Advance in ligation techniques for peptide and protein synthesis

Since its discovery 20 years ago, native chemical ligation (NCL) remains the most widely spread chemical technique to achieve the coupling of peptide fragments.¹  From the first protein obtained by NCL, human interleukin-8 (IL-8),²  to the forefront of total protein synthesis with the recent preparation of the fully synthetic erythropoietin (EPO) glycoprotein,³  it is evident that in two decades the field has witnessed impressive advances.

The method originally developed by Dawson et al.,²  allows the formation of a native amide bond between the C-terminal thioester and the N-terminal Cys of two unprotected peptide segments. The reaction proceeds through an initial chemoselective thiol-thioester exchange or capture step, followed by a spontaneous and irreversible intramolecular S,N-acyl shift yielding the desired peptide bond (Scheme 1).

Notwithstanding its many practical aspects; highly chemoselective and high yielding reaction, use of neutral aqueous media ideal for the solubility of unprotected peptide segments, absence of racemisation at the C-terminus; the original NCL technique presents important limitations. Notably, the incompatibility of the thioester functionality with automated Fmoc solid-phase peptide synthesis (SPPS) has forced the preferable use of the Boc SPPS strategy for the preparation of thioesters. However, the instability of post-translational modifications (PTM) to the strong acidic cleavage conditions of Boc SPPS and the difficult handling of HF have rendered the preparation of the thioester peptide fragments challenging. These restrictions and the low abundance of Cys in naturally occurring proteins have prompted researchers to focus their efforts on the development of a more general approach to NCL but also on alternatives to Dawson's NCL.

To face the requirement for Cys residues, various backbone acyl-transfer auxiliaries⁴  as well as sugar-assisted ligation (SAL)^5,6  and side-chain-assisted ligation (SCAL)⁷  techniques were engineered (Fig. 1).

Although these methods extend the scope of NCL beyond the Cys residue, they present their own inherent restrictions. Firstly, unlike the N-terminal Cys, the N^α-acyl transfer auxiliaries generate a secondary amine which, due to increased steric hindrance, has been found to only react efficiently at non-sterically demanding ligation sites. In practice these N^α-acyl transfer auxiliaries are mostly limited to Gly-Gly and Gly-Ala ligation sites. On the other hand, the use of SAL is restricted to the synthesis of glycopeptides, since the introduction of a sugar unit as removable auxiliary would be both expensive and tedious. Finally, difficulty in the cleavage of the SCAL-auxiliary have been reported which considerably hamper the reaction.⁸  Alternatively, with the introduction of the Ala NCL through the ligation-desulfurisation strategy by Yan and Dawson⁹  and the development of a more environmentally-friendly metal-free dethiylation (MFD) protocol by Wan and Danishefsky,¹⁰  attention turned towards mercaptoamino acids as Cys-surrogates. In recent years, exciting progress has been made in the synthesis of thiolated amino acids and their application to the preparation of polypeptides. In addition to cis and trans γ-thiol Pro and penicillamine which are commercially available, an important repertoire of synthetic routes to β-thiol Phe,¹¹  γ-thiol Val,¹²  γ- and δ-thiol Lys,^13–16  γ-thiol Thr,¹⁷  β-thiol Leu,^18,19  γ-thiol Pro,²⁰  γ-thiol Gln,²¹  β-thiol Arg,²²  β-thiol Asp^23,24  and γ-thiol Glu²⁵  have also been described. The structures of the Cys surrogates which have already been used in Cys-free NCL are depicted in Fig. 2.

