Chemical synthesis of proteins using hydrazide intermediates

Protein chemical synthesis offers useful and otherwise-difficulty-to-obtain biomacromolecules for biological and pharmaceutical studies. Recently, the hydrazide chemistry has drawn attentions in this field as peptide or protein hydrazides can be used as key intermediates for different synthesis and modification purposes. Besides being a traditional bioorthogonal chemical handle, a hydrazide group can serve as a readily accessible precursor of a thioester. This strategy significantly improves the efficiency and scope of native chemical ligation for protein chemical synthesis. Here we review the chemical transformations of peptide or protein hydrazides and total/semi/enzymatic protein synthesis methods involving peptide or protein hydrazides. Several examples of protein chemical synthesis using peptide hydrazides as key intermediates are described.


INTRODUCTION
Proteins play a central role in the cellular processes of living organisms. The biochemical function carried out by a specific protein relies on its 3D architecture. This folded structure is determined by its primary polypeptide chain, a unique sequence usually composed of 20 canonical amino acids. To study the functions and mechanisms of protein molecules, we need to obtain sufficient quantities of homogeneous proteins of interest to perform biochemical and biophysical experiments [1]. Such experiments are needed to provide detailed understanding of structure-function relationship of proteins or protein complexes in the context of their biological or pathological processes.
Standard molecular biology relies on molecular cloning and recombinant expression to produce proteins, for instance from Escherichia coli. However, traditional recombinant expression often has difficulty in the production of proteins containing (1) post-translational modifications (PTMs) that serve as critical modulators of protein functions [2] and (2) non-proteineous groups such as fluorophores [3]. To gain access to these proteins that are of increasing interest to the biomedical stud-ies, researchers have exploited chemical synthesis to produce these otherwise-difficult-to-obtain proteins in vitro [4]. Our own work in the field indicates that protein or peptide hydrazides are useful intermediates for protein chemical synthesis.
In the present paper, we describe recent applications of protein or peptide hydrazides in protein chemistry, with an emphasis on hydrazide-based native chemical ligation (NCL).

HYDRAZIDES IN PROTEIN CHEMISTRY
Alkyl or aryl hydrazides are usually obtained by hydrazinolysis of the corresponding acyl halides or esters. Because hydrazide is a potent nucleophile with a much lower pK a value than an ordinary amine due to the α effect (e.g. pK a 's of glycylhydrazide = 2.38 and 7.69) [5], this functional group is able to undergo some unique chemoselective reactions. In protein chemistry, two types of reactions based on hydrazide reactivity are usually employed, (1) Schiff base/imine formation to yield acyl hydrazone for protein labeling and (2) hydrazide oxidation to yield acyl azides for condensation of peptide segments.

