Novel protein science enabled by total chemical synthesis

Abstract

Chemical synthesis is a well‐established method for the preparation in the research laboratory of multiple‐tens‐of‐milligram amounts of correctly folded, high purity protein molecules. Chemically synthesized proteins enable a broad spectrum of novel protein science. Racemic mixtures consisting of d‐protein and l‐protein enantiomers facilitate crystallization and determination of protein structures by X‐ray diffraction. d‐Proteins enable the systematic development of unnatural mirror image protein molecules that bind with high affinity to natural protein targets. The d‐protein form of a therapeutic target can also be used to screen natural product libraries to identify novel small molecule leads for drug development. Proteins with novel polypeptide chain topologies including branched, circular, linear‐loop, and interpenetrating polypeptide chains can be constructed by chemical synthesis. Medicinal chemistry can be applied to optimize the properties of therapeutic protein molecules. Chemical synthesis has been used to redesign glycoproteins and for the a priori design and construction of covalently constrained novel protein scaffolds not found in nature. Versatile and precise labeling of protein molecules by chemical synthesis facilitates effective application of advanced physical methods including multidimensional nuclear magnetic resonance and time‐resolved FTIR for the elucidation of protein structure–activity relationships. The chemistries used for total synthesis of proteins have been adapted to making artificial molecular devices and protein‐inspired nanomolecular constructs. Research to develop mirror image life in the laboratory is in its very earliest stages, based on the total chemical synthesis of d‐protein forms of polymerase enzymes.

Introduction

The past 60 years has seen the elucidation of the covalent structures of protein molecules and experimental determination of the folded (tertiary) structures of numerous proteins by X‐ray crystallography and multi‐dimensional nuclear magnetic resonance (NMR) techniques.1 Recombinant DNA methods have been used to express proteins for structural studies, and to systematically vary their amino acid sequences in order to probe the relationship of a protein molecule's structure to its functional properties.2 Despite these achievements, we do not yet have the fundamental understanding that is needed to design novel proteins from scratch. Chemical synthesis provides for unlimited variation of protein structure, and thus has a unique capacity to help us further understand and control the principles by which the amino acid sequence of the polypeptide chain determines the folded structure of the protein molecule and its consequent properties.

Chemical Protein Synthesis

General and efficient methods for the total chemical synthesis of proteins have been available for more than 20 years.3, 4, 5, 6 Total chemical synthesis is based on the central dogma of protein science, Anfinsen's thermodynamic hypothesis,7 which states that the sequence of amino acids in the polypeptide chain determines the folded structure and properties of the protein molecule. Total chemical synthesis of a protein comprises two key steps: preparation of the linear polypeptide chain; and, folding the polypeptide to give the synthetic protein molecule [Fig. 1(A)]. Synthesis of the protein's polypeptide chain is based on the “chemical ligation” principle: chemoselective covalent condensation of two unprotected peptide segments equipped with unique, mutually reactive function groups, enabled by formation of a non‐peptide bond at the ligation site.3 The most effective condensation reaction is native chemical ligation4 and its extensions. In native chemical ligation an initial thioester‐linked ligation product is formed that undergoes rapid, spontaneous rearrangement to give a condensation product that has a native peptide bond at the ligation site. Native chemical ligation has enabled the straightforward synthesis of polypeptide chains containing more than two hundred amino acids, and is applicable to the condensation of polypeptide chains regardless of size.[Link]

Figure 1.

Synthesis and characterization of a synthetic protein. (A) General synthetic scheme. Unprotected peptide segments, prepared by solid phase peptide synthesis (SPPS) are condensed by (native) chemical ligation to give the full‐length polypeptide chain which is folded to give the protein molecule. (B) Direct infusion electrospray mass spectrum of [Lys24, 38, 83]EPO prepared by total chemical synthesis. Each of the major peaks corresponds to a different charge state (number of excess protons) of the synthetic protein molecule. Note the absence of minor peaks on the low m/z side of each major peak, showing the absence of microheterogeneity in the synthetic protein product. (The minor peaks on the high m/z side of each peak are Na⁺ and other metal ion adducts.)

The synthetic polypeptide chain is folded to give the functional structure of the target protein molecule. Protocols for solubilizing the inclusion bodies that are formed when proteins are over‐expressed in recDNA‐engineered bacteria, and for the efficient folding of these ribosomally translated polypeptides to yield functional protein molecules,9 work equally well for the full‐length polypeptide chains prepared by chemical synthesis.10, 11, 12 A key aspect of chemical protein synthesis is rigorous analytical characterization, in accord with the accepted standards of organic chemistry: homogeneity; covalent structure; and—uniquely for proteins—validation of the correctly folded “tertiary” structure of the synthetic protein molecule13 [Fig. 1(B)].

Chemical protein synthesis has been the subject of recent comprehensive14, 15 and contemporary reviews,16, 17, 18, 19 including chemical ligation strategies not based on the mechanism of native chemical ligation.17

Novel Protein Science

This article will describe examples selected from the scientific literature to illustrate key aspects of the impacts that total chemical synthesis has had on protein science, and to give a sense of the potential future contributions that chemical synthesis will make.

