From Chemical Mutagenesis to Post‐Expression Mutagenesis: A 50 Year Odyssey

Abstract Site‐directed (gene) mutagenesis has been the most useful method available for the conversion of one amino acid residue of a given protein into another. Until relatively recently, this strategy was limited to the twenty standard amino acids. The ongoing maturation of stop codon suppression and related technologies for unnatural amino acid incorporation has greatly expanded access to nonstandard amino acids by expanding the scope of the translational apparatus. However, the necessity for translation of genetic changes restricts the diversity of residues that may be incorporated. Herein we highlight an alternative approach, termed post‐expression mutagenesis, which operates at the level of the very functional biomolecules themselves. Using the lens of retrosynthesis, we highlight prospects for new strategies in protein modification, alteration, and construction which will enable protein science to move beyond the constraints of the “translational filter” and lead to a true synthetic biology.


Introduction
Thed etailed, mechanistic interrogation of protein function with residue-level precision requires methods for the conversion of one amino-acid side chain into another (one residue into another) and the assembly of proteins which contain such precise changes.T od ate,s ite-directed gene mutagenesis has provided the primary technology for achieving this goal. [1] Gene-level mutagenesis is founded on the central dogma of Francis Crick, which describes the directional flow of information in biological macromolecules from DNAt op rotein. [2] By using this form of site-directed mutagenesis,i nformation is modified at the nucleotide level to ultimately effect achange in protein sequence,and hence at the functional biomolecule level. Through the suppression of certain codons [3] with tRNAs loaded with unnatural amino acids,u nnatural amino acids can be introduced at as pecific position of ap rotein, for example,b y stop codon positioning [4] if as pecific, paired tRNA-tRNAsynthetase system can be engineered. [5] This approach is an exciting and strikingly powerful method for certain structural motifs which are well-tolerated by the corresponding biosynthetic systems (e.g.,s ynthetases,r ibosome). However,o ne shortcoming of such (stop)codon suppression methodology is the lack of chemical diversity in the residues beyond those that can be incorporated with current technology,l argely reflecting the evolved stringencyo ft he aminoacid tRNAs ynthetase proteins.M ost amino acids that have been incorporated thus far resemble lysine or tyrosine (by stop-codon) or methionine (by Met-codon) derivatives,t hus reflecting the natural propensities of the most common synthetases that have been exploited. [3] Notably,t here have been no synthetases reported for the direct incorporation of certain residues that are central to much current biological exploration. Such residues are the methylated derivatives of lysine,glutamine,and arginine,which are widely implicated in epigenetics or for any glycosylated amino acids.Indeed, whilst there have been some elegant indirect strategies adopted, [6] the difficulty of engineering as ynthetase that can either discriminate as ingle methyl group and solely charge the modified residue,o ra ccommodate even as ingle sugar on aside chain (let alone the larger complex structures found in glycobiology) is daunting.D espite these shortcomings,t he Site-directed (gene) mutagenesis has been the most useful method available for the conversion of one amino acid residue of ag iven protein into another.Until relatively recently,this strategy was limited to the twenty standardamino acids.T he ongoing maturation of stop codon suppression and related technologies for unnatural amino acid incorporation has greatly expanded access to nonstandard amino acids by expanding the scope of the translational apparatus.H owever,t he necessity for translation of genetic changes restricts the diversity of residues that may be incorporated. Herein we highlight an alternative approach, termed post-expression mutagenesis,w hichoperates at the level of the very functional biomolecules themselves.Using the lens of retrosynthesis,weh ighlight prospects for new strategies in protein modification, alteration, and construction whichwill enable protein science to move beyond the constraints of the "translational filter" and lead to at rue synthetic biology.
success of these technologies is illustrated by their daily use in countless labs around the world.
In some respects,a pproaches such as these,w hich are transmitted (transcripted and translated = expressed) from the nucleotide level, are afundamentally circuitous means of achieving what is really sought after-a change in as pecific side chain at the protein level. In principle,amethod for posttranslational or post-expression mutagenesis would be able to accomplish this transformation directly,a tt he level of the protein itself.Such amethod would not be constrained by the evolutionary "filter" of the translational apparatus and would enable site-selective installation of any desired side chain, natural or designed, straight into the biomolecules that actually directly determine function in biology (the proteins and even the sugars and lipids).
