WO2004042034A2

WO2004042034A2 - Method for the linear amplification of complementary dna

Info

Publication number: WO2004042034A2
Application number: PCT/US2003/035350
Authority: WO
Inventors: Era L. Pogosova-Agadjanyan; Derek L. Stirewalt
Original assignee: Fred Hutchinson Cancer Center
Current assignee: Fred Hutchinson Cancer Center
Priority date: 2002-11-04
Filing date: 2003-11-04
Publication date: 2004-05-21
Anticipated expiration: 2005-05-04
Also published as: AU2003291325A8; WO2004042034A3; AU2003291325A1

Abstract

In one aspect, the present invention provides methods for linearly amplifying cDNA from an RNA molecule. The methods comprise the steps of: (a) synthesizing an antisense cDNA molecule using an RNA molecule as a template; (b) synthesizing an extended antisense cDNA molecule using a template-switching oligonucleotide as a template; (c) synthesizing multiple copies of sense cDNA molecules using the extended antisense cDNA strand as a template; and (d) generating double-stranded cDNA molecules using the multiple copies of sense cDNA molecules as templates. Another aspect of the invention provides methods for linearly amplifying RNA, comprising the steps o£ (a) using an RNA molecule as a template for synthesizing an antisense cDNA molecule; (b) using a template-switching oligonucleotide as a template for synthesizing an extended antisense cDNA molecule; (c) using the extended antisense cDNA molecule as a template for synthesizing multiple copies of sense cDNA molecules; (d) using the sense cDNA molecules as templates for generating double-stranded cDNA molecules; and (e) using the double-stranded cDNA molecules as templates for synthesizing RNA molecules.

Description

METHOD FOR THE LINEAR AMPLIFICATION OF COMPLEMENTARY DNA

FIELD OF THE INVENTION The present invention relates to methods for linearly amplifying complementary DNA (cDNA) from small quantities of RNA.

BACKGROUND OF THE INVENTION The sequencing of the human genome has provided researchers with the DNA blueprint that controls the biological process in humans (Lander et al. (2001) Nature 409:860-921 ; Ventner et al. (2001) Science 291 :1304-51), and has also lead to the development of large scale microarray-based gene expression profiling techniques to examine the expression patterns of the numerous genes making up this DNA blueprint. Standard microarray expression profiling techniques are relatively easy to perform, allowing investigators to examine the expression of thousands of genes within days.

Two major microarray platforms exist: (1) oligonucleotide microarrays (see, e.g., Lockhart et al. (1996) Nat. Biotechnol: 14:1675-80; McGall et al. (1996) Proc. Natl. Acad. Sci. U.S.A. 93:13555-60); and (2) cDNA microarrays (see, e.g., Schena et al. (1995) Science 270:467-70). There are advantages and disadvantages to either platform that have recently been reviewed in more detail (Ramaswamy & Golub (2002) J. Clin. Oncol. 20: 1932-41), but both methods require between 5 and 40 micrograms of high quality RNA (Duggan et al. (1999) Nat. Genet. 21 :10-14; Lockhart et al. (1996) Nat. Biotechnol. 14:1675-80; McGall et al. (1996) Proc. Natl. Acad. Sci. U.S.A. 93:13555-60; Schena et al. (1995) Science 270:467-70; Ramaswamy & Golub (2002) J. Clin. Oncol. 20: 1932-1941).

The limited availability of cells, and thus RΝA, often prohibit the use of microarrays and severely restrict the ability to perform the replicate experiments necessary to ensure valid results (Lee et al. (2000) Proc. Natl. Acad. Sci. U.S.A. 97:9823- 9). The standard Affymetrix protocol for producing RΝA that can be used for microarray analyses involves reverse transcribing the starting RΝA to generate double-stranded complementary DΝA (cDΝA), followed by an in vitro transcription reaction using the cDΝA as a template to yield labeled complementary RΝA (cRΝA, antisense RΝA). Labeled RΝA produced using the standard Affymetrix protocol yields reliable and reproducible microarray results, whenever the amount of starting RΝA is at least about 1 microgram. Several protocols have been developed to produce labeled RNA in situations where the amount of starting RNA is less than about 1 microgram. For an amplification technique to be reliable, the labeled RNA produced using an amplification protocol (i.e., RNA from amplified cDNA) must provide a similar expression profile in microarray analyses as labeled RNA produced using the standard protocol (i.e., RNA from unamplified cDNA).

One type of amplification protocol, often referred to as in vitro transcription amplification protocol (IVT amplification protocol, e.g., the NIH protocol or the Lemischka protocol), makes double-stranded cDNA from messenger RNA (mRNA) using reverse transcriptase, followed by synthesis of a sense cDNA with mixture of E. coli DNA polymerase I and DNA ligase. The double-stranded cDNA is used to generate labeled cRNA using, for example, methods described by Phillips & Eberwine (1996), Van Gelder et al. (1990), and Baugh et al. (2001) (Phillips & Eberwine (1996) Methods 10:283-8; Van Gelder et al. (1990) Proc. Natl. Acad. Sci. U.S.A. 87:1663-7; Baugh et al. (2001) Nucleic Acids. Res. 29:E29). IVT amplification protocols have been shown to provide adequate amplification from as little as 2 nanograms of starting RNA. Although the IVT amplification protocols have been shown to be reproducible, it has been shown that multiple rounds of IVT amplification results in sequence-dependent biases for the 3- prime end of RNAs (Baugh et al. (2001) Nucleic Acids. Res. 29:E29). IVT amplification protocols are believed to be linear, which may account for the high reproducibility found with this method (Baugh et al. (2001) Nucleic Acids. Res. 29:E29). IVT amplification protocols are probably the method that is most frequently used by investigators working with small amounts of starting RNA.

Another type of amplification protocol uses a template-switching primer at the 5- prime end of the sense cDNA strand (SMART PCR cDNA synthesis kit, Clontech, Palo Alto, CA), followed by polymerase chain reactions (Wang et al. (2000) Nat. Biotechnol. 18:457-9). Unlike IVT amplification protocols, the SMART PCR technique produces exponential amplification of the starting material. The SMART PCR technique can reproducibly amplify small amounts of starting RNA (starting with as little as 50 nanograms of total RNA)(Puskas et al. (2002) Biotechnigues 32:1330-4) to generate labeled RNA for the performance of microarray studies. The SMART PCR technique eliminates some of the 3 -prime amplification biases of the IVT amplification protocols, and provides high rates of amplification. However, when INT amplification protocols and SMART PCR amplification protocols were compared to the standard Affymetrix protocol using 2 micrograms of poly(A)⁺ starting RNA, the SMART PCR protocol had a lower reproducibility and higher variation in transcript levels (Puskas et al. (2002) Biotechnigues 32:1330-4). Thus, there is a need in the art for amplification protocols that will permit the use of even lower amounts of starting RNA. In particular, there is a need for amplification protocols that combine the advantages of higher reproducibility and less variation of transcript levels of the linear amplification protocols (such as IVT amplification protocols) with the advantages of greater amplification and less 3 -prime amplification bias of exponential amplification protocols (such as SMART PCR protocols).

SUMMARY OF THE INVENTION In one aspect, the present invention provides methods for linearly amplifying cDNA from an RNA molecule. The methods comprise the steps of: (a) synthesizing an antisense cDNA molecule using an RNA molecule as a template; (b) synthesizing an extended antisense cDNA molecule using a template-switching oligonucleotide as a template; (c) synthesizing multiple copies of sense cDNA molecules using the extended antisense cDNA strand as a template; and (d) generating double-stranded cDNA molecules using the multiple copies of sense cDNA molecules as templates. In some embodiments, the methods comprise the steps of: (a) incubating an RNA molecule with a first primer under suitable conditions for synthesizing an antisense cDNA molecule, wherein the first primer comprises a sequence complementary to a region of the RNA molecule and a first defined nucleic acid sequence; (b) incubating the antisense cDNA molecule with a second primer under suitable conditions for synthesizing an extended antisense cDNA molecule, wherein the second primer comprises a template-switching oligonucleotide and a second defined nucleic acid sequence, and wherein the extended antisense cDNA molecule comprises a sequence complementary to the second defined nucleic acid sequence; (c) incubating the extended antisense cDNA molecule with a third primer under suitable conditions for synthesizing multiple copies of sense cDNA molecules, wherein the third primer comprises the second defined nucleic acid sequence; and (d) incubating the sense cDNA molecules with a fourth primer under suitable conditions for synthesizing double-stranded cDNA, wherein the fourth primer comprises the first defined nucleic acid sequence. The amplified cDNA may be used, for example, as a template for synthesizing labeled RNA for use in microarray assays. Another aspect of the invention provides methods for linearly amplifying RNA, comprising the steps of: (a) using an RNA molecule as a template for synthesizing an antisense cDNA molecule; (b) using a template-switching oligonucleotide as a template for synthesizing an extended antisense cDNA molecule; (c) using the extended antisense cDNA molecule as a template for synthesizing multiple copies of sense cDNA molecules;

(d) using the multiple copies of sense cDNA molecules as templates for generating double-stranded cDNA molecules; and (e) using the double-stranded cDNA molecules as templates for synthesizing RNA. In some embodiments, the methods comprise the steps of: (a) incubating an RNA molecule with a first primer under suitable conditions for synthesizing an antisense cDNA molecule, wherein the first primer comprises a sequence complementary to a region of the RNA molecule and a first defined nucleic acid sequence; (b) incubating the antisense cDNA molecule with a second primer under suitable conditions for synthesizing an extended antisense cDNA molecule, wherein the second primer comprises a template-switching oligonucleotide and a second defined nucleic acid sequence, wherein at least one of the first defined nucleic acid sequence and the second defined nucleic acid sequence comprises a promoter sequence for an RNA polymerase, and wherein the extended antisense cDNA molecule comprises a sequence complementary to the second defined nucleic acid sequence; (c) incubating the extended antisense cDNA molecule with a third primer and under suitable conditions for synthesizing multiple copies of sense cDNA molecules, wherein the third primer comprises the second defined nucleic acid sequence; (d) incubating the sense cDNA molecules with a fourth primer under suitable conditions for synthesizing double-stranded cDNA, wherein the fourth primer comprises the first defined nucleic acid sequence; and

(e) incubating the double-stranded cDNA with the RNA polymerase under suitable conditions for synthesizing RNA. In some embodiments, the methods may be used to provide labeled RNA for use in microarray assays.

