Convergent Chemical Synthesis and High-Resolution X-Ray Structure of Human Lysozyme

Convergent chemical synthesis and high-resolution x-ray structure of human lysozyme

Thomas Durek, Vladimir Yu. Torbeev, and Stephen B. H. Kent*

Institute for Biophysical Dynamics, Department of Biochemistry and Molecular Biology, Department of Chemistry, University of Chicago, 929 East 57th Street, Chicago, IL 60637

Edited by Brian W. Matthews, University of Oregon, Eugene, OR, and approved January 9, 2007 (received for review November 30, 2006) In this article, we report the total chemical synthesis of human relatively small (Ͻ50-residue) peptide segments, which span the lysozyme. Lysozyme serves as a widespread model system in entire sequence of the polypeptide chain of the protein. These various fields of biochemical research, including protein folding, unprotected segments then are linked by chemical ligation reactions enzyme catalysis, and amyloidogenesis. The 130-aa wild-type to give the full-length polypeptide, which then is folded in vitro to polypeptide chain of the human enzyme was assembled from four form the active protein. Critical to this synthetic strategy was the polypeptide segments by using native chemical ligation in a fully development of chemical ligation approaches, which made possible convergent fashion. Key to the assembly strategy is the application the chemoselective linking of unprotected peptide segments in good of the recently developed kinetically controlled ligation method- yield (17). Native chemical ligation (NCL) (18) is the most suc- ology, which provides efficient control over the ligation of two cessful chemoselective reaction developed so far and has enabled peptide ␣thioesters to yield a unique product. This result enables the synthesis of a number of proteins, which often were equipped the facile preparation of a 64-residue peptide ␣thioester; this with nonnative features (such as biophysical probes, backbone segment is joined by native chemical ligation to a 66-aa Cys modifications, D-amino acid residues, or glycan mimetics) to ad- peptide, to yield the target 130-aa polypeptide chain. The synthetic dress specific experimental questions (19–22). NCL involves the polypeptide chain was folded in vitro into a defined tertiary reaction of an unprotected peptide ␣thioester with another unpro- structure with concomitant formation of four disulfides, as shown tected peptide carrying an N-terminal cysteine. Initial reversible by 2D TOCSY NMR spectroscopy. The structure of the synthetic transthioesterification between the sulfhydryl group of the N- human lysozyme was confirmed by high-resolution x-ray diffrac- terminal cysteine and the peptide ␣thioester gives a thioester-linked tion, giving the highest-resolution structure (1.04 Å) observed to intermediate, which spontaneously rearranges in a rapid second date for this enzyme. Synthetic lysozyme was obtained in good step to form a native peptide bond (18). yield and excellent purity and had full enzymatic activity. This facile Most proteins synthesized so far by NCL have been constructed and efficient convergent synthesis scheme will enable preparation from merely two peptide segments, thus limiting chemical access to of unique chemical analogs of the lysozyme molecule and will target proteins of Ϸ100 or fewer amino acids (13). To gain synthetic prove useful in numerous areas of lysozyme research in the future. access to longer polypeptide chains, ligation of a larger number of peptide segment building blocks must be used. To date, essentially chemical protein synthesis ͉ kinetically controlled ligation ͉ native chemical all three-segment syntheses have been performed in a rather ligation ͉ peptide thioester ͉ protein folding inflexible fashion by sequential ligations starting from the C- terminal peptide segment with extension toward the N terminus. ysozyme is possibly one of the best studied enzymes. The x-ray Multiple rounds of ligation and intermediate product purification Lstructure of hen egg-white lysozyme, initially reported in 1965, typically result in substantial losses. This problem has been mini- was the first high-resolution 3D structure of an enzyme molecule mized by carrying out several ligations in a one-pot manner (23), but (1). Since then the protein has served as a model system for the the rapid build up of impurities effectively limits such one-pot study of protein folding and misfolding, enzyme catalysis and syntheses to only three segments. For these reasons, a more efficient mechanism, x-ray crystallography, enzyme evolution, and protein convergent synthetic strategy is needed. engineering (2–9). Moreover, human lysozyme recently has at- We recently introduced the concept of kinetically controlled tracted considerable interest because certain mutations in the ligation (KCL) (24), which enables the reaction of a peptide enzyme were shown to render the protein amyloidogenic (10, 11). thioarylester and a Cys–peptide thioalkylester to yield a single Despite extensive genetic, structural, and physico-chemical stud- product. This process enables the synthesis of a protein in a fully ies carried out over the last 50 years, many questions regarding convergent fashion (24). In a convergent synthesis (Scheme 1), each lysozyme folding, catalysis, and amyloid fibril formation remain starting peptide segment is approximately the same number of unsolved, unsatisfactorily explained, or controversial. This deficit is chemical transformations away from the final product (25). This at least in part attributable to the limited means that could be used to modify the chemical structure of the lysozyme molecule. More powerful and versatile control over the structure of the enzyme is Author contributions: T.D. and S.B.H.K. designed research; T.D. and V.Y.T. performed required for a detailed understanding of the properties of the research; T.D., V.Y.T., and S.B.H.K. analyzed data; and T.D. and S.B.H.K. wrote the paper. protein on the molecular or atomic scale. Chemical protein syn- The authors declare no conflict of interest. thesis has emerged as a powerful tool in this respect, especially This article is a PNAS direct submission. because it grants nearly absolute control over the covalent structure Abbreviations: SPPS, solid-phase peptide synthesis; NCL, native chemical ligation; KCL, of an enzyme molecule (12–14). Given the widespread and long- kinetically controlled ligation; MPAA, 4-mercaptophenylacetic acid; Thz, 1,3-thiazolidine- 4-carboxylic acid; MESNA, sodium 2-mercaptoethanesulfonate; obsd., observed; calcd., lasting interest in lysozyme research, it is not surprising that for calculated; TCEP, Tris(2-carboxyethyl)phosphine hydrochloride; LC, liquid chromatogra- exactly these reasons chemical synthesis of the full-length protein phy; RP-HPLC, reversed-phase HPLC; ESI, electrospray ionization; TFA, trifluoroacetic acid. has been envisioned (and attempted) as early as the 1970s (15, 16). Data deposition: Structure factors and coordinates described in this paper have been However, these early experiments were unsuccessful, and despite deposited in the Protein Data Bank, www.pdb.org (PDB ID code 2NWD). significant progress made in the field, robust chemical access to the *To whom correspondence should be addressed. E-mail: [email protected]. Ϸ130-aa family of lysozyme proteins has not been established so far. This article contains supporting information online at www.pnas.org/cgi/content/full/ Modern approaches to the chemical synthesis of proteins involve 0610630104/DC1. the synthesis by stepwise solid-phase peptide synthesis (SPPS) of © 2007 by The National Academy of Sciences of the USA