In nature too,p ost-translational modification (PTM) is the key means by which protein side-chain diversity is extended beyond the proteinogenic amino acids.C hemically simple modifications such as methylation, and more complex structures,s uch as the attachment of sugars to glycoproteins, have profound consequences for biological processes in all kingdoms of life.T he biochemical study of PTMs has been hampered by the lack of general and robust methods for their selective installation to ap rotein. Thep ost-translationally modified side chains pose ad istinct challenge for gene-level mutagenesis.A na lternative approach to an expanded mutagenesis could draw synthetic inspiration from this natural strategy of PTM (Scheme 1, disconnection I) and modify proteins directly,without genetic mutation, in atest tube and ultimately inside cells and organisms.
An umber of biochemists,i nt he mid-to-late 1960s,w ho were working quietly before the biotechnological revolution of molecular biology and amongst the more vocal backdrop of peptide synthesis,e nvisaged as ynthetic approach to mutagenesis as av aluable tool for probing protein function. This work, which led to the first far-sighted conversion of one amino acid into another through the chemical alteration of aresidue side chain (from Ser-to-Cys), [7] demonstrated akey principle.Inthis way,apowerful conceptual legacy for amore general method that could be applied directly to biomolecules was established. Indeed, this approach enabled the first point (site-directed protein) mutations of an enzyme,b ya ny method, by using just such as ynthetic, chemical approach. Polgar and Bender used the term "simulated mutations" and described ah ypothetical method:" ……by chemical or enzymatic means it is possible to substitute one amino acid of the protein molecule by another,s imulating the effect of mutation." [7b] 2. The Early Years: b,g-Carbon-Heteroatom Bond Formation We can trace the first roots of the chemical mutagenesis concept to a1 965 paper from Wilchek et al.,i nw hich serine residues were quantitatively converted into cysteines in polypeptides (Scheme 2a). [7a] Serine tosylation and subsequent S N 2displacement enabled the synthesis of both cysteine (reaction with thioacetate followed by hydrolysis) and unnatural analogues (reaction with other thiol nucleophiles), thus demonstrating the advantages of the chemical approach for rapid diversification of side-chain structures.The powerful strategic leap embodied in this work can be analyzed through simple retrosynthetic analysis of ag eneric polypeptide structure.I tw as the case then (and remains the case now, some would argue) that the dominant disconnection of protein synthesis has been that of the peptidic bond (Scheme 1, disconnection Fand see Scheme 3). This disconnection reflects not only natures primary synthetic method (amino-acid ester + peptide amine = amide) but also chemical methods through other activated carboxy derivatives of amino acids. [8] Wilchek et al. (Scheme 1, disconnection E) ushered in the notion that other disconnections would and should prove powerful, and would offer additional synthetic flexibility in the creation of proteins. [7a] Building on this prescient work, in 1966, the groups of Koshland and Bender independently reported the chemical transformation of one specific amino-acid side chain into another on an intact protein, that is,the first point mutation, by any method (Scheme 2b). [7b,c] In both reports,t he activesite serine of subtilisin was chemically converted into ac ysteine residue.B yc omparison, the first mutation of ap rotein achieved by DNA-level site-directed mutagenesis was not to be achieved until 1978. [9] Sulfonylation of the active-site serine with phenylmethanesulfonyl fluoride,t o create au seful leaving group,e nabled displacement by thioacetate and subsequent acetyl hydrolysis,l ikely through the inherent activity of the enzyme.This reaction allowed the creation of athiol-subtilisin, bearing acysteine residue as the active-site nucleophile instead of Ser.T he presence of the desired cysteine was confirmed by amino-acid analysis and colorimetric titration.