Acyl hydrazone formation
The condensation of aldehyde/ketone compounds with aminooxy/hydrazine derivatives to form oximes and hydrazones is a well-known reaction in classical organic chemistry (Fig. 1A). In an effort to make acyl hydrazone formation irreversible, reduction with sodium cyanoborohydride can be carried out [6]. For protein modification, several commercially available hydrazide probes have been developed to label proteins and other biomolecules both in vitro and in vivo (Fig. 1B). These probes include Alexa Fluor hydrazides, BODIPY hydrazides, biotin hydrazides and PEG-derived hydrazides.
In 2003, the hydrazide method was applied to quantitative profiling of N-linked glycoproteins ( Fig. 2A) [7]. First, sodium periodate is used to oxidize cis-diols within a carbohydrate into dialdehydes. Next, aldehydes covalently react with hydrazidefunctionalized resins to enrich glycoproteins of interest. Immobilized N-linked glycoproteins are then proteolyzed into short peptide segments and resins are washed extensively. Finally, the remaining N-linked glycopeptides are isotope labeled, released through enzymatic cleavage and quantitatively ana- lyzed by using LC-MS (liquid-phase chromatography/mass spectroscopy).
Another protein modification study taking advantage of hydrazide-aldehyde condensation was reported by Bertozzi et al. (Fig. 2B) [8]. They introduced a conserved six-amino-acid motif -Leu-Cys-Thr-Pro-Ser-Arg-into the protein of interest. This tag was termed 'aldehyde tag' because Cys in this sequence can be selectively oxidized to formyl glycine as catalyzed by exogenously added or endogenously coexpressed formylglycine-generating enzyme. After that, in vitro site-specific protein labeling can be achieved with different hydrazide probes. For example, aldehyde-tagged maltose-binding protein was condensed with biotin hydrazide and purified. The acyl hydrazone linked protein was then exchanged with a thermodynamically more stable aminooxybased FLAG probe for immunoblot assays.
In addition to protein side-chain modifications, acyl hydrazone chemistry has also been found useful for protein linear assembly [9,10], terminal modification [11] and cyclization [12]. Recently, Cotton et al. carried out site-specific C-terminal PEGylation using recombinant protein hydrazides and ketonemodified PEG polymers [11]. In the case of protein cyclization, the C-terminal hydrazide can selectively react with the N-terminal aldehyde to give cyclic proteins at pH 5 [12].  Overall, the acyl hydrazone forming reaction is bioorthogonal. The two functional groups are biologically inert and can forge a stable linkage at an acceptable rate when the reaction is performed under physiological pH and temperature. This bioorthogonal reaction is simple and many improved protocols have recently been introduced such as aniline catalysis [13] and tuning of substrate reactivity [14].

Acyl azide formation
Oxidative conversion of hydrazides to acyl azides was first observed by Curtius as early as the birth of synthetic peptide chemistry [15]. Amino acid azides and peptide azides are reactive to nucleophilic reagents and can be directly used in amide coupling to extend the peptide chain during both solution phase peptide synthesis and solid-phase peptide synthesis (SPPS).
The peptide coupling between an acyl azide and the amino group on resin or in solution was tested around 1970 [16,17]. Importantly, this method can retain the chiral integrity of the amino acyl azide because the cyclized oxazolone is generally not formed with the acyl azide [18]. Nonetheless, the utility of azide coupling in the process of solidphase peptide synthesis or in-solution peptide segment ligation is limited because (1) direct aminolysis of acyl azide is slow; (2) acyl azides are unstable at high temperature giving rise to Curtius rearrangement/degradation above 0 • C to produce isocyanate-derived byproducts [18].
To overcome the problems, Ramage et al. described a method by converting the in situ formed acyl azide into the more stable and isolable active ester or thioester (Fig. 3A) [19,20]. Such (thio)esters do not exchange with other carboxylic acid groups and can participate in stoichiometric coupling with external amines in organic solvents in a one-pot fashion. The main limitation of Ramage's approach is the use of organic solvents such as DMF and the need for protection of the amine and thiol groups. Many protected peptides have limited solubility in organic solvents as well as aqueous solutions, which can cause slow and incomplete couplings, together with purification and characterization difficulties.
Acyl azide chemistry was applied to synthesis of ubiquitin-based inhibitors of ubiquitin-activating enzymes and deubiquitinases in aqueous buffer [21,22]. An ubiquitin acyl azide was generated in situ from an ubiquitin hydrazide and underwent direct aminolysis with a large excess amount of amino nucleophile in a high overall yield. Use of excess nucleophilic amine is crucial to prevent unproductive intra-or intermolecular side-chain nucleophilic attack and hydrolysis of the acyl azide.

