WO2006018277A1

WO2006018277A1 - Method for rational combinatorial synthesis

Info

Publication number: WO2006018277A1
Application number: PCT/EP2005/008885
Authority: WO
Inventors: Jean-Louis Reymond
Original assignee: Universitaet Bern
Current assignee: Universitaet Bern
Priority date: 2004-08-17
Filing date: 2005-08-16
Publication date: 2006-02-23
Anticipated expiration: 2007-02-17

Abstract

The invention describes a method for preparing combinatorial libraries using a modification of the split-and-mix procedure, wherein the building blocks are so chosen as to allow deducing the sequence of building blocks from the determination of the amount and type of building blocks after total cleavage. The method is particular suitable for the synthesis of combinatorial libraries of dendrimers.

Description

Method for rational combinatorial synthesis

Background of the invention

Chemically synthesized molecules are used in many different products of daily life, including drugs, food additives, dyestuffs, polymers and plastics. Currently, many research groups are looking for ways to identify molecules that display the desired properties for a particular application. Synthetic chemistry allows one to synthesize molecules with almost any structure by stepwise synthesis. Historically, each molecule used to be prepared one at a time and then tested for its properties. This approach is time consuming and inefficient. In search for a more efficient method the combinatorial synthesis method which allows one to prepare large numbers of molecules simultaneously was developed. Each of these molecules is then tested for its properties and identified. This process greatly speeds up the discovery of active molecules because many more molecules can be tested for a particular property than through conventional step-by-step synthesis.

In the so-called "split-and-mix" protocol, the molecules being synthesized are attached to a solid support. The solid support generally consists of small polymer beads. Combinatorial libraries of compounds are obtained by growing the molecules by stepwise attachment of building blocks through chemical reactions. IfN different building blocks are used at each step in a sequence comprising S steps where variable building blocks can be attached, the resulting number of molecules is then N in conventional synthesis, but N^s for a combinatorial split-and-mix synthesis (K. S. Lam, S. E. Salmon, E. M. Hersh, V. J. Hruby, W. M. Kazmierski, R. J. Knapp, Nature 1991, 354, 82-84). The N^s molecules are then tested for a particular property either directly on the solid support, or after cleavage from this support.

Once an active molecule has been found in a combinatorial library, its structure must be identified. This can be achieved by using different strategies. One approach are so-called encoding methods which include binary encoding (M. H. J. Ohlmeyer, R. N. Swanson, L. W. Dillard, J. C. Reader, G. Asouline, R. Kobayashi, M. H. Wigler, W. C. Still, Proc. Natl. Acad. ScL USA 1993, 90, 10922-10926), whereby a molecular code is attached after each step, or radiofrequency tags, whereby each solid-support bead is equipped with a microchip into which the step information can be written and read electronically, and the tea-bag method, whereby labeled tea-bags containing portions of beads are moved between synthesis vessels.

In the case of peptide molecules, it is also possible to identify the structure or sequence of the molecule by stepwise Edmann degradation directly on the beads, although the method is expensive (K. S. Lam, S. E. Salmon, E. M. Hersh, V. J. Hruby, W. M. Kazmierski, R. J. Knapp, Nature 1991, 354, 82-84).

Another method for identifying the structure of active molecules in a combinatorial library is to carefully choose the building blocks so that each molecule has a different molecular weight. Thus determination of the molecular weight of an active molecule by mass spectrometry (MS) allows one to know its structure (WO 97/08190). This MS determination is only possible if precise rules defining the exact molecular weights of the building blocks are used in the library, which limits the diversity of building blocks that can be used.

Summary of the invention

The invention describes a method for preparing combinatorial libraries, characterized in that beads are equally split in n portions, each portion reacted with one building block of a first group of n building blocks, reacted beads thoroughly mixed and equally split in m portions, each portion reacted with one building block of a second group of m building blocks, the steps of thoroughly mixing, equally splitting in portions and reacting with one building block of a further group of building blocks repeated for 1 to 20 times; the resulting beads with the sequence of building blocks, or the sequence of building blocks after cleavage from the beads, analyzed for desired properties and selected based on the desired properties; selected beads with the sequence of building blocks, or the selected cleaved sequence of building blocks, cleaved into individual building blocks, the amount and type of building blocks determined, and the sequence of building blocks deduced from the amount and type of building blocks determined; wherein one building block in a group of building blocks may be absent; and wherein the building blocks in each group of building blocks are so chosen as to allow deducing the sequence of building blocks from the determination of the amount and type of building blocks after cleavage.

The method is particular suitable for the synthesis of combinatorial libraries of dendrimers. Such combinatorial libraries are also a subject of the invention.

