High throughput synthesis of peptides and peptidomimetics

Abstract

Peptide synthesis has been developed into one of the most efficient synthetic procedures in organic chemistry. The problems of orthogonal functional group protection and amide bond formation without racemization have been developed in a number of ingenious strategies. Optimization, in particular, has been achieved in stepwise solid phase synthesis. This in turn made possible the development of combinatorial synthesis allowing the synthesis of millions of peptide compounds of high purity in a few days. A variety of methodologies and strategies have been developed and continue to be developed to determine structures and to evaluate peptides and peptidomimetics. The development of methods for solid phase synthesis of a variety of organic and inorganic structures using similar strategies as in peptide synthesis are being vigorously pursued. However, existing instrumentation and technology is not sufficient to cover current demands for peptides, and thus new approaches and technologies for cost-effective synthesis of peptide arrays are needed.

INTRODUCTION

Peptides and proteins constitute the most ubiquitous and important natural products, and are central mediators of bioactivity for all living systems from single cell organisms to highly complex vertebrates. Despite their central importance in all living systems, their organic synthesis in the laboratory developed more slowly than many other natural products due to both scientific and sociological aspects. Scientifically the most important was the difficulty of synthesizing amide bonds of α-amino acids without racemization. In Nature, 19 (20) α-amino acids of the L-configuration and glycine are utilized to construct polypeptides proteins (we ignore peptides which are synthesized by non-chromosomal mechanisms). Thus the synthesis of even a rather small peptide of 4 or 5 amino acid residues can quickly lead to a complex mixture of diastereoisomers that is very difficult to separate when racemizations occurs. In addition most a-amino acid possesses different side chain functional groups, and generally these have to be protected to avoid a wide variety of side reactions. The racemization problem was partially solved when it was realized that whereas N^α-acyl protected α-amino acids readily racemized during activation of the carboxyl group for amide bond formation, the use of N^α-urethane protecting groups (e.g. benzyloxycarbonyl, Cbz,Z) greatly reduced or eliminated racemization. As for the side chain groups, protecting groups developed for various functional groups (OH, SH, CO₂H, NH₂, etc.) in many different areas of synthetic organic chemistry could be incorporated in peptide synthesis (For early background see the three volume treatise of Greestein and Winitz (1), and the book by Bodanszky, Klausner and Ondetti (2). The culmination of this early work was the virtually simultaneous reports of duVigneaud and co-workers on the structure and total synthesis of the cyclic nonapeptide hormone (and neurotransmitter) oxytocin (3). This monumental accomplishment dramatically changed the views of chemists of what could be synthesized, and a new understanding of the central roles of peptides and proteins in biology and medicine, and was recognized almost immediately with the Nobel Prize in 1955.

PEPTIDE SYNTHESIS STRATEGIES

Solution Phase Synthesis

The vast majority of synthetic methods that have been developed by organic chemists have been done in the solution phase. It was soon realized that formation of the peptide (amide) bond was best accomplished by “activation” of the carboxyl group to make it an excellent electrophile that could be reacted with the α-amino group of an amino acid or of a peptide. It also was quickly determined that activation of the carboxyl group of a peptide lead to greater racemization in general than activation of a N^α-urethane protected α-amino acid. Hence the major synthetic strategy developed was to synthesize peptides from the C-terminus to the N-terminus one amino acid at a time, or when peptide fragment condensation was necessary for the total synthesis of a peptide, to utilize C-terminal amino acid residues for activations that generally were less easily racemized (e.g. Gly, Pro, Ala, Phe, etc.). The most widely used strategies are built around the chemical properties of the N^α protecting group which generally needs to be removed without removal of side chain or C-terminal protecting groups. Three major strategies, therefore, have been developed for peptide synthesis. The first developed around the benzyloxycarbonyl (Cbz) amino protecting group which could be removed either by hydrogenation or by strong acic (HBr/HOAc, etc.). This strategy, though highly successful in the original synthesis of oxytocin, is not in common use today.

A second widely used strategy is the Boc strategy after the t-butyloxycarbonyl group introduced by Carpino et al. (4). The group is readily cleaved in minutes by acids such as trifluoroacetic acid (TFA). In this strategy, side chain protecting groups for the amino acids need to be reasonably stable to such acid conditions as must the C-terminal protecting group. It generally has been found that benzyl or substituted benzyl groups are the best protecting groups for most of the side chain protecting groups in native peptides. Thus this strategy in peptide synthesis often is referred to as the Boc/benzyl strategy (5). A third widely used strategy is the Fmoc procedure using N^α-fluorenylmethyloxycarbonyl protecting group (6). The group is readily cleaved with organic bases such as secondary amines and is generally stable to acids. Side chain functional group protection utilizes readily and labile protecting groups such as t-butyl and Boc. In Table 1 we list the most widely used protecting groups (7).

Table 1.