BRIEF DESCRIPTION OF THE DRAWINGS The foregoing aspects and many of the attendant advantages of this invention will become more readily appreciated as the same become better understood by reference to the following detailed description, when taken in conjunction with the accompanying drawings, wherein:

FIGURES 1 A-D show the steps in a single-stranded linear amplification protocol (SLAP) according to the present invention, as described in EXAMPLE 1. FIGURE 1A shows that reverse transcription of total RNA generates antisense cDNA with the incorporation of T7 promoter and SMART II A sequences; FIGURE IB shows that PCR IIA primer anneals to the antisense cDNA, and linear amplification using DNA polymerase produces multiple copies of sense DNA strands; FIGURE 1C shows that DET7F anneals to the multiple copies of sense DNA strands, and DNA polymerase makes double-strand cDNA; FIGURE ID shows biotin-labeling, fragmentation, and hybridization to microarrays.

FIGURES 2A-C show the overlap of false negative detection call changes compared to the Affymetrix protocol for SLAP (FIGURE 2A), the Lemischka protocol (FIGURE 2B), and the NIH protocol (FIGURE 2C), using different amounts of starting RNA, as described in EXAMPLE 2. FIGURE 2A: solid line surrounds positives obtained using the Affymetrix protocol at 10 micrograms of starting RNA (N = 2989); thick dashed line surrounds false negatives obtained using SLAP at 1.00 microgram of starting RNA (N = 308); thin dashed line surrounds false negatives obtained using SLAP at 0.10 micrograms of starting RNA (N ^•= 200); stippled line surrounds false negatives obtained using SLAP at 0.01 micrograms of starting RNA (N = 229). FIGURE 2B: solid line surrounds positives obtained using the Affymetrix protocol at 10 micrograms of starting RNA (N = 2989); thick dashed line surrounds false negatives obtained using the Lemischka protocol at 1.00 microgram of starting RNA (N = 185); thin dashed line surrounds false negatives obtained using the Lemischka protocol at 0.10 micrograms of starting RNA (N = 907); stippled line surrounds false negatives obtained using the Lemischka protocol at 0.01 micrograms of starting RNA (N - 1649). FIGURE 2C: solid line surrounds positives obtained using the Affymetrix protocol at 10 micrograms of starting RNA (N = 2989); thick dashed line surrounds false negatives obtained using the NIH protocol at 1.00 microgram of starting RNA (N = 236); thin dashed line surrounds false negatives obtained using the NIH protocol at 0.10 micrograms of starting RNA (N = 305); stippled line surrounds false negatives obtained using the NIH protocol at 0.01 micrograms of starting RNA (N = 1102).

FIGURES 3A-C show the overlap of false negative detection call changes compared to the Affymetrix protocol for SLAP, the Lemischka protocol, and the NIH protocol using 1 microgram of starting RNA (FIGURE 3 A), 100 ng of starting RNA (FIGURE 3B), and 10 ng of starting RNA (FIGURE 3C), as described in EXAMPLE 2. FIGURE 3 A: solid line surrounds positives obtained using the Affymetrix protocol at 10 micrograms of starting RNA (N = 2989); thick dashed line surrounds false negatives obtained using SLAP at 1.00 microgram of starting RNA (N = 308); thin dashed line surrounds false negatives obtained using the Lemischka protocol at 1 microgram of starting RNA (N = 185); stippled line surrounds false negatives obtained using the NTH protocol at 1 microgram of starting RNA (N = 236). FIGURE 3B: solid line surrounds positives obtained using the Affymetrix protocol at 10 micrograms of starting RNA (N = 2989); thick dashed line surrounds false negatives obtained using SLAP at 0.10 micrograms of starting RNA (N = 200); thin dashed line surrounds false negatives obtained using the Lemischka protocol at 0.10 micrograms of starting RNA (N = 907); stippled line surrounds false negatives obtained using the NIH protocol at 0.10 microgram of starting RNA (N = 305). FIGURE 3C: solid line surrounds positives obtained using the Affymetrix protocol at 10 micrograms of starting RNA (N = 2989); thick dashed line surrounds false negatives obtained using SLAP at 0.01 micrograms of starting RNA (N = 229); thin dashed line surrounds false negatives obtained using the Lemischka protocol at 0.01 micrograms of starting RNA (N = 1649); stippled line surrounds false negatives obtained using the NIH protocol at 0.01 micrograms of starting RNA (N = 1102).

FIGURES 4A, B show the overlap of false positive detection call changes compared to the Affymetrix protocol for SLAP (FIGURE 4A) and the NIH protocol (FIGURE 4B), using different amounts of starting RNA, as described in EXAMPLE 2. FIGURE 4A: solid line surrounds positives obtained using the Affymetrix protocol at 10 micrograms of starting RNA (N = 3785); thick dashed line surrounds false positives obtained using SLAP at 1.00 microgram of starting RNA (N = 33); thin dashed line surrounds false positives obtained using SLAP at 0.10 micrograms of starting RNA (N = 23); stippled line surrounds false positives obtained using SLAP at 0.01 micrograms of starting RNA (N = 31). FIGURE 4B: solid line surrounds positives obtained using the Affymetrix protocol at 10 micrograms of starting RNA (N = 3785); thick dashed line surrounds false positives obtained using the NIH protocol at 1.00 microgram of starting RNA (N = 37); thin dashed line surrounds false positives obtained using the NIH protocol at 0.10 micrograms of starting RNA (N = 88); stippled line surrounds false positives obtained using the NIH protocol at 0.01 micrograms of starting RNA (N = 20). DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT Unless specifically defined herein, all terms used herein have the same meaning as they would to one skilled in the art of the present invention.

In one aspect, the present invention provides methods for linearly amplifying cDNA from an RNA molecule. The methods comprise the steps of: (a) synthesizing an antisense cDNA molecule using an RNA molecule as a template; (b) synthesizing an extended antisense cDNA molecule using a template-switching oligonucleotide as a template; (c) synthesizing multiple copies of sense cDNA molecules using the extended antisense cDNA strand as a template; and (d) generating double-stranded cDNA molecules using the multiple copies of sense cDNA molecules as templates.

In some embodiments, the methods comprise the steps of: (a) incubating an RNA molecule with a first primer under suitable conditions for synthesizing an antisense cDNA molecule, wherein the first primer comprises a sequence complementary to a region of the RNA molecule and a first defined nucleic acid sequence; (b) incubating the antisense cDNA molecule with a second primer under suitable conditions for synthesizing an extended antisense cDNA molecule, wherein the second primer comprises a template-switching oligonucleotide and a second defined nucleic acid sequence, and wherein the extended antisense cDNA molecule comprises a sequence complementary to the second defined nucleic acid sequence; (c) incubating the extended antisense cDNA molecule with a third primer under suitable conditions for synthesizing multiple copies of sense cDNA molecules, wherein the third primer comprises the second defined nucleic acid sequence; and (d) incubating the sense cDNA molecules with a fourth primer under suitable conditions for synthesizing double-stranded cDNA, wherein the fourth primer comprises the first defined nucleic acid sequence. In the first step of the methods of this aspect of the invention, an RNA molecule is incubated with a first primer and an enzyme possessing reverse transcriptase under conditions suitable to permit the template-dependent extension of the primer to synthesize an antisense cDNA molecule (see FIGURE 1A). The methods of the invention are not limited to any particular method of RNA preparation. A large number of well-known methods for isolating and purifying RNA are suitable for use in the methods of the invention (see, e.g., Sambrook et al. (1989) Molecular Cloning - A Laboratory Manual (2d ed.) v.l, ch. 3 (Cold Spring Harbor Laboratory, Cold Spring Harbor, NY). Most eukaryotic messenger RNAs (mRNAs) have 3' poly(A) tails. It is often desirable to isolate mRNA from RNA samples. Both total RNAs or poly(A)-containing RNAs are suitable for use in the methods of the invention.