4846–4851 ͉ PNAS ͉ March 20, 2007 ͉ vol. 104 ͉ no. 12 www.pnas.org͞cgi͞doi͞10.1073͞pnas.0610630104 Downloaded by guest on September 25, 2021 Scheme 1. Convergent synthesis of human lysozyme. The 130-aa polypeptide is assembled from four segments of comparable length in a symmetrical fashion. Key to the synthetic strategy used is the KCL of [Lys1-Trp(CHO)64]-␣thioarylester and [Cys30-Trp- (CHO)64]-␣thioalkylester and the temporary protection of Cys65.(Inset) The 130-aa target (wild-type) sequence of human lysozyme with cysteines underlined. Steps: (i) transthioesterification; (ii) KCL; (iii) transthioesterification; (iv) NCL; (v) Cys deprotection; (vi) NCL; and (vii) Acm/CHO removal, disulfide formation, and folding. R ϭ alkyl. BIOCHEMISTRY fact becomes particularly significant when multiple analogs of a arylester (28). Formation of the C-terminal half of human lysozyme given target have to be prepared and the sites of modification are corresponding to residues 65–130 can be accomplished by ligating scattered across the entire sequence. Convergent synthesis, in [Thz65-Ala94]-␣thioalkylester with [Cys95-Val130]. In this case, the principle, also will increase final yields when compared with se- N-terminal Cys65 needs to be protected temporarily, which can be quential assembly tactics (25). achieved by using 1,3-thiazolidine-4-carboxylic acid (Thz) (23, 29). In this article, we report the reproducible convergent chemical In the convergent synthetic scheme, the full-length product [Lys1- 130