Brief description of the figures

Figure 1: The known combinatorial split-and-mix vs. the classical synthesis protocol

Step 1): Beads are split in 4 portions and reacted with building blocks Al - A4 in 4 different vials. In the classical synthesis protocol, the compounds in the vials are again treated with building blocks Al - A4 in each of steps 2) and 3), giving 4 final products (right lane). Mixing the obtained compounds from step 1) and equally redistributing

("splitting") in 4 vials for step 2) gives 4² = 16 products. Mixing the obtained compounds from step 2) and splitting in 4 vials for step 3) gives 4³ = 64 products.

Figure 2: The combinatorial group split-and-mix of the invention Different building blocks from groups A, B or C are used at each of the three steps 1), 2) and 3). In the group protocol, building blocks A1-A4, then B1-B4 and finally C1-C4 are coupled in reactions vials 1-4, respectively. This combinatorial group split-and-mix also leads to the formation of 64 different molecules.

Figure 3: The combinatorial group split-and-mix with one "no-building-block"

Again different building blocks from groups A, B or C are used at each of the three steps 1), 2) and 3) as in Figure 2, however, one building block is absent, i.e. a "no-building- block". In the group protocol (shown on the left hand side), A4, B4 and C4 are absent. Combinatorial group split-and-mix leads to the formation of 63 different molecules with sequence length of 1 to 3 building blocks. Figure 4: Dendrimer architecture suitable for group split-and-mix procedure A combinatorial split-and-mix library of 65'536 different peptide dendrimers with the structure (((A8-A7)₂B-A6-A5)₂B-A4-A3)₂B-A2-Al-spacer-suρport. Al to A8 denote variable α-amino acid positions, dots • (shown with three bonds) indicate a branching unit B (e.g. a diaminoalkanoic acid), spacer is e.g. an co-aminoalkanoic acid allowing undisturbed peptide coupling, and the support is a bead suitable for peptide coupling.

Detailed description of the invention

The present invention describes a method for preparing combinatorial libraries where the building block information determines the structure. The molecules are designed in advance in such a manner that the structure of each molecule in the library can be determined by quantitative analysis of the building blocks used in the synthesis. By using exactly defined and different building blocks at each variable step during synthesis and thus restricting the size of the library, the structure of a product can be determined by analyzing the original building block composition and by assigning the found compounds to the specific position in the molecule design. One of the advantages of this method is that combinatorial libraries can be prepared without the need to attach an extra molecule for encoding the steps or being restricted by the need to use molecules of different molecular weight. In this manner, the synthesis can be simplified and accelerated. For example, the use of combinatorial libraries for the discovery of active molecules of which the structure can be determined is greatly improved. The concept can be used for any type of library and any type of analysis method and is not restricted to the examples described in the following.

"Combinatorial libraries" are collections of macromolecules of related structure comprising sequences of building blocks.

"Beads" are solid carriers in shapes convenient for handling in chemical reactions, e.g. balls, rods or the like. Solid means that the beads are solid under the usual conditions of reactions with building blocks, i.e. do not dissolve in aqueous solvents or organic solvents such as hydrocarbons, chlorinated hydrocarbons, ethers, alcohols, ketones, and, in particular, dipolar aprotic solvents such as dimethylformamide, dimethylacetamide, N- methylpyrrolidone, dimethyl sulfoxide or hexamethyl phosphoric acid triamide. Beads may be of different sizes, preferably having an average diameter of between 0.05 mm and 5 mm, depending on the reaction vessels contemplated. Preferred as solid carrier material are polystyrene cross-linked with 1-2% divinylbenzene, and polystyrene-polyethylene glycol copolymer, but any other type of resin suitable for solid phase synthesis may be used.

"Building blocks" are small organic compounds that can be attached in a chemical reaction to a functionalized solid carrier, and are preferably such building blocks that are conventionally used in solid phase synthesis. Examples of such building blocks are nucleotides, carbohydrates and amino acids, but also other di-and polyfunctional compounds attachable to a functionalized solid carrier. Preferred building blocks of the invention are carbohydrates and amino acids, in particular amino acids. Amino acids may be naturally occurring (L)-α-amino acids such as Ala, Arg, Asn, Asp, Cys, GIn, GIu, GIy, His, He, Leu, Lys, Met, Phe, Pro, Ser, Thr, Trp, Tyr, and VaI, but also corresponding (D)- α-amino acids or other (e.g. homologous) α-amino acids, β-amino acids, γ-amino acids, δ- amino acids, ε-amino acids, di-, tri-, tetra- and penta-amino acids, and the like.