Chirality

One of the most striking aspects of the natural protein world is homochirality: all natural proteins are made up of l‐amino acids and the achiral amino acid glycine.20 Chemical synthesis of proteins can be used to manipulate protein chirality in novel ways, in order to facilitate the determination of protein structure, to generate novel therapeutic lead molecules, and to increase our understanding of the molecular basis of protein function.

Anfinsen's thermodynamic hypothesis7 implied that total chemical synthesis using d‐amino acids and glycine could be employed to make mirror image “d‐proteins”: that is, protein molecules whose folded structures are mirror images of natural proteins. Pioneering chemical syntheses of d‐proteins confirmed this assumption.21, 22 Access to d‐proteins provided by total chemical synthesis enables several novel kinds of protein science.

Racemic protein X‐ray crystallography

There are two critical aspects of the determination of the structure of a protein molecule by X‐ray crystallography: formation of highly ordered protein crystals; and (after acquiring X‐ray diffraction data) obtaining the phase information necessary to solve the structure of the protein. Protein racemates can be crystallized and used for X‐ray diffraction studies.22 This application of chemically synthesized d‐protein molecules can facilitate both the formation of diffraction‐quality crystals and phasing of the diffraction data for structure solution.23

A case in point is the snow flea antifreeze protein (sfAFP). Although it was straightforward to prepare sfAFP by total chemical synthesis,24 the synthetic protein proved to be difficult to crystallize. After exploring thousands of crystallization conditions over several months, crystals were eventually obtained and synchrotron X‐ray diffraction data were acquired to 0.98 Å resolution. However, since no related globular protein was known at that time the structure of the protein could not be solved by molecular replacement, and it was not possible to obtain crystals of selenomethionine‐containing sfAFP for the anomalous dispersion approach to phasing.25

In a letter to Nature in 1989,26 Alan Mackay pointed out that a centrosymetric crystal of a protein racemate would greatly simplify the phase problem, because the possible phases are severely restricted: off‐diagonal phases cancel, so all phases are related by pi radians; for example, in the space group P1 < bar>, phases will be either 0 or pi radians. In an effort to simplify phasing and use direct methods for structure solution, the mirror image (d‐protein) form of sfAFP was prepared by total chemical synthesis using d‐amino acids and glycine. Surprisingly, a racemic mixture of l‐sfAFP and d‐sfAFP protein enantiomers gave diffraction quality crystals within a matter of days in more than half of the conditions used in two standard Hampton Index screens. Furthermore, crystallization of a mixture of selenium‐containing l‐sfAFP and d‐sfAFP that did not contain selenium also gave diffraction quality quasi‐racemic crystals (discussed below), and enabled determination of the structure of sfAFP using phase information from multiple wavelength anomalous dispersion.25

Experience from the structural genomics programs over the past 15 years has shown that only about one‐third of recombinantly expressed, purified globular proteins can be crystallized, and that little more than half of those protein crystals are of diffraction quality.27 That is, there is an ~80% failure of globular protein molecules to give crystals suitable for structure determination by standard X‐ray diffraction methods. In many cases, the inability to crystalize a globular protein molecule can be surmounted by racemic protein crystallography, in which a racemic protein mixture is used, that is, one containing equal amounts of l‐protein and d‐protein forms of the same protein molecule (Fig. 2). Facilitated crystallization of racemic protein mixtures was predicted in 1995 by Wukovitz and Yeates on theoretical grounds.28 That prediction was vindicated by the facile crystallization of the sfAFP protein racemate.25 Crystallization of protein racemates is now being used in an increasing number of instances to elucidate the structures of protein molecules. In several dozen examples to date, most of which involved recalcitrant proteins, more than 80% success obtaining diffraction‐quality crystals was reported.29

Figure 2.

Racemic protein crystallography. Adapted from Reference 29.

Racemic protein crystallization both facilitates crystallization and simplifies the phase problem. The Rv1738 gene is the most up‐regulated and transcribed when M. tuberculosis enters the persistent dormant state.30 However, until recently neither the structure nor the function of the predicted Rv1738 protein molecule was known. Rv1738 protein expressed in E. coli failed to give diffraction quality crystals in extensive trials over a several year period, employing the full range of alterations to the protein molecule known to facilitate crystallization, together with robotic screening of a wide range of crystallization conditions.31 The l‐protein and d‐protein forms of Rv1738 were prepared by total chemical synthesis. A racemic mixture of l‐Rv1738 and d‐Rv1738 gave centrosymmetric crystals that diffracted to 1.5 Å resolution, and the structure of Rv1738 was solved using a novel ab initio phasing method enabled by the simplified phases (Fig. 3). The unit cell contained mirror image homodimers related through a center of inversion. The l‐homodimer was made up of two l‐Rv1738 proteins, and the d‐homodimer was made up of two d‐Rv1738 proteins. A structure‐based search revealed similarity between the homodimeric Rv1738 protein molecule and bacterial “hibernation promoting factor” proteins that shut down ribosomal protein synthesis, thus suggesting the functional role of Rv1738 in M. tuberculosis persistent dormancy.31

Figure 3.