Research groups involved in the preparation of these Cys surrogates have investigated the potential of these mercaptoamino acids to effect NCL with an assortment of peptide thioesters. To do so, they often use model peptide thioester sequences similar to the LYRAX originally employed by Hackeng et al.²⁶  for Cys NCL. In general, these thiolated amino acids demonstrated comparable ligation rate to Cys under standard NCL conditions, providing complete and fast ligation with unhindered amino thioesters such as Gly, Ala and Phe, whereas slower ligations occurred at β-branched thioesters (i.e. Thr, Val, Ile) still affording good yields after multiple hours of reaction. However, the Xaa-Pro ligation, known to be challenging in standard cysteine ligation, often requires few days to reach an acceptable yield. Notably, the ligation between hindered mercaptoamino acids with sterically demanding thioesters still proceeds as exemplified by the successful ligation between γ-thiol Val with C-terminal Pro p-nitrophenyl ester, which yielded the coupling product in 55% yield after 10 h.¹² 

Beyond model peptides, this new strategy for the Cys-free NCL has been employed for the synthesis of complex peptides, as well as for peptide post-translational modifications and the preparation of proteins and glycoproteins (Scheme 2). The Brik group has been remarkably active in this field and have notably employed the γ-thiol Gln for their synthesis of the full-length YAP65 WW domain(1–40).²¹  Furthermore, the Danishefsky group also employed a Leu ligation. They developed a kinetic controlled ligation (KCL) based on the slower coupling rate of β-thiol Leu compare to Cys to achieve the synthesis of erythropoietin(95–120) in a three-segment coupling strategy.¹⁸  The human galanin-like peptide(1–60) (hGALP), a potential therapeutic for the treatment of obesity, was recently prepared by Guan et al.²⁴  They used the β-thiol Asp ligation in order to disconnect hGALP at the Gln-Asp site, so that difficulties related to eventual aspartimide formation during Fmoc SPPS could be avoided. Longer proteins have also been prepared. Using a Leu ligation, the Brik group successfully achieved the synthesis of the 86 amino acids’ protein HIV tat, which they had failed to access through the SCAL technique.¹⁹  Both the 84-mer human parathyroid hormone (hPTH)²⁷  and the 141-mer human parathyroid hormone related-protein (hPTHrP)²⁸  were assembled by the Danishefsky group using Cys surrogates at ligation sites. While, hPTH required a four-segment approach comprising a single standard Cys ligation and a β-thiol Leu as well as a γ-thiol Val ligation, the first attempt for the synthesis of hPTHrP using a Pro ligation failed. Nonetheless, hPTHrP could be successfully synthesised via a new disconnection relying on a Leu ligation and two traditional NCL. Furthermore, γ- and δ-mercapto-Lys have also been employed as tools for the site-specific instalment of PTM and biochemical tags such as ubiquitylation and biotinylation.^13,14  After an initial chemoselective ligation step, the mercapto-Lys-containing peptide was subjected to a desulfurisation step under the mild dethyilation conditions of the MFD, thus restoring the natural Lys residue while leaving the newly ligated ubiquitin or sulphur-containing biotin moieties intact. Notably, Yang et al.¹³  reported a dual ligation at both the α- and ɛ-amine of a single Lys building block, making use of the same sulfhydryl handle to form the native and isopeptide bonds.

The compatibility of the ligation-desulfurisation technique with the presence of PTM was demonstrated by various research groups. The Payne group reported the synthesis of a three 20-residue mucin 1 (MUC1) variable-number tandem repeat (VNTR) peptide decorated with O-linked monosaccharides using β-thiol Arg in conjunction with the MFD protocol.²²  A remarkable use of the Cys-free NCL as an approach to glycopeptide synthesis was made by the Danishefsky group who prepared homogeneous hEPO(79–166) presenting a N-linked chitobiose and a O-linked glycophorin unit via a four-segment strategy utilising two Ala- and one Pro ligation followed by the mild dethyilation procedure they initially reported.²⁹ 