The antisense cDNA molecule generated in this first step is a single-stranded DNA molecule that is complementary the RNA molecule. The antisense cDNA molecule can be complementary to the entirety of the RNA molecule, or to a portion thereof. For example, the antisense cDNA molecule may be complementary to a portion of the RNA molecule extending 5-prime from the poly(A) tail of the RNA. Alternatively, the antisense cDNA molecule may be complementary to an internal portion of the RNA molecule. Any enzyme possessing reverse transcriptase activity may be used to synthesize the antisense cDNA molecule. Examples of enzymes possessing reverse transcriptase activity are the reverse transcriptase from Moloney murine leukemia virus (MMLV-RT), avian myeloblastosis virus (AMV-RT), bovine leukemia virus (BLV-RT), Rous sarcoma virus (RSV), and human immunodeficiency virus (HIV-RT). In some embodiments, the enzyme possessing reverse transcriptase used is a MMLV-RT with a single amino acid substitution that eliminates RNase H activity while leaving the polymerase activity intact (e.g., POWERSCRIPT RT, BD Biosciences Clontech Laboratories, Inc., Palo Alto, CA). This allows the reverse transcriptase to synthesize a higher percentage of full-length antisense cDNA molecules. The reverse transcriptase molecule may be a thermostable enzyme so that the first strand synthesis reaction can be conducted at as high a temperature as possible.

The first primer can be a single-stranded oligonucleotide or a double-stranded oligonucleotide with a single-stranded portion. In all cases, the first primer comprises a single-stranded priming region that binds to a complementary region of the RNA. The priming region may bind to any region of the RNA molecule, including the poly(A) tail. A suitable oligonucleotide primer is typically in the range of about ten to about 60 bases in length. An oligonucleotide primer can be DNA, RNA, chimeric mixtures, or derivatives or modified versions thereof, so long as it is still capable of priming the desired reaction. The oligonucleotide primer can be modified at the base moiety, sugar moiety, or phosphate backbone, and may include other appending groups or labels, so long as it is still capable of priming the desired reaction.

In some embodiments, the priming region comprises an poly(dT) sequence complementary to the poly(A) tail of mRNAs. The poly(dT) sequence typically consists of from five to 25 nucleic acid residues, such as from 15-25 nucleic acid residues. In some embodiments, the synthesis of the antisense cDNA molecules is primed using a first primer mixture comprising a multiplicity of first primers, wherein each of the first primers includes a random nucleic acid sequence. The random nucleic acid sequence comprises a random sequence of nucleic acid residues, and typically consists of from four to 20 nucleic acid residues.

In some embodiments, the synthesis of the first DNA molecules is primed using a mixture of primers, wherein the mixture includes poly(dT) primers that each comprise a poly(dT) sequence. In some embodiments, the poly(dT) primers are used with a first primer mixture comprising a multiplicity of first primers, wherein each of the first primers comprises a random sequence.

In some embodiments, the first primer additionally comprises a first defined nucleic acid sequence in addition to the sequence that is required to prime the synthesis of the antisense cDNA molecule. The first defined nucleic acid sequence is typically located 5-prime to the priming region (e.g., the random sequence or the poly(dT) sequence) of the primer. Typically, all the first primers in a first primer mixture include the same first defined nucleic acid sequence so that every antisense cDNA molecule generated has an identical first defined nucleic acid sequence.

The first defined nucleic acid sequence of the first primers comprises a selected sequence of nucleic acid residues, and may include the minimum amount or more of nucleic acid residues of an RNA polymerase promoter region to allow an RNA polymerase to bind and direct the synthesis of RNA molecules. The first defined nucleic acid sequence is incorporated into the sequence of the antisense DNA molecules. Thus, if the first defined nucleic acid sequence includes an RNA polymerase promoter, the RNA polymerase promoter can subsequently be used for the synthesis of RNA molecules that are complementary in sequence to the antisense cDNA molecules. Any RNA polymerase promoter sequence can be included in the first defined nucleic acid sequence of the first primers. Representative examples of useful RNA polymerase promoters include a T7 RNA polymerase promoter and an SP6 RNA polymerase promoter. A representative first defined nucleic acid sequence of a first primer that includes a T7 RNA polymerase promoter sequence is 5' - GGC CAG TGA ATT GTA ATA CGA CTC ACT ATA GGG AGG CGG - 3' (SEQ ID NO: 1), which is the defined nucleic acid sequence of representative first primer, T7(dT)₂₄ 5' - GGC CAG TGA ATT GTA ATA CGA CTC ACT ATA GGG AGG CGG TTT TTT TTT TTT TTT TTT TTT TTT - 3' (SEQ ID NO:2).

The conditions suitable for synthesizing an antisense cDNA molecule by reverse transcription of an RNA molecule include the presence at appropriate temperatures and for sufficient lengths of time of effective amounts of a reverse transcriptase and effective amounts other reagents, such as buffers, dithiothreitol, RNase inhibitors, and a deoxynucleotide triphosphate mixture. Typically, the RNA is denatured at a first temperature (e.g., at around 65°C for about 2 minutes), followed by annealing of the primer to the RNA at a second temperature (e.g. , at around 42°C for about 2 minutes), followed by synthesis of antisense cDNA at a third temperature (e.g., at around 42°C for about 1 hour). Exemplary conditions for the synthesis of an antisense cDNA molecule from an RNA molecule are provided in EXAMPLE 1.

In the second step of the methods of this aspect of the invention, the antisense cDNA molecule is incubated with a second primer comprising a template-switching oligonucleotide and a second defined nucleic acid sequence and an enzyme possessing reverse transcriptase activity and terminal transferase activity under conditions suitable to synthesize a template-dependent extension to the antisense cDNA that is complementary to the second defined nucleic acid sequence (see FIGURE 1A). In the course of antisense cDNA synthesis, the reverse transcriptase stops at the 5-prime end of the mRNA template, which typically includes a 7-methylguanosine CAP structure present on the 5- prime ends of all eukaryotic mRNAs. The terminal transferase activity of the enzyme then adds a few additional nucleotides, typically deoxycytidine, to the 3 -prime end of the newly synthesized antisense cDNA strand. The second primer comprises a second defined nucleic acid sequence at its 5-prime end and, at the 3-prime end, a template- switching oligonucleotide comprising an oligo(rG) sequence. The oligo(rG) sequence of the template-switching oligonucleotide base-pairs with the deoxycytidines at the 3-prime end of the antisense cDNA to create an extended template. The reverse transcriptase then switches templates and continues synthesis of the antisense cDNA strand to incorporate a sequence that is complementary to the second defined nucleic acid sequence. The second defined nucleic acid sequence comprises a pre-selected arbitrary sequence suitable for serving as a primer in the next step of the methods of the invention, as described below. In some embodiments, the second defined nucleic acid sequence comprises an RNA polymerase promoter, as described above. An exemplary template-switching oligonucleotide is the CAPswitch SMART II A oligonucleotide 5' - AAG CAG TGG TAT CAA CGC AGA GTA CGC GGG - 3' (SEQ ID NO:3) (provided in the SMART PCR kit, Clontech, Palo Alto, CA). Template-switching oligonucleotides are further described in U.S. Patent No. 5,962,271 and U.S. Patent No. 5,962,272. In some embodiments, the second primer comprising the template-switching oligonucleotide may be added at the time of performing the first step of the methods of the invention.

The conditions suitable for synthesizing a template-dependent extension of the antisense cDNA molecule by reverse transcription of an RNA molecule include the presence at appropriate temperatures and for sufficient lengths of time of effective amounts of a reverse transcriptase and other reagents, such as buffers, dithiothreitol, RNase inhibitors, and a deoxynucleotide triphosphate mixture, as described above for the synthesis of antisense cDNA molecules. Exemplary conditions for the synthesis of an extended antisense cDNA strand from an RNA molecule are provided in EXAMPLE 1.

In the third step of the methods of the invention, the extended antisense cDNA molecule is incubated with a third prime under conditions suitable to synthesize multiple copies of sense cDNA molecules and thereby result in the linear amplification of the informational content present in the original RNA sequence (see FIGURE IB). The third primer comprises the second defined nucleic acid sequence and is, therefore, complementary to the 3-prime end of the antisense cDNA molecule generated in the second step. A DNA polymerase is used to synthesize the sense cDNA molecules. In some embodiments, multiple copies of sense cDNA are synthesized using the polymerase chain reaction (PCR) and a thermostable DNA polymerase.

The conditions suitable for synthesizing multiple copies of sense cDNA molecules using antisense cDNA molecules as templates include the presence at appropriate temperatures and for sufficient lengths of time of effective amounts of a DNA polymerase and other reagents, such as buffers and a deoxynucleotide triphosphate mixture. Typically, the antisense cDNA molecules are denatured at a first temperature (e.g., at around 95°C for about 1 minute in the first cycle, at around 95°C for about 15 seconds in subsequent cycles), followed by primer annealing at a second temperature (e.g., at around 60°C for about 15 seconds), and elongation of sense cDNA molecules at a third temperature (e.g., at around 68°C for about 6 minutes). Conditions for amplification of nucleic acid sequences using PCR are well-known in the art. Exemplary conditions for the linear amplification of sense cDNA strand from an antisense cDNA molecule using PCR are provided in EXAMPLE 1.

In the fourth step of this aspect of the invention, the sense cDNA molecule is incubated with a fourth primer and a DNA polymerase under conditions suitable to synthesize one molecule of antisense cDNA for each sense cDNA molecule in order to generate double-stranded cDNA (see FIGURE 1C). The fourth primer comprises the first defined nucleic acid sequence and is, therefore, complementary to the 3-prime end of the sense cDNA generated in the third step.

