved in the RNA replication. Box2 encloses the genes that were up 20 regulated by all the #11 and tl2 treatments. Box3 encloses the genes which showed no response by #11 -L treatment but were down regulated by the others. Box4 encloses the genes that were highly repressed by low concentration herbal treatment but were less repressed by high concentration herbal tr6atnient. Finally, in Box1 and Box3, the expression profiles of #11 treated cells are different from the profiles generated by the other 3 treatments. This result is 25 consistent with the finding depicted in Figure 14B. Combining the data sets of the gene expression profiles of the 3 single-elements and the 2 preparations of multi-element he-bal medicines together, a couple features are noted as an illustration. The KIAA0212 ,ene (Clone ID 310497, containing ATP/GTP-binding site motif A) was highly induced by all the high concentration herbal treatments except that it was only 30 mildly induced by the #11-t and the CSM treatments. Two genes, an anonymous gene, (Clone ID 510908) and Proteasomjie chain' precursor (Clone ID 70088) were highly up-regulated on all the treatments except dlowii-regulated by the CSM treatment.

WO 01/66803 PCT/USO1/07608 -84 We next worked on the crux to cluster analyze the gene expression profiles of the 5 different types of herbal medicine treatments. The data pre-processing procedures were performed as aforementioned and 500 genes were selected for cluster analysis. A hierarchical clustering was performed by the program "Cluster" described above. The hierarchical tree was 5 cut at the position where the range of distance between clusters is the greatest (Romesburg, 1989) and the result is shown in Figure 16A. The 3 single-element herbal medicines, CSM, ST024, and ST 117 are clustered together. Out of the 2 different batches of the multi-element herbal medicine, #12-H, #12-L and #11H are clustered together and the #11-L stands by itself. The result suggests that higher similarity exists between the #11 and #12 as compared with that 10 of CSM, ST024, and ST 117. In order to better classify the different herbal medicines, the data analysis algorithm was improvedby standardizing all the data sets so that the expression level of each gene across the different data sets has zero-mean and unit-variance (Tavazoie et al., 1999; Chen et al., 1999). This yields the transformed variables: 15 X~i =(xi-p'X)/ (YX p x: mean expression levels in th&data set a(: standard deviation of the expression levels in the data set x : un-transformed gene expression level 20 Xj: transformed gene expression level After standardizing the data set, #11 and #12 are clustered together as shown in Figure 16B. The ST024 and ST117 are blustered together and the CSM is in an independent cluster. Furthermore, the clustering also suggpsts that CSM is more similar to #11 and #12 than to the 25 ST024 and ST1 17. Another clustering algorithm, self-organizing maps, was performed with the same standardized data sets and yields the same result as the hierarchical clustering (Figure 16C). Class predictors for discrimirating #11 and #12 herbal treated expression profiles. The above cluster analyses for the#1 1and #12 show that they are similar and further classification 30 is difficult by the hierarchloal clustering or self-organizing maps methods with the data set containing the 500 genes of the highest P(i) values. We then modified the algorithm to select genes with larger expression ratio difference between #11 and #12 herbal treated cells, but WO 01/66803 PCT/USO1/07608 -85 smaller variation in the two herbal treated cells. The T(i) value is defined to score this feature as following: T(i)= log(p1) - log(p12) / (ani+a 1 2 ) 5 p: Mean expression ratios in three time experiments for #11 treated cells (pr) or #12 treated cells (p12) a: Standard deviation of the expression ratios for #11 treated cells (all) or #12 treated cells (a,) 10 We calculated the T(i) value for each gene and selected 50 genes with the highest scores as class predictors,(Figure 17). 18 genes were up-regulated by the #11 treatment and down-regulated by the #12,treatment. The rest of the genes were up-regulated by the #12 treatment and down-regulated by the #11 treatment. We then used these class predictors to 15 classify two test herbal prepaiaions based on a modified method described by Golub et al., 1999. Two different batches of Huang Chin Tang preparations, PHYO 10401 (#16) & PHYO10402* (#17) were obtained from Sun Ten Pharmaceutical Co. and were used for the class prediction test. The gene expression profiles of #16 and #17 preparations were 20 normalized with the expression profile of the untreated control cells and standardized with the class predictors. Each predictor gi votes for either #11 or #12 herbal preparation depending on whether its expression level xiis closerto #11 or #12. The vote for each gene is given by vi Ix, - (gmpe)/2|, where 25 p: Mean expression ratio in three repeat experiments for #11 (p) or #12 herbal treated cells (p12) The average votes 1 and 12 were collected from the predictor genes correlated with the predictor on #11 and #12, respectively. The prediction strength (PS) reflects the margin of 30 victory and it was defined as PS= (V 1