The use of thiolated proteinogenic amino acids in a ligation-desulfurisation strategy has therefore considerably broaden the scope of possible Xaa-Xaa ligation site and is thus emerging as a powerful method to achieve Cys-free NCL, however certain issues remained to be addressed. First of all it has appeared that both diastereoisomers of the thiolated amino acids do not react with the same efficiency. While the diastereoisomers of γ-thiol Val, β-thiol Leu and γ-thiol Pro have shown significant disparity in ligation rates,^12,18,30  no difference was observed between the erythro and threo β-thiol Asp.²³  Despite the removal of the chiral centre at the sulfhydryl handle in the desulfurisation step, the influence of the stereochemistry of the Cys surrogates on the ligation rate precludes the use of diastereoisomeric mixtures of mercaptoamino acids. The preparation of these precious building blocks thus requires the development of complex asymmetric synthetic routes. Unfortunately, these lengthy syntheses (most of all counts 7 to 16 steps) and the low overall chemical yields considerably restrain Cys surrogates from finding wide utility among peptide chemists. Furthermore, although the MFD protocol is compatible with a wide range of sulphur-containing amino acids such as Met and other sulphur-containing moieties such as thiazolidine and thioesters, native Cys must be protected with a temporary acetamidomethyl protecting group to prevent their unwanted conversion to Ala.¹⁰  If this limitation does not prohibit the use of the Cys-free NCL through ligation-desulfurisation strategy for fully synthetic peptides, this method is not suited for the multitude of semi-synthetic peptides now accessible through expressed-protein ligation (EPL).³¹ 

Recent efforts of the Payne group have focused on resolving these issues. In addition to the short and scalable routes to both β-thiol Asp and γ-thiol Glu they developed,^23,25  they reported a ligation at the Trp residue relying on the late-stage sulfenylation of Trp-containing peptides.³²  The 2-thiol Trp model peptide was either prepared by in-solution reaction with 2,4-dinitrophenylsulfenyl chloride (DNPS-Cl) followed by thiolysis of the newly formed thioether or by sulfenylation of the resin-bound peptide and subsequent cleavage from the resin prior to final thiolysis. Although the first attempt to carry out the ligation of the 2-thiol Trp-containing peptide with aryl thioester formed in situ failed, the use of pre-formed Gly-, Ala-, Met- and Phe-thiophenyl thioesters under optimised conditions afforded the ligation products in good yields (i.e. 65–81%) in 24 h reaction time. Ligation with Pro-thiophenyl thioester required extended reaction time (i.e. 30 h) to yield the desired peptide in 58% yield. Even though, the use of metal-based conditions proved necessary to effect the removal of the sulfhydryl handle, this strategy allowed the successful synthesis of CXCR1(1–28) peptide containing an N-linked N-acetylglucosamine (GlcNAc) and multiple Met residues. The progresses towards Trp ligation enabled by either in-solution or on-resin late-stage sulfenylation strategies are summarised in Scheme 3.

Remarkably, the Payne group also made significant progress towards the selective desulfurisation in presence of unprotected Cys by furnishing the first example of chemoselective dethyilation of thiol-derived proteinogenic amino acids. Using tris-(2-carboxyethyl)phosphine (TCEP) in combination with dithiothreitol (DTT) under acidic pH, they achieved the selective desulfurisation of β-thiol Asp-containing peptides.²³  A one-pot process for the ligation-desulfurisation approach was also reported.²³  After completion of the Cys-free NCL in the presence of thiophenol in the solution of chaotropic and reducing agents, the aryl thiol was extracted in order to prevent it from hindering the following desulfurisation reaction. This simple extraction step alleviates the need for an intermediate RP-HPLC purification and thus increases the yield of the final product. Payne and co-workers successfully applied their one-pot Asp ligation-chemoselective desulfurisation strategy to assemble the extracellular N-terminal domain of the chemokine receptor CXCR4 bearing an unprotected Cys, a Tyr sulfation and a N-linked glycosylation.²³  Unfortunately, in the case of γ-thiol Glu, the chemoselective desulfurisation failed, a one-pot Glu ligation-MFD strategy was adopted instead.²⁵  Encouraged by the selective deselenisation reported by Dawson and co-workers,³³  the Danishefsky and Payne group, prepared the trans-selenol Pro³⁴  and the β-selenol Phe,³⁵  respectively. These selenol amino acids were introduced into model peptides and were shown to successfully promote the ligation reaction. The chemoselective deselenisation of peptides containing unprotected mercaptoamino acids such as Cys and γ-thiol Pro was achieved yielding the desired peptides in good yields with no trace of dethyilated side-products.^34,35  The general selenol ligation-selective deselenisation approach is represented in Scheme 4. Although selenocysteine (Sec); owing to their inherent properties; are considerably more difficult to handle than Cys, in light of the recent advances in the selenol ligation-selective deselenisation strategy, it is expected that researchers will set a goal of extending the Cys surrogate approach to Sec.³⁶ 