The conditions suitable for generating double-stranded cDNA molecules include the presence at appropriate temperatures and for sufficient lengths of time of effective amounts of a DNA polymerase and other reagents, such as buffers and a deoxynucleotide triphosphate mixture. Typically, the sense cDNA molecules are denatured at a first temperature (e.g., at around 95°C for about 1-2 minutes), followed by primer annealing at a second temperature (e.g., at around 60°C for about 15 seconds), and elongation of antisense cDNA molecules at a third temperature (e.g., at around 68°C for about 10 minutes). Conditions for the synthesis of an antisense cDNA strand using a sense cDNA strand as a template are well-known in the art. Exemplary conditions for generating double-stranded cDNA from sense cDNA are provided in EXAMPLE 1.

In some embodiments, the methods further comprise using the double-stranded cDNA molecules as templates for synthesizing RNA molecules, which may be used, for example, for microarray expression analyses.

A second aspect of the invention further provides methods for synthesizing RNA from the double-stranded cDNA. Thus, a second aspect of the invention provides methods for amplifying RNA, comprising the steps of: (a) using an RNA molecule as a template for synthesizing an antisense cDNA molecule; (b) using a template-switching oligonucleotide as a template for synthesizing an extended antisense cDNA molecule; (c) using the extended antisense cDNA molecule as a template for synthesizing multiple copies of sense cDNA molecules; (d) using the multiple copies of sense cDNA molecules as templates for generating double-stranded cDNA molecules; and (e) using the double- stranded cDNA molecules as templates for synthesizing RNA molecules.

The first four steps of the methods of this aspect of the invention are essentially similar to the four steps of the methods of the first aspect of the invention. Thus, in the first step of this aspect of the invention, an RNA molecule is incubated with a first primer under suitable conditions for synthesizing an antisense cDNA molecule, as described above. The first primer typically comprises a sequence complementary to a region of the RNA and a first defined nucleic acid sequence, for example, an RNA polymerase promoter. In the second step, the antisense cDNA molecule is incubated with a second primer under suitable conditions for synthesizing an extended antisense cDNA molecule, as described above. Typically, the second primer comprises a template-switching oligonucleotide and a second defined nucleic acid sequence. In this aspect of the invention, at least one of the first defined nucleic acid sequence and the second defined nucleic acid sequence comprises an RNA polymerase promoter, such as the T7 promoter described above. In the third step, the extended antisense cDNA molecule is incubated with a third primer under suitable conditions for synthesizing multiple copies of sense cDNA molecules, wherein the third primer comprises the second defined nucleic acid sequence, as described above. In the fourth step, the sense cDNA strands are incubated with a fourth primer under suitable conditions for generating double-stranded cDNA, wherein the fourth primer comprises the first defined nucleic acid sequence, as described above. Exemplary primer sequences and conditions for performing the first four steps of this aspect of the methods of the invention are as described above for the first aspect of the invention.

In the fifth step of the second aspect of the invention, the double-strand cDNA molecules are incubated with an RNA polymerase under suitable conditions for synthesizing RNA. If the first defined nucleic acid sequence comprises an RNA polymerase promoter, antisense RNA may be synthesized using the sense cDNA strand as a template (see, e.g., EXAMPLE 1). If the second defined nucleic acid sequence comprises an RNA polymerase promoter, sense RNA may be synthesized using the antisense cDNA strand as a template. Successive rounds of transcription from either cDNA template results in amplified RNA.

The conditions suitable for synthesizing RNA using double-stranded cDNA molecules as templates include the presence at an appropriate temperature (e.g., around 37°C) and for a sufficient length of time (e.g., for about 3-4 hours) of effective amounts of an RNA polymerase and other reagents, such as buffers, RNase inhibitors, dithiothreitol, and ribonucleotides (e.g., biotin-labeled ribonucleotides). Methods for in vitro transcription are well known to those of skill in the art (see, e.g., Van Gelder et al. (1990) Proc. Natl. Acad. Sci. U.S.A. 87:1663-1667; Eberwine et al. (1992) Proc. Natl. Acad. Sci. U.S.A. 89:3010-14). In some embodiments, the in vitro transcription reaction may be coupled with labeling of the resulting RNA, for example with biotin. An exemplary method for in vitro transcription and labeling of RNA is provided in EXAMPLE 1. The RNA provided by the methods of this aspect of the invention may be used, for example, as a hybridization target in microarray assays. Before hybridization, the RNA may be fragmented. Methods for using RNA for hybridizing microarray assays are well-known in the art. An exemplary methods is provided in EXAMPLE 2. The methods of the invention provide efficient amplification of small amounts of starting RNA, as described in EXAMPLE 2. For example, the methods of the invention produce more than 15 micrograms of RNA (the recommended amount for microarray assays) from as little as 1 nanogram of starting RNA, as shown in Table 1. In comparison, other amplification protocols (e.g., the Lemischka and NIH protocols) require at least 0.05 micrograms of starting RNA to produce 15 micrograms of RNA, as shown in Table 1. Moreover, the methods of the invention provide a 5-prime to 3-prime signal conservation that is from about 3 -fold to about 30-fold higher than that obtained using other amplification protocols with less than 0.05 micrograms of starting material, as shown in Table 2.

The RNA produced according to the methods of the invention provide highly reproducible microarray assay results, compared to the standard Affymetrix protocol and the IVT amplification protocols (Lemischka and NIH), as described in EXAMPLE 2. With at least 1 microgram of starting RNA, the standard Affymetrix protocol exhibits the highest reproducibility (Kappa range = 0.826-0.908), but the reproducibility rapidly decreases with lower amounts of starting RNA (see Table 3). The RNA produced according to the methods of the invention yields the highest reproducibility with amounts starting RNA of less than 0.05 micrograms (Kappa range = 0.783-0.860).

In addition, the RNA produced according to the methods of the invention provide highly reliable microarray data, compared to the standard Affymetrix protocol and INT amplification protocols (Lemischka and ΝIH), as described in EXAMPLE 2. With at least 1 microgram of starting RΝA, the standard Affymetrix protocol exhibits the highest reliability, as measured by differences in log2 (signals) and Kappa measures of agreement

(Kappa range = 0.858-0.876), but the RΝA produced according to the methods of the invention yields the overall highest reliability with less starting RΝA (Kappa range = 0.744-0.821) (see Table 4). Moreover, the methods of the invention produced the fewest false negative calls with nanogram quantities of starting RNA, as shown in Table 5 and FIGURES 2-4.

The invention provides methods for amplifying RNA that are extremely easy to perform. For example, the methods of the invention can be performed in only two days, compared to the IVT amplification protocols, which require an average of three days, as shown in EXAMPLES 1 and 2. Also, the methods of the invention require fewer nucleic acid clean-up steps than the NIH and Lemischka IVT amplification protocols, as shown in EXAMPLES 1 and 2. The following examples merely illustrate the best mode now contemplated for practicing the invention, but should not be construed to limit the invention.

EXAMPLE 1 This Example describes an exemplary method of the invention for amplifying cDNA (single-stranded linear amplification protocol, SLAP). A. RNA EXTRACTION AND INTEGRITY ASSAY

Total RNA was isolated from the same stock of Bio-Rad Amplicheck Positive control cells (Bio-Rad Laboratories Diagnostics Group, Irvine, CA) using the standard RNeasy Mini RNA extraction protocol (Qiagen, Valencia, CA). RNA was analyzed on HP 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA) to assess the integrity of total RNA. The integrity of RNA was excellent based on the electropherograms, low baseline noise, and 28S/18S rRNA ratio. RNA concentration was determined by OD₂₆o reading on Eppendorf BioPhotometer (Brinkmann Instruments, Inc., Westbury, NY).

B. SINGLE-STRANDED LINEAR AMPLIFICATION PROTOCOL (SLAP) 1. Reverse Transcription to Make Antisense cDNA (FIGURE I A) cDNA synthesis was performed in Techne Progene hot-lid thermocycler (Jepson

Bolton, Watford Herts, United Kingdom). The POWERSCRIPT RT kit was used for cDNA synthesis (BD Biosciences Clontech Laboratories, Inc. Palo Alto, CA).

Total RNA was mixed with T (dT)₂₄ oligonucleotide (84 pmol, HPLC-purified, Operon, Alameda, CA) (SEQ ID NO:2) and SMART II A oligonucleotide (84 pmol, BD Biosciences Clontech Laboratories, Inc.) (SEQ ID NO:3). The RNA mixture was denatured for 2 minutes at 65°C in a total volume of 64 microliters. The reaction mix was then incubated for 2 minutes at 42°C, allowing the T (dT) ₄ oligonucleotide to anneal to the poly(A) region for reverse transcription. After 2 minutes, 5x first-strand buffer, 0.2 micromol dithiothreitol (DTT), 100 units RNase inhibitor (Brinkmann Instruments, Inc.), lOO nmol deoxynucleotide triphosphate mixture (dNTP) (Invitrogen, Carlsbad, CA), 500 units POWERSCRIPT Reverse Transcriptase (BD Biosciences Clontech), and molecular biology grade (MBG) water (Brinkmann Instruments, Inc.) were added to the mixture (total volume 106 microliters). Antisense cDNA synthesis was performed at 42°C for one hour, allowing the incorporation of SMART II A oligonucleotide at the 3' end of the transcript. The reaction was stopped with 1 micromol EDTA (pH 8). Primer sequences are: T₇(dT)₂₄ oligonucleotide 5' - GGC CAG TGA ATT GTA ATA CGA CTC ACT ATA GGG AGG CGG - (dT)₂₄ - 3' (SEQ ID NO:2); SMART II A oligonucleotide 5'-AAG CAG TGG TAT CAA CGC AGA GTA CGC GGG-3' (SEQ IDNO:3)