Structure of the protein Rv1738 from M. tuberculosis determined by racemic protein crystallography. (A) Racemic crystal. (B) X‐ray diffraction pattern. (C) Mirror image structures of the Rv1738 homodimer. The blue image is the l‐homodimer and the red image is the d‐homodimer. The inversion center is shown as a cyan dot. (D) 2Fo–Fc electron density map for a Trp side chain. Adapted from Ref. 31.

Another example of the utility of racemic protein crystallography involved a nanomolar affinity d‐protein binder for VEGF‐A that was developed by mirror image phage display (see below). To determine how the d‐protein binder interacted with VEGF‐A, a solution was prepared that contained four chemically synthesized proteins: the protein enantiomers d‐VEGF‐A and l‐VEGF‐A together with two equivalents of each enantiomer of the protein binder. Racemic protein crystals were obtained and synchrotron diffraction data were collected to a resolution of 1.6 Å.32 The structure of the racemic heterochiral protein complex was determined by molecular replacement. A concise account of this work can be found here:29

Quasi‐racemic protein X‐ray crystallography

Facilitated crystallization of racemic protein mixtures is a property of the mirror image shapes of the protein enantiomers. This important phenomenon also applies to protein enantiomorphs—near‐mirror image protein molecules that are not true enantiomers, but which have mirror image shapes. The first example of quasi‐racemic protein crystallography was the determination of the structure sfAFP recounted above. In that work, a selenium‐containing l‐sfAFP analogue (in which an Asn residue was replaced by an alkylated SeCys residue) was mixed with d‐sfAFP that did not contain selenium in order to facilitate the formation of diffraction‐quality quasi‐racemic crystals for phase determination by the anomalous dispersion method.[Link] 25 Quasi‐racemic crystallography was combined with racemic crystallography to solve the structure of the plant anti‐microbial protein snakin‐1 by radiation‐damage‐induced phasing.33

An important feature of the quasi‐racemic crystal structure of an l‐protein analogue with the d‐protein form of the corresponding native protein is that, by digital inversion of the d‐protein structure, a reference native l‐protein structure is determined in the same experiment. This feature of quasi‐racemic protein crystallography was used to good effect in the determination of the crystal structure of a novel topological analogue of the protein crambin and its comparison with native crambin [described under Topology below (see Fig. 5)].34

Figure 5.

Quasi‐racemic X‐ray crystallography determination of the structure of an interpenetrating linear‐loop topological analogue of crambin. (A) Crystallographic unit cell. (B) Cartoon representation of the backbone structure. The new amide link is shown as CPK spheres. C. electron density 2Fo–Fc map with the fitted structure of the new amide bond shown as sticks. (D) Superimposition of the crambin topological analogue structure (green) and the structure of native crambin (cyan) generated by inverting the structure of d‐crambin. Both protein structures were ontained in the same experiment. Adapted from Ref. 34.

Facilitated crystallization of quasi‐racemic protein mixtures consisting of individual l‐protein analogues and the d‐protein enantiomer of the native protein molecule can greatly speed up determination of analogue protein structures by X‐ray diffraction methods. The utility of this application of quasi‐racemic crystallography was illustrated in a study of the effects of replacing α‐amino acids by β‐amino acids in a small protein molecule,35 and in the determination of structures of protein side chain diastereomer analogues (see Protein diastereomers below).36

Glycoproteins can be difficult to crystallize because of the flexibility of the complex glycan moieties. In order to explore facilitated structure determination of a glycoprotein by quasi‐racemic crystallography, the chemokine CCL1 was chemically synthesized as the l‐protein, glycoyslated l‐protein, and d‐protein forms.37 Crystals were obtained from the true racemate {d‐CCL1 plus l‐CCL1}, and also from the quasi‐racemate {d‐CCL1 plus glyco‐l‐CCL1}. Both crystalline forms diffracted to a resolution of 2.15 Å. The crystal structure of CCL1 was determined by molecular replacement, using diffraction data obtained from the true racemate. The crystal structure of glycosylated CCL1 was determined by molecular replacement using diffraction data from the quasi‐racemate.38 Despite facilitated crystallization of the CCL1 glycoprotein as the quasi‐racemate, the complex glycan moiety was disordered and only the sugar directly attached to Asn29 was well defined in the resulting electron density map.