One potentially attractive factor missing in these early reports was the element of synthetic divergence from as uitable precursor.T he work of Clark and Lowe [10] on chemical mutations of papain provided the first example of mutational divergence in amino-acid side-chain conversion through the creation of both Gly and Ser from acommon Cys precursor (Scheme 2c). Then ucleophilic cysteineo fp apain was chemically mutated via the intermediate formylglycine.
Notably,f ormylglycine has itself since been recognized as adesirable residue,both as anatural PTM (derived enzymatically from Cys or Ser) mediating sulfate ester hydrolysis at the active site of type Is ulfatases,a nd as au niquely reactive aldehyde tag for further modification. [11] Synthetically,chemical mutation of papain by Clark and Lowe commenced with selective alkylation of the reactive active-site Cys using phenacylb romide.R epeated photolysis was necessary to convert the phenacyl-inhibited protein into the desired thioaldehyde product. Thet hioaldehyde slowly hydrolyzed, thus forming formylglycine with the release of hydrogen sulfide.F rom formylglycine,reduction with NaBH 4 provided Ser while incubation at pH 9.0 for prolonged periods led to Gly by ar etroaldol reaction. [10] Whilst undeniably prescient work, it should be noted that the chemistry performed by the groups of Koshland, Bender, and Lowe exploited the enhanced nucleophilicity of the active-site nucleophiles in subtilisin (Ser) and papain (Cys). This approach provides asalient illustration of two attributes which can together create synthetic utility:b oth chemo-and regioselectivity.A lthough, therefore,apowerful strategy for achieving regioselectivity,i ts necessarily limited scope (to as ingle active-site residue in certain enzymes) does not provide ag eneral solution with free-ranging control of site selectivity. [12] 3

. Retrosynthetic Analysis of Protein Modification
To overcome the limitations in generality of early strategies for chemical approaches to mutagenesis (lack of generality in the side chains which can be accessed and also sometimes harsh reaction conditions), the application of retrosynthetic logic, as ac onvenient tool for allowing one to escape occasional subjective narrowness of design, [13] helps to identify further disconnections and thence synthons.F or example,t hrough the analysis of the early work on chemical mutagenesis described in the section above,wesee that these E-type disconnections (Scheme 1) deliver,i nb oth cases, electrophilic protein synthons at Cb matched with suitable Xg nucleophiles.T he candidate precursors that are the synthetic equivalents of such synthons are typically called tags.Wehave used the terminology of tag-and-modify [14] to highlight how we feel ag eneral disconnective strategy can be developed more broadly.Such an analysis often helps to highlight useful strategic similarities in apparently different protein chemistries.Thus,for example,whilst other Cb electrophile tag-andmodify reactions have now been explored more broadly in proteins (e.g., Cb-Seg:n ew nucleophile [15] or Dha:a saC b electrophile for Sg, [16] and Seg [17] )t he potential for aC b nucleophile to react with ah eteroatom electrophile (an inverted polarity disconnection) has not yet been described. In this Minireview,weseek to explore the potential generality of future disconnections (and putative yet unrealized reactions) further. Perhaps strangely,t he tools of retrosynthetic analysis have rarely been applied to protein chemistry in as ystematic fashion, at least in print. We believe that undertaking such an analysis clarifies the challenges and opportunities inherent in chemical protein modification. By examining ag eneric polypeptide backbone two broadly defined disconnections are immediately apparent (Scheme 1): amide backbone (disconnections F-H) and side-chain disconnections (disconnections A-E and I).