Protecting Groups for Peptide Synthesis

Functional Group	Protecting Group
	Boc/benzyl	Fmoc/lost-butyl
Hydroxyl (Sor, Thr, Tyr)
Thiol (Cys)
Amino (Lys)
Carboxyl (Asp, Glu)
Guanidino (Arg)
Amido (Asn, Gln)
Imidazolo (His)

Solid Phase Synthesis

Arguably the most important advance in the synthesis of peptides and proteins indeed in all of synthetic chemistry has been the development of solid phase peptide chemistry (the Merrifield method (8]). This development was at first met with skepticism, indeed hostility, by virtually all synthetic peptide chemists and organic chemists, but this only stimulated the Merrifield group and alumni, and a few other groups (Manning, Rivier, Hruby and others) to optimize both the synthetic chemistry and purification methods. N^α-Boc removal and coupling occurred with greater than 99-100 percent yield (9-11) (simultaneous developments in solid phase nucleic acid synthesis will not be discussed here). At the same time the development of modern separation methods (HPLC, CZE, gel filtration, and partition chromatography) greatly accelerated the process as it became possible to separate complex mixtures of minor byproducts such as diastereoisomers and deletion peptides from the desired products. Despite these and other advances in detailed analytical analysis of peptide structure, analysis by highly sensitive sequencing methods and a variety of mass spectrometry methods, most synthetic chemists did not take advantage of these new breakthroughs in synthetic chemistry either because they were unaware of these developments or didn’t appreciate their significance. It wasn’t until the development of solid phase combinatorial chemistry (12-14) that most synthetic chemists began to make use of the advantages of solid phase chemistry in a wide variety of synthetic strategies.

Combinatorial Chemistry

Though often utilized in the same context, combinatorial chemistry and parallel synthesis are two different things. Combinatorial chemistry is the application of mathematical statistics to chemical synthesis to produce “libraries” of compounds according to the statistics of the chemical reactions used. Parallel synthesis is running a series of reactions (usually related) in parallel, often in a format conducive to rapid chemical or biological analysis (e.g. 96 well “plates”).

In principle one can prepare combinatorial libraries such as to maximize to number of different compounds produced (e.g. Poisson statistics) or to bias the statistics in various ways (e.g. Gaussian statistics). Though the genetic code has a built in bias in terms of the incorporation of amino acids into proteins (further modified by evolution), most combinatorial chemistry has concentrated on maximizing the number of compounds obtained even when a “biased” library has been produced. From this perspective the application of combinatorial mathematics to combinatorial chemistry and its applicability to the biological and medical sciences is still largely unexplored. Combinatorial chemistry has been utilized in primarily two different modes: a) to produce diverse mixtures; or b) to produce single molecular species specially segregated (one bead, one peptide (compound)) on a single bead. In the former case structure identification requires some deconvolution strategy, and assays are run on mixtures. In the latter use, assays can be run in mixtures of beads or in single beads with detection of activity generally dependent on the spectroscopic or radioactivity readout at the specific bead level. Structural analysis can then be determined at the single bead level. Both of these approaches require highly sensitive analytical and bioassay methods. Fortunately these methods have become available during the past 20 years, and increasingly sensitive methods of analysis continue to be developed both for structural elucidation and for detection of binding and a wide variety of biochemical and biophysical “responses.” Single beads contain approximately 100 pmol of peptide, although larger beads are available. This amount is at the detection limit that can be reliably sequenced by Edman degradation. However, the sensitivity of FT-MS or MALDI allows to detect and fragment (read the peptide sequence) quantities more than two orders of magnitude lower. This ability to detect low femtamolar amounts permitted the development of encoding strategies. The binding structure, located at the surface of the bead, may be encoded with an internal code, buried inside the bead and inaccessible to proteins or receptors. The code, for example of unnatural peptides, defines a sequence of building blocks used for construction of the “binding” structure, similar as a natural DNA-protein coding system. The code allows one to “read” isobaric and structurally complicated building blocks which may be thus assembled. These and many other aspects of combinatorial chemistry have been often reviewed and several books have appeared on the subject (e.g. 15-18).