Antisense cDNA was purified using NucleoSpin Extraction Kit (BD Biosciences Clontech). All spins, unless otherwise noted, were for 1 minute at 14,000 rpm. Three volumes of NT2 Buffer were added to each cDNA mixture, which was then transferred into a NucleoSpin Extraction Spin column and centrifuged. The column with bound antisense cDNA was transferred into a clean collection tube. Antisense cDNA was washed with 500 microliters of NT3 Buffer for a total of 3 times. The column was transferred into a fresh collection tube and centrifuged to remove traces of ethanol. The column was then transferred into a 1.5 mL microcentrifuge tube and antisense cDNA was eluted with two aliquots, 50 microliters and 35 microliters, of MBG water. 2. Linear Amplification of the Sense Strand (FIGURE IB)

Sense strand was synthesized using linear amplification and the Advantage 2 PCR kit (BD Biosciences Clontech). A mix containing lOx Advantage 2 PCR buffer, 20 nmol dNTP mix, 15 pmol 5' PCR II A oligonucleotide 5' - AAG CAG TGG TAT CAA CGC AGA GT - 3' (SEQ ID NO:4) (FHCRC Biotech Center, Seattle, WA), 50x Advantage 2 Polymerase, and MBG water, was added to the purified antisense cDNA (total volume 100 microliters). The antisense cDNA was denatured for 1 minute at 95°C, followed by 23 cycles of 95°C for 15 seconds (denaturation), 60°C for 15 seconds (primer annealing) and 68°C for 6 minutes (elongation). GeneAmp PCR System 2400 (Applied Biosystems, Foster City, CA) was used for this and subsequent amplification. The amplified sense cDNA was then purified using the NucleoSpin protocol described above. 3. Synthesis of Double-Stranded cDNAfrom Amplified Sense cDNA (FIGURE 1C)

A mix containing lOx Advantage 2 PCR buffer, 20 nmol dNTP, 15 pmol DET7F oligonucleotide 5* - GG C CAG TGA ATT GTA ATA CGA CTC A - 3' (SEQ ID NO:5) (Fred Hutchinson Cancer Research Center Biotech Center), 50x Advantage 2 Polymerase mix, and MBG water, was added to the purified amplified sense cDNA strand (total volume 100 microliters). The amplified cDNA was denatured at 95°C for 1.25 minutes, followed by primer annealing at 60°C for 15 seconds and elongation step at 68°C for 10 minutes. Upon completion, the double-stranded (ds) cDNA was purified using the NucleoSpin protocol described above and eluted in 15 microliters MBG water.

EXAMPLE 2 This Example describes a comparison of the reliability and reproducibility of microarray data obtained using labeled cRNA produced using the Affymetrix protocol, the SLAP protocol, the Lemischka protocol, and the NIH Nanogram-Scale protocol. A. METHODS

1. RNA EXTRACTION AND INTEGRITY ASSAY

Total RNA was isolated and analyzed as described in EXAMPLE 1, above.

2. AFFYMETRIX PROTOCOL All reagents unless otherwise noted are from Invitrogen Life Technologies,

Carlsbad, CA. α. Reverse Transcription to Make Antisense cDNA

Total RNA and 100 pmol T (dT)₂₄ oligonucleotide (Operon, Alameda, CA, 5' - GGC CAG TGA ATT GTA ATA CGA CTC ACT ATA GGG AGG CGG - (dT)₂₄ - 3' (SEQ ID NO:2) were denatured for 10 minutes at 70°C (Precision 180 Series Water Bath, Chicago, II). The reaction mix was then incubated on ice for 5 minutes to allow the T₇(dT)₂₄ oligonucleotide to anneal. A mix, containing 5x first strand buffer, 0.2 micromol DTT, and 10 nmol dNTP, was then added, bringing the total volume to 18 microliters. The reaction mix was warmed to 42°C (Precision Dual Chamber Water Bath). Superscript RT (400 units) was then added to the mix and antisense cDNA was synthesized at 42°C for one hour in the water bath, followed by a 5-minute incubation on ice. b. Second Strand Synthesis

After the antisense-strand synthesis, a mix, containing 5x second strand buffer, 30 nmol dNTP, 10 units E. coli DNA ligase, 40 units E. coli DNA polymerase, and 2 units RNase H, was added to the cDNA reaction mix (total volume 150 microliters). Second strand (sense cDNA) was generated during a 2-hour incubation at 16°C (Precision Water Bath Model 183). Ends of the ds cDNA were polished with 10 units T4 DNA polymerase at 16°C for another 5 minutes. Reaction was stopped with 5 micromol EDTA (pH 8).

Double-stranded cDNA was purified using Phenol: Chloroform extraction and phase-lock gel tubes (PLG) (Brinkmann Instruments, Inc.) as per the Affymetrix GeneChip Protocol. The pellet was rehydrated in 12 microliters of MBG water (Brinkmann Instruments, Inc.). c. Biotin-Labeling ofcRNA

The entire volume of the purified ds cDNA was used as per the standard protocol for the Enzo BioArray High Yield RNA Transcript Labeling Kit (Affymetrix, Santa Clara, CA). Biotin-labeled cRNA was purified using the RNeasy Mini Protocol for RNA cleanup and quantified using Eppendorf BioPhotometer. d. Fragmentation of Biotin-Labeled cRNA

Fragmentation was accomplished using the standard Affymetrix protocol for GeneChip Expression Analysis. Twenty micrograms of labeled RNA (unless otherwise noted) was fragmented for each sample. Fifteen micrograms of fragmented cRNA (unless otherwise noted) was hybridized to the HuGene FL 6800 chips (Affymetrix). 3. SLAP PROTOCOL

Double-stranded (ds) cDNA was synthesized from RNA as described in EXAMPLE 1, above, and shown in FIGURE lA-C. a. Biotin-Labeling of a cRNA via IVT (FIGURE ID)

The entire volume of the purified ds cDNA was used as per the standard protocol for the Enzo BioArray High Yield RNA Transcript Labeling Kit (Affymetrix, Santa Clara, CA). Biotin-labeled cRNA was purified using the RNeasy Mini Protocol for RNA cleanup and quantified using Eppendorf BioPhotometer. b. Fragmentation of Biotin-Labeled cRNA (FIGURE ID)

Fragmentation was accomplished using the standard Affymetrix protocol for GeneChip Expression Analysis. Twenty micrograms of labeled RNA was fragmented for each sample. Fifteen micrograms of fragmented cRNA was hybridized to the HuGene FL 6800 chips (Affymetrix). 4. LEMISCHKA PROTOCOL

All incubations were performed in Techne Progene hot-lid thermocycler. a. First Cycle: Reverse Transcription to Make Antisense cDNA Complementary DNA was synthesized using SUPERSCRIPT II Reverse Transcriptase system (Invitrogen Life Technologies). T₇(dT)₂ oligonucleotide (SEQ ID NO:2) (100 pmol) was added to total RNA. The mixture was first denatured for 10 minutes at 70°C. T (dT)₂₄ oligonucleotide was annealed to the template during 5 minute incubation at 42°C. Antisense cDNA was synthesized in a reverse transcription reaction using a total volume of 20 microliters, containing 5x first strand buffer, 0.2 micromol DTT, 100 units RNase inhibitor, 10 nmol of dNTP mix, and 200 units Superscript II RT. Synthesis of antisense cDNA was performed at 42°C for one hour. b. First Cycle: Second Strand Synthesis After a 5 -minute incubation on ice, a mixture containing 5x second strand buffer,

30 nmol dNTP, 10 units E. coli DNA ligase, 40 units E. coli DNA polymerase, 2 units RNase H (all from Invitrogen Life Technologies), and MBG water to bring the total volume to 150 microliters, was added to the cDNA reaction mix. Second strand (sense cDNA) was generated during a 2-hour incubation at 16°C. Ends of the ds cDNA were polished with 10 units T4 DNA polymerase (Invitrogen Life Technologies) during a 10- minute incubation at 16°C. The reaction was stopped with 5 micromol EDTA (pH 8).