Side-Chain Heteroatom-Carbon Disconnections:F urther Developments
In nature,b oth of these amide backbone and side-chain disconnections are exploited and accomplished synthetically, essentially solely by carbon-heteroatom bond formation (Scheme 1, disconnections Fand I). Hence,b oth nature and the early work in chemical mutagenesis exploit(ed) useful natural polarities and an oft-touted [18] cue to reactivity.T his polarity can also provide benefits in chemoselectivity by allowing matched selectivities which are highly valuable in the sea [19] of competing functional groups that can be found in nature (or even in asingle biomolecule). Such an intent to use carbon-heteroatom bond-forming chemistry drives the extensive use of the natural nucleophilicity of cysteine thiol groups [20] in many bioconjugations (and indeed in natural PTM). Thed evelopment of site-directed mutagenesis thus broadened the generality (site selectivity,r egioselectivity) of protein chemical modification by allowing natural residues with particular reactivity,s uch as cysteine,t ob ei ntroduced flexibly at predetermined sites.T oday,s equential use of genetic mutagenesis to install/position areactive cysteine tag residue,and then thiol-nucleophile chemistry (in some form) to modify the cysteine tag is perhaps,n ext to nonselective lysine conjugation, the most common of all synthetic protein modification procedures.Innature,cysteine is itself subject to an umber of PTMs,s ome of which are amenable to installation by direct reaction of the Sg nucleophile with aC d electrophile (e.g.,p renylation [21] ), and many which are not (e.g., phosphorylation [22] ).
Mimicry of atarget side-chain structure (such as aPTM) has been oft-employed in protein science when the desired residue is inaccessible by current chemical methods.T his approach too,h as often exploited heteroatom-carbon bond formation to create thia-Xxx analogues such as thia-Lys, [23] thia-Arg, [24] or even thia-homo-Glu, [25] thereby allowing sometimes unique mechanistic insight (into for example, pK a or geometry) beyond that offered by the limited palette of classical mutagenesis.Insome cases,the trace Cys left from native chemical ligation (NCL;s ee below) has been advantageously exploited, [25b] and even the degree of the mimicry has been investigated. [26] Notably,however, careful control of reaction conditions may prove necessary to prevent unwanted over-reaction on non-Cys residues (e.g., His.M et [27] ).
Such side-chain mimicry has not been restricted to the use of Cys as an ucleophile.T he same principle of exploiting inherent polarity also underpinned the creation of what might be viewed as an ear-chemical mutagenesis,t hat is,t he conversion of Lysi nto homoArg using isoureas through reaction of an Nw nucleophile with aC w electrophile, [28] an unnatural disconnection which shares close similarity to natural disconnections (e.g.,L ys acylation).
Notably,d espite retrosynthetic analysis of C À X( and all) bonds allowing,i nt heory,t hree different synthons from this type of disconnection [two heterolytic (positive-negative and vice versa) and one homolytic],todate most approaches have focused on the use of charged synthons with natural polarity. [29] An important and usefully provocative exception is the use of thiol-ene (strictly thiyl-ene) chemistry,w hich takes advantage of the single-electron character of at hiyl-radical for alkylation with somophiles (somophiles or SOMO-philes are used here as acatch-all for chemical species with affinity for 'open-shell' free-radical species that possess singularly occupied MOs (SOMOs) such as terminal alkenes.I nt his system, inherent biases of polarity (electrophilic thiyl) allow Angewandte Chemie regioselectivity (terminal thiylation), but unlike heterolytic disconnections do not dictate the mode of disconnection per se. [30] This feature has thus allowed disconnection not only to atypically unreactive double-bond Cd somophile, [30] but also access to ar are example of aC -type disconnection (Scheme 1). Theinstallation of an N-acetyl thialysine mimetic of acetyllysine (using an Sg radical), which is inaccessible by direct alkylation (using an Sg nucleophile), [31] also exploits the altered selectivity of radicals in protein chemistry.Afascinating recent report from the group of Payne provides atantalizing hint at the potential use of Cradicals in carbonheteroatom bond formation in protein modification. [32] Generation of an alanyl radical on ap eptide substrate by deselenization of synthetically introduced selenocysteine and subsequent trapping with an excess of oxone provides access to Ser through aCb radical and an Og somophile.This reaction allows an alternative homolytic access to an E-type disconnection (Scheme 1).
It should be noted that in recent years an umber of reactions,o ften formal cycloadditions that are carbonheteroatom bond-forming,h ave been developed for protein modification, and were largely inspired by the click-chemistry concept. [18] An umber of different goals have informed the design of such reactions,a nd often those of "labeling" have dominated many aspects of synthetic protein research in chemical biology.T he use of mutually selective,c ompatible reagent pairs with rapid kinetics for heteroatom bond formation has evolved as ap otent solution for this application. However,the limited range of natural structures that can be accessed by such an approach still necessitates the investigation of different disconnections to allow access to structures that might more usefully mimic those needed to recapitulate function.