Parallel Synthesis

Parallel synthesis, as opposed to random library synthesis (e.g.OBOC), is characterized by defined position of a compound in a library. Obviously, the solid support allows easy separation of products and automatization of the whole process. The support has to be distributed in reaction vessels, and reagents are delivered into the individual reactors. This can be achieved by simple distribution of the resin, for example, in 96-well polypropylene plate. The liquid reactant or solvent can be removed from the top, by needle aspiration, or by filtration from the bottom of the plate. The liquid aspiration requires well sedimented resin beads and slow “surface” suction of liquid. The needle has to be immersed into the plate to the level of liquid meniscus then slowly remove only a top part of liquid without disturbing the sedimented resin. This process has been automated (Trega), but none of these instruments are available today. Besides aspiration, centrifugation can effectively remove liquid from the top of plate. The filtration plate approach, removing liquid by filtration through filter in the bottom of plate, is almost exclusively the technology of choice for commercially available plate synthesizers (e. g. Zinnser Analytical, JKem, Advanced Chem Tech). Parallel synthesis is not limited to plate resins. A flexible approach was developed by Houghten (19). This so called “tea-bag” technology utilizes distribution of the resin in small packets (polypropylene mesh bags). The synthesis is carried out simultaneously on many resin tea-bags without contamination. The tea-bag approach pioneered automated sorting technology. The tea-bags were redesigned to permeable microreactor (NanoKans, DPI) equipped with a radiofrequency tag or a 2D bar code. Sorting technology substantially improves efficiency of parallel synthesis by decreasing the number of reaction vessels needed in synthesis. NanoKans are sorted and pooled before each reaction and thus the number of reactions defines the number of vessels. This approach resembles one-bead-one compound strategy, except that the compound structure information is tracked through synthesis. Another technique of parallel resin processing is removal of liquid from the plate (or other reaction vessel) by centrifugation. The plates with the resin slurry are rotated in centrifuges tilted slightly towards the center axis of the centrifuge. This creates a packet were the resin is collected. The possibility of well-to-well contamination was disproved either theoretically or by LC-MS analysis of libraries. The centrifuge synthesizers also are available in oligonucleotide versions (Illumina and Spyder Instruments).

Alternatively, other supports than the Merrifield resin have been developed such as paper, cotton, rigid polymer pins or lanterns grafted with hydrophilic polymers, and glass chips. The major disadvantage of the Merrifield polymer lies in the gel-like properties of the resin, excellent for synthesis, but mechanically fragile. The polystyrene support, despite many attempts to prepare larger beads, can carry out solid-phase synthesis only in the form of small ca 100 micrometer beads. Alternative support research was pioneered by Frank & Doering (20), who first published parallel nucleotide synthesis on paper disks, and later applied the same strategy to peptide synthesis. Currently, the SPOT technique utilizes paper as a support for high-throughput peptide synthesis. The reagents are spotted on the circle compartments of paper which allows soaking in limited space. The excess of reactants is removed by immersion in washing reservoirs. Paper sheets (ca 800 peptides per sheet) may be tailored into small compartments and individual peptides cleaved off the paper, but more often the paper sheets are screened directly and the compound is tracked by the position of spot on the sheet. An automated SPOT synthesizer is commercially available (Intavis AG). Peptide array synthesis on pins, developed by Geysen et al. (21), utilizes rigid polyethylene rods or crowns grafted with gel-like polymers, typically a soft polystyrene layer, where the synthesis takes place. The rigid scaffold allows manufacturing of reactors of any shape and size; pin combs are arranged in a grip mapping the plate, or in a small polypropylene mold resembling the shape of lantern. Unfortunately, the supports commonly used in solid phase organic synthesis, still need to be improved to be suitable for peptide synthesis. Recently, a new way of capturing the resin in the reaction media was developed by Bradley et al. (22). The Merrifield resin was mixed with high-density polyethylene (HDPE) and warmed to melt the HDPE only. Formed in the shape of a plug, they can be used for solid phase synthesis. Plugs are commercially available (Polymer Laboratories), but only a few peptide synthetic applications have been published. Finally, another promising peptide chip technology was introduced by Fodor et al. (23). Peptide microarays of high density (40,000 peptides per cm2) are synthesized on the surface of a glass slide by a photolithographic method. Although the photolithographic method looks very promising and is successful in the DNA array industry, the technology needs to be further developed for peptide applications. Parallel synthesis of compounds has emerged as a powerful tool in drug discovery. Since its debut in the early 1990s, parallel synthesis went through sine curves of undeserved popularity or rejection. Finally, the methodology became an indispensable tool for drug discovery for basic academic research. High efficiency parallel synthesis, together with improved SAR, knowledge of functional mechanisms, stability against proteolysis, new delivery systems, and improved manufacturing techniques has led to renaissance of peptide therapeutics. The need for peptides, peptidomimetics, and novel technologies for their synthesis, driven by genomics and proteomics research, steadily grows. There now are more than 40 peptides in the marketplace, and close to 300 in various stages of clinical testing and development, and there are reports that the peptide share of the drug market will grow further. Peptides will not substitute for current small compound drugs, but rather enhance the market for drugs with applications where the major advantages of peptide therapeutics (high efficiency and specifity, low toxicity, simple metabolism) are needed. A few technologies for parallel synthesis are available today (Table 2). Existing instrumentation and technology is not sufficient to cover current demands, and thus new approaches and technologies for cost-effective synthesis of peptide arrays are needed.

Table 2.

Technologies for parallel solid phase synthesis

Support and distribution	Method	Company
Polyslyrene	Parallel synthesis	Advanced ChemTech, CEM, CSPS, Metter Toledo, Protein Technologies. Spyder Instruments, Torviq, Zinsser Analytics.
Microreactors (PS)	Encoding	Discover Partners International.
Grafted polymers	Directed split-and-pool	Mimotopes, Torviq
Cellulose	SPOT	Intavis AG, Jerini
PS/HDPE	Plugs	Polymer Laboratories
Glass chip	Photolithography	Xeotron