In parallel to the impressive advances in Cys-free NCL, considerable progresses have also been made towards the development of improved procedures for the preparation of thioester peptides. Although Fmoc SPPS has outstrip the Boc strategy by providing a safer alternative, compatible with acid-labile PTM, Boc SPPS has remained the preferred technique for the preparation of peptide thioesters. Indeed, unlike Boc chemistry which allows the synthesis of peptide thioesters to be carried out on-resin, the sensibility of the thioester functionality to repeated piperidine treatment has cursed the Fmoc SPPS of peptide thioesters. However, a large repertoire of techniques has been examined in order to circumvent the problem. While highly acid-labile linkers have been used for the cleavage of fully protected peptides followed by in-solution thioesterification, various methods including backbone-amide linkers (BAL), O,S- or N,S-acyl shifts and safety-catch linkers (SCL) have been developed for the on-resin Fmoc-based synthesis of peptide thioesters (Scheme 5).³⁷ 

Despite the numerous strategies available, each of these techniques presents limitations which have so far prevented the breakthrough of one preferred method for the Fmoc SPPS of thioesters among the peptide chemist's community. Notably, the in-solution thioesterification after elongation by Fmoc SPPS raises solubility issues of the fully protected peptide. Furthermore, epimerisation of the C-terminus has also been reported. The main drawbacks of the BAL technique are the low reactivity of the secondary amine of the linker towards the coupling of the second amino acid as well as diketopiperazine formation. However, by masking the thioester as a trithioortho ester Brask et al.³⁸  have been able to considerably reduce this side-reaction. While the interesting method reported by Botti et al.³⁹  based on an O,S-acyl shift was hampered by hydrolysis of the ester, N,S-acyl shift auxiliaries suffer from various shortcomings including low yield of coupling of the first amino acid onto the N,S-acyl shift devices, epimerisation, thioester hydrolysis as well as diverse side-reactions.^40–44  Techniques following the safety-catch linker strategy present similar drawbacks. In addition, the reaction conditions employed to increase the reactivity of the acyl group to thiolysis are susceptible to alter the desired peptide thioesters.

Recently, new methods and improvements to the existing techniques to effect the synthesis of peptide thioesters have been reported. The Alewood group notably reported a high-throughput approach to peptide thioesters using a safety-catch linker strategy in Boc SPPS allowing for parallel HF cleavage.⁴⁵  By employing a SCL stable to both trifluoroacetic acid (TFA) and HF, and compartmentalising the peptide-bound resin beads with tea bags, they conducted the simultaneous HF removal of the side-chain protecting groups of six peptide thioesters without undesired cleavage of the peptide from the solid-support. Each peptide was next subjected to NH₄I/dimethyl sulphide (DMS) in order to activate the SCL linker, thus becoming labile in TFA. Although this technique affords good yields for the preparation of peptide thioesters in a library-style operating manner, therefore avoiding repeated handling of hazardous HF, it remains inconciliable with PTM. A Fmoc-based approach relying on the in-solution post-SPPS thioesterification of unprotected peptides was reported by Okamoto et al.⁴⁶  The selective activation of an amide bond present at the N-terminus of a Cys residue was achieved through Cys thiocarbonylation. Subsequent treatment with N-acetylguanidine in DMF or DMSO afforded the peptidyl-N-acetylguanidine. The latter underwent thiolysis in the presence of sodium 2-mercaptoethanesulfonate (MESNa), thus yielding the desired thioesters which could be used in NCL (Scheme 6).