synthesis of human lysozyme, a 130-aa residue protein molecule Val ] is obtained by ligating the two halves of the target polypep- CHEMISTRY containing four disulfides. The structure of the synthetic protein tide, after which folding and disulfide formation are used to yield was confirmed by mass spectrometry, 2D TOCSY NMR, and the native enzyme molecule. high-resolution (1.04 Å) x-ray diffraction. The synthetic enzyme had full catalytic activity. Synthesis of Peptide Segment Building Blocks. Peptide segments were prepared by using in situ neutralization Boc chemistry Results and Discussion SPPS protocols as described in ref. 26. Segments 1–29, 30–64, Designing a Synthesis of Human Lysozyme. Our convergent strategy and 65–94 were prepared on modified TAMPAL resins gen- for the total chemical synthesis of the human lysozyme molecule is erating C-terminal ␣thioalkylesters upon HF cleavage (27). shown in Scheme 1. Human lysozyme contains eight cysteines, Segment 95–130 carrying a free ␣carboxyl group was synthe- which are distributed evenly across the polypeptide chain and form sized on ϪOCH2-Pam resin. All five tryptophans were incor- four disulfide bonds in the folded enzyme molecule (Scheme 1). All porated as Trp(CHO), and His78 was incorporated as cysteines could serve as potential ligation sites. We envisioned a His(Dnp). As expected, both the Trp(CHO) and His(Dnp) final ligation of two large peptide segments to give the 130-residue side-chain protecting groups were unaffected by the HF/p- polypeptide chain. To optimize the synthesis, we used four peptide cresol treatment used for deprotection and cleavage from the segments of approximately equal length: peptides 1–29, 30–64, resin. These protecting groups typically are removed under 65–94, and 95–130. These peptides are of convenient length conditions that are not fully compatible with thioesters and (Ϸ29–35 aa), and their preparation by optimized stepwise Boc thus were removed at later stages of the synthesis. chemistry SPPS (26) was expected to be straightforward. In terms During our initial studies on ligation reactions between of reaction rates, the chosen ligation sites (Met29-Cys30, Trp64-Cys65, Cys-containing lysozyme peptides, we observed unusually slow and Ala94-Cys95) were selected to be straightforward, i.e., product reaction rates (Ͼ24 h until completion), which we attributed to formation was expected to occur rapidly (27) (unlike with Ile/Val/ the large number of unprotected internal cysteines forming Thr/Pro-Cys ligation sites). unproductive internal thioesters (thiolactones), which slow According to Scheme 1, generation of the N-terminal half of the down product formation. This finding seemed to be even more enzyme (corresponding to residues 1–64) can be achieved by KCL pronounced when preformed ␣thioarylesters, which are more of [Lys1-Met29]-␣thioarylester with [Cys30-Trp(CHO)64]-␣thioalky- activated toward nucleophilic attack than standard ␣thioalkylester to form [Lys1-Trp(CHO)64]-␣thioalkylester. The peptide lesters are, were used (28, 30). In one particular case, incu- ␣thioarylester was generated by transthioesterification from the bation of a peptide ␣thioester containing a single ‘‘internal’’ corresponding ␣thioalkylester by using an excess of 4-mercaptophe- cysteine residue under ligation conditions led to almost quan- nylacetic acid (MPAA). The latter compound recently was shown titative thiolactone formation within seconds. Although such to be a superior thiol catalyst to the established and widely used a side reaction would only slow down a normal NCL reaction, thiophenol, and it is particularly useful for the facile conversion of we reasoned that the effect of thiolactone formation on a KCL a peptide ␣thioalkylester to the corresponding peptide ␣thio- would be more deleterious because kinetic control relies on the