-V

12 )/ (VI 1

+V

12 ). If the PS was greater than 0, it indicated that the herbal preparation was more similar to #11 and less similar to #12. The results obtained from the analyses on,#16 and #17 indicated that #16-H was similar to #11 WO 01/66803 PCT/USO1/07608 -86 (PS=0.1) and #16-L, #17-H, aid #17-L were similar to #12 (PS=-0.29, -0.21 and -0.2, respectively). Based on the information of #16 and #17 preparations, this test failed to correctly identify #16-L as more similar to #11. Discussion. 5 Characteristic gene. expression profiles for herbal medicines treated cells. The predictor genes are selected from the differentially expressed genes in two herbal treated cells. These genes represent the cellular responses to the herbal medicine treatments. In this study, we have identified some interesting genes based on their responses across different herbal treatments (Figures 13, 15, and 17). These genes are valuable assets in studying the signaling pathways of 10 cells in response to the herbal stimulation and in deciphering the molecular pharmacological mechanism of herbal medicines. Classification of hei-bal'midicines by nucleic acid microarray analysis. In summary, a two-step classification procedure is proposed. An initial classification procedure based on the standardized data sets and the clustering algorithms is performed and followed by a final 15 classification procedure with the class predictors. All the procedures can be integrated in a computer program. In these pielininary studies, all the genes have the same contribution for classification. When the data sets afe large enough, the weight for each gene (or predictor) can be acquired from the linear correlation coefficient (Golub et al., 1999, Chen et al., 1999). The #11-L failed to be clustered with #11-H by significant association. A similar 20 preparation of #11, the #16 preparation yielded the same results. It was interesting to discover that no matter what clustering algorithms were applied, the #11 and the #16 preparations did not yield the expected result. Even with independent experiments, the results remained the same. The reason behind thefailure will be investigated with detailed information of #11 and #16 preparations. 25 Quality control and analysis in microarray system. The quality of acquired microarray data and the choice of the statistical analysis methods are both important factors for achieving meaningful results. Wd haveredognized that the variation among arrays contributes to errors in measuring gene expression levels. Based on the data in this report, we have found that for every herbal preparation, the high concentration treated expression profiles always cluster with 30 its lower concentration counterpart (Figures 16B and 16C) and we could classify the STl 17, ST024, CSM and Huang Chin-'Tang with two different clustering methods. In the past three months, the array quality has been- improved to have less than 7% CV. We have also set up a standard procedure for assessing'the quality of every batch of arrays fabricated in the lab. All WO 01/66803 PCT/USO1/07608 -87 these experimental findings and improvements in the microarray technologies suggest that classification and characterization of the Chinese herbal medicines by microarray system are feasible. References for Example 5 Bertucci-F; Bernard-K; Loriod-B; Chang-YC; Granjeaud-S; Birnbaum-D; Nguyen-C; Peck-K; Jordan-BR (1999) Sensitivity issues in DNA array-based expression measurements and performance of nylon micio ays'fir small samples Human Mol. Genetics 8(9): 1715-1722. Bittner-L, Trent-J, Meltzer-P (1999) Data analysis and integration: of steps and arrows. Nature 10 Genet. 22, 213-215 Brown-PO; Botstein-D (1999) Exploring the new world of the genome with DNA microarrays. Nature genetics 21 (1) supplement, 33-37. 15 Chen-JJ; Wu-R; Yang-PC; Huang-JY; Sher-YP; Han-MB; Kao-WC; Lee-PJ; Chiu-TF; Chang F; Chu-YW; Wu-CW; Peck-k(i 8) expression patterns and isolating differentially expressed genes by cDNA microarrdy systie'with colorimetry detection. Genomics. 51: 313-24. Chen-Y, Bittner-M, Dougherty-ER (1999) Issues associated with microarray data analysis and 20 integration. Information supplementary to article by Michael Bittner, Jeffrey Trent and Paul Meltzer (Nature Genet. 22,213-215). Duggan-DJ; Bittner-M; Chen-Y; Meltzer-P Trent -JM (1999) Expression profiling using cDNA microarrays.Natur geiietics 21 (1) supplement, 10-14. 25 Eisen-M (1999) Cluster and Teeview manual.'(rana.stanford.edu/software) Eisen-M, Spellman-PT, Brovn-P , Botstein-D. (1998) Cluster analysis and display of gene wide expression patterns. Proc. Natl. Acad. Sci. USA 99:14863-14868 30 Golub-TR, Slonim-DK Taiiayo-P, Huard-C, Gaasenbeek-M, Mesirov-JP, Coller-H, Loh-ML, Downing-JR, Caligiuri-MA, Bloomfield-CD, and Lander-ES (1999) Molecular classification WO 01/66803 PCT/USO1/07608 -88 of cancer: class discovery and class prediction by gene expression monitoring. Science Oct 15: 531-537 Lander-ES (1999) Aray of hope. Nature genetics 21 (1) supplement, 3-4. 5 Romesburg-HC (1989) Cluster analysis for researchers. Chapter 16: How to make classifications. P203-216, Kriegei Publishing Co. Malabar, Florida, USA Slonim-DK, Tamayo-P, Mesirov-JP, Golub-TR, Lander-ES (1999) Class prediction and 10 discovery using gene expression data. (www.genome.wi.mit.edu/MPR) Tavazole-S, Hughes-JD Canpbell-MJ, Cho-RJ, Church-GM. (1999) Systemic determination of genetic network architecture. Nature genetic 22: 281-285 Example 16. Identify characteristic gene expression profiles induced by an herbal 15 medicine As stated in Example 15, the prescription of the Chinese herbal medicine Scute and Licorice combination (Huang Chin Tang) stops diarrhea, relieves spasms and clears fever. The ingredients of Huang Chin Tang are Scute, Peony, Licorice and Jujube. In this study we used the nucleic acid microarray technology to study the gene expression profiles induced by herbal 20 medicines in mammalian cells. To investigate the characteristic expression profiles induced by the Huang Chin Tang, Jurkati cells were ttated with 5 batches of Huang Chin Tang (PHY01040; #16, PHY010402; #17, PHY03061; #18, PHY03062; #19 and PHY02231; #20 obtained from Sun Ten Pharmaceutical Co.) by 5 concentrations (1/2, 1/2.5, 1/5, 1/10, and 1/20 of IC 50 ). 