In addition to the synthesis of model peptides, using this technique Okamoto et al.⁴⁶  also prepared a glycopeptide thioester bearing a nonasaccharide. Attempts to carry out NCL directly on the peptidyl-N-acetylguanidine compound in the presence of 4-mercaptophenylacetic acid (MPAA) failed, but the one-pot thioesterification/NCL proceeded with MESNa, thus illustrating the reactivity of the N-acetylguanidine as a leaving group. The specific reactivity of the N-acetylguanidine group was utilised for the preparation of a 35-amino acid glycopeptides by KCL. Despite the advance that represents the thioesterification of unprotected peptide, it remains necessary to orthogonally protect other Cys present in the peptide sequence as well as to carry out selective N-Boc protection of Lys residues prior to reaction with the N-acetylguanidine. Nevertheless it is the use of latent thioesters which is currently drawing peptide chemist's attention. Indeed, the N-sulfanylethylanilide (SEAlide) peptides developed by the Otaka group,^43,47  the bis(2-sulfanylethyl)amino (SEA) peptides independently reported by the Melnyk⁴⁸  and the Liu⁴⁹  groups as well as the hydrazide peptides from the Liu group⁵⁰  have recently emerged as efficient thioester surrogates for NCL ligation. These crypto-thioesters can be readily prepared by Fmoc SPPS using appropriate resin-linker and have proven to be useful tools for the one-pot multiple fragments assembly of proteins thanks to their easily tuned reactivities. Unlike standard KCL which depends upon the difference of reactivity of alkyl and aryl thioesters and is therefore limited to unreactive alkyl thioester, the thioester surrogates can be turned on and off to avoid undesired oligomerisation and therefore afford greater synthetic flexibility. While the activation of SEAlide and SEA peptides rely on N,S-acyl transfer triggered by the use of phosphate buffer and TCEP respectively, hydrazide peptides undergo in situ conversion to thioesters in presence of NaNO₂ and an external thiol at pH 3–4 (Scheme 7).

Otaka's SEAlide strategy was notably combined to Kent's KCL for the synthesis of monoglycosylated GM2 activator protein analogue GM2AP(1–162) through a 5 segments ligation approach.⁵¹  Similarly, Ollivier et al.⁵²  published the synthesis of the K1 domain of hepatocyte growth factor (HGF)(125–209) using SEA peptides. In details, sequential N to C ligation of peptide thioester fragment 125–148 with fragment 149–176 bearing a N-terminus Cys and a C-terminus SEA^off in the absence of TCEP yielded SEA^off segment 125–176, which was subsequently subjected to a second NCL with fragment 177–209 in the presence of TCEP. Adding TCEP to the reaction mixture allowed the reduction of the SEA^off disulfide bond, thus affording SEA^on fragment 124–176 which simultaneously reacted with N-terminus Cys segment 177–209. The total syntheses of GM2AP(1–162) and K1 HGF(125–209) are summarised in Scheme 8.

The peptide C-terminus thioester 125–148 was prepared by Fmoc SPPS by converting the corresponding SEA peptide to 3-mercaptopropionic acid (MPA) thioester. As reported by Dheur et al.,⁵³  SEA peptides are also practical N,S acyl transfer devices for the preparation of peptide thioesters. Using MPA at pH 4 Gly, Ala, Tyr and Val SEA peptides were converted to their parent thioesters without epimerisation. It should be noted that the preparation of SEA^off peptides required orthogonally protected Cys(SStBu) residues to be used during the elongation of the peptide sequence.