Durek et al. PNAS ͉ March 20, 2007 ͉ vol. 104 ͉ no. 12 ͉ 4847 Downloaded by guest on September 25, 2021 superior reactivity of ␣thioarylesters versus ␣thioalkylesters.† Therefore, we decided to use the S-Acm group to protect the thiol-containing side chains of all five nonligation site cysteines.‡

Assembly of the N-Terminal Half. For KCL of [Lys1-Met29]-␣thioester and [Cys30-Trp(CHO)64]-␣thioester, the segment 1–29 first was transthioesterified with MPAA in a separate reaction (Scheme 1i) (24, 28). The resulting purified [Lys1-Met29]-␣thio(4-carboxymethyl)phenyl ester 1 was reacted with [Cys30-Trp(CHO)64]- ␣thioalkylester 2 in 6 M guanidine hydrochloride/0.2 M sodium phosphate under KCL conditions, i.e., in the absence of added thiol catalyst (Scheme 1ii). Liquid chromatography (LC) analysis (Fig. 1) indicated that most starting material was consumed within 5 h, and a heterogeneous compound mixture consisting mainly of [Lys1- Trp(CHO)64]-Cys30-␣thiolactone 3 and the branched thioester [Lys1- Cys30([Lys1-Met29])-Trp(CHO)64]-␣thioester 4 had been formed (Fig. 1B). The product mixture was resolved by adding sodium 2-mercaptoethanesulfonate (MESNA), which effectively reversed all undesired side products by transthioesterification to give the N-terminal half [Lys1-Trp(CHO)64]-␣MESNA thioester 5 in good yield [51% after purification; LC-MS observed (obsd.) mass, 7,799.6 Ϯ 0.8 Da; calculated (calcd.) mass, 7,798.7 Da (average isotopes)] (Fig. 1C). The formation of such unwanted thioester species is typical of KCL and represents an undesirable side reaction. Especially formation of the branched thioester 4 can be expected to be the principal reason for nonquantitative KCL (aside from oligomerization and cyclization) because it dissipates precious activated peptide ␣thioarylester.