25 Nucleic Acid microarray with a two color detection method was employed to measure the expression profiles. The miRNA extracted from herbal treated cells was labeled with digoxigenin and the mRNA extracted from untreated cells was labeled with biotin. Arrays of 9600 features were employed and the procedures described by Chen et al. (Genomics, 51, 313 324, 1998) were adopted in the experiments. For data pre-processing, only array spots of high 30 data quality were selected. The 'selection was based on signal to background ratio greater than 2.5 and 1.5X differential expression ratio. By these criteria, 1081 genes were selected for further statistical analysis Non-hieiarchical cluster analysis programs such as the Genecluster program developed in Massadhusetts Institute of Technology (Tamayo et al., 1999) was WO 01/66803 PCT/USO1/07608 -89 employed to categorize the expression profiles. The Genecluster program is based on self organizing map (SOM) principle. A'6X4 SOM clustering of expression profiles are shown in Figure 18A. The -details of gene expression profiles for the selected clusters are shown in Figure 18B. In these clusters,'clusters 3 and 20 (labeled c3 and c20) were selected for that the 5 gene expression levels increase with higher herbal concentrations. Similarly, clusters 5 and 9 for that the gene expression levels decrease with higher herbal concentrations. Cluster 23 collects genes whose expression levels were up regulated compared with that in untreated cells and cluster 0 for down regulated genes. These expression profile clusters are further condensed into two major groups, A & B. Group A collects genes up regulated by herbal treatment and 10 Group B collects.genes down regulated by herbal treatment. The expression profiles in Group A & B form the basis of a characteristic expression data set for an herbal preparation. The same procedures were repeated for 5 different batches of Huang Chin Tang, and 952 genes were selected to establish tlie -characteristic expression profile database of the Huang Chin Tang. 15 As shown in Figure 19, byth& aforementioned procedures, a gene can be categorized as Group A, B or none (non-A and non-B) and its expression profile can be represented by 1, -1, and 0 respectively. The number of different gene expression profiles between batch #1 and batch #2 are 3 in Group- A (Gene 6,'7, and 8) and 2 in Group B (Gene 15 & 16). By the same principle, the number of different expression profiles between batch #1 and #3 are 10 in Group 20 A and B and the number is 11 between batch #2 and batch #3. These numbers indicate that batch #1 and #2 are more. similar than batch #3. This principle was applied to classify 5 different batches of herbal preparations. The following algorithm is designed to calculate the distance between a pair ofheibal preparation btches, i and j. dij= Z S(Xi, X ) (Hamming distance) 25 The gene X in i batch of preparation is assigned to Group A, B or none if Xi<>X , 8(XiX ) 1: if Xi=Xj, S(X,X)=0. We calculated all the d value between pairs of herbal preparations for cluster analysis. The analysis programs, Kitschluster was based on hierarchical clustering principle and was 30 written by Dr. Joseph Felsenstein of Washington University (http://evolution.genetics.washiigton.edu/phylip.html). From the Hamming distance table (Figure 20), one can clearly identify that the shortest distance lie between batch #17 and batch #18 and that batch #17 is similar to #18. Batch #16 also similar with batch #17 and #18 but WO 01/66803 PCT/USO1/07608 -90 batch #19 is dissimilar to the-nest ofbatches. The results were confirmed by HPLC analyses as described below. By HPLC, the chemical composition of the 5 batches of herbal preparations was analyzed. Four major peaks (BG, B, Gly, and Pf) in the chromatograms were selected for statistical 5 analyses. Two additional paranieters, BG+B and BG/B, were included in plotting the 6 coordinate radar graph as shovn in Figure 21. The distance on each coordinate is the integrated intensity of that particular chemical constituent in the chromatogram. In general, #16, #17 and #18 were similar in their constituent content (baicalin and baicalein) of Scutellariae Radix, which are within 33.55-36.08; while the amount of the same constituents are higher in #19 and 10 #20 (42.49 and 44.96, respectively). The resemblance of #16, #17, and #18 can be seen from the coincident radar plots in Figure 21B. In order to identify the unknown herbal medicine based on the characteristic expression profile database established as described as above, Jurkat T cells were treated with a tester sample #17 in 5 concentratidis to set up the characteristic expression data set for the tester. 15 The Hamming distances betie~n the tester and each of the data sets (#16, #17, #18, #19 and #20) in the characteristic expression database were calculated and the scores are: #16: 502, #17: 405, #18: 402, #19: 699, and #20: 531. These data show that the tester is most similar to #17 having a lowest Hamming distance score of 405. The example demonstrates that this invention teaches a method to-identify unknown herbal medicine based on the gene expression 20 profiles induced by the herbal'medicine inmammalian cells. The identity of the unknown herbal medicine can be infoted by aligning the characteristic expression profiles with a collection of characteristic expression profiles of herbal medicines in an HBR Array. Based on the characteristic expression database, marker genes and signature expression profiles can be deduced for an herbal medicine for studying its pharmacological mechanisms 25 and for optimizing the formuaftion of a'complex herbal preparation. For this example, 5 different batches of Huang Chin Tang preparations (#16, #17, #18, #19 and #20) were obtained from Sun Ten Pharmaceutical Co. and a characteristic expression profile database was constructed based on aforementioned procedures. For each gene, the consistency of expression profiles in the database was scored by the coefficient of variation (CV value): 30 CV=a/(E pi/n) pi: Mean expression ratios fo #i treated cells. n : Number of the data set, n 5 in this case. c7: Standard deviation of the expression ratios for #16, #17, #18, #19 and #20.