An ingenious extension of quasi‐racemic crystallography was recently reported by Lei Liu (Tsinghua University) and his collaborators.39 Several different branched oligoubiquitin constructs of natural l‐protein chirality were co‐crystallized with the d‐protein form of ubiquitin monomer. Facilitated crystal formation was observed, giving centrosymmetric quasi‐racemic crystals in which multiple copies of d‐ubiquitin monomer had arranged themselves to form quasi‐racemates with each l‐ubiqutin in the branched covalent oligomers. X‐ray diffraction data was collected to good resolution, and structures of each of the branched oligomeric ubiquitins were determined.39 This useful {d‐monomer+l‐oligomer} method suggests that even more radical quasi‐racemates can be envisioned for facilitated crystallization of recalcitrant protein molecules, such as the use of maltose binding protein (MBP) fusion proteins plus the d‐protein form of MBP.

Screening chiral peptide, protein, and natural product libraries

Natural proteins, naturally occurring peptides, and almost all low MW secondary metabolite natural products found in nature are chiral compounds.40 It can be desirable to make mirror image forms of these naturally occurring molecules, in the case of proteins and peptides in order to improve in vivo stability and reduce immunogenicity, and for low MW natural products in order to provide novel lead compounds for the development of pharmaceuticals.

Mirror image proteins are the key component of the ingenious “mirror image phage display” method for systematically developing high affinity d‐peptide ligands for natural proteins.41 The d‐protein form of the natural l‐protein is prepared by total chemical synthesis, and is used as bait in the selection of binders from phage‐displayed peptide libraries. Once a high affinity peptide has been identified, a d‐peptide of the same amino acid sequence is synthesized. For reasons of symmetry,42 the d‐peptide has the same high affinity for the native l‐protein target. This powerful method for developing stable, high affinity d‐peptide ligands is currently being used for a variety of therapeutic target proteins.43, 44, 45, 46, 47, 48, 49 In many cases, it is sufficient to make a single domain of the target protein in mirror image form, rather than the entire protein molecule, thus simplifying the synthetic challenge.

Mirror image protein phage display is a powerful method for systematic engineering of interactions between left and right‐handed protein molecules. In order to develop a d‐protein antagonist of the angiogenic protein hormone vascular endothelial growth factor (VEGF‐A),50 the 22 kDa mirror image protein d‐VEGF‐A was prepared by total chemical synthesis and was used to screen large numbers of phage‐displayed variants of a protein scaffold. A nanomolar affinity d‐protein binder was identified. Data from the racemic crystal structure of the heterochiral complex of VEGF‐A and the d‐protein binder (described above)32 were used to guide the refinement of the properties of this d‐protein antagonist of VEGF‐A.50

Chirality is the dominant factor in the interaction of low MW natural product compounds with protein molecules.51 There is usually great overlap in drug discovery research in the intensely competitive field of natural product‐based pharmaceutical development. Typically the same proteins are identified as currently important targets, and it is frequently the case that several drug development programs come up with similar lead compounds for further refinement by medicinal chemistry. For that reason, a method for identifying unique lead compounds would be of great value. Very early on it was realized that screening libraries of chiral natural product (NP) compounds against the chemically synthesized mirror image form of a protein target, followed by making the mirror image of a selected NP by conventional organic synthesis, could provide unique leads not found by conventional screening methods52 (Fig. 4).

Figure 4.

Screening a chiral natural product library against the mirror image forms of an enzyme can provide novel hits. Synthesis of the mirror image of a hit compound then gives a novel natural product‐related molecule active against the native enzyme.

Recently, Nobutaka Fujii et al. reduced this concept to practice.53 They screened a chiral library of ~22,000 natural product compounds against mirror image forms of the p53‐binding domain of MDM2 protein. A chiral α‐tocopherol derivative was identified as an inhibitor of the d‐MDM2/d‐p53 interaction. The mirror image of the identified α‐tocopherol derivative, prepared by conventional organic synthesis, inhibited l‐MDM2/l‐p53 interaction but had no effect on d‐MDM2/d‐p53 interaction.53

Protein diastereomers

Diastereomers are proteins that contain a mixture of d‐amino acids and l‐amino acids,54 or that have inverted Thr or Ile side chain stereochemistries.55 These stereochemical hybrids can be useful probes of the molecular basis of protein properties.

The thermodynamics of C‐terminal “helix capping,” the geometrically constrained region where an alpha‐helical stretch of polypeptide transitions to a more extended secondary structure, were investigated in the ubiquitin protein molecule.56 Total chemical synthesis was used to replace the conserved Gly residue immediately after the alpha‐helix with d‐ and l‐enantiomers of the amino acids Ala and Val, thus creating two pairs of protein diastereomers. Measured energy differences between the stabilities of the diastereomers showed that the dominant thermodynamic effect was the necessity to adopt a left‐handed backbone conformation for the residue after the alpha‐helix, a conformation that can only be adopted by Gly of all the proteinogenic amino acids or by a d‐amino acid.56