Assembly of the C-Terminal Half. The C-terminal half of lysozyme representing residues 65–130 was obtained by ligating [Thz65- Ala94]-␣thioester and [Cys95-Val130]. Segment 65–94 contains a C-terminal ␣thioester and an N-terminal cysteine. To prevent undesired side reactions (cyclization, oligomerization, etc.), during the envisaged first ligation reaction with segment 95–130 we protected Cys65 as the Thz. Conversion of Thz to Cys is achieved readily by treatment with alkoxyamines at pH 4, generating the desired free N-terminal cysteine, which can be used in another Fig. 1. Assembly of the N-terminal half of human lysozyme by KCL. HPLC round of ligation (23, 29). analysis of the ligation of [Lys1-Met29]-␣thioarylester with [Cys30-Trp(CHO)64]- 65 94 ␣ Before ligation, [Thz -Ala ]- thioalkylester was transthioesteri- ␣thioalkylester in the absence of added thiol catalyst. (A) Reaction mixture at fied with 120 mM 2-mercaptophenylacetic acid in 6 M guanidine t ϭ 1 min. (B) Reaction mixture after 90 min. (C) Reaction mixture after5hof hydrochloride and 0.2 M sodium phosphate at pH 6.8. This pro- ligation followed by 30 min of transthioesteriﬁcation with MESNA. Chromato- cedure not only generated the more reactive [Thz65-Ala94]-␣thio(4- graphic separations were performed by using a linear gradient (5–65% of carboxymethyl)phenyl ester 6 but also led to rapid and efficient buffer B in buffer A over 18 min after an initial isocratic phase of 5% buffer B in buffer A for 3 min). Buffer A: 0.1% (vol/vol) TFA in water; buffer B: 0.08% dinitrophenyl (DNP) removal from His78, thus simplifying the (vol/vol) TFA in acetonitrile. R, OCH2COO(Arg)3⅐amide. ligation reaction product mixture (Scheme 1iii). Ligation of purified [Thz65-Ala94]-␣thio(4-carboxymethyl)phenyl ester 6 and [Cys95- Val130] 7 was carried out in 6 M guanidine hydrochloride, 0.2 M Ligation of N- and C-Terminal Halves and Removal of Protecting sodium phosphate, 30 mM MPAA, and 20 mM Tris(2-carboxyeth- Groups. The final ligation of [Lys1-Trp(CHO)64]-␣MESNA thio- yl)phosphine hydrochloride (TCEP) at pH 6.8 (Scheme 1iv). The ester 5 and [Cys65-Val130] 8 was performed under the same condi- reaction was complete within 2 h (based on LC analysis) (Fig. 2) tions as described above for the synthesis of [Cys65-Val130] (Scheme after which methoxyamine⅐HCl was added to 0.4 M and the pH was 1vi; see Fig. 3 for analytical data). The reaction was complete within adjusted to 4.0 to effect the conversion of Thz65 to Cys65 (Scheme 12 h. After solid-phase extraction and elution into 50% (vol/vol) 1v) (23). Deprotection was quantitative within 6 h, and the resulting aceonitrile-water containing 0.1% (vol/vol) trifluoroacetic acid 66-residue peptide 8 was purified by reversed-phase HPLC (RP- (TFA), the S-Acm side-chain protecting groups subsequently were HPLC) [obsd. electrospray ionization (ESI)-MS mass, 7,539.8 Ϯ 0.7 removed by treatment with a large excess of AgOAc (35 equiv. per Da; calcd., 7,539.2 Da]. S-Acm group) over 10 h. Silver thiolates were converted to the free thiols by adding a large excess of DTT; a precipitate was formed immediately (consisting of Ag–DTT complexes). In previous at- †Thiolactone formation of this type could preclude the further C-terminal activation, as the tempts to prepare lysozyme by total chemical synthesis, the removal thioaryl ester, of the product of a KCL. of S-Acm protecting groups had proved problematic (15). Modern

‡This use of the S-Acm group, depending on the number and location of Cys residues, may MALDI-MS allowed us rapidly to optimize removal of the Acm preclude use of the S-Acm for directed disulﬁde formation in the ﬁnal product protein. groups. Acm deprotection initially was studied on model peptides,