WO 01/66803 PCT/USO1/07608 -91 Since the CV reflects the variation of data, the marker genes for an herbal medicine were selected based on the CV score. The top 50 genes with the minimum CV scores were selected. Figure 22 shows 25 markegenes with up regulated signature profiles and 25 marker genes 5 with down regulated signature profiles for Huang Chin Tang. The characteristic expression profile database can be used to infer the expression profiles of individual chemical constituents in a mixture as complex as an herbal medicine if the amount of the chemical constituents can be semi-quantitatively determined. In this example, the chemical composition of.an herbal medicine is determined by high performance liquid 10 chromatography. The integrated intensities of 4 chemical constituents in five batches of Huang Chin Tang preparation were quantified by HPLC analysis. The gene expression ratios for each batch of herbal preparation were calculated by taking the median of the expression ratios induced by 5 concentrations ofherbal preparation. The correlation between a constituent and a gene expression profile was quantified by the Pearson correlation coefficient. The Pearson 15 correlation coefficient forgenx and the constituent y is: R = (1/n) Y(xi-px)( yi-y) / ax y , i= 1 to n n : Number of the herbal breparaion,. n= 5 in this case. pxmg: Mean expression ratios in five herbal preparation for gene x. pty: Mean integrated intensity in five batches of herbal preparation for constituent y. 20 xi: Gene expression ratios in #i 'herbal preparation for gene x. yi: Integrated intensity in #i herbal preparation for constituent y. a: Standard deviation of the expression ratios (a,) or integrated intensities (ay) for five herbal preparations. 25 Several genes whose expression levels highly correlated (with IRI>0.99) with the amount of chemical constituents in Huang Chin Tang were identified for each constituent. For example, the R value between the gene (clonelID: 67185) and Glycyrrhizin was 0.998 (Figure 23A). One the other hand, the gene (cloieID: 344720) whose expression levels increase with the decrease of Wogonin(WG) has an R value of -0.997 (Figure 23B). In addition to the above two 30 examples, 191 and 170 genes were highly correlated with individual constituents with R value > 0.9 and R value < -0.9, respectively. For instance, 17 and 18 genes were positively and negatively, respectively, correlated with Albiflorin (Af) (Figure 24). This example teaches a WO 01/66803 PCT/USO1/07608 -92 method to profile gene expression for individual constituents in a mixture without isolating them to perform the expression analyses one constituent by another. References for Example 16. US Patent documents 5 Stoughton-Roland and Friend-S H. USP# 5965352: Methods for identifying pathways of drug action Brown-PO and Shalon-TD USP#5807522: Methods for fabricating microarrays of biological samples 10 Lockhart-DJ, Brown-EL, Wong7GG Chee-MS, and Gingeras-TR. USP#6040138: Expression monitoring by hybridizatio.fihigli density oligonucleotide arrays Ladunga-I. USP#59$7390: Methods and systems for identification of protein classes 15 Mahant-S, Shivaling-S, Vie-G. USP#5951711: Method and device for determining hamming distance between two nulti-bit digital words Foreign Patent documents 20 Brown-PO and Shalon-TD EP#913485A1: Method and apparatus for fabricating microarrays of biological samples Other Publications Bertucci-F; Bernard-K; Loriod-B; Chang-YC; Granjeaud-S; Birnbaum-D; Nguyen-C; Peck-K; 25 Jordan-BR (1999) Sensitivity issues in DNA array-based expression measurements and performance of nylon microagrays for small samples Human Mol. Genet. 8(9): 1715-1722. Bittner-L, Trent-J,- Meltzer-P (1999) Data analysis and integration: of steps and arrows. Nature Genet. 22, 213-215 30 Brown-PO; Botstein-D (1999) Exploring the new world of the genome with DNA microarrays. Nature genet. 21 (1) supplement, 33-37.