"Reacting with a building block" comprises all steps required in solid phase synthesis to attach the building block to the growing sequence of building blocks on a bead. Such reaction steps and reaction conditions are well known in the art and depend on the type of bead, the type of building block, and the length of sequence on the bead. Particular reaction steps that may be necessary are activating a functional group on the bead or the growing sequence on the bead by mixing with an activating reagent, mixing beads with building blocks in protected or partially protected form, shaking or stirring reactions mixtures containing beads and reagents in particular solvents for a predetermined amount of time at a particular reaction temperature or a reaction temperature gradient, filtering or otherwise separating beads from reagents and solvents, washing beads by shaking or stirring in particular solvents or wash solutions, cleaving or otherwise manipulating protecting groups on building blocks connected to the beads using particular reagents, and the like.

"Macromolecule" refers to a compound synthesized by a series of steps wherein building blocks are attached in sequence to a growing structure. Preferred examples of macromolecules are those composed of the preferred building blocks defined above, and are e.g. polynucleotides, carbohydrate chains or polypeptides, linear or branched, or also combinations thereof such as polypeptides bearing carbohydrate groups, and polypeptides comprising artificial building blocks.

"Variable position" refers to a substructure of a macromolecule in a combinatorial library constituted of one copy or multiple copies of a building block attached to the growing macromolecule chain during a defined step of its synthesis, wherein a mixture of different building blocks are used for that step creating such substructure (position) during library synthesis.

"Conserved position" refers to a substructure of a macromolecule in a combinatorial library constituted of one copy or multiple copies of a building block attached to the growing macromolecule chain during a defined step of its synthesis, wherein a single building block is used for that step creating such substructure (position) during library synthesis.

"Copy number" refers to the number of copies of a building block attached at a given "position" of a macromolecule in a combinatorial library during its synthesis.

For the group split-and-mix procedure of the invention beads are split in equal portions, e.g. n or m portions. The number of portions (including n and m) may be any number between and including 2 and 20, preferably between 2 and 6, for example 3 or 4 portions. The number of portions may be the same or different in any step of "mixing / equally spitting / reacting with building block" as defined hereinbefore. "Groups of building blocks" contain corresponding numbers of building blocks, i.e. between and including 2 and 20, preferably between 2 and 6, for example 3 or 4 building blocks, creating a

"variable position" as defined hereinbefore. One building block within a group of building blocks may be absent, i.e. the actual number of building blocks in a group may be one lower than indicated above, and as a consequence one portion of beads may not react with a building block in a particular step of "mixing / equally spitting / reacting with building block". At any stage of the group slit-and mix procedure as defined hereinbefore the beads may be reacted with a single building block in place of a group of building blocks, creating a "conserved position".

Analysis for desired properties and selection of members of a combinatorial library based on such an analysis is well known in the art, and is applied accordingly. The analysis may be performed directly on the bead carrying a member of the combinatorial library, or also on the macromolecules cleaved from the bead, i.e. the sequence of building blocks after cleavage, taking into account that all macromolecules cleaved from a bead have an identical structure.

The present invention is based on the following two basic conditions:

1) Use of an analytical method to cleave the selected molecules of the library back into their constitutive building blocks, and identifying and quantifying these building blocks. In the case of peptides, a method of choice would be cleaving the peptides by acidic hydrolysis and analyzing individual amino acids by HPLC of the corresponding phenylisothiocyanate (PITC) derivatives.

2) The building blocks should be partitioned into separate groups in such a way that when a building block and its copy number are identified by analysis after cleavage, the position of the building block in the sequence is unequivocally deducible (within the accuracy of the quantification possible by the analysis method chosen).

The invention describes a method of distributing a given series of building blocks to a series of variable positions in a macromolecule synthesis such that the structure of the macromolecule can be deduced from knowing the synthesis sequence, the list of building blocks used at each position, and a quantitative analysis of the building blocks that have been used at the variable positions. In particular, the invention describes such a method wherein any building block used for variable positions can be used at more than one variable position, in particular 2 or 3 positions, during the synthesis. The invention further concerns combinatorial libraries obtainable by the method of the invention, in particular a combinatorial library of dendritic compounds as defined hereinbelow. Such a combinatorial library is e.g. a collection of compounds of the formula (((A8-A7)₂B-A6-A5)₂B-A4~A3)₂B-A2-A1, wherein Al and A2, A3 and A4, A5 and A6, and A7 and A8 are each a group of different building blocks as defined hereinbefore, and B is a single building block as defined hereinbefore with two functional groups for reaction with a group of building blocks. For the sake of clarity, "different" means that the building blocks in position Al and A2 are different, the building blocks in position A3 and A4 are different, the building blocks in position A5 and A6 are different, and the building blocks in position A7 and A8 are different, but a building block in Al (or A2) may be identical with a building block in A3 (or A4); and/or identical with a building block in A5 (or A6); and/or identical with a building block in A7 (or A8); likewise a building block in A3 (or A4) may be identical with a building block in A5 (or A6); and/or identical with a building block in A7 (or A8); and likewise a building block in A5 (or A6) may be identical with a building block in A7 (or A8). Such building blocks are e.g. amino acids, for example naturally occurring α-amino acids. B is e.g. a diamino acid.