The propensity of SEA peptides to promote NCL at difficult ligation sites was also reported.^54,55  SEA peptides were either used as intermediates en route to novel thiazolidine thioester scaffold capable of accelerating the ligation rate at hindered ligation sites⁵⁴  or were used as it is for Pro ligation.⁵⁵  SEA peptides were efficiently converted into thioesters featuring a thiazolidine on the thiol handle by reaction with zinc at pH 1 followed by addition of glyoxylic acid. Both hindered Tyr and Val peptide thiazolidine thioesters showed a significant increase in NCL rate compared to their MPA thioester equivalents (e.g. 60- and 4.4-fold, respectively). Notably, NCL at Val afforded the desired ligated product in 47% after 48 h for the thiazolidine thioester instead of 33% after 7 days for the traditional MPA thioester.⁵⁴  Furthermore, the use of Pro-SEA peptides at reduced pH but increased temperature was shown to considerably decreased side-product formation observed during Pro-MPA thioester NCL, thus leading to higher isolated yields of the targeted products (Scheme 9).⁵⁵ 

Interestingly, the Melnyk group demonstrated the use of Fmoc-Asp(SEA^off)-OH and Fmoc-Glu(SEA^off)-OH building blocks for the synthesis of tail-to-side-chain cyclic peptides as well as branched peptides.⁵⁶  The building blocks were incorporated in the peptide sequence using standard Fmoc SPPS. The resulting peptides were either reacted with an intermolecular N-terminal Cyst peptide or intramolecularly with their N-Cys tails in the presence of TCEP and MPAA to afforded branched or cyclic peptides, respectively (Scheme 10).

The chemoselective hydrazide ligation described by the Liu group is of particular interest as it can be employed for both the ligation of synthetic peptide fragments and the ligation of expressed peptides and proteins.⁵⁰  Indeed recombinant peptide hydrazides were easily obtained through hydrazinolysis of the protein intermediate in the standard EPL. In addition, synthetic peptide hydrazides featuring all twenty genetically encoded amino acids except Asp, Glu and Gln at C-terminus were readily prepared and subjected to hydrazide ligation. Ligation occurred at rates comparable to standard NCL. Furthermore, the reaction conditions were found to be compatible with the functional groups of fully unprotected side-chains and no significant racemisation was detected. Interestingly, this novel technique is suitable for total synthesis of protein by sequential N to C and C to N fragments assembly relying solely on hydrazide ligation. Indeed, following the oxidation step, the peptide azide newly formed is reacted in situ with MPAA to form the reactive thioester intermediate. The excess MPAA quenches the oxidant initially introduced in the reaction mixture. Conducting the activation of the first hydrazide fragment prior to addition of the second N-Cys C-terminus hydrazide peptide therefore allows for hydrazine ligation to take place at the second peptide N-Cys terminus without affecting its hydrazide C-terminus, which is thus available for a subsequent activation-ligation cycle. In addition, the peptide hydrazides also are compatible with desulfurisation conditions, thus allowing them to be used in conjunction with Cys surrogates. Fang et al.⁵⁰  reported the successful synthesis of the 66-amino acids protein CssII through a four segment N to C sequential hydrazide ligation approach. The same group also described the convergent synthesis of 142-amino acids ribosomal protein S25 (RpS25).⁵⁷  Although their initial strategy relying on the three segments sequential N to C ligation of peptide fragment 1–69 failed due to the unwanted formation of thiolactone side-products in significant yields, the issue could be overcome by using a sequential C to N ligation approach instead. Thus, both 1–69 and 70–142 peptide fragments were prepared through three segments sequential C to N ligation using 2-(tert-butyldisulfanyl)ethyloxycarbonyl (Tbeoc) protected (R)-1,3-thiazolidine-4-carboxylic acid (Thz) N-terminus Cys C-terminus hydrazide peptide fragments. Next, fragment 1–69 was subjected to desulfurisation conditions prior to thioester conversion. The isolated thioester in turn underwent final hydrazide ligation to N-terminus Cys peptide fragment 70–142 yielding the desired RpS25 (Scheme 11).