WO 01/66803 PCT/USO1/07608 -93 Chen-JJ; Wu-R; Yang-PC; Huang-JY; Sher-YP; Han-MH; Kao-WC; Lee-PJ; Chiu-TF; Chang F; Chu-YW; Wu-CW;'Peck-K (1998) expression patterns and isolating differentially expressed genes by cDNA microarray system with colorimetry detection. Genomics. 51: 313-24. 5 Chen-Y, Bittner-M, Dougherty-ER (1999) Issues associated with microarray data analysis and integration. Information supplementary to article by Michael Bittner, Jeffrey Trent and Paul Meltzer, Nature Genet. 22, 213-215. 10 Duggan-DJ; Bittner-M; Chei-Y; Meltzer-P Trent -JM (1999) Expression profiling using cDNA microarrays. Nature genet. 21'(1) supplement, 10-14. Eisen-M (1999) Cluster and Treeview manual. (ftp://rana.stanford.edu/software) 15 Eisen-M, Spellman-PT, Brown-PO, Botsteiii-D. (1998) Cluster analysis and display of gene wide expression patterns. Proc. Natl. Acad. Sci. USA 99:14863-14868 Golub-TR, Slonim-DK, Taiayo-P, Huard-C, Gaasenbeek-M, Mesirov-JP, Coller-H, Loh-ML, Downing-JR, Caligiuri-MA, Bloomfield-CD, and Lander-ES (1999) Molecular classification 20 of cancer: class discovery and class prediction by gene expression monitoring. Science Oct 15: 531-537 Lander-ES (1999) Array of hope. Nature genet, 21 (1) supplement, 3-4. 25 Tamayo-P, Slonim-D, Merirov-J, Zhu-Q, Kitareewan-S, Dmitrovsky-E, Lander-ES, and Golub-TR (1999) Interpietii patterns of gene expression with self-organizing maps: Methods and application'to henatopoietic differentiation. Proc. Natl. Acad. Sci. USA 99:2907-2912 Romesburg-HC (1989) Cluster analysis for researchers. Chapter 16: How to make 30 classifications. P203-216, Krieger Publishing Co. Malabar, Florida, USA Scherf-U, Ross-DT, Waltham-l, Smith-LH, Lee-JK, Tanabe-L, Kohn-KW, Reinhold-WC, Myers-TG, Andrews-DT, Scudiero-DA, Eisen-MB, Sausville-EA, Pommier-Y, Botstein-D, WO 01/66803 PCT/USO1/07608 -94 Brown-PO, Weinstein-JN (2000) A gene expression database for the molecular pharmacology of cancer. Nature genet. 24: 236-44. Slonim-DK, Tamayo-P, Mesirov-JP, Golub-TR, Lander-ES (1999) Class prediction and 5 discovery using gene expression data. (http://www.genome.wi.mit.edu/MPR) Tavazole-S, Hughes-JID, CanpbellbMl, Cho-RJ, Church-GM. (1999) Systemic determination of genetic network architecture. Nature genet. 22: 281-285 Example 17. Identification of the bioresponses and the signature genes of an 10 herbal composition. To further investigate the expression profiles induced by the.Huang Chin Tang, Jurkat T cells were treated with Huang Chin Tang (PHY906#2 obtained from Sun Ten Pharmaceutical Co. Taiwan) by 5 concentrations (1/20, 1/10, 1/5, 1/2.5, and 1 of IC 50 ). Nucleic acid microarray with dual-color detection was employed to measure the expression profiles. The 15 mRNA extracted from herbal treated and untreated cells were labeled with Biotin-16-dUTP and Dig-1 1-dUTP, respectively. A control group was established by using mRNA extracted from untreated cells and labeled the mRNA with biotin and dig respectively in equal proportions. Five sample groups with different concentrations of herbal treatment and one control group were employed for the study by the procedures described by Chen et al. (1998) .20 with minor modifications. For data pre-processing, only array spots of high data quality were selected. The selection criteria were spots with signal to background ratio greater than 2.5-fold and more than 1.5-fold in differential expression ratio. By these criteria, 1044 genes were selected for further statistical analysis. The gene expression profiles of the control group were highly correlated with only 48'genes listed as statistical outliers that lie beyond the 2-fold 25 differential expression range (Figure 25A). For the sample group, many differentially expressed genes are evident as illustrated in Figure 25B. The number of genes whose differential expression "ratio greaterthan 2-fold increases with concentration of herbal treatment as shown in Figure 25C. These results prove that the identified differential expressed genes are truly induced byHuang Chin Tang treatments. 30 To identify the genes that were specifically induced by PHY906 (signature genes for PHY906), the expression profiles were clustered by a non-hierarchical cluster analysis programs "GeneCluster" developed'by Tamayo et al., 1999. The computer program is based on self-organizing map (SOM) principle and the clusters of expression profiles are shown in WO 01/66803 PCT/USO1/07608 -95 Figure 26. The X-axis represents the-herbal concentration from low to high and the Y-axis is the gene-expression ratio. Th signature genes were selected from the expression profiles which exhibit dosage response to the PHY906#2. The induced and repressed genes were selected from cluster 3 &A and cluster 18 & 19, respectively. In order to identify signature 5 genes for PHY906, anotherbatch6f Huang Chin Tang, PHY906#3, with the same formula and manufacturing process Were performed as described for PHY906#2. The induced and repressed genes commonly founding both batches are shown in Figure 27. Score similarity of bioresponses by self-organizing map (SOM) To differentiate herbal medicines of similar compositions, a scoring method is developed 10 and the score S represents the' difference in bioresponses of a biosystem to two different herbal compositions. S =ZPYWY, Where Pg is the number of the common genes induced both by herbal prep. A and herbal prep. B in cluster i and in clusteij. For example, the SOM clustering results for the expression 15 profiles of both batches of PIHY906 are shown in Figure 28A. In cluster C13 and C14, 17 and 25 genes share the same expression profiles for both batches of PHY906, respectively. In addition, 10 genes whose Gexpresion profiles induced by PHY906#2 are clustered in C13 but are clustered in C14 for PHY906#3. Therefore, P 1313 =17, P 1414 = 25, and P 1314 =10. A weighing factor, Wy, describes the ditance between the cluster i andj to indicate the similarity of the 20 two expression profile clustdrd. In the case of C 13 and C 14, these 10 genes have similarly response to PHY906#2 and PHY906#3 (Figure 28B). The weighing factor is defined as: Wy = 1 - Eu /Max(E), where Eu is the Euclidean distance between the cluster i and] and the value is normalized by F' /Max(EU). When i =j, Wy is 1. The number decreases as cluster i and clusters become more different (Figure 28C). 25 Classification of 5 batches of herbal medicines To test how well the a6bV- method performs in classifying 5 batches of similar herbal preparations, Jurkat T 'cells weieteated with 5 batches of Huang Chin Tang (PHY01040; #16, PHY010402; #17, PHY0306:1 ;#8, PHY03062; #19 and PHY02231; #20 obtained from Sun Ten Pharmaceutical Co.) by 5 concentrations (1, 1/2.5, 1/5, 1/10, and 1/20 of IC 50 ). The Sy 30 scores were calculated betwe en pairs of herbal preparations in cluster analysis (Figure 29). The analysis programs, Kitsch Cluster was based on hierarchical clustering principle and was written by Dr. Joseph Felsenstein of Washington University (http://evolution.genetics.washington.edu/phylip.html). The S scores (distance) are tabulated WO 01/66803 PCT/USO1/07608 -96 (Figure 29A), one can clearly identify that the shortest distance lie between batch #17 and batch #18 and that batch #17 issimilar to #18. Batch #16 also similar with batch #17 and #18 but batch #19 is dissimilar to the rest of batches. The results were confirmed by HPLC analyses. 