The scorpion venom toxin ShK protein contains two Ile and four Thr residues. Based on predicted relative stabilities from molecular dynamics calculations, three ShK diastereomers were chemically synthesized as l‐proteins that contained a single residue with inverted side chain stereochemistry: [l‐allo‐Ile⁷]ShK; [l‐allo‐Thr¹³]ShK; and [l‐allo‐Thr³¹]ShK.[Link] 36 The rates of folding and the relative stabilities of the folded diastereomers were in agreement with calculation. Quasi‐racemic crystallography, in which each l‐ShK diastereomer was co‐crystallized with d‐ShK, was used to determine the structure of the diastereomers [l‐allo‐Ile⁷]ShK and [l‐allo‐Thr¹³]ShK. Interestingly, the least stable diastereomer [l‐allo‐Thr³¹]ShK spontaneously crystallized, alone, from the quasi‐racemic mixture with d‐ShK.36

Topology

In structural protein science, the term “topology” is used to describe the secondary and tertiary structures of the folded protein molecule. In its mathematical sense,57 “topology” differentiates the linear polypeptide chains usually found in proteins from branched, circular, linear‐loop, and numerous other possible polypeptide chain configurations. Here we will use the term topology in both of these senses.

Secondary structure

Preformed components of secondary structure, or secondary structure‐inducing moieties, can be introduced into a protein molecule's peptide chain by chemical synthesis. The Gly¹⁶–Gly¹⁷ sequence in a reverse turn in the HIV‐1 protease protein molecule was replaced by BTD, a non‐peptide type II’ β‐turn mimic of fixed geometry.58 The folded BTD containing protein had full enzymatic activity, unchanged substrate specificity, and enhanced thermal stability. A similar approach was used to study the protein GB1 domain of Staph. Aureus.59 More recently, lactams formed between the side chain of an amino acid and a neighboring peptide bond have been used to control protein secondary structure.60

Another approach to controlling protein secondary structure can be thought of as “Ramachandran space engineering.” The homodimeric HIV‐1 protease protein molecule has two identical mobile flap structures that close down over the bound substrate. A covalent dimer form of the HIV‐1 protease with a polypeptide chain of 203 amino acids was prepared by total chemical synthesis, in order to enable asymmetric changes to be made in the covalent structures of the flaps.61 Residues Gly⁵¹ and Gly^51’ were replaced by d‐Ala, l‐Ala, or Aib (α‐amino‐isobutyric acid) in one or both flaps, thus generating a series of analogue protein molecules in which the accessible backbone conformational space at the tip of each flap was restricted in distinct ways. Symmetric substitution of both Gly⁵¹ and Gly^51’ with either d‐Ala or l‐Ala resulted in substantially reduced enzymatic activity. The [Ala⁵¹,Aib^51’]HIV‐1 protease, in which the accessible flap conformations were severely restricted, had dramatically reduced enzymatic activity. Interestingly, the asymmetric analogue [d‐Ala⁵¹, l‐Ala^51’]HIV‐1 protease had full enzymatic activity, suggesting that distinct backbone geometries observed for Gly⁵¹ and Gly^51’ in the native enzyme are functionally relevant.62

Novel covalent topology

Synthesis can be used to create uniquely chemical variations on the simple linear, N‐terminus‐to‐C‐terminus topology of the polypeptide chains in most natural proteins.

The nuclear proteins cMyc and Max regulate gene expression by forming a non‐covalent heterodimer that binds to a specific nucleotide sequence in double‐stranded DNA.63 A 172 residue heterodimer construct was designed, in which the distinct b/HLHZ domains of cMyc and Max were covalently linked at the C‐terminal region of each domain, to give a protein molecule of unprecedented covalent structure: the polypeptide chain changes direction at the mid‐point and has two N‐terminal residues, but no C‐terminus. The target covalent heterodimer was prepared by condensation of four synthetic peptide segments by using two distinct, mutually compatible ligation chemistries. This uniquely chemical cMyc‐Max analog protein had the same binding specificity as the non‐covalent heterodimer. Interestingly, CD spectroscopy of the covalent heterodimer cMyc‐Max protein synthetic construct showed that it was pre‐folded in the absence of the dsDNA ligand, whereas the b/HLHZ domains of native cMyc and Max only form a folded protein (non‐covalent) dimer in the presence of target dsDNA.64

Ubiquitin is a highly conserved protein molecule that in nature is attached to other proteins to alter their biochemical properties.65 The attachment of ubiquitin to a lysine side chain in another protein molecule can be used to target that protein for degradation by the cell's proteasome machinery.66 Covalent modification of histone proteins by ubiquitin modifies gene expression.67 In both these biochemical systems, an attached ubiquitin can be further modified by covalent attachment of one or more ubiquitins to form ubiquitin chains.68 Topologically, ubiquitinated proteins are comprised of branched polypeptide chains. Several research groups have used chemical protein synthesis to make branched ubiquitin‐protein constructs with defined covalent structures, and to then investigate their biochemical properties.69, 70, 71, 72, 73 Novel methods for the chemical synthesis of oligo‐ubiquitins have provided branched protein constructs of defined covalent structure with masses up to 51 kDa,74 and even larger multi‐ubiquitins by controlled native chemical ligation polymerization.75