Double-stranded cDNA was purified using Phenol: Chloroform extraction and Microcon-50 columns (Millipore Corporation, Bedford, MA). One volume phenol:chloroform:isoamyl alcohol (25:24:1) (Ambion Inc., Austin, TX), saturated with Tris (pH 8), was added to the sample, mixed well and centrifuged at 14,000 rpm for 2 minutes to separate the two phases. The upper aqueous phase was transferred to Microcon-50 column and overlaid with 350 microliters MBG water. Double-stranded (ds) cDNA was adhered to the membrane during a 7-minute spin at 11,000 rpm. The column was washed with 500 microliters MBG water and centrifuged for 8 minutes at 11,000 rpm. It was then inverted into a clean collection tube and centrifuged for 5 minutes at 3,500 rpm to elute the ds cDNA. The total volume collected ranged between 10-15 microliters. If volume exceeded the expected yield, ds cDNA was concentrated using Automatic Environmental SpeedVac System with VaporNet (Thermo Savant, Holbrook, NY). c. First Transcription Step cRNA was transcribed from ds cDNA using the Ambion MEGAscript kit (Ambion Inc.). Incubation was performed for 3 hours at 37°C as per the kit protocol. The generated cRNA was purified using the phenol: chloroform extraction with Microcon columns as outlined above. d. Second Cycle: cDNA synthesis cRNA and 1 micrograms random hexamers (Invitrogen Life Technologies) were denatured for 10 minutes at 70°C, followed by 10-minute holds at 4°C and 25°C. To generate sense cDNA, 5x first strand buffer, 0.2 micromol DTT, 10 nmol dNTP mix, 100 units RNase Inhibitor, and 200 units SUPERSCRIPT RT, were added to the initial mix of cRNA and random hexamers. Sense cDNA was synthesized at 37°C for 1 hour. Two units RNase H were added to degrade cRNA component of the cDNA RNA hybrids during a 20-minute incubation at 37°C. SUPERSCRIPT RT was inactivated by a 2-minute hold at 94°C, followed by a 10-minute hold at 4°C. e. Second cycle: Second Strand synthesis

T (dT)₂₄ oligonucleotide (SEQ ID NO:2) (100 pmol) was added to the sense cDNA and the mixture was denatured at 70°C for 5 minutes, followed by primer annealing at 42°C for 10 minutes. Following a 5-minute incubation, a mixture containing 5x second strand buffer, 30 nmol dNTP, 10 units E. coli DNA ligase, 40 units E. coli DNA polymerase, 2 units RNase H, and MBG water to bring the volume to 150 microliters, were added to the cDNA reaction mix. Antisense cDNA was generated during a 2-hour incubation at 16°C. Ends of the ds cDNA were polished with 10 units T4 DNA polymerase during a 10-minute incubation at 16°C. The reaction was stopped with 10 micromol EDTA (pH 8). Double-stranded cDNA was purified using Phenol: Chloroform extraction and Microcon-50 columns using the procedure outlined above.

Biotin labeling and fragmentation of cRNA was accomplished using procedures described above.

5. NIH NANOGRAM-SCALE PROTOCOL

A more detailed protocol is available at http://dc.nci.nih.gov/protocols. a. First Cycle: Reverse Transcription to Make Antisense

All incubations were performed in Techne Progene hot-lid thermocycler, unless otherwise noted. cDNA was synthesized using SUPERSCRIPT II RT system (Invitrogen Life Technologies).

Total RNA and T₇(dT)₂₄ oligonucleotide (SEQ ID NO:2) (100 pmol) were denatured for 10 minutes at 70°C. The reaction mix was then incubated for 5 minutes at 4°C to allow for T₇(dT)₂₄ oligonucleotide annealing. A mix, containing 5x first strand buffer, 0.2 micromol DTT, 100 units RNase inhibitor, 10 nmol dNTP, and 200 units SUPERSCRIPT II RT, was then added (total volume 20 microliters). Antisense cDNA was synthesized at 42°C for one hour, followed by a 4°C hold for 5 minutes. b. First Cycle: Second Strand Synthesis

After the antisense-strand synthesis, a mix, containing 5x second strand buffer, 30 nmol dNTP, 10 units E. coli DNA ligase, 40 units E. coli DNA polymerase, and 2 units RNase H, was added to the cDNA reaction mix (total volume 150 microliters). Second strand (sense cDNA) was generated during a 2-hour incubation at 16°C, followed by a 4°C hold for 5 minutes. Ends of the ds cDNA were polished with 10 units T4 DNA polymerase at 16°C for another 5 minutes. Reaction was stopped with 5 micromol EDTA (pH 8).

Double-stranded cDNA was purified using Phenol: Chloroform extraction and phase-lock gel tubes (PLG) (Brinkmann Instruments, Inc.) as per NCI/NIH protocol. The pellet was rehydrated in 8 microliters of MBG water. c. First Transcription Step

Using the Ambion MEGAscript kit, cRNA was transcribed from ds cDNA. Incubation was performed for 5 hours at 37°C as per protocol. The generated cRNA was purified using the phenolxhloroform extraction and PLG tubes as per protocol. The cRNA was rehydrated in 10 microliters of MBG water. d. Second Cycle: cDNA Synthesis cRNA and 250 ng random hexamers were denatured for 10 minutes at 70°C, followed by a 4°C hold for 5 minutes. A mix containing 5x second strand buffer, 30 nmol dNTP, 10 units E. coli DNA ligase, 40 units E. coli DNA polymerase, and 2 units RNase H, was added to the cDNA reaction mix (total volume 150 microliters). 5x first strand buffer, 0.2 micromol DTT, 10 nmol dNTP mix, and 400 units Superscript was added to cRNA mixture and incubated for 10 minutes at 25°C. cDNA was synthesized at 42°C for 1 hour, followed by a 4°C hold for 5 minutes. e. Second cycle: ds cDNA Synthesis T₇(dT)₂₄ oligonucleotide (SEQ ID NO:2) (100 pmol) was added to the cDNA reaction mix, which was incubated at 70°C for 10 minutes to denature cDNA, followed by a 4°C hold for 5 minutes. A mixture containing 5x second strand buffer, 30 nmol dNTP, 10 units E. coli DNA ligase, 40 units E. coli DNA polymerase, and 2 units RNase H were added to the cDNA reaction mix (total volume 150 microliters). Second strand (antisense cDNA) was generated during a 2-hour incubation at 16°C, followed by a 4°C hold. Ends of the ds cDNA were polished with 10 units T4 DNA polymerase at 16°C for 10 minutes. The reaction was stopped by the addition of 5 micromol EDTA (pH 8). Double-stranded cDNA was purified using Phenol: Chloroform extraction and

PLG tubes as per protocol. Double-stranded cDNA was rehydrated in 22 microliters of MBG water.

Biotin labeling of cRNA was accomplished using procedure described above. Labeled cRNA was purified using a modified form of the RNeasy Mini Protocol for RNA cleanup protocol. The centrifugation speed was decreased while the duration of the spins was increased. Fragmentation of labeled cRNA was performed according to the method described above. Comparison of the NIH and Lemischka Protocols

The RT for the NIH protocol is similar to the Lemischka protocol. The major differences included an annealing temperature of 4°C for 5 minutes after the denaturing of the total RNA and T₇(dT) ₄ oligonucleotide and an incubation step at 4°C rather than on ice. Second Strand Synthesis was the same in the NIH protocol as in the Lemischka protocol, except that the cDNA ends are polished at 16°C for another 5 minutes. The purification of double-stranded cDNA was different between the two protocols. The NIH protocol uses phenokchloroform extraction and phase-lock gel tubes (PLG) (Brinkmann Instruments, Inc.), and the pellet was rehydrated in 8 microliters of MBG water.

The first transcription step is exactly the same in the NIH protocol as the Lemischka protocol, except the incubation was performed for 5 hours at 37°C as per protocol. The generated cRNA was purified using the phenokchloroform extraction and PLG tubes as per protocol. The cRNA was rehydrated in 10 microliters of MBG water. cDNA synthesis of the second cycle was performed similarly in the NIH protocol compared to the Lemischka protocol. Major differences included 1) the addition of 250 ng (rather than 1 microgram) random hexamers, 2) the addition of the second reaction mixture prior to increasing the temperature to 25°C for 5 minutes for random hexamer annealing, and 3) cDNA synthesis at 42°C (instead of 37°C) for 1 hour, followed by a 4°C hold for 5 minutes. Double-stranded cDNA synthesis had only a single 4°C hold for 5 minutes after the denaturing step in the NIH protocol, but the rest of the double- stranded cDNA synthesis was the same. Again, the clean-up was different in the NIH protocol, using the phenol: chloroform and PLG system. Double-stranded cDNA was rehydrated in 22 microliters of MBG water in the NIH protocol.

6. TARGET HYBRIDIZATION AND SCANNING

As per recommendation of Affymetrix, all hybridization procedures were performed with 15 micrograms of fragmented, biotin labeled RNA, unless otherwise noted. Some of the protocols did not produce 15 micrograms of RNA for labeling. For these samples, all the biotin labeled RNA was used for hybridization, and this is noted in the results. All samples were prepared for hybridization as per the Affymetrix standard eukaryotic target hybridization protocol (Gene Expression Monitoring; NIH Expression Analysis, Technical Manual - 701021 rev 1 p. 2.3.5 - 2.4.18). Samples were hybridized to HuGene FL 6800 chips (Affymetrix).