5 Characterize an unknown herbal medicine based on the expression profiles To identify an unknown herbal medicine based on the characteristic expression profile database established as described as above, Jurkat T cells were treated with a tester sample #17 in 5 concentrations to set up the characteristic expression data set for the tester. The S score between the tester and each-of the data sets (#16, #17, #18, #19 and #20) in the characteristic 10 expression database were calculated and the S scores are: #16: 0.78, #17: 0.85, #18: 0.84, #19: 0.77, and #20: 0.79. These data show that the tester is most similar to #17 having a higher S score of 0.84. The example demonstrates that one can apply the method to identify an unknown herbal medicine based on the gene expression profiles induced by the herbal medicine in mammalian tells. The identity of the. unknown herbal medicine can be inferred by 15 aligning the characteristic expression profiles with a collection of characteristic expression profiles of herbal medicines in 'an HBR Array. The property of an herb can be described by four natures and five flavors (in Chinese Herbal Phramaceuticals, Ed. Zheng Hua Yen, People's Health publications, Beijing, China, 1997; Book of Ben Cao Gan Mu by Shi Zeng Li, Ming Dynasty, China). Each of the four 20 herbs in PHY906 may relate to another set of herbs with similar property (see Table 19). Or herbs with similar propertyniayexhibit similar bioresponse. HBR Arrays may be used to determine or measure the relatedne'ss in terms of the property of herbs. Such information may be useful in creating a new herbal formulation. 25 30 WO 01/66803 PCT/USO1/07608 -97 Table 19: Herbs with PHY906 Properties Herb Properties Drugs with similar properties Scutellariae Radix. Bitter tasting, Coptidis Rhizoma, Phellodendri Cortex, Gentianae Radix, cold natured Gardeniae Fructus Paeoniae Radix Bitter and sour Canarii Fructus (sweet and sour tasting, moderate tasting, mildly, natured), Potulacae Herba (sour tasting, cold natured), cold in nature. Fraxini. Cortex( bitter tasting, cold natured), Sophorae Flos (bitter tasting, mild cold in nature), Bletillae Tuber (bitter, sweet and harsh in taste, and mildly cold in nature). Glycyrrhizae Sweet tasting, Lycii Fructus, Polygonati Officinalis Rhizoma, Polygonati Radix moderate natured Rhizoma Zizyphi Fructus Sweet tasting and- Saccharum Granorum, Juglandis Semen warm natured Properties: Four natures-cold, hot, warm, cool. 5 Five flavors-acrid, bitter,, sweet, sour, bland. Example 18. Evaluation of an herbal medicines by HBR Array. As stated in Example T6, the component herbs of Huang Chin Tang are Scute, Peony, Licorice and Jujube. The gene expression profiles induced by five batches of Huang Chin Tang in mammalian cells Were characterized. A standard formula for Huang Chin Tang can be 10 defined and characterized with animal studies or with clinical studies. For example, the #17 was used as the standard formula for Huang Chin Tang based on the quality control and other standards set up by Sun TenPharmaceutical Co. The bioresponses of #17 were used to build the HBR Array for Huang Chin Tang. The marker genes in the HBR Array Were selected to evaluate other preparations of Huang Chin Tang composition. A tester Huang Chin Tang may 15 contain the same herb compositions but the component herbs may be grown under various environmental characteristics.. Coruparing the bioresponses of the tester with the marker genes of standardized HBR Arr ay;,the biological activities of the tester were evaluated. Furthermore, the marker 'genes whose expression levels are highly correlated (with |RI>0.99) with the dosage of component herbs in Huang Chin Tang (as stated in Example 16 20 and Figure 25) are selected foi evaluation purpose. The tester Huang Chin Tang can be evaluated by compaiing'the specific bioresponses or expression levels of the selected set of marker genes with the HBR Array. If the expression levels or bioresponses of the selected marker genes are beyond'the acceptable variation region, the amount or characteristics of the WO 01/66803 PCT/USO1/07608 -98 component herbs are adjusted or modified to meet the acceptable variation. The process is repeated until the bioresponses induced by the revised herbal composition are within the acceptable variation range by comparing with the standard HBR Array. Example 19. Predicting biological activity and therapeutic applications of an 5 herbal composition. According to the identified marker genes for PHY 906 (Figure 27), these genes can be used to predict the biological acti ities of the herbal composition. For example, the following underlined marker genes of PHY906 have been reported to involve in the following biological activities and therapeutic effects. The only effective drug against ALL is to inhibit the 10 asparagine synthetase due to increased cellular apoptosis (Nandy et al., 1998). Long-acting drug somatostatin analogs are applied in the treatment of neurofibroma for their tumor growth inhibitory effect because they induce antiproliferative action mediated by the inhibition of G6PD, transketolase, or bofh (Boros et al., 1998). Ephrin-A1 is a new melanoma growth factor and is highly expressed during melanoma progression (Easty et al., 1999). Mitogen 15 activated protein kinase (MAPK) family members have been recently reported to have opposing effects on apoptosis (Dabrowski et al., 2000). The expressions of asparagine synthetase, transketolase, ephrin-Al and MAPK are repressed with the higher concentration of PHY906 treatments. The down-regulation of these genes are involved in cell apoptosis. The expression of the enzyme argininosuccinate synthetase, cathepsin G and chemokine RANTES 20 are highly induced in inflammatory mechanism. By the PHY906 treatment the inflammatory involved genes are suppressed.,. These literature reports provide a basis for predicting the biological activities or therapeutic effects of an herbal composition. References for Example 19. Nandy-P; Periclou-AP; Avramis-VI (1998) The synergism of 6-mercapopurine plus 25 cytosine arabinoside followed by PEG-asparaginase in human leukemia cell lines (CCRF/CEM/0 -and CCRF/01M/ara-C/7A) is due to increased cellular apoptosis. Anticancer Research 18: 727-737 Boros-LG; Brandes'JL; Yusuf-FI; Cascante-M; Williams-RD; Schirmer-WJ (1998) 30 Inhibition of the oxidative aidnonoxidative pentose phosphate pathways by somatostatin: a possible mechanism of antitumor action. Medical Hypotheses 50: 501-506 Easty-DJ; Hill-SP; Hsu-MY; Fallowfield-ME; Florenes-VA; Herlyn-M; Benett-DC WO 01/66803 PCT/USO1/07608 -99 (1999) Up-regulation of ephrin-A1 during melanoma progression. Int. J. Cancer 84:494-501 Dabrowski-A; Tribillo-I; Dabrowska MI; Wereszczynska-SU; Gabryelewicz-A (2000) Activation of mitogen-activated protein kinases in different models of pancreatic acinar cell 5 damage. Z-Gastroenterol.- 38: 469-481 The foregoing detailed description has been given for clearness of understanding only and no unnecessary limitations should be understood therefrom as modifications will be obvious to those skilled in the art. 10 While the invention has been described in connection with specific embodiments thereof, it will be understood that itis apable of further modifications and this application is intended to cover any variations; uses; Ior adaptations of the invention following, in general, the principles of the invention and including such departures from the present disclosure as come within known or customarypiactic6within the art to which the invention pertains and as may 15 be applied to the essential features hereinbefore set forth and as follows in the scope of the appended claims.