In the cell, some linear polypeptide chains produced by ribosomal translation undergo post‐translational biochemical processing to give proteins that have circular polypeptide chains: that is, protein molecules in which the N‐terminus and C‐terminus of the polypeptide chain are joined by a peptide bond. Circular proteins were initially isolated from plants and called “cyclotides.”76 In the research laboratory, chemical synthesis is a preferred way of making these circular protein molecules for structure–function studies.77 Circular proteins have been identified from a wide variety of natural sources, including bacteria.78 The Bode group recently reported the total chemical synthesis of the 70 residue circular protein bacteriocin As‐48.79

Catenanes are molecules in which two macrocyclic structures interpenetrate and are thus mechanically tethered to each other. Chemical synthesis has been used to design and build protein catenanes that contain interlocked macrocyclic polypeptide chains.80

Chemical synthesis was also used to prepare a unique crambin analogue in which an ion pair interaction between the side chain guanidinium of residue Arg10 and the α‐carboxylate of the polypeptide chain was replaced by a covalent bond, to give a linear‐loop polypeptide chain covalent topology. Folding gave a protein molecule of novel topology in which the N‐terminus of the polypeptide chain penetrates the macrolactam ring and is held in place by disulfide bonds, in a manner reminiscent of lasso peptides.[Link] 34 The structure of this novel protein topological analogue was determined by quasi‐racemic X‐ray crystallography (Fig. 5).

Chemical engineering

Synthesis enables use of the entire repertoire of organic chemistry to generate uniquely chemical variants of protein covalent structure. There is a vast literature of “peptidomimetic chemistry,”82 all of which can be applied to protein molecules using modern chemical synthesis methods. Correlation of changes in the covalent structure of protein molecules with effects on protein folded structure and properties provides new insights into the molecular basis of protein function, and enables novel forms of protein (re)design.

Backbone engineering

An early example of uniquely chemical protein variants was “backbone engineering,” by total chemical synthesis, of the HIV‐1 protease protein molecule. Two peptide bonds were replaced by thioester bonds, thus precluding the formation of hydrogen bonds to carbonyls on the peptide substrate.83 The resulting chemically engineered enzyme had k _cat (turnover number when the enzyme is saturated with substrate) lower by ~3,000‐fold compared with the native backbone enzyme molecule. That catalytic rate reduction of ~5 kcal/mole corresponds to the energy of two hydrogen bonds, suggesting that these interactions between the peptide substrate and the protein backbone are catalytically relevant.84 Recent examples of protein backbone engineering by chemical synthesis include the use of “click” chemistry condensation of peptide segments to give protein molecules containing a triazole isostere in place of a peptide bond.85, 86 Chemical synthesis has been used to carry out systematic C‐alpha and amide nitrogen methyl scans of a protein's polypeptide chain, to identify backbone conformations that affect protein structure and function.87

Side chain translocation

In natural protein molecules, amino acid side chains are pendant from the α‐carbons in the polypeptide backbone. Chemical protein synthesis can be used to shift the side chain of an amino acid to the peptide bond nitrogen atom. In the model protein bovine pancreatic trypsin inhibitor (BPTI), residue Cys38 was replaced by a peptoid moiety bearing an N^α‐ethanothiol moiety. The uniquely chemical protein analogue [(N^α(CH₂)₂SH)Gly³⁸)]BPTI had the same folded structure and thermal stability as native BPTI, as shown by multidimensional NMR.88 More recently, semi‐synthesis was used to effect a series of peptoid translocations of the side chains of individual Lys, Glu, and His residues in the N‐terminal region of the polypeptide chain of the enzyme ribonuclease A, and to study the effects on the enzymatic activities of the resulting analogues.89

Medicinal chemistry

Using synthetic chemistry, the principles of medicinal chemistry can be applied to protein molecules, in order to optimize their properties. For the chemokine RANTES, this resulted in development of the uniquely chemical protein analogue AOP‐Rantes, which binds to the chemokine receptor CCR5 and inhibits HIV‐1 entry into peripheral blood cells.90, 91 Recently, the properties of insulin for use as a human therapeutic have been enhanced by incorporation of halogen‐modified amino acids.92 Application of medicinal chemistry to insulin is greatly facilitated by the development of an efficient total synthesis of the insulin protein molecule via “ester insulin,” a chemical surrogate for proinsulin, in which the two polypeptide chains of mature insulin are linked by an ester bond between the side chains of Glu^A4 and Thr^B30. Effectively this replaces the 35 residue C peptide of proinsulin by a single covalent bond between two side chains, The 51 residue depsipeptide chain folds at physiological pH and the resulting ester insulin protein can be chemically converted to human insulin in excellent yield.93, 94, 95 Backbone ester depsiproteins can be prepared by KAHA ligation chemistry, which can directly generate ester‐linked polypeptide chain ligation products.96

Glycoprotein engineering

In a tour de force of synthetic glycoprotein science, Kajihara et al. reported the total chemical synthesis, folding, and biological activities of five distinct glycoforms of erythropoietin (EPO). Each glycoform contained one or more human‐type biantennary sialyloligosaccharides at defined glycosylation positions, and they were prepared in high purity and with defined chemical structure97 (Fig. 6).