7. STATISTICAL METHODS a. Absolute Analyses

DAT files were used to generate CHP files for individual samples using Affymetrix MAS 5.0 software. Individual samples underwent absolute analyses. Scale factor was set at 500 for all analyses. No probe mask was used. The detection algorithm was based upon the default settings as per the recommendations provided by Affymetrix. Each probe pair has a Discrimination score (R) calculated, where R = (PM - MM)/(PM+MM). The R is compared to a predefined threshold called Tau. Probes with R greater than Tau are called Present, while probes with R less than Tau are absent. The greater the R, the smaller the p value is, and vice versa. For all analyses, the default Tau setting, Tau = 0.015 was used. The Detection ? value was generated using One-sided Wilcoxon's Signed Rank test via the MAS 5.0 software. The p value cut-offs are predetermined by the user and based upon αl and α2, such that/, values less than αl is assigned a Present call (P), p values greater α2 assigned a Absent call (A), and p values between αl and α2 are assigned a Marginal call (M). The default alpha settings for all analyses, αl = 0.04, and α2 = 0.06 were used (for additional details see Affymetrix, Statistical Algorithms Reference Guide, http://www.affymetrix.com/support/technical/ technotesmain.affx). b. Comparison Analyses

Comparison Measures. Comparison of replicates within a protocol allows an assessment of the reproducibility of protocols. By treating expression values obtained from the standard Affymetrix protocol using 10 micrograms of starting RNA as the gold standard, the reliability of other protocols at varying amounts of starting total RNA can be assessed. Assessment of transcript abundance results in quantitative assessment of reproducibility and reliability. In contrast, comparison of detection calls or p values results in qualitative assessments of reproducibility and reliability. Qualitative Assessment of Detection Calls: The Kappa statistic was used to measure reproducibility and reliability of detection calls. Essentially, Kappa measures the degree of concordance in detection calls between pairs of replicates or protocols. A

Kappa of 1 indicates perfect agreement between paired vectors, beyond the role of chance. In contrast, a Kappa of 0 implies merely a chance agreement. Qualitative Assessment ofp values: p values were transformed into a logit scale via ln[p/(l-p)]. The differences in the transformed values for each pair of replicates were normalized to have mean 0. Standard deviations (SD) of the normalized differences provided a qualitative assessment of transformed p values. The smaller the value of the

SD, the greater the reproducibility or reliability. Quantitative Assessment of Transcript Abundance: Signal intensities were transformed into log2(intensity). The differences in the transformed values for each pair were normalized to have mean 0. The SDs of the normalized differences then provided a quantitative measurement of reproducibility. The smaller the value of the SD, the greater the reproducibility or reliability. B. RESULTS

1. QUANTITY OF AMPLIFICATION VERSUS PROTOCOL

Affymetrix recommends that at least 15 micrograms of biotin labeled RNA be used for hybridization. Therefore, the minimal amount of starting RNA for standard Affymetrix protocol and the three IVT amplification protocols that would successfully produce 15 micrograms of biotin-labeled RNA was determined. All protocols demonstrated significant amount of RNA amplification, including the standard Affymetrix protocol (Table 1). Using the Affymetrix standard protocol and starting with as little as 1 microgram of total RNA, 30 micrograms of biotin-labeled RNA were produced, enough to perform hybridization. However, the production of biotin-labeled RNA rapidly decreased when the starting RNA amount decreased below 1 microgram, such that suboptimal amounts of RNA were obtained for hybridization. Using nanograms of RNA starting material, SLAP, Lemischka, and NIH protocols generated significantly greater amounts of biotin-labeled RNA than the standard Affymetrix protocol (Table 1). SLAP exhibited the most robust RNA amplification, yielding adequate amounts of biotin- labeled RNA with as little as 0.0001 micrograms of starting RNA, while the NIH and Lemischka protocols required a minimum of 0.050 micrograms of starting RNA to produce 15 micrograms of biotin-labeled RNA (Table 1).

Table 1. Amount of Biotin-Labeled Product Obtained Using Each Protocol

Starting RNA Affymetrix SLAP Lemischka NIH

(micrograms) (micrograms) (micrograms) (micrograms) (micrograms)

10.000 66 ND ND ND

5.000 90 ND ND ND

1.000 30 76 97 50

0.500 9 ND ND ND

0.100 1 53 48 67

0.050 ND 75 59 77

0.010 ND 54 12 3

0.002 ND 35 10 5

0.001 ND 28 ND ND

0.0001 ND 6 ND ND

ND - assays were not done.

However, IVT amplification protocols consistently resulted in elevated 3-prime to

5-prime ratios for control genes, such as Actin and GAPDH, emphasizing the 3-prime bias of such methods (Table 2). Actin and GAPDH are two "housekeeping" genes with probe sets spanning the entire length of their transcript. To assess the quality of the biotin-labeled product, 5-prime signals from both genes were divided by the 3-prime signals, determining the relative conservation of signal throughout the entire transcript. SLAP demonstrated excellent conservation of the 5-prime to 3-prime signals for both genes, mimicking that of the standard Affymetrix protocol using 10 micrograms of starting RNA, as shown in Table 2. In contrast, the NIH and Lemischka protocols rsulted in low percentages of signal conservation, indication loss of 5-prime expression signal for both genes. Table 2. Conservation of 5-prime to 3-prime Signals of Actin and GAPDH

Actin Signal Conservation GAPDH Signal Conservation

Starting RNA Affymetrix NIH SLAP Affymetrix NIH SLAP

(micrograms)

10.000 92% 105%

5.000 89% 100%

2.500 90% 85%

1.000 52% 9% 89% 101% 52% 93%

0.500 68% ND ND 68% ND ND

0.100 9% 16% 103% 29% 26% 99%

0.050 12% 108% 35% 97%

0.010 1% 112% 10% 85%

0.002 3% 102% 13% 79%

0.001 ND 124% ND 99%

0.0001 ND 71% ND 96%

ND - assays were not done.

2. REPRODUCIBILITY OF AMPLIFICATION PROTOCOLS

To investigate the reproducibility of each protocol, replicates were performed for the different amounts of starting RNA amounts. The HuGene FL 6800 chip contains 7129 probe sets, of which 59 are Affymetrix control probe sets. These control proble sets were not used for analyses. Reproducibility of a particular protocol at a given starting amount of RNA was determined by comparing these replicates. The detection call change is defined by genes for there was: (1) a Change call from Absent (A) or Marginal (M) to present (P), or (2) a Change call from Present (P) to Absent (A) or Marginal (M). The percentage of genes with a detection call change was assessed by standard deviation of log2 (signal), standard deviation of logit (p value), and Kappa (Table 3). A Kappa equal to 1 indicates complete agreement between the replicates without evidence of chance, while a Kappa equal to 0 suggests no agreement beyond that of chance. The standard Affymetrix protocol displayed the highest reproducibility at amounts of starting total RNA 1 microgram or more (Table 3). However, the reproducibility of the standard Affymetrix decreased as the starting amount of total RNA fell below 1 microgram, paralleling which the overall drop in the amount of biotin-labeled RNA produced (Table 1). The reproducibility for the SLAP, Lemischka, and NIH protocols is reasonably good, with the Lemischka protocol displaying the poorest reproducibility with low levels of starting RNA (Table 3). The reproducibility seems to vary slightly for each protocol, depending on the starting amount of RNA.

Overall, the reproducibility data indicate that the standard Affymetrix protocol is as good as, if not better than, the IVT amplification protocols with an amount of 1 microgram of starting total RNA. However, the results also confirm that the standard Affymetrix protocol looses this advantage as the amount of starting material decreases to less than 1 microgram. All four protocols, even the standard Affymetrix protocol at 10 micrograms, introduced gene expression biases, which was expected given that all four protocols utilized IVT amplification methods for labeling RNA.

Table 3. Reproduciblity of Replicates for Each Protocol

3. RELIABILITY OF AMPLIFICATION PROTOCOLS.

The standard Affymetrix, SLAP, Lemischka, and NIH protocols were compared to the standard Affymetrix protocol at 10 micrograms of starting RNA, which was chosen as the gold standard for comparison because Affymetrix recommends starting with 5 to 15 micrograms of RNA. The reliability of the experimental protocols was assessed as the amount of starting RNA decreases compared to the gold standard Affymetrix protocol. Each experimental sample was compared to each of the two 10 microgram Affymetrix RNA chips separately. The results for each gene were then averaged. The detection call change was defined as described above and reliability was assessed by standard deviation of log2 (signal), standard deviation of logit (p value), and Kappa (Table 3).

With amounts of starting total RNA of equal to or more than 1 microgram, the standard Affymetrix protocol displayed the highest level of reliability (Table 4). With 1 microgram of starting RNA, the standard Affymetrix protocol had a Kappa measure of agreement of 0.874 (Table 3). Similar to the reproducibility results, the reliability of the standard Affymetrix protocol decreased dramatically as the amount of starting total RNA fell below 1 microgram, and at 0.1 micrograms, the standard Affymetrix protocol demonstrated the lowest level of reliability. Because of the high reliability of the standard Affymetrix protocol using 1 microgram of starting RNA, the amplification protocols were not examined with starting RNA amounts of more than 1 microgram.

With respect to the three amplification protocols (SLAP, NIH, and Lemischka), there seem to be little difference in reliability among the three protocols starting with 1 microgram of total RNA (Table 4). For example, all three amplification protocols had a consistently lower Kappa measures of agreement than the standard Affymetrix protocol (0.874) with 1 microgram of starting RNA (Table 4). These results suggest that all three amplification protocols introduce an amplification bias beyond the amplification bias introduced by the standard Affymetrix protocol. However, as the amount of starting RNA decreases, there does appear to be a need for amplification in order to obtain enough biotin-labeled RNA for microarray assays. Overall, the SLAP protocol had the highest measure of agreement as the starting RNA amount decreased to below 0.05 micrograms (Table 4).

Table 4. Reliability of Replicates for Each Protocol

As a more qualitative measure of reliability, the percentage of detection call changes (gold standard Affymetrix protocol Present (P) — > experimental protocol Absent (A) or Marginal (M) or gold standard Affymetrix protocol A/M --» experimental protocol P) was determined using the Affymetrix protocol at 10 micrograms of starting RNA as the gold standard. Since the baseline and experimental groups were performed in replicates, all four comparisons for an experimental protocol at a given amount of starting RNA were averaged to obtain the overall percentage of detection call changes. Comparisons between Affymetrix 10 microgram vs. Affymetrix 10 microgram identified that 4.6% of genes had detection call changes between these replicates (Table 5). As suggested by the statistical analyses for reliability, the three amplification protocols initially displayed similar percentages of detection call changes, but the detection call changes for the NIH and Lemischka protocols dramatically increased as the starting RNA amount fell below 0.05 micrograms. By far, the most common call change (60% to 98%) was the gold standard Affymetrix P — experimental A /M call change (range = 60% - 98%), suggesting that the amplification protocols more frequently call genes inappropriately absent (false negative) than present (false positive).