PEPTIDE HYDRAZIDES IN NCL
In 1994, Kent et al. in their seminal paper reported a native amide forming reaction known as NCL [23], which transformed the history of protein synthesis. NCL is essentially a chemoselective conjugation between a thioester and an N-terminal cysteine residue. In mechanism, the equilibrium between the starting thioester and the intermolecular thioester of the N-terminal cysteine is followed by a rapid irreversible S→N acyl shift at cysteine through a fivemembered ring transitional state. The most significant improvement of NCL is that it can take place between two unprotected peptide segments in aqueous buffer at low concentrations. Since its discovery, NCL has become a powerful method for protein total synthesis, semisynthesis and modification [24,25].
The preparation of thioesters by Fmoc-SPPS was one of the challenges in NCL, due to the sensitivity of thioesters to piperidine during Fmoc deprotection. We envisioned that unprotected peptide hydrazides may be converted to peptide thioesters by combination of nitrite oxidation and thiolysis. We anticipated that the main potential problem may be the reduction of azides by the phosphine additives or thiols in the ligation system. We also worried about the oxidation of side chains of unprotected amino acids such as Cys, Trp and His. Fortunately, our experiments showed that the above problems did not occur. In fact, peptide azides could be quantitatively converted to peptide thioesters with thiols. Although cysteine can be oxidized by REVIEW nitrite in the activation step, and the oxidized cysteine can be cleanly reduced back to cysteine by thiols. Thus, peptide hydrazides are good thioester surrogates for total chemical protein synthesis and protein semisynthesis.

Thiolysis of fully unprotected peptide hydrazides
A peptide hydrazide can be chemoselectively converted to a peptide thioester in a racemization-free manner (Fig. 3B) [26] in aqueous buffers [27]. When an unprotected peptide hydrazide, typically 0.5-2 mM in 6 M guanidinium chloride buffer, is treated with 5-20 mM NaNO 2 at low pH (3.0-4.0) and low temperature (−20 to −10 • C), the hydrazide can be oxidized to the acyl azide smoothly within 30 min [28]. The only affected residue is cysteine that can be oxidized to S-nitroso cysteine.
Once an external thiol such as 100 mM MPAA (4-mercaptophenylacetic acid) is added and the reaction mixture is kept at pH 5.0-6.0 at room temperature for 10 min, the acyl azide can be quantitatively converted to the corresponding thioester. The Snitroso cysteine can be cleanly reduced back to cysteine in the presence of MPAA. This hydrazide-tothioester conversion is fully compatible with all 20 amino acids located in the middle of the peptide sequence. For the C-terminus, three amino acids Asp, Asn and Gln are not appropriate, as they may lead to unproductive intramolecular cyclization products [28]. The thioester product can either be purified by using semipreparative HPLC (high-resolution liquid-phase chromatography) or used directly for NCL in a one-pot oxidation-thiolysis-ligation manner [28,29].
To examine the robustness of the above method, we synthesized a model peptide antibiotic trifolitoxin [Ala23Cys] by ligating two segments H-[Met 1 -Ala 22 ]-NHNH 2 and H-[Cys 23 -Ala 42 ]-OH. The ligation was performed using a one-pot protocol in an excellent 97% HPLC yield [28]. This example provides convincing evidence that the one-pot reaction is chemoselective because an internal Cys and nine Lys residues did not interfere with the peptide fragment condensation.

Multiple segment ligation of peptide hydrazides
Because typical proteins contain several hundred amino acid residues and typical SPPS can only produce peptides with ∼50 amino acids, multiple peptide segments ligation is required to achieve efficient synthesis of proteins [30]. Sequential ligations in either the N-to-C or the C-to-N direction re-quire temporary protection of the thioester or the N-terminal cysteine, respectively. Thz protection of Cys is the most popular way to fulfill C-to-N sequential NCL [31]. However, use of protected thioesters or crypto-thioesters can also be considered [32]. A peptide hydrazide turned out to be a desirable choice [28,33] among different crypto-thioesters [34][35][36][37][38][39][40].
We tested the idea with a 142 residue ribosomal protein S25 (RpS25) [33]. This protein was divided into six segments and synthesized through convergent ligations. We found that Thz was unstable to the hydrazide oxidation condition but Tbeoc (2-(tertbutyldisulfanyl)ethyloxycarbonyl) protection [41] of Thz could overcome this problem. Thus, Tbeoc-Thz protected N-terminal Cys C-terminal hydrazide peptides were synthesized and used in the convergent NCL protocol towards Rps25. This example showed that the application of masked Cys and masked thioester functional groups, in this case Tbeoc-Thz and hydrazide, enables flexibility for the design of efficient sequential or convergent total synthetic route towards proteins.