The invention will now be explained in detail for the preparation of a peptide library using the standard split-and-mix protocol, in particular a ramified, "dendritic" peptide library, whereby the quantitative analysis is carried out by total acidic hydrolysis of the peptide, followed by analysis of the amino acid building blocks by HPLC.

The combinatorial split-and-mix and group split-and-mix syntheses described below comprise variable steps creating a variable position, in which different building blocks are used, and conserved steps creating a conserved position, where common operations are carried out and/or common building blocks are attached to the molecule undergoing synthesis. The description focuses on the variable steps where different building blocks are used.

State of the art combinatorial split-and-mix synthesis applies to multi-step syntheses in which different building blocks are bound together to form the target molecule. For example, amino acid building blocks can be used to form a linear peptide in a solid- supported synthesis. If 4 different amino acids are used in each of 3 successive coupling steps in 4 different reaction vessels, the simple linear synthesis gives only 4 products. In the combinatorial split-and-mix protocol, the four portions of solid support in each of the four reaction vessels are mixed together after each coupling step and re-separated into four equal portions for the next coupling steps. This protocol delivers 64 products for the series of 3 steps, representing all possible sequences. In Figure 1 the split-and-mix (left) vs. the classical synthesis protocol (right) is shown. Building blocks A1-A4 are added in vials 1 to 4. Mixing the synthesis beads between operations allows for the synthesis of all possible 64 molecules. In the classical protocol without mixing and splitting, only four different molecules are obtained. Since each bead of the polymer support is present in only one of the four reaction vessels during the coupling steps, each bead contains only a single product.

The split-and-mix protocol thus allows the rapid preparation of many different compounds in a separated manner. However the structure of each compound on each synthesis bead is not known since the beads are not identified during the synthesis but simply mixed randomly. In particular, analyzing the composition of the bead in terms of building block does not give the structure unequivocally. For example, if a composition analysis indicates one copy of each Al, A2 and A3 as building blocks for a product on one of the beads, there are six different possible structures for this product: A1-A2-A3, A1-A3-A2, A2-A3- Al, A2-A1-A3, A3-A1-A2, A3-A2-A1.

In the combinatorial group split-and-mix synthesis, which is the object of the present invention, different unique building blocks (for example amino acids or carbohydrates) at each variable step can be used. In one embodiment of the invention as shown in Figure 2 different building blocks from groups A, B or C are used at each of the three steps 1, 2 and 3. In the protocol shown in Figure 2, building blocks A1-A4, then B1-B4 and finally Cl- C4 are coupled in reactions vials 1-4 respectively. Combinatorial group split-and-mix leads again to the formation of 64 different molecules, while the classical protocol gives only four molecules. With this procedure the structure of a product on a single bead is obtained by analyzing the building block composition. For example, if a composition analysis gives one copy of Al, one copy of B2 and one copy of C3, the structure of the product must be A1-B2-C3 as based on the original design. It is the only product containing building blocks Al, B2 and C3 in the library. In a further embodiment using the group split-and-mix synthesis the preparation of length- variable libraries by using a "no-building-block" variable at each variable step as shown in Figure 3 can be performed (three building blocks per group and an omission). In the example shown in Figure 3 the synthesis with three groups of three building blocks A, B and C results in a combinatorial group split-and-mix library of 63 different compounds which can be identified by composition analysis as described above.

In a further particular embodiment the invention is directed to the preparation of structures in which certain variable positions occur in multiple copies, meaning that one of the coupling steps attaches precisely one, two, or more copies of a building block to the molecule. Such a synthesis results in ramified molecules, i.e. dendrimers. Table 1 lists the combinations of copy numbers acceptable according to the invention, i.e. the combinations which still allow to deduce the correct structure of the sequence from the cleaved building blocks. One building block of different groups of building blocks can be used for the synthesis at several different variable positions (with different copy numbers), provided that all possible sum combinations of copy numbers of these different variable positions are different from the copy number used in the synthesis. The combinations allowed according to this system are shown in the left part of Table 1, and the combinations where the system does not apply is shown in the right part. If, for example, a position has one copy number, and another one two copy numbers, the same group of building blocks can be used at both positions, because the quantification of the building block will indicate at which position it was attached.