Figure 6.

EPO glycoforms prepared by total chemical synthesis. (A) Modular convergent synthetic scheme. The bottom row shows cartoon representations of the five glycoforms. (B) Direct infusion electrospray mass spectra of the synthetic glycoproteins. Observed masses of the synthetic glycoproteins were 20,540 Da, 22,746 Da (three isomeric glycoforms), and 24,952 Da. Adapted from Ref. 97.

In earlier work, synthetic erythropoiesis protein (SEP)—a chemical “glycoprotein mimetic” variant of EPO—was designed and synthesized. SEP had two‐tetra‐branched oligo(ethyleneoxide‐amide) moieties each carrying four negative charges that replaced the complex sialoglycans of the native protein. The 50,825 Da SEP protein molecule had full biological activity and an extended duration of action in vivo.98, 99

A priori protein design

Design of protein molecules with predetermined structures and novel functions is the “holy grail” of protein science.100 Total chemical synthesis allows the designer to go beyond the 20 amino acids normally found in proteins, and to circumvent the linear polypeptide chain paradigm imposed by ribosomal translation.101 The current state of the art for a priori design of protein molecules with novel tertiary structure is exemplified by the designed tricyclic protein “CovCore 3H2.” The 6280 Da protein was prepared by total chemical synthesis and contains a circular polypeptide chain. The protein was locked in what would otherwise have been a high energy conformation by unnatural covalent constraints. Its structure was determined by X‐ray crystallography102 (Fig. 7).

Figure 7.

X‐ray structure of an a priori designed protein molecule. The circular polypeptide chain is shown in purple with cartoon representation of the helical regions. The aromatic ring used to covalently join the three helices and constrain the folded structure is shown in cyan.102

A similar approach was used to construct a 24 kDa trimeric analogue of an amyloid aggregate.103

Advanced physical techniques

Chemical synthesis provides for exquisitely precise introduction of isotope probe nuclei into the protein molecule and for site‐specific attachment of other spectroscopic probes. This has enabled novel applications of advanced spectroscopic methods to elucidation of the molecular basis of protein structure and function.

Nuclear magnetic resonance

NMR spectroscopy is widely used to determine protein structures and to probe protein dynamics. Assignment of resonances to individual nuclei and determination of molecular structure by multi‐dimensional NMR methods can be challenging for multi‐domain proteins, and is particularly challenging for integral membrane proteins because of the spectral overlap of resonances from hydrophobic amino acid residues in transmembrane helices.104 Chemical ligation can be used to simplify these problems, by incorporation of NMR isotope labeled segments and unlabeled segments into a protein's polypeptide chain.105, 106, 107 NMR spectroscopy is particularly useful for determining the ionization properties of functional groups in protein molecules. The unambiguous determination of the pKa values of individual side chain functionalities is important for understanding enzyme catalysis, and is greatly facilitated by site‐specific incorporation of NMR probe nuclei‐containing amino acids into the protein molecule.108

Fluorescence

Fluorescent dyes are useful probes of protein dynamics and can be used to measure distances within proteins and protein complexes. Chemical synthesis can incorporate fluorescent dyes at specific sites in a protein molecule for distance measurements by Förster Resonance Energy Transfer (FRET).109 Site‐specifically labeled fluorescent toxin proteins were prepared by total chemical synthesis and used for lanthanide resonance energy transfer (LRET) measurements of conformational changes in the voltage‐gated sodium channel Na_V1.4 in resting and inactivated states, and in live cells110 (Fig. 8).

Figure 8.

Measurement of distances in the voltage gated sodium channel Nav1.1 protein molecule. Lanthanide resonance energy transfer (LRET) measurements were enabled by the total chemical synthesis of small protein toxins site‐specifically labeled with a fluorophore dye. Adapted from Ref. 110.

Ensemble measurements average the conformations and dynamics of sub‐populations of proteins, and thus obscure the behavior of individual protein molecules. Fluorescent probes can be used for single molecule studies of enzyme catalysis. A covalent homodimer HIV‐1 protease equipped with a PEG‐Biotin tether was prepared by total chemical synthesis and anchored to a streptavidin‐coated glass slide. Using a peptide substrate labeled with one member of a FRET dye pair at the N‐terminus and the other dye at the C‐terminus, single molecule time‐dependent observations of substrate binding and cleavage were recorded.111 Single‐molecule FRET can also be carried out with chemically synthesized protein molecules containing multiple fluorescent dyes. Convergent total chemical synthesis combined with regiospecific cysteine side chain protection was used for the site specific introduction of three different fluorophores into the α‐synuclein protein molecule. Single‐molecule three‐color FRET experiments were carried out in solution on freely diffusing α‐synuclein molecules.112