One-pot chemical ligation of peptide hydrazides
A frontier of protein chemical synthesis is the onepot ligation approach that permits condensation of three or more peptide segments [30]. Compared with traditional sequential ligation approach, onepot experiment only needs one final purification step that can save the time and labor cost in the handling of intermediate peptides. Many recent examples have demonstrated the practical efficiency and operational simplicity of one-pot chemical ligations [31,42,43].
Mambalgin-1 is a 57-residue cysteine-rich toxin peptide isolated from the venom of mamba snake. The absence of its structure and the low yield and purity through natural isolation called for chemical synthesis of mambalgin-1. Initial attempts using stepwise SPPS of Mambalgin-1 failed as almost no product was detected. Multiple ligation-purification approach was not efficient either with a low yield (7.3%). By using an azide protected Cys [44] as the middle segment, three-segment ligation was successful with a good yield (35%) (Fig. 4) [45]. The middle segment and the C-terminal segment were ligated first followed by TCEP-induced azide cleavage to expose a free Cys for the second ligation. The second ligation was then performed by adding an N-terminal thioester segment pre-converted from the corresponding peptide hydrazide. Functional studies of folded mambalgin-1 and NMR structural determination revealed that mambalgin-1 is a typical member of the three-finger toxin family.  Post-translationally modified histones are intriguing targets for chemical protein synthesis. Total synthesis enables production of histones with multiple modified residues. In 2014, we reported a one-pot milligram-scale synthesis of two histones, methylated H3 and acetylated H4 [46]. We prepared a new Cys protecting group called Dobz (p-boronobenzyloxycarbonyl) [47] and incorporated the building block Dobz-Cys(Trt)-OH during SPPS. Dobz can be deprotected through selective H 2 O 2 oxidation. This treatment does not make any pH change and releases water as the only side product. In our model test, H 2 O 2 deprotection is fast and does not affect other oxidation-sensitive amino acids such as Cys, Met and Trp. By using this one-pot ligation method, we synthesized Lys4-trimethylated H3 (H3K4me3) and Lys16-acetylated H4 (H4K16ac) [46]. High-resolution LC-MS and in vitro nucleosome reconstitution experiments confirmed the structural homogeneity and integrity of synthetic H3K4me and H4K16ac.
More recently, our group reported a one-pot four-segment ligation method for protein chemical synthesis [48]. This method introduces a new Tfacm (trifluoroacetamidomethyl) group to protect Cys temporarily. Deprotection can be performed by adjusting reaction pH to 11. As no external electrophilic or oxidative reagent is added, the reaction mixture is clean, which is beneficial for multiple NCL to be carried out at mM concentrations of peptide segments. In a typical onepot four-segment synthesis of a model protein crambin, thioester segments H-[Thr 1 -Ala 15 ]-SR, Tfacm-[Cys 16 -Gly 31 ]-SR, Tfacm-[Cys 32 -Thr 39 ]-SR were synthesized and purified from their hydrazide precursors. Then crambin was pieced together in the C-to-N direction starting from H-[Cys 40 -Asn 46 ]-OH. HPLC analysis showed that in each step of the one-pot process, the anticipated intermediate was produced as the predominant peptide species. Assembly of the full-length crambin was completed within 30 h in 40% yield. X-ray structure of the folded synthetic crambin was determined to be identical to that of the previous study. In another example, a human chemokine hCCL21 was synthesized using the similar one-pot procedure.