Furthermore, the hydrazide ligation technique was recently employed for the head-to-tail cyclisation of 5- to 42-amino acids linear peptides.⁵⁸  Interestingly, an alternative route to peptide hydrazides was described by the Macmillan group.⁵⁹  Using the selective fragmentation approach at Gly-Cys and His-Cys they previously reported for the preparation of thioesters⁶⁰  in conjunction with hydrazinium acetate as a hydrazine additive, they achieved the hydrazinolysis of both synthetic and recombinant peptides and proteins. In summary, these novel techniques not only represent powerful tools for the reliable preparation of transient thioesters through mild Fmoc SPPS chemistry but also introduce an increased flexibility in the total synthesis of proteins.

Despite NCL being the most popular technique for the assembly of peptide fragments, other approaches have participated to enrich the ligation toolbox. The α-ketoacid-hydroxylamine (KAHA) ligation reported by the Bode group notably represents a viable alternative to NCL for the formation of native peptide bonds.⁶¹  This technique relies on the chemoselective ligation of unprotected peptide fragments based on the condensation of a C-terminus α-ketoacid and an N-terminus N-hydroxylamine and is therefore not limited by the Cys requirement and the difficulty in thioester synthesis which hamper NCL. Nevertheless, KAHA ligation presents its own drawbacks. Although the α-ketoacid segment can be readily prepared on-resin by oxidation of a cyanosulfur-ylide based linker,⁶²  the incorporation of the N-hydroxylamine moiety in the peptide remained troublesome. Furthermore, a general strategy for KAHA ligation compatible with the presence of aqueous solvents failed to be established, thus leading to solubility issues. Recent advances in KAHA ligation have been published by Pattabiraman et al.⁶³  which address these latter issues and thus are likely to expand the use of KAHA ligation for the total synthesis of proteins. In this report the traditional N-hydroxylamine has been replaced by 5-oxaproline (Opr) which not only is readily incorporated in the peptide fragment by Fmoc SPPS but also allows ligation to be conducted in aqueous solvents, affording a homoserine (Hse) residue at the ligation site (Scheme 12).

Analogues of the 63-mer prokaryotic-ubiquitin-like protein (Pup) and 66-mer probable cold shock protein A (cspA) were successfully prepared using the 5-oxaproline KAHA ligation in a two fragments strategy. Building on this discovery, Ogunkoya et al.⁶⁴  described the synthesis of C-terminal variants of ubiquitin-fold modifier 1 (UFM1) (2–83) by sequential C to N KAHA ligation of three peptide fragments. First, N-terminus Opr segment 61–83 was coupled with the middle segment Fmoc-Opr-(31–60)-α-KA. The fragment Fmoc-Opr-(31–83) thus formed was subsequently subjected to in situ Fmoc deprotection prior to RP-HPLC purification. A second KAHA ligation between Opr-(31–83) and C-terminus α-KA-(2–29) afforded the desired UFM1 (2–83) bearing Hse residues at position 30 and position 61 (Scheme 13).

Another method which expands the scope of peptide ligation beyond the Cys residue was recently described by the Li group.⁶⁵  They reported the chemoselective peptide bond forming reaction between N-terminal Ser/Thr peptides and C-terminal salicaldehyde ester peptides. The reaction proceeds through the formation of an N,O-benzylidene acetal intermediate followed by O,N-acyl transfer with a final acidolysis step affording the native amide bond (Scheme 14).