4848 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0610630104 Durek et al. Downloaded by guest on September 25, 2021 Fig. 3. Linking the N- and C-terminal halves to give the full-length lysozyme polypeptide. HPLC analysis of the ligation of [Lys1-Trp(CHO)64]-␣MESNA thio- Fig. 2. Assembly of the C-terminal half of human lysozyme by NCL. HPLC ester with [Cys65-Val130]. (A) Reaction mixture at t ϭ 1 min. (B) Reaction BIOCHEMISTRY analysis of the ligation of [Thz65-Ala94]-␣thioarylester with [Cys95-Val130]. (A) mixture after 12 h of ligation. Chromatographic separations were performed Reaction mixture shortly (1 min) after mixing of the peptide segments. (B) as described in the Fig. 1 legend. Reaction mixture after2hofligation, followed by6hofmethoxyamine⅐HCl treatment at pH 4.0. Chromatographic separations were performed by using a linear gradient (1–49% of buffer B in buffer A over 11 min after an initial isocratic phase of 1% buffer B in buffer A for 2 min). Buffer A: 0.1% (vol/vol) tifiable compound mixture probably representing nonnative disul- TFA in water; buffer B: 0.08% (vol/vol) TFA in acetonitrile. fide isomers of the lysozyme molecule. ESI-MS analysis of the principal peak revealed a mass of 14,693.4 Ϯ 0.7 Da (Fig. 4). This value corresponds to a loss of 6.8 Ϯ 2.2 Da, which is in excellent CHEMISTRY comparing AgOAc and Hg(OAc)2 deprotection protocols under agreement with the formation of four disulfides and the concom- otherwise similar conditions. In our hands, AgOAc-mediated Acm itant loss of eight protons (calcd., 14,692.7 Da; average isotopes). removal was significantly slower than Acm deprotection by Folded synthetic lysozyme was purified from the crude folding Hg(OAc)2. However MALDI-MS analysis indicated that decom- solution by RP-HPLC and was recovered in good yield (40%). position of the formed Hg-thiolate compounds turned out to be ESI-MS, MALDI-MS, and RP-HPLC analysis indicated high pu- more challenging and required longer reaction times and a larger rity of the synthetic material (Fig. 4A and SI Figs. 6C and 7). excess of DTT when compared with the corresponding Ag-thiolates The conformational homogeneity of the synthetic protein was (data not shown). For this reason, we chose to stick to the AgOAc demonstrated by 1D 1H-NMR and 2D TOCSY 1H-1H NMR protocol, which also has been reported to be milder in the presence spectroscopy under solution conditions. SI Fig. 8 A and B shows the of sensitive amino acid residues (31, 32). dispersion of chemical shifts in the amide/aromatic and aliphatic Tryptophan formyl protecting groups (five in total) then spectral regions of the 1D 1H-NMR spectrum. The 2D TOCSY were removed by treating the crude peptide with 20% (vol/vol) 1H-1H NMR spectrum of the aliphatic spin systems is shown in Fig. piperidine and 38% (vol/vol) 2-mercaptoethanol in aqueous 4B. The observed chemical shifts are identical to those assigned in guanidine hydrochloride solution for 40 min at 0°C. The the spectrum of biosynthetic human lysozyme (33, 34) and indicate HPLC-purified fully deprotected and reduced 130-aa polypep- native folding and the absence of detectable misfolded conforma- tide [supporting information (SI) Fig. 6A] was obtained in tions in the synthetic sample. reasonable yield [28% overall yield for (ligation ϩ removal of The synthetic enzyme was crystallized, and its 3D structure was Acm and formyl protecting groups ϩ HPLC purification), determined by x-ray diffraction. Data were collected to a resolution based on the starting reactants 5 and 8; ESI-MS obsd. average of 1.04 Å (SI Table 1). The structure was solved by rigid-body mass, 14,700.2 Ϯ 1.5 Da; calcd., 14,700.7 Da]. approximation followed by restrained mixed anisotropic/isotropic refinement to an R factor of 0.136 and an Rfree of 0.151 for 50–1.04 Folding, Disulfide Formation, and Product Characterization. The pu- Å data by using a previously reported structure as a starting model rified 130-residue polypeptide chain was folded by dilution into (PDB ID code 1JSF) (Fig. 5) (35). Superposition of the obtained buffer containing a redox system consisting of 5 mM oxidized structure with previous structures obtained from biosynthetic hu- glutathione and 2 mM DTT at pH 8 (final peptide concentration man lysozyme (35) revealed identical covalent structures and highly after dilution was 0.175 mg/ml). After 14 h, formation of the similar native folds. correctly folded structure was revealed on LC analysis by the To unequivocally demonstrate anticipated enzymatic func- appearance of a sharp peak eluting at earlier retention time (SI Fig. tion, the synthetic protein product was tested for lysozyme 6). Such a shift in retention time has been observed in numerous activity by using a well established assay (36). In this assay, the cases upon (re)folding of disulfide cross-linked globular proteins clearance of a bacterial cell-wall suspension as a result of and reflects the burial of hydrophobic residues within the protein lysozyme-mediated cell-wall degradation is measured spectro- core. The sharp peak was preceded by a poorly resolved uniden- photometrically. Addition of a sample of the synthetic protein to