Applications of peptide hydrazides to protein chemical synthesis
Several studies utilizing the chemical ligation of peptide hydrazides were reported in recent years. The synthetic procedures employed the thioestermasking property of hydrazides to achieve sequential or convergent multiple ligations.
In 2013, Allis et al. reported their identification of an asymmetric dimethylation site Arg42 of histone H3 (H3R42me2a) in mammalian cells [49]. They found that H3R42me2a is modified by a methyltransferase coactivator and an arginine methyltransferase in vitro and in vivo. To study the effect of this PTM on DNA transcription, semisynthesis provides a perfect solution because enzymatic methylation of H3 is incomplete and no homogeneous H3R42me2a can be prepared in this way. The location of Arg42 in the middle of the sequence leads to a three-segment ligation strategy. First, H-[Ala 1 -Ser 28 ]-SR and H-[Cys 29 -Val 46 (R42me2a)]-NHNH 2 were chemically synthesized while the third segment H-[Cys 47 -Ala 135 ]-OH was expressed. N-to-C sequential ligations were performed to afford the desired full-length H3R42me2a in good purity and yield. Then the authors managed to incorporate H3R42me2a into histone core octamers with recombinant H2A, H2B and H4. A direct stimulatory effect of arginine methylation in transcription was observed. This is the first evidence that methylation of 'non-tail-region' arginine residues can exert positive effects on transcription.
Later, Unverzagt et al. described a robust semisynthesis of glycosylated human interleukin-6 (IL-6) [50]. Because the glycosylation site is in the middle of the polypeptide sequence, the authors resorted to a three-segment ligation strategy in which the N-terminal and C-terminal segments were both produced through intein-mediated recombinant expression. The middle segment was chemically synthesized as a GlcNAc-modified short peptide hydrazide. GlcNAc modification did not REVIEW cause any side reactions in the process of sequential ligations. The crude ligation mixture was dialyzed under reductive conditions and then went through oxidative folding and gel filtration. The optimized refolding gave highly pure full-length glycoyslated IL-6 based on LC-MS, CD (circular dichroism) and SDS-PAGE analysis. Another glycoform of IL-6 with a biantennary nonasaccharide was also synthesized. Activities of both glycoforms of IL-6 were found to be equal to the recombinant IL-6 with no sugar modification. Key to this study is the use of chemically synthesized glycopeptide hydrazides that provided rapid access to diverse glycoforms of human IL-6.
Another paper by Kay et al. focused on the ability of chaperones to fold mirror-image proteins. Mirrorimage proteins consist of D-amino acids and are promising tool molecules of chemical biology. They synthesized a 312-residue bacterial protein, DapA, in both L-and D-form, the largest protein ever totally synthesized [51]. Retrosynthetic design split DapA into seven segments including six peptide hydrazides and one peptide amide. Using optimized Fmoc SPPS and convergent assembly operations, they successfully synthesized the full-length DapA in L-and D-chirality. Although microheterogeneity was detected according to high-resolution LC-MS and enzymatic assays, DapA was sufficiently pure to be used for chaperone-mediated folding assay. The authors found that chaperone pairs GroEL/ES can fold both enantiomeric DapA. This experiment supported a previous mechanism that GroEL/ES assists folding mainly through hydrophobic interactions with no bias against the protein's chirality.
Our group recently reported a potentially general approach for chemical synthesis of membrane proteins. Although chemical synthesis is an important alternative to produce otherwise low-yielding membrane proteins, difficulty in preparing hydrophobic membrane peptides sets a bottleneck. We developed a reversible Arg 4 -tagged backbone protecting group for membrane peptides and combined it with the use of peptide hydrazide-based NCL to allow practical synthesis of membrane proteins [52]. The removable backbone tag ensures that membrane peptides behave like the ordinary peptides in the process of SPPS, purification and ligation. This method demonstrates good efficiency and robustness in the synthesis of Ser64 phosphorylated influenza A virus M2 proton channel and a human potassium channel Kir5.1. M2 was incorporated into lipid bilayers and produced channel activity in acidic buffers. Kir5.1 was reconstituted into a homogeneous functional tetramer to measure its channel conductance. Both synthetic membrane proteins exhibited their expected functions in vitro.
In collaboration with Qi et al., we used chemically one-pot synthesized two-photon-activatable chemokine CCL5 to study lymphocyte migration [53]. To ensure fast and tissue-penetrable light irradiation to uncage the activity of CCL5, we chose a two-photon-labile protecting group to mask Ser1 and Ser5 in the polypeptide sequence of CCL5. One-pot ligations went smoothly and the synthetic CCL5 could be uncaged under 720-nm irradiation with high spatiotemporal precision (<0.1 μm 2 ; <5 s). Its application in artificial lymphocyte repositioning was established. Besides we showed that on the single-cell level, T cells do not require PI3K to perceive the directional hint. This is the first work displaying the application of totally synthetic photocaged protein applied to in vivo immunology studies.
In a more recent paper, we developed a Dpeptide antagonist blocking the PD-1/PD-L1 interaction which is an important cancer immunotherapy target [54]. Hydrolysis-resistant D-peptide antagonists were obtained through mirror-image phage display [55]. First, we synthesized the D-form IgV domain of PD-L1 termed D IgV PD-L1 using hydrazide-based NCL. This D-form protein was then applied to classical phage display screening to obtain L-peptide antagonists of D IgV PD-L1 . In principle, D-versions of the selected L-peptide binders should bind L IgV PD-L1 . As expected, a D-peptide antagonist D PPA-1 was found to inhibit PD-1/PD-L1 interaction both in vitro and in vivo. This study suggested the potential of proteolysis-resistant Dpeptides as therapeutics for cancer immunotherapy.