Amide Bond Disconnections:State-of-the-Art and Prospects
Ther ibosome,f or many the most remarkable and useful biological machine in nature,r elies upon the amide bond disconnection F( Scheme 3). At the ribosomal center the condensation reaction of an amino-acid ester with an amine is enhanced and controlled. Thelogic of designed processes for synthetic manipulation at the translational level (stop codon suppression, auxotrophic incorporation, etc.) thus relies on the amide bond disconnection approach applied through any associated biosynthetic stringency. Thus the fidelity of the system responsible for error-free protein synthesis,necessary for cellular functioning, also places grave constraints on such protein engineering because of the evolved stringencyofboth the ribosome and associated amino-acid/tRNAs ynthetases. Whilst cell-free systems and those with removed or modulated error mechanisms may benefit from greater plasticity, this tension is ever present in direct co-translational methods for the incorporation of artificial amino acids.N onetheless, some very useful PTM examples (nitro-Tyr, [33] sulfo-Tyr, [34] various acylatedLys, [35] phosphoSer [36] )have been successfully accomplished by exploiting associated plasticities and/or selectivities.A tt heir best these successful systems for incorporation can, of course,b enefit from the self-same translational stringencya sam ode of selectivity and quality control, but looser systems which create plasticity may also generate unwanted infidelity. [37] Moreover,stop-codon suppression can of course allow the smuggling of a 'stump' for further 'grafting' as at ag for chemical modification. Opportunities exist in this regard for synthetic divergence by encoding residues which act as precursors to multiple modifications.F or example, p-boronophenylalanine can be chemically transformed, post-expression, to either phenylalanine or tyrosine. [38] Scheme 3. Peptide bond disconnectionf or inserting an amino acid of choice. As the dominant mode of protein synthesis (chemically and biosynthetically), strategic methods for creating sources of fragments are widespread. Here are two contrastingmethods for generating an amino acid of choice. Both exploit the nucleophilic C-terminal synthetic equivalent fragment which is revealed by disconnection F: a) A"charged" tRNA used during N-to-C ribosomal biosynthesis (and so must be tolerated by the ribosome). b) AC-terminal peptide fragmentthat bears athiol auxiliary which may participateinthioester-mediated NCL with asuitably activated N-terminal carboxyd onor fragment.

Angewandte Chemie
Thebiocatalytic manipulation of amides is,ofcourse,not restricted to ribosomal catalysis:p roteolytic activity and its reversal can in principle be harnessed to achieve similar ends. Such as eries of enzymatic reactions has,i nf act, been suggested as ag eneralized means for the conversion of asingle amino-acid precursor into anumber of different side chains. [39] Thus,m utation from Arg to Lyso nt he protein soybean trypsin inhibitor through incubation with trypsin (leading to selective cleavage of the Arg64-Ile65 peptide bond) with subsequent addition of an excess of af ree amino acid and use of carboxypeptidase Ba sac atalyst (or use of chemical activation) allowed the ligation of free Lystothe Cterminus of the N-terminal fragment. Theu se of chemical methods valuably broadened substrate scope to any suitably activated amino acid, albeit with the danger of loss of selectivity.
These and other protease-catalyzed ligations [40] as enzymatic mutations can be seen as aforerunner to contemporary chemical protein ligation methods,t hus directly altering the protein backbone by ac ut-and-paste approach. Creative application of such an approach, aided by the rapid maturation of ligation technologies,has significant potential for new approaches in chemical mutagenesis.N ative-chemical ligation (NCL;Scheme 3b) [8c] is perhaps the seminal example of the utility of chemical approaches to protein synthesis (and thus the incorporation of modifications). Although our focus is on varied methods,s ome key features that lie behind the success of the reaction are worth considering as potential guidelines for the design of future reactions that rely on other disconnections.A mongst these,t he requirement for two chemical recognition events,the initial capture by afree thiol followed by nucleophilic attack of the N-terminal amine,lies behind the notable selectivity of the reaction. Thus,d ual motifs (if found in suitably productive reaction manifolds) may prove beneficial.