Interestingly, the Ser/Thr ligation of unprotected peptide fragments is racemisation-free and compatible with thioester functionality. The same group published a convenient preparation for the salicylaldehyde ester precursor by post-Fmoc SPPS derivatisation via phenolysis of peptide N-acyl-benzimidazolidinone (Nbz) with salicylaldehyde dimethyl acetal.⁶⁶  The Li group demonstrated the utility of the Ser/Thr ligation with the total synthesis of 98-mer human erythrocyte enzyme acylphosphatase and 44-mer human growth hormone-releasing hormone (hGH-RH).^66,67  Both proteins were prepared through three fragment C to N sequential ligation. While the Gly-Thr and Gly-Ser ligation sites chosen for the assembly of acylphosphatase allowed the salicylaldehyde ester segments to be prepared by direct coupling without epimerisation, the Nbz strategy had to be employed for the preparation of the C-terminal Leu- and Met-salicylaldehyde ester segments required for the synthesis of hGH-RH. In order to avoid self-condensation of the middle fragment during the first ligation step a temporary N-terminus protecting group was necessary. Although the use of Fmoc was found to give good results during the preparation of acylphosphatase, it was not compatible with the Nbz approach and partial loose of Fmoc was observed likely resulting from the use of diisopropylethylamine during the cyclisation of the 3,4-diaminobenzoic acid (Dbz) linker. p-(Methylsulfinyl)benzyloxycarbonyl (Msz) was used instead and afforded the desired N-protected C-terminus salicylaldehyde ester middle fragment. However it was found that the Met residue in the peptide sequence had to be replaced with Met sulfoxide in order to avoid Met-assisted reduction of Msz to its acid-labile p-(methylthio)benzyloxycarbonyl (Mtz) form. The Met sulfoxide was readily converted back to methionine during the Msz removal step prior to the second Ser ligation. The general strategy for the three-fragment synthesis of peptides through Ser/Thr ligation is summarised in Scheme 15.

Besides enabling the synthesis of long proteins, the Ser/Thr ligation also found use in the preparation of short cyclic peptides.⁶⁸  Indeed the challenging head-to-tail cyclisation of tetrapeptides which do not contain turn-inducing elements was successfully achieved through intramolecular Ser/Thr ligation. A Boc SPPS strategy relying on the use of a salicylaldehyde linker-resin was devised for the preparation of the salicylaldehyde ester peptides. The resin-bound peptides were readily cleaved by ozonolysis thus avoiding the use of HF. Nevertheless, Cys, Met and Trp residues were found to suffer oxidation under these conditions. Cyclisation of the tetrapeptides was shown to proceed through an imine-induced ring closure initially forming 16-membered rings, thus decreasing the cyclisation energy barrier compare to the direct intramolecular lactamisation of 12-membered peptides. The following contraction to 12-membered rings occurred through O,N-acyl transfer. The N,O-benzylidene thus formed was finally converted to Ser/Thr by acidolysis (Scheme 16).

The syntheses of nine cyclic tetrapeptides were investigated. At a concentration of 1 mM of linear peptides, no epimerisation took place during the cyclisation and the ratio of cyclomonomer versus cyclodimer was found to be comprised between 3:2 to 9:1, depending on the linear sequences. Notably, for cyclo-(SYIA) only the monomeric product was observed.

In conclusion, powerful alternatives to the traditional NCL technique are now being developed which not only allow for the total synthesis of complex proteins but also are viable methods for the preparation of small cyclic peptides otherwise difficult to access. Furthermore these methods are relieved from the restrictions inherent to standard NCL. In summary, critical advances have been made in recent years in the synthesis of peptides and proteins by ligation techniques. While attention is now turning towards Cys and/or Sec surrogates and crypto-thioesters as efficient tools to expand the scope of standard NCL and bring more flexibility to the synthetic design, novel ligation methods are also emerging which complement the peptide chemist's toolbox. These progresses are expected to tremendously facilitate and accelerate the access to complex peptides and proteins and thus stimulate important discoveries in biochemical sciences

The work in the laboratory of the authors is partially funded by the CICYT (CTQ2012-30930), the Generalitat de Catalunya (2009SGR1024), and the Institute for Research in Biomedicine Barcelona (IRB Barcelona).