SYNTHESIS OF PEPTIDE HYDRAZIDES
Because the peptide hydrazide is a useful reagent in protein chemistry, to increase the flexibility of protein synthesis, there is a need to develop versatile methods to produce this important synthon.

Chemical synthesis of peptide hydrazides
Peptide hydrazides can be obtained through both Boc and Fmoc SPPS. Starting from hydroxylfunctionalized resin (PAM resin for Boc SPPS and Wang resin for Fmoc SPPS), one can treat it with p-nitrophenyl chloroformate and then hydrazine hydrate to produce the hydrazine carbonyloxymethyl resin [28,56]. This resin can be used fresh to synthesize peptide C-terminal hydrazides. A major drawback is the overcoupling of glycine residues onto the proximal amine of the hydrazine resin. In this regard, we then used a hydrazine 2-chlorotrityl chloride (2CTC) resin with improved yield and purity of crude peptide hydrazides [57]. Huang et al. 113 However, hydrazine hydrate replacement of the commercially available 2CTC resin causes partial hydrolysis that can lower the loading capacity of the resin. Although anhydrous hydrazine might be used, this reagent is hazardous and dangerous. Therefore, we designed a third-generation hydrazine resin called Fmoc-hydrazine 2CTC resin with high loading capacity [58]. First, the 2CTC resin was swollen in CH 2 Cl 2 and reacted with Fmoc hydrazine hydrochloride and base in DMF/CH 2 Cl 2 (1/1) at 0 • C to room temperature. The remaining free chloride was treated with methanol to block its activity. Extensive washing and drying under oil pump vacuum gave light-yellow Fmoc-hydrazine 2CTC resin. An additional advantage of our thirdgeneration hydrazine resin is that it can be stored at fridge for several months owing to Fmoc protection. Other methods to obtain peptide hydrazides include on-resin hydrazinolysis of C-terminal esterlinked polypeptides [59] and post-chain-assembly photolysis [60].

REVIEW
In 2013, Macmillan group presented a complementary route to peptide hydrazides. They found that peptides containing -Gly-Cys-and -His-Cysmotifs can undergo direct hydrazinolysis at the amide bond before Cys under native or denatured conditions (Fig. 5A) [61]. For example, human erythropoietin [A 1 -C 29 ] terminated with a -Gly-Cysmotif was exposed to hydrazinolysis in denatured buffer. No hydrazinolysis occurred across the internal -Ile-Cys-motif and a glycosylated residue was not affected. More recently, Otaka et al. developed a new methodology to prepare peptide hydrazides. They noted a previous report that peptide motif -Ser/Thr-Xaa-His-Zaa-(Xaa and Zaa can be any amino acids except Pro) can be hydrolyzed at the peptide bond before Ser/Thr, a process called Ni(II)-catalyzed hydrolysis (Fig. 5B) [62]. Their speculation was that the same peptide bond should also be cleaved off by alcoholysis in an aqueous alcohol system to generate peptide esters. As anticipated, this Ni(II)-catalyzed alcoholysis followed by simple one-pot hydrazinolysis can release peptide hydrazides in moderate to good yields [63]. Importantly, both Macmillan and Otaka's hydrazinolysis protocols can be applied to recombinant peptide/protein substrates (See next section) because their enzymatic recognition motifs comprise native residues [61,63].