Infrared spectroscopy

Time‐resolved 2D Fourier transform infrared (FTIR) protein spectroscopy can provide information on protein conformational dynamics on picosecond to millisecond time scales.113 In order to use FTIR to probe the role of specific hydrogen bonds in the formation of insulin dimers, total chemical synthesis was used to prepare insulin molecules containing arrays of ¹³C=¹⁸O isotope probe nuclei at the carbonyl moieties of specific peptide bonds in the protein's two polypeptide chains.114

Novel vibrational probes that are accessible to observation through the transparent window between 1880 and 2500 cm⁻¹, can be introduced at specific sites to act as reporters of electrostatics and other features of the physicochemical environment within the protein molecule.115 The vibrational Stark effect, the shift of the frequency of an infrared resonance in response to changes in an applied electric field, is a sensitive probe of local electrostatics within a protein molecule.116 Vibrational probe moieties, such as nitrile and thiocyanate, have been introduced into enzyme protein molecules on Cys residue side chains. In pioneering experimental studies, nitrile vibrational Stark effect probes were used to elucidate the role of electrostatic effects in enzyme catalysis.117 These and similar Stark effect studies support a significant contribution to catalysis by electric fields within the enzyme protein molecule complexed with its substrate.118 A current perspective on the role of electrostatics in enzyme catalysis has recently been published.119 Chemical synthesis would be a more versatile and controlled way of introducing vibrational Stark effect probe moieties into protein molecules. Total chemical synthesis will play an important role in further experimental tests of the role and magnitude of pre‐oriented H‐bond dipoles in enzyme catalysis.

Summary and Prospects

Over the past two decades, robust methods for the total synthesis of protein molecules based on the chemical ligation principle have enabled a broad spectrum of novel protein science, some examples of which are described in this article. New methods continue to be developed. Current improvements in methods for chemical protein synthesis include novel thioester precursors,120 expanded auxiliary‐mediated native chemical ligation,121 new post‐ligation desulfurization conditions,122 milder conversion of hydrazides to thioesters for convergent synthesis,123 and refinements of reactive moieties used in KAHA chemical ligation.124 Flow chemistry is being adapted to the rapid synthesis of peptide building blocks,125 and to the continuous production of chemically synthesized protein molecules.126

Methods for the preparation of chemical analogues will continue to be extended. Divergent chemical synthesis can be used to generate multiple protein analogues from a library of variants of the same peptide segment.127 A novel approach to post‐translational side chain modification of proteins by attachment of any of a range of chemical structures at a uniquely reactive dehydroalanine moiety in a protein molecule has been described.128 Facile creation of chemically diverse protein libraries can be imagined by insertion of diverse sets of peptide segments or other chemical moieties into the backbone of the polypeptide chain in a protein scaffold. Conceptually, such an insertion reaction is the reverse of intein‐mediated protein splicing in which a central peptide segment is excised from a polypeptide chain,129 and could be achieved by chemoselective reaction at a unique artificial chemical functionality in the backbone of the chemically synthesized polypeptide chain.

Chemistries used for the total synthesis of protein molecules are well suited to the construction of protein‐inspired nano‐ and other molecular devices.130 Engineered coiled‐coil 100 nm long covalently linked peptide nanotubes of defined structure were assembled by native chemical ligation.131 With the goal of developing novel bioresponsive materials, covalent concatenation of multiple different protein domains was performed by iterative intein‐mediated native chemical ligation.132 A rotaxane‐based molecular machine was designed that carries out the synthesis of peptides of pre‐defined structure by sequential native chemical ligation reactions.133, 134

All living organisms are homochiral at the level of DNA molecules which are comprised exclusively of d‐ribose units, and proteins which are comprised exclusively of l‐amino acids and the achiral amino acid glycine. Chemical synthesis uniquely provides access to a world of mirror image protein molecules which have chiral properties, including chiral specificity of enzyme catalysis,21 that are the reciprocal of natural l‐proteins. As a first step toward the ultimate aim of “chemically synthesizing an alternative, mirror‐image form of life in the laboratory,” the total chemical synthesis of two mirror image DNA polymerase enzymes has been reported.135, 136 These mirror image polymerases were shown to be capable of replicating l‐DNA, and transcribing l‐DNA into l‐RNA. The d‐protein form of the 352 amino acid residue Sulfolobus solfataricus P2 DNA polymerase IV was prepared by convergent total chemical synthesis and could be used in mirror image polymerase chain reaction amplification.137

In the future, we can foresee even more ingenuity applied to the development of novel methods for the total chemical synthesis of protein molecules, and continued expansion of the manifold applications of synthetic chemistry to the world of proteins. The fundamental understanding of the molecular basis of protein function generated in this way will ultimately enable us to step beyond the boundaries of the proteinogenic amino acid building blocks/linear polypeptide chain topology used by nature, and to use the full repertoire of organic chemistry to engineer designed proteins, particularly enzymes, with predetermined structures and novel properties.

Conflict of Interest

Stephen Kent is listed as an inventor on patents related to several of the topics discussed in this review. He is a co‐founder of and holds stock in Reflexion Pharmaceuticals.