Claims (24)

1. A method of establishing a standardized Herbal BioResponse Array (HBR Array) for an 5 herbal composition comprising: d) selecting a characterized herbal composition; e) exposing a biosystem to a-batch of the characterized herbal composition and collecting data 10 on two or more markers, wherein one of the markers is a change in gene expression determined through the use of a nucleic acid microarray, produced by the steps comprising: iv) producing a cell banking system; 15 v) profiling the gene expression pattern of cells from the cell banking system before and afterkexposure to the herbal composition; vi) selecting as markers those genes whose expression levels are changed by exposure to the herbal composition; 20 f) storing the marker data of step b) as a standardized HBR array.

2. The method of claim 1, further comprising: 25 g) repeating steps b) and c) for one more batches of the herbal composition using two or more of the same or differeiit iatkers than used in step b); h) combining the HBR Arrays obtained in steps c) and d); and 30 i) analyzing the combined HBR Array of step e) to generate a standardized HBR Array for the characterized herbal composition. WO 01/66803 PCT/USO1/07608 -101

3. The method of claims 1 or 2, wherein the characterized herbal composition has at least one known BioResponse.

4. The method of claims 1 or 2, wherein one or more of the following is known for the 5 characterized herbal composition: chemical testing, the part of the plant used, the growing conditions of one or more of the individual herbs in the characterized herbal composition, the pre-harvest treatment of one or more of the individual herbs in the characterized herbal composition, the post-harvest treatment of one or more of the individual herbs in the characterized herbal composition, the post-harvest treatment of the characterized herbal 10 composition, and the relative proportions of the individual herbs in the herbal composition.

5. The method of claims Ior 2, wherein the cell banking system comprises a master cell bank and a working cell 'bank. 15

6. The method of claim 5, wherein'the cells of the working cell bank are obtained from the master cell bank.

7. The method of claim 5, wherein the step of profiling the gene expression pattern of cells from the cell banking system before and after exposure to the herbal composition is 20 performed using cells from the working cell bank.

8. The method of claims 1 or,2, wherein the change in gene expression is determined using a nucleic acid microarray. 25

9. The method of claim 8, wherein the said genes whose expression levels are changed by exposure to the herbal coi-nposition are selected based on the criteria of having a signal to noise ratio of about 2.51or greater in the nucleic acid microarray and having an about 1.5 or greater change in the differential expression ratio. 30

10. The method of claim 8, herein data regarding between about 10 and about 20,000 genes whose expression levels are changed is stored as part of the IBR Array. WO 01/66803 PCT/USO1/07608 -102

11. The method of claim 10, wherein data between about 10 and about 1,500 genes whose expression levels are, changed is stored as part of the HBR Array.

12. A method of evaluating an herbal composition comprising: 5 a) exposing a biosystem toa batch of the herbal composition and collecting data on two or more markers, wherein one of the markers is a change in gene expression determined through the use of a nucleic acid microarray, produced by the steps comprising: 10 i) producing a cell banking system; ii) profiling the gene expression pattern of cells from the cell banking system before and after exposure to the herbal composition; 15 iii) selecting: as markeis those genes whose expression levels are changed by exposure to the herbal composition; b) comparing the colldited'niarker data with a standardized HBR Array for the same or a substantially same herbal composition as that of the batch herbal composition, wherein the 20 standardized HBR Array contains one of the markers data on gene expression.

13. A method of determining if an herbal composition meets a standard specification comprising: 25 a) exposing a biosystem to a batch of the herbal composition and collecting data on two or more markers, wherein one fthe markers is a change in gene expression determined through the use ofa nucleic'adid microarray, produced by the steps comprising: i) producing a cel banking system; 30 ii) profiling the gene expression pattern of cells the cell banking system before and after exposure to the herbal composition; WO 01/66803 PCT/USO1/07608 -103 iii) selecting as markers those genes whose expression levels are changed by exposure to the herbal composition; b) comparing the collected marker data with a standardized HBR Array for the same or a 5 substantially same herbal composition as that of the batch herbal composition, wherein the standardized HBR Array contains as one of the markers data on gene expression; and c) determining which herbal compositions have marker data that is similar to that of the standardized HBR Array within an acceptable level. 10

14. The method of claim 113, wherein said determining which herbal compositions have marker data which is similar to that of the standardized HBR Array within an acceptable level is determined quantitatively or'qualitatively.

15 15. The method of claims 13 or 14, wherein the standardized HBR Array includes an acceptable range of variation for each marker.

16. A method of adjusting the components of a herbal composition so that it meets a standard specification for the same or substantially the same herbal composition, comprising: 20 a) exposing a biosystem to a batch of the herbal composition and collecting data on two or more markers, wherein 'one of the markers is a change in gene expression determined through the use of a nucleic acid:microarray, produced by the steps comprising: 25 i) producing a cell banking system; ii) profiling the gene expression pattern of cells the cell banking system before and after exposure to the herbal composition; 30 iii) selecting as makers those genes whose expression levels are changed by exposure to 'the herbal composition; WO 01/66803 PCT/USO1/07608 -104 b) comparing the collected marker data with a standardized HBR Array for the same or a substantially same herbal composition as that of the batch herbal composition, wherein the standardized HBR Array contains as one of markers data on gene expression, and wherein the standardized HBR Array also includes an acceptable range of variation for each 5 marker; c) detennining whether the herbal composition has marker data that is within the acceptable level of variation for the standardized HBR Array; and 10 d) if the marker data is not within the acceptable level of variation for the standardized HBR Array, adjusting the components of the herbal composition.