Recombinant synthesis of peptide hydrazides
We envisioned that recombinant peptide hydrazides would be useful substrates for protein semisynthesis. Traditionally, recombinant peptide thioesters are produced through intein-mediated acyl transfer in the presence of an external thiol. The expressed thioester can then be used in NCL also termed as expressed protein ligation. To our delight, the acyl transfer step can be interrupted by hydrazine as an external nucleophile that generates a peptide hydrazide (Fig. 6A) [28]. Inspired by this thioester-hydrazinolysis mechanism, we figured that an oxoester-hydrazinolysis should also produce peptide hydrazides. Indeed, by incorporating an αhydroxy acid into the protein backbone and followup hydrazinolysis, recombinant protein hydrazides can be readily obtained (Fig. 6B) [64]. We used these two approaches to prepare two representative proteins, LC3 and HdeA.

Enzymatic synthesis of peptide hydrazides
As there is a need for developing specific inhibitors of ubiquitination or deubiquitination process, Wilkinson et al. explored the use of ubiquitin derivatives with C-terminal artificial modifications to interrupt the ubiquitin pathway. Ubiquitin C-terminal hy- drazide is an attractive molecular module that can be further functionalized to prepare diverse inhibitors. Ubiquitin is a 76-residue superstable globular protein that is susceptible to trypsin cleavage only at -Arg 74 /Gly 75 -site. When trypsin digestion was conducted in the presence of high concentrations of an external reagent glycylglycine ethyl ester, ubiquitin [Met 1 -Gly 76 ] C-terminal ethyl ester can be isolated (Fig. 7A) [65,66]. This ethyl ester can then be converted to the ubiquitin hydrazide almost quantitatively [21].
Our group developed another enzymatic reaction to prepare peptide hydrazides (Fig. 7B) [67]. Sortase recognizes a specific sequence -Leu-Pro-Xaa-Thr-Gly-, cleaves the peptide bond after Thr and produces a thioester intermediate. This intermediate can react with an external nucleophile containing Nterminal Gly, forming the ligation product -Leu-Pro-Xaa-Thr-Gly'-. We reported a new sortase-mediated hydrazinolysis reaction in which the thioester intermediate was captured by hydrazine irreversibly to produce protein hydrazides. Besides, an external acyl hydrazide can be used as the trapping nucleophile to give the diacyl hydrazide product. This is potentially favorable to protein C-terminal labeling and modification. Although an additional sequence '-Leu-Pro-Xaa-Thr-' has to be incorporated in the protein, main advantages of this sortase-mediated hydrazinolysis approach include high yield and simple operation.

CONCLUSIONS AND PERSPECTIVE
Hydrazides are interesting bioorthogonal reagents in chemical biology and have found increasing applications in synthetic protein chemistry. Hydrazides are convenient to prepare and transform. They can be used directly through traditional acyl hydrazone forming reaction or indirectly as crypto-thioesters for single and multiple NCL leading to efficient protein chemical synthesis. Improved protein chemical synthesis is expected to open new opportunities for the fundamental studies on protein PTMs as well as protein pharmaceutics [68,69]. In addition to providing a useful tool in protein biochemistry and biophysics, protein chemical synthesis is also a long-standing question of synthetic chemistry. The mere wonder of synthesizing a complex protein with correct tertiary structure and full biological activity continues to challenge the discipline of chemical synthesis.

FUNDING
This work was supported by the National Natural Science Foundation of China (21532004 and 21225207) and the Specialized