Moreover,t he wonderful suitability of this dual motif means that, despite more than thirty years of research, NCL typically relies on this reactivity alone (notwithstanding variations of homologated or altered thiol/selenol etc). Therefore,i np ractice,t he strategy is still largely limited to either aX aa-Cys or Xaa-Ala backbone disconnection although numerous dechalcogenative variations from bespoke thiol/selenol amino acids provide other disconnections in principle. [41] Thus,t he chief obstacle to extending the approach lies also in the reliance on this dual motif,that is,the requirement for an N-terminal thiol peptide moiety as well as the C-terminal thioester of as econd peptide,w hich can be challenging to synthesize.W hile many new thiolated aminoacid building blocks have been reported for peptide synthesis, few are commercially available and their syntheses tend to be involved (although are ever-improving). Thec onversion of cysteine into alanine by desulfurization is currently the only extension to the original method that has been widely applied to protein substrates. [41] Furthermore,e xpressed protein ligation (EPL), [42] one of the most useful applications of NCL for the study of medium to large proteins,r emains limited to Xaa-Cys and Xaa-Ala disconnections when performed with an expressed C-terminal fragment. New ligations at non-Cys sites (e.g., Met [43] ), as well as new methods for converting cysteine residues into other natural residues postligation (e.g., akin to prior methods used to create mimics, [25b] using other disconnections shown in Scheme 1) would thus significantly extend the usefulness of ligation strategies.Also, examples wherein an N-terminal thiol/selenol auxiliary is installed post-synthetically,a re particularly noteworthy [44] since they too may prove accessible on recombinant proteins, thus extending access to the C-terminal partner fragment in the EPL methodology.

Prospects in Chemical Mutagenesis:Overlooked Disconnections
Notably,i ns eeking to highlight current methods,t his Minireview has touched on but ah andful of the possible disconnections identified in Scheme 1: essentially disconnections E, F, and I. Theo thers remain largely absent or populated by rare examples.T odate,side-chain modification has also largely been limited to carbon-heteroatom linkages, and driven by the perceived need for fast labeling kinetics in applications such as fluorescent labeling in cells.I nt hese applications,w here bolting-on of al abel is paramount, the precise nature of the linkage in the side chain used for modification, and thus the identity of the resultant product, are not necessarily of interest. Such carbon-heteroatom chemistry is thus understandably used because it is rapid and operationally simple.Perhaps an even greater limitation is that currently,o nly av ery few carbon-carbon forming reactions which are potentially protein compatible can be performed in aqueous media. However,t he chemical construction of the carbon-rich native amino-acid side chains in situ as well as most of their post-translationally modified variants,i natarget-driven fashion, will necessitate carboncarbon bond disconnections (such as disconnections Ao rB (Scheme 1), amongst others).
TheC a ÀC b disconnection (A) results in ag lycine-type species as the protein-associated synthon. At present, no reactions have been reported for the functionalization of glycine in proteins.H owever, glycine modification is wellestablished in the field of peptide chemistry,w ith reported approaches utilizing glycine cations,a nions/enolates,a nd radicals,i ncluding asymmetric variants of the reaction. Unfortunately,f ew of the reported reactions are sufficiently benign to be readily transferred to ap rotein context, with most requiring organic solvent, strong base,o re levated temperatures well beyond those either accessible in water or in proteins.F urthermore,a sg lycine is one of the most abundant amino acids in proteins,o pportunities for sitespecific modification are likely to be limited in part by the sheer statistics of design and/or selectivity (a problem that also often plagues Lysmodification methods). In this context, functionalization at N-terminal glycine residues or the use of sequon-based approaches to position am etal catalyst or activating reagent may provide some requisite chemoselectivity.