Table 1. Combinations of copy numbers allowed for variable positions using the same group of building blocks.

Allowed combinations: Combinations not allowed: copy no. sums copy no. sums

1, 2 1, 2, 2+1 1, 2, 3 1, 2, 1+2 = 3^υ, 3+1, 3+2, 3+2+1

1, 3 1, 3, 3+1 2, 3, 5 2, 3, 2+3 = 5²⁾, 5+2, 5+3, 5+2+3

1, 4 1, 4, 4+1

2, 3 2, 3, 3+2

2, 5 2, 5, 5+2

1, 2, 4 1, 2, 2+1, 4, 4+1, 4+2, 4+2+1

1, 2, 5 1, 2, 2+1, 5, 5+1, 5+2, 5+2+1

^!) If quantification of building blocks after cleavage indicates 3 building blocks, one cannot decide whether these stem from the position with three copies or from both the positions with one and with two copies.

²⁾ If quantification of building blocks after cleavage indicates 5 building blocks, one cannot decide whether these stem from the position with five copies or from both the positions with two and with three copies.

The key of the invention is that the analysis by quantification of building blocks allows to know how much of a building block is present, and one can assign it then to two different positions with different copy numbers, for example 1 and 2 copies, because all possible sums are unique. This is useful for dendritic sequences. One given building block may be used more than once, in particular 2 times or 3 times, if the copy number of the corresponding positions are as described in Table 1.

Using each building block 2 times allows one to construct a large library with only few building blocks, which is not possible if each building block is assigned to one position only. For example, 16 building blocks assigned once to 8 positions gives 2 building blocks per position, and a 2⁸ = 256 member library. When each building block is used twice, 4 building blocks are availbale per position, which gives a 4 = 65536 member library. The general protocol for applying group split-and-mix synthesis for making a library of molecules is described in the Flowchart. The library and synthetic schemes are designed for a group split-and-mix library synthesis (steps 1, 2 and 3 in the Flowchart). The split- and-mix synthesis is then carried out (step 4 in the Flowchart) and the library is tested for active molecules (step 5 in the Flowchart). The sample of an active molecule, which is either the polymer bead with the molecule attached to it, or a solution containing this molecule, is then analyzed for identification and quantification of the building blocks (step 6 and 7 in the Flowchart). Knowledge of the building block composition then allows to deduce the actual structure of the molecule by taking the library design into consideration (step 8 in the Flowchart). The known molecule can then be resynthesized in pure form and its activity confirmed and investigated in more detail.

Flowchart: Discovering active molecules using a group split-and-mix combinatorial library.

In de novo protein design one attempts to create artificial proteins with defined structure and function, usually with the help of trial-and-error procedures that scan a large number of possible amino acid sequences. The approach to de novo protein design exemplified hereinbelow is based on peptide dendrimers, and represents a particular embodiment of the present invention. Dendrimers are tree-like structures that adopt a globular or disk-shaped structure as a consequence of topology rather than folding. Since peptide dendrimers are synthesized on solid support, it should be possible to apply the principles of combinatorial peptide synthesis, which have been exploited successfully for peptide-based and small molecule catalysts. To achieve this goal, a dendrimeric architecture containing eight variable positions connected by three successive branching diamino acid units is chosen (Figure 4).

It is known that peptide dendrimers bearing surface histidine residues catalyze ester hydrolysis in water. In the present example a combinatorial approach to peptide dendrimers based on group split-and-mix synthesis and on-bead screening is used. The method is exemplified by the discovery of catalytic and binding peptide dendrimers in a 65'536-member library. A split-and-mix library of peptide dendrimers is designed by distributing sixteen proteinogenic amino acids into four groups of four amino acids each. The resulting 65'536-member library requires splitting in four portions at each variable sequence step, whereby each group of four amino acids would be used at two positions in different branches of the dendrimer. The single-bead sequencing problem is solved by taking advantage of the dendrimeric structure in which amino acids are present in one, two, four or eight copies depending on their placement in the different branches. Since each amino acid is present at most at two positions in different branches, the sequence can be unambiguously assigned by quantitative amino acid analysis of the dendrimer.

The invention is now further illustrated in the following examples without limiting it to these particular conditions:

Example 1: Synthesis of a peptide dendrimer library of a structure as shown in Figure 4 using the group split-and-mix protocol.