Table 5. Detection Call Changes for Each Protocol

Starting RNA Affymetrix SLAP Lemischka NIH

(micrograms)

1.0 6.10 9.40 11.02 8.94

0.1 29.45 9.70 19.55 10.00

0.05 12.36 14.00 9.63

0.01 8.70 29.10 22.47

0.002 11.13 35.80 29.57

It was of interest to investigate whether the gene expression differences between the standard Affymetrix and amplification protocols were due to random or nonrandom amplification events. Since the majority of detection change calls were false negative calls, these analyses were limited to the 2989 genes that were consistently called P by the standard Affymetrix protocol at 10 micrograms of starting RNA. The 2989 genes were sorted to identify the genes that had a detection call change in both replicates of the amplification protocols.

False negative calls (gold standard Affymetrix P — > experimental A/M) were the most common type of error. With respect to SLAP, a significant amount of overlap existed between the genes that were false negatives, with 24% (99/419) of the false negatives occurring at all three starting amounts of RNA (FIGURE 2A). Approximately 50% (200/419) of the false negatives were unique to a single amount of starting RNA. The SLAP results suggest that although some of the false negatives are due to nonrandom events, approximately 50% of these false negatives may be due to random amplification biases.

The amount of overlap of false negatives appeared greatest for the NIH and Lemischka protocols (FIGURES 2B and 2C), but much of this overlap may be secondary to the dramatic increase in the number of false negatives with low amounts of starting RNA. However, it has been demonstrated that IVT amplification protocols systematically cause 3-prime amplification biases, leading to nonrandom amplification biases (Baugh et al. (2001) Nucl. Acids Res. 29:E29). The extent of overlap of false calls suggests that both random and nonrandom (i.e., systematic) amplification biases may be contributing to some of the false calls. Performing multiple replicates could minimize false calls due to random amplification events, but systematic amplification biases cannot be addressed this way. Rather, systematic amplification biases must be acknowledged and affected genes eliminated from analyses. To make such corrections feasible, the false calls need to be relatively small and constant, as is seen using SLAP.

Although false negative calls for SLAP were relatively constant (range 200 to 308), the false negative calls for the NIH and Lemischka protocols varied dramatically, increasing from 236 to 1102 and from 185 to 1649, respectively, as the amount of starting RNA decreased (FIGURE 2A-C). Overall the false negatives obtained using SLAP were far less than those found in the NIH and the Lemischka protocols (FIGURE 2B and 2C). For example, comparing the numbers of false negative calls between the experimental protocol using the McNemar test (Parker et al. (2002) Obstet. Gynecol. 100:277080), SLAP displayed more false negative calls at 1 microgram of starting RNA than the NIH protocol (308 vs 236,/? < 0.0001), but the NIH protocol produced significantly more false negative calls at 0.1 micrograms of starting RNA (200 vs 305, ? < 0.0001) and at 0.01 micrograms of starting RNA (229 vs 1102,/. < 0.0001). The amount of overlap of false negative detection call changes for each amount of starting RNA for SLAP, the Lemischka protocol, and the NIH protocol is shown in FIGURE 3. False negative calls for SLAP were relatively constant (range from 200 to 308), while the number of false negatives for the NIH and Lemischka protocols decreased dramatically with decreasing amount of the starting of RNA. With respect to false positive calls (gold standard Affymetrix A/M - experimental P), the frequencies were relatively similar for SLAP and the NIH and Lemischka protocols. For example, there were 23-33 false positive calls for SLAP compared to 20-88 false positive calls for the NIH protocol, with the NIH protocol producing slightly more false positive calls at 0.1 micrograms of starting RNA (88 vs 23, p < 0.0001), and both protocols demonstrated a relatively high percentage of overlap among the false detection calls (FIGURE 4A, B).

Recently, a modified PCR protocol using poly(A) tailing (Global RT-PCR) was found to preserve global expression of selected genes on spotted arrays (1718 distinct PCR-amplified human cDNAs) better than a single-round of IVT amplification (Iscove et al. (2002) Nat. Biotechnol. 20:940-3). The Global RT-PCR protocol identified 76/104 true outliers (false negative rate 25%) starting with nanogram amounts of RNA, which was superior to a single round of IVT (false negative rate 68%). Unlike the SLAP protocol, both the Global RT-PCR and the SMART PCR protocols exponentially amplify from both ends of transcripts (Puskas et al. (2002) Biotechnigues 32:130-4, 1336, 1338, 1340; Iscove et al. (2002) Nat. Biotechnol. 20:940-3). The results with SLAP compare favorably with those produced using Global RT-PCR and SMART PCR, with SLAP having a false negative rate of about 7-10% and a false positive rate of <1%. In conclusion, multiple different amplification protocols have been developed but many of these protocols create amplification biases that make them unreliable, preventing comparisons across different RNA starting amounts. Compared to the NIH and Lemischka protocols, SLAP had the most robust amplification, the best preservation of 5- prime signal of the transcript, and the highest level of reliability when using nanogram amounts of starting RNA. In addition, this improved reliability translated into fewer false negative calls, displaying far fewer false negative detection calls. SLAP also displayed the highest reproducibility with amounts of starting RNA of less than 0.05 micrograms. Moreover, SLAP is extremely easy to perform, taking only 2 days, while the other amplification protocols take more than 3 days and contain multiple clean-up steps that can introduce additional variables in the process.

While the preferred embodiment of the invention has been illustrated and described, it will be appreciated that various changes can be made therein without departing from the spirit and scope of the invention.

Claims

The embodiments of the invention in which an exclusive property or privilege is claimed are defined as follows:

1. A method for linearly amplifying sense cDNA from an RNA molecule, comprising the steps of:

(a) synthesizing an antisense cDNA molecule using an RNA molecule as a template;

(b) synthesizing an extended antisense cDNA molecule using a template- switching oligonucleotide as a template;

(c) synthesizing multiple copies of sense cDNA molecules using the extended antisense cDNA strand as a template; and

(d) generating double-stranded cDNA molecules using the multiple copies of sense cDNA molecules as templates.

2. The method of Claim 1, wherein step (a) comprises incubating the RNA molecule with a first primer under suitable conditions for synthesizing an antisense cDNA molecule, wherein the first primer comprises a sequence complementary to a region of the RNA molecule and a first defined nucleic acid sequence.

3. The method of Claim 2, wherein the first defined nucleic acid sequence comprises a promotor for an RNA polymerase.

4. The method of Claim 3, wherein the RNA polymerase promoter comprises a T7 promoter.

5. The method of Claim 1, wherein step (b) comprises incubating the antisense cDNA molecule with a second primer under suitable conditions for synthesizing an extended antisense cDNA molecule, wherein the second primer comprises a template-switching oligonucleotide and a second defined nucleic acid sequence and wherein the extended antisense cDNA molecule comprises a sequence complementary to the second defined nucleic acid sequence.

6. The method of Claim 5, wherein the second defined nucleic acid sequence comprises a promoter for an RNA polymerase.

7. The method of Claim 5, wherein the RNA polymerase promoter comprises a T7 promoter.

8. The method of Claim 1, wherein step (c) comprises incubating the extended antisense cDNA molecule with a third primer under suitable conditions for synthesizing multiple copies of sense cDNA molecules, wherein the third primer comprises the second defined nucleic acid sequence.

9. The method of Claim 1, wherein step (d) comprises incubating the sense cDNA molecules with a fourth primer under suitable conditions for synthesizing double- stranded cDNA, wherein the fourth primer comprises the first defined nucleic acid sequence.

10. A method for amplifying RNA, comprising the steps of:

(a) incubating an RNA molecule with a first primer under suitable conditions for synthesizing an antisense cDNA molecule, wherein the first primer comprises a sequence complementary to a region of the RNA and a first defined nucleic acid sequence;

(b) incubating the antisense cDNA molecule with a second primer under suitable conditions for synthesizing an extended antisense cDNA molecule, wherein the second primer comprises a template-switching oligonucleotide and a second defined nucleic acid sequence, wherein at least one of the first defined nucleic acid sequence and the second defined nucleic acid sequence comprises a promoter for an RNA polymerase, and wherein the extended antisense cDNA strand comprises a sequence complementary to the second defined nucleic acid sequence;

(c) incubating the extended antisense cDNA molecule with a third primer under suitable conditions for synthesizing multiple copies of sense cDNA molecules, wherein the third primer comprises the second defined nucleic acid sequence;

(d) incubating the sense cDNA molecules with a fourth primer under suitable conditions for synthesizing double-stranded cDNA, wherein the fourth primer comprises the first defined nucleic acid sequence; and

(e) incubating the double-stranded cDNA molecules with RNA polymerase under suitable conditions for synthesizing RNA.

11. The method of Claim 13, wherein the first defined nucleic acid sequence comprises a T7 promoter.

12. The method of Claim 13, wherein the RNA polymerase is T7 RNA polymerase.