TheC b ÀC g disconnection (B) results in consideration of an alanine-type species.A lanine itself could conceivably be modified by selective C sp3 ÀHactivation. This activation would Angewandte Chemie likely also require either the identification or design of ap olypeptide sequence to facilitate metal coordination and specific C À Hactivation. Recent work on palladium-catalyzed transformations,s uch as Suzuki-Miyaura couplings, [45] has shown that transition-metal-catalyzed reactions can be used for protein modification, and more applications of these powerful metal catalysts can be envisioned, although it is likely that CÀCb ond formation in the creation and manipulation of non-aromatic side chains will prove at ougher challenge than those shown so far for C(sp 2 ) À C(sp 2 )g eneration.
Particularly,r esearch into the prospects for CÀCb ond formation using such an approach should be ah igh priority and make use of the full gamut of disconnective pathways from new and existing tags as synthetic equivalents.W hilst C sp3 À C sp3 formation would provide useful direct access, approaches such as metathesis could conceivably provide indirect access by C sp2 ÀC sp2 formation and functional-group interconversion (i.e.r eduction) to exploit the same disconnection. Although efficient cross-metathesis on proteins has, to date,e xploited chalcogen-mediated relay effects [17,46] (and hence access to longer side chains), multiply relayed methods could be envisaged, for example,for the C b À C g disconnection (B).
One tantalizing opportunity (amongst many) awaits the discovery of synthetic methods for these CÀCdisconnections. Building the carbon scaffold that dominates amino-acid side chains will allow the proper synthesis of not only many natural amino acids but also their modified variants.Interestingly,f unctional chemical recapitulation of the decoration of proteins with PTMs by writer enzymes can be demonstrated with, for example,t hia-analogues of either methylated Lys [22, 23d] or GlcNAcylated [22,47] or phosphorylated Ser [48] . However,n ative (e.g., C À C) linkages would allow true emulation using tag-and-modify chemistry,inwhich auniquely reactive functional group is introduced into the protein (the synthon tag), which is then modified by selective chemical functionalization to bring in, through the reagent, not only the amino-acid side chain but also any chosen attached modification. Ready methods for diversely creating many variants at one site from as ingle tag/synthon protein intermediate can then be envisaged. Thes ynthetic challenges for any method that addresses disconnections Aa nd B( Scheme 1) are significant and exciting in their own right. Thus,a ny method that allows these disconnections,either AorB ,will also face multiple challenges:1 )sufficient reactivity in aqueous/biological milieu;2 )selective bond formation;3 )sufficiently mild reaction conditions under which both protein function and modification function are retained after reaction. Many existing reactions that one can envisage,fail. Thes olution to such ac hallenge will likely entail dual development of entirely new reactions with retrofitting of partially forgotten approaches (as highlighted by this Minireview).

Conclusion
Thef ield of chemical mutagenesis stands poised for ar esurgence.T he explosion of interest in protein modifica-tion chemistry,c oupled with the rapid maturation of ligation and carbon-heteroatom bond-forming approaches,i ndicates the field is ripe for anew challenge.Natural PTMs are one of the outstanding challenges in protein modification, with the current methods to access these residues in at arget-driven fashion still lacking in key respects.C urrently,i td oes not seem possible to efficiently introduce very small (such as monomethylation of Lys) or overly large (such as GlcNAcylation of Ser) modifications by intervention in the translational machinery.This range leaves enormous opportunity for the discovery of new chemical synthetic methods in the mechanistic dissection of biology.Wehope that the somewhat simplistic (but perhaps nonetheless informative) retrosynthetic analyses suggested herein highlight just how many opportunities remain unexplored, and will encourage the investigation of new methods.
Theutility of the products of such synthetic methodology seem immediately apparent:n ot only the creation of an ew generation of artificial protein drugs, [49] but it may not be too much to envisage the development of av ibrant, long-lived, associated chemical industry,m uch as current chemistry still bears the strategic hallmarks of the utilization of coal tar.I n our own opinion, although discussion of synthetic biology is currently dominated by the exploitation of indirect methods from nucleotides,o nly when the vision of chemical mutagenesis is achieved and when we can flexibly reprogram the functional molecules of life (proteins,s ugars,l ipids), can we speak of atrue synthetic biology to match the development of synthetic chemistry in the last century.