A combinatorial split-and-mix library of 65'536 different peptide dendrimers is synthesized on solid support with the structure (((AcNH-A8-A7)₂B-A6-A5)₂B-A4- A3)₂B-A2-Al-spacer-support, where Al to A8 denote variable α-amino acid positions, the branching unit B is (5)-2,3-diaminopropanoic acid, the spacer is 6-aminohexanoyl- glycine, and the support is Tentagel bearing the photolabile 4-(4-hydroxymethyl-2- methoxy-5-nitrophenoxy)butanoic acid at 0.28 mmol/g loading. The peptide dendrimer library is prepared starting with a 400 mg resin batch, which ensures that each sequence would be present in approximately 15 beads. Aromatic, hydrophobic, positively charged, negatively charged and small amino acids are distributed evenly among the inner positions A1-A4 and the outer positions A5-A8. Histidine is placed in the outermost layer (A6 and A8) since surface placement of this residue is known to favor catalysis. The sequence starts with an 6-aminohexanoyl-glycine spacer to minimize interactions with the solid support. All coupling steps are quantitative, as indicated by the negative TNBS-staining test. Side-chain protecting groups are removed with trifluoroacetic acid after Fmoc deprotection, resulting in a functional dendrimer library on beads. For hydrolysis studies of Example 2, N-termini were acetylated.

The following groups of amino acids are chosen: Al and A3: Asp, Ala, Thr, VaI A2 and A4: Lys, Phe, He, Tyr A5 and A7: Pro, Ser, Arg, Leu A6 and A8: His, GIu, Trp, GIy

Coupling of the Fmoc protected amino acids: The resin was washed and swelled inside the reactor (e.g. polypropylene syringes from which the resin beads could be easily washed out for the mixing operation) with dichloromethane (DCM, 2x5 ml) and dimethylformamide (DMF, 1x5 ml). The Tentagel resin (0.23 mmol/g) was acylated with 2.5 equivalents of JV- Fmoc amino acid in the presence of 2.5 equivalents of benzotriazol-1-yl-oxy-tris- (dimethylamino)-phosphoniumhexafluorophosphate (BOP, Novabiochem) and 6 equivalents of JV,JV'-diisopropylethylamine (DIEA) in DMF. After 30 min (first generation), 1 h (second generation), 2 h (third generation) the resin was washed (3><each) with DMF, DCM and MeOH and controlled with the trinitrobenzenesulfonic acid (TNBS) test. Double coupling and longer coupling times were necessary for completion toward the end of the synthesis, in particular after addition of the third branching diamino acid.

Fmoc-amino acids were purchased from Novabiochem (Fmoc-Cys(Trt)-OH), Bachem (Fmoc-His(l -Boc)-OH, Fmoc-Gly-OH, Fmoc-DAP(Fmoc)-OH) or Senn Chemicals (Fmoc-Ala-OH, Fmoc-Ile-OH, Fmoc-Leu-OH, Fmoc-Val-OH, Fmoc-Pro-OH, Fmoc-Phe- OH, Fmoc-Arg(Pbf)-OH, Fmoc-Trp(Boc)-OH, Fmoc-Lys(Boc)-OH, Fmoc-Asρ(OtBu)- OH, Fmoc-Glu(OtBu)-OH, Fmoc-Tyr(tBu)-OH, Fmoc-Thr(tBu)-OH, Fmoc-Ser(tBu)-OH).

Cleavage of Fmoc protecting group: The Fmoc protecting group was removed with 5 ml of a solution of piperidine in DMF (1 :4) for 10 min. After filtration, the procedure was repeated and then washed (3x each) with DMF, DCM and MeOH.

Capping of the N-terminus. At the end of the synthesis, the resin was acetylated with a solution of acetic anhydride in DCM (1:1) for 30 min, washed with DCM, DMF and methanol. The resin was dried under vacuum and stored at -20°C.

TFA cleavage of protecting group. The cleavage was carried out using trifluoroacetic acid / 1, 2-ethanedithiol / H₂O/ triisopropylsilane 94/2/2/2 solution for 3 h. Then the library was washed with water, methanol and DMF, dried under vacuum and stored at -2O⁰C.

Example 2: Screening of the dendrimer library for catalytic hydrolysis of the fluorogenic ester 8-butyryloxypyrene-l , 3, 6-trisulfonate.