Durek et al. PNAS ͉ March 20, 2007 ͉ vol. 104 ͉ no. 12 ͉ 4849 Downloaded by guest on September 25, 2021 Fig. 5. X-ray structure of synthetic human lysozyme. (A) Ribbon represen- tation of the x-ray structure of synthetic human lysozyme. (B) Final 2Fo Ϫ Fc electron density map around the active site at 1.04-Å resolution. (C) Super- position of the synthetic lysozyme structure (red) with lysozyme structures obtained from biosynthetic sources (PDB ID codes: 1JSF, green; 1IWT, blue).

a suspension of Micrococcus lysodeikticus cells led to a rapid decrease in turbidity, revealing that enzymatically active protein indeed had been obtained (Fig. 4C). The specific activity of the synthetic material was determined as 72,250 Ϯ 1,000 units/mg of sample, which is comparable to reported values for commercially available recombinant material of high purity. Based on these biochemical and structural data, we conclude that chemically synthesized human lysozyme is a homogeneous preparation identical to human lysozyme isolated from biological sources. A typical preparation of synthetic lysozyme yielded Ϸ2.5mgof native enzyme.§ This amount of enzyme was more than sufficient to perform all biophysical and biochemical experiments described Fig. 4. Characterization of synthetic human lysozyme. (A) LC analysis of puriﬁed above. and folded synthetic lysozyme. Chromatographic separations were performed as described in the Fig. 1 legend. (Inset) ESI-MS spectrum. The calculated mass is Conclusions 14,692.7 Da. Deconvolution of the ESI-MS spectrum yields an observed mass of In this article, we report on efficient and robust total chemical 14,693.4 Ϯ 0.7 Da. (B) 2D TOCSY 1H-1H NMR spectrum showing aliphatic spin synthesis of human lysozyme. The 130-aa residue protein molecule systems. (C) Clearance of a bacterial cell-wall suspension by synthetic human lysozyme (curve i). The arrowhead indicates the time when enzyme or buffer was added to the cuvette. Negative control (buffer blank, curve ii). The calculated §An overall yield for the synthesis of 5.6% can be calculated based on the limiting reactants, speciﬁc enzymatic activity was 72,250 Ϯ 1,000 units/mg of sample, comparable peptides 30–64 and 95–130. The effective yield is reduced 2-fold further if the conversion with recombinant human lysozyme preparations. of the peptide thioalkyl to thioaryl esters is taken into account.

4850 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0610630104 Durek et al. Downloaded by guest on September 25, 2021 was assembled from four peptide segments in a fully convergent concentration of Ϸ1 mM.¶ MPAA was added to 120 mM, and the manner by using modern chemical ligation methods. The size (Ϸ30 pH was adjusted to 6.5–6.8. The mixture was stirred for 2–6 h at aa) of the peptide segments used makes them easily accessible by room temperature. Yields were typically Ϸ80–90% (based on SPPS combined with standard HPLC purifications, even in the HPLC analysis of the crude reaction mixture). After the reaction mixture reached equilibrium, the product was purified by pre- hands of only moderately experienced peptide chemists. Our con- ␣ vergent synthetic design is especially well suited for the preparation parative RP-HPLC. The obtained lyophilized thioarylesters of various lysozyme analogs because any desired chemical modifi- were stable for several months at 4°C. cations or biophysical labels can be introduced at any position of the molecule without the need for a change in strategy. Because of a NCL and KCL. NCL was carried out in degassed ligation buffer (6 M high degree of sequence conservation within the c-type lysozyme guanidine hydrochloride, 0.2 M sodium phosphate, 20 mM TCEP, and 30 mM MPAA) at a peptide concentration of 1–2 mM. The pH family and a highly conserved location of cysteine residues in ␣ was kept at 6.8–7.0 at all times. KCL of [Lys1-Met29]- thio(4- particular (7), we also believe that the synthetic approach will be carboxymethyl)phenyl ester and