17. The method of claim: 16, wherein steps (a) through (d) are repeated until the marker data of the herbal composition is within the acceptable level of variation of the standardized HBR 15 Array.

18. A method of changing the components of a herbal composition so that it meets a standard specification of anotherherbal composition, comprising: 20 a) exposing a biosystem to a batch of the herbal composition and collecting data on two or more markers, wherein oiie of the markers is a change in gene expression determined through the use of a nucleic acid microarray, produced by the steps comprising: i) producing a cellbdiking system; 25 ii) profiling.the gene expression pattern of cells the cell banking system before and after exposure to the herbal composition; iii) selecting as markers those'genes whose expression levels are changed by 30 exposure to the herbal composition; b) comparing the collected marker data with a standardized HBR Array for the other herbal compositions as that of the.batch herbal composition, wherein the standardized HBR Array WO 01/66803 PCT/USO1/07608 -105 contains as one of markers data on gene expression, and wherein the standardized HBR Array also includes an acceptable range of variation for each marker; c) determining whether the herbal composition has marker data that is within the acceptable 5 level of variation for the standardized HBR Array; and d) if the marker data is not within the acceptable level of variation for the standardized HBR Array, changing the components of the herbal composition. 10

19. The method of claim 18, wherein steps a) through d) are repeated until the marker data of the herbal composition is within the acceptable level of variation of the standardized HBR Array.

20. A method for predicting the biological activity of an herbal composition comprising: 15 a) exposing a biosystem to a batch of the herbal composition and measuring the differential responses of two or more markers, wherein one of the markers is a change in gene expression determined-through the use of a nucleic acid microarray, produced by the steps comprising; 20 i) producing a cell banking system; ii) profiling the gene expression pattern of cells the cell banking system before and after exposure to the herbal composition; 25 iii) selecting as markers those genes whose expression levels are changed by exposure to 'the herbal composition; wherein the set of differential response measurements constitute an Herbal BioResponse Array 30 (HBR Array) data set; WO 01/66803 PCT/USO1/07608 -106 b) comparing the HBR Array of the batch herbal composition to at least one previously obtained HBR Array of a characterized herbal composition, wherein the previously obtained HBR Array contains as one of the markers data on gene expression; and 5 c) predicting the biological activity of the batch herbal composition based on the HBR Array comparison made in step b).

21. A method of measuring the relatedness of a herbal composition to a characterized herbal composition comprising: 10 a) exposing a biosystem to a batch of the herbal composition and measuring the differential responses of two or iore markers, wherein one of the markers is a change in gene expression determined through the use of a nucleic acid microarray, produced by the steps comprising; 15 i) producing a cell banking system; ii) profiling the gene expression pattern of cells the cell banking system before and after exposure to the herbal composition; 20 iii) selecting as markers those genes whose expression levels are changed by exposure to the herbal composition; wherein the set of diffei-ential response measurements constitute an Herbal BioResponse Array 25 (HBR Array) data set; b) comparing the HBR Array of the batch herbal composition to at least one previously obtained HBR Array of a characterized herbal composition, wherein the previously obtained HBR Array contains as one of the markers data on gene expression; and 30 c) determining the relatedness of the herbal composition to the characterized herbal composition based on the HBR Array comparison made in step b). WO 01/66803 PCT/USO1/07608 -107

22. A method for predicting new therapeutic applications of an herbal comprising: a) exposing a biosystem to a batch of the herbal composition and measuring the differential responses of two or more markers, wherein one of the markers is a change 5 in gene expression determined through the use of a nucleic acid microarray, produced by the steps comprising; i) producing a cell banking system; 10 ii) profiling the gene expression pattern of cells the cell banking system before and after exposure to the herbal composition; iii) selecting as markers those genes whose expression levels are changed by exposure to the herbal composition; 15 wherein the set of diffei-eritial response measurements constitute an Herbal BioResponse Array (HBR Array) data set; b) predicting the new therapeutic applications based on the predicted biological activity of the markers in the HBR Array. 20

23. A method for determining the gene expression profile induced by individual chemical entities in an herbal composition comprising: 25 a) producing a cell banking system; b) profiling the gene expression pattern of cells from the cell banking system before and after exposure to the herbal composition; 30 WO 01/66803 PCT/USO1/07608 -108 c) selecting as markers those genes whose expression levels are changed by exposure to the herbal composition and placing into an HBR Array; 5 d) comparing the HBR Array generated in step (c) with a standardized HBR Array for a similar or modified herbal composition; e) determining the relative amounts of the individual chemical entities of the herbal 10 composition; and f) comparing the'ariouni of the individual chemical entities to the result of step (b) to identify those genes\*hose expression levels change as the amount of the individual chemical entity in the herbal composition changes. 15

24. A method for determining the gene expression profile induced by individual chemical entities in a complex mixture without extracting the chemicals from the complex mixture such as an herbal composition, comprising 20 a) producing a cell banking system; b) profiling the gene eipixeion pattern of cells from the cell banking system before and 25 after exposure to th6 herbal composition; c) selecting as markers those genes whose expression levels are changed by exposure to the herbal composition; 30 d) comparing the collected marker data with a standardized HBR Array for a substantially same or modified herbal composition; WO 01/66803 PCT/USO1/07608 -109 e) characterizing'the chemical components of the said herbal compositions; 5 f) comparing.the -identified chemical compositions to identify the differential levels of individual chemical components in herbal compositions; g) correlating the differential chemical component amounts with the differential bioresponses in the HBR Array to identify the characteristic bioresponses for each 10 chemical entity.