Beads containing N-acetylated dendrimers of Example 1 are soaked in 20 mM aqueous Bis-Tris buffer containing 80 μM 8-butyryloxypyrene-l, 3, 6-trisulfonate at 25°C for 30 min. Hits are green-fluorescent under UV light at 356 nm. Approximately 60% of all beads show a low level fluorescence, among which very few beads (10 beads/40 mg resin) show an intense green fluorescence indicating ester hydrolysis. These active beads are picked, washed thoroughly with buffer and water, and subjected to total hydrolysis with hydrochloric acid and chromatographic quantification of the amino acids constituents by HPLC. Analytical HPLC columns: Chromolith Performance RP- 18e, 0.46 x 10 cm, flow 3 mLmin^"1; detection by UV at 214 nm, solvent systems: A = 0.1% trifluoracetic acid (TFA) in H₂O; B = 40% H₂O / 60% CH₃CN, 0.1% TFA. 10 sequences were analyzed (Table 2). In all cases the dendrimer sequence could be unambiguously assigned from the amino acid integration data. Six out often dendrimers contained histidine, known to be a catalytic residue, at position A⁸ (60 % vs. 25% random) and three of these contained histidine at positions A⁸ and A⁶ (30 % vs. 6.1 % random). The four other sequences did not contain any histidine residues. Sequencing of beads not selected for the fluorogenic reaction gave very different sequences with no over-representation of either histidine or basic amino acids, confirming that the sequences observed in fluorescent beads in the assay with the fluorogenic ester reflected an activity selection and not a possible composition bias that might have occurred during synthesis. The amino acid composition analysis proved very reliable for sequence determination. Additional mass spectroscopy analysis of the dendrimers after photodeprotection of the photolabile linker was therefore not required.

Two consensus sequences (denoted CIl and C 12) and two original sequences (8 and 10) are re-synthesized as single sequences as above and purified by preparative HPLC. Preparative RP-HPLC was performed with HPLC-grade acetonitrile and MiIIiQ deionized water in a Waters prepak cartridge 500 g (RP-Cl 8 20 mm, 300 A pore size) installed on a Waters Prep LC4000 system from Millipore (flow rate 100 mLmin^"1, gradient 1% min^"1 CH₃CN). HPLC conditions for dendrimers 8, 10, CI l, C12: A/B = 85/15 to A/B = 60/40 in 35 min. Dendrimer 10, which did not contain a histidine residue, was not active and represented a false positive. All three histidine-containing dendrimers 8, Cl 1 and C12 showed enzyme-like catalysis with the substrate, with substrate binding ZM ~ 0.1 mM and catalytic rate constant &_cat ~ 0.1 min^"1 and specific rate acceleration k_cjk_uncai ~ 2¹OOO- lO'OOO. Catalysis was proportional to dendrimer concentration, and showed multiple turnovers. There was no detectable product inhibition or catalyst inactivation upon reaction, as evidenced by the fact that addition of fresh substrate after completion of a dendrimer catalyzed reaction resulted in further hydrolysis of the substrate at the same initial rate as in the first reaction.

Dendrimer 8: From Tentagel Rink amide resin (160 mg, 0.25 mmol/g), D8 was obtained as colorless foamy solid after preparative HPLC purification (6.4 mg, 12.8%); RP-HPLC: t_κ = 22.4 min; MS (ES+): calcd. for C₁₇₈H₂₈₃N₇₅O₅₅: 4351.16, found: 4352.50 Dendrimer 10: From Tentagel Rink amide resin (160 mg, 0.25 mmol/g), DlO was obtained as colorless foamy solid after preparative HPLC purification (3.8 mg, 7.6%); RP-HPLC: tR = 23.8 min; MS (ES+): calcd. for C₁₇₀H₂₉₅N₇₁O₄₇: 4083.28, found: 4085.63 Dendrimer CIl: From Tentagel Rink amide resin (160 mg, 0.25 mmol/g), the dendrimer CI l was obtained as colorless foamy solid after preparative HPLC purification (4.2 mg, 8.4%); RP-HPLC: t_κ = 22.5 min; MS (ES+): calcd. for C₁₇₀H₂₇₃N₆₃O₅₉: 4130.95, found: 4132.13 Dendrimer Cl 2: From Tentagel Rink amide resin (160 mg, 0.25 mmol/g), the dendrimer C12 was obtained as colorless foamy solid after preparative HPLC purification (19.7 mg, 39.4%); RP-HPLC: t_R = 21.9 min; MS (ES+): calcd. for C₁₉₉H₃₀₅N₇₁O₅₆: 4585.32, found: 4586.88

Table 2. Peptide dendrimer sequences identified by amino acid analysis of active beads from the combinatorial library, and the corresponding consensus sequences.

No. A⁸ A⁷ A⁶ A⁵ A⁴ A³ A² A¹

1 W S G R K V I A

2 H L H S Y A I D

Hits for hydrolysis of 8-butyryl- 3 H L G L Y T I V oxypyrene- 1 ,3 ,6-trisulfonate 4 H P G P K T I A

(Example 2) ^a) 5 E R G S I V I V

6 G R W R I V I A

7 H S H L F A F D

8 H S G R / A I V

9 H S H P K V F V

10 G R G P I V I V consensus sequences CIl H S H L K V I V

C12 H S G S / V I V

13 W L H S / A K A

14 E P G R Y T Y D

Hits for binding to vitamin B₁₂ 15 W P E S Y A Y D

(Example 3) ^b) 16 G P W P Y V K V

17 G R E R / T I D

18 H L G R K V K D

19 W R E S / V I V consensus sequences C20 W P G R Y V Y D

C21 G R E S / T K V

^a) All N-termini N-acetylated. ^b^ N-termini are free amine. Example 3: Screening of the dendrimer library for binding to vitamin Bn

Beads containing dendrimers of Example 1 are equilibrated in aqueous phosphate-buffered saline (PBS₅ 10 mM phosphate, 160 mM NaCl, pH 7.4) containing 400 μM cyano- cobalamin (vitamin B₁₂) for 30 min., followed by washing with PBS and water. Hits are bright orange. These active beads are picked, washed thoroughly with PBS and water, and subjected to total hydrolysis with hydrochloric acid and chromatographic quantification of the amino acids constituents by HPLC. 10 sequences were analyzed (Table 2).

Two consensus sequences (denoted C20 and C21) and two original sequences (14 and 19) are re-synthesized as single sequences and purified by preparative HPLC as described in Example 2. Screening for binding to vitamin B₁₂ confirms selective binding to vitamin B₁₂ as cyano or aquo complex, but no binding affinity to cobinamide.

Claims

1. A method for preparing combinatorial libraries, characterized in that beads are equally split in n portions, each portion reacted with one building block of a first group of n building blocks, reacted beads thoroughly mixed and equally split in m portions, each portion reacted with one building block of a second group of m building blocks, the steps of thoroughly mixing, equally splitting in portions and reacting with one building block of a further group of building blocks repeated for 1 to 20 times; the resulting beads with the sequence of building blocks, or the sequence of building blocks after cleavage from the beads, analyzed for desired properties and selected based on the desired properties; selected beads with the sequence of building blocks, or the selected cleaved sequence of building blocks, cleaved into individual building blocks, the amount and type of building blocks determined, and the sequence of building blocks deduced from the amount and type of building blocks determined; wherein one building block in a group of building blocks may be absent; and wherein the building blocks in each group of building blocks are so chosen as to allow deducing the sequence of building blocks from the determination of the amount and type of building blocks after cleavage.

2. The method of claim 1 further comprising reacting all beads at any stage with a single building block.

3. The method of claim 1 or 2 wherein building blocks are amino acids or carbohydrates.

4. The method of claim 2 wherein a single building block is a building block introducing 2, 3, 4 or 5 reactive centers.

5. The method of any of claims 1 to 4 wherein the building blocks in the groups of building blocks are amino acids.

6. The method of claim 4 or 5 wherein the single building block is a di-, tri-, tetra- or penta-amino acid.

7. The method of any of claims 1 to 6 wherein the mixing and splitting is repeated for 1 to 10 times.

8. The method of any of claims 1 to 7 wherein n and m is 4.

9. The method of any of claims 1 to 8 wherein one of the building blocks in at least one group of building blocks is absent.

10. The method of any of claims 1 to 9 wherein the building blocks in each group of building blocks are different.

11. The method of any of claims 4 to 9 wherein the single building block is a building block introducing 2, 3, 4 or 5 reactive centers, and building blocks in groups of building blocks before and after the single building block(s) may be identical, with the proviso that building blocks in groups of building blocks may not be present in combinations of 1, 2 and 3 copy numbers or of 2, 3 and 5 copy numbers.

12. The method of any of claims 4 to 9 wherein the single building block is a building block introducing 2 reactive centers and building blocks in groups of building blocks before and after the single building block(s) may be identical.

13. A combinatorial library of dendritic compounds of the formula (((A8-A7)₂B-A6-A5)₂B-A4-A3)₂B-A2-A1, wherein Al and A2, A3 and A4, A5 and A6, and A7 and A8 are each a group of different building blocks, and B is a single building block.

14. The combinatorial library of claim 13 wherein A is an α-amino acid and B is a diamino acid.

15. The combinatorial library of claim 14 wherein A is a naturally occurring α-amino acid and B is 2,3-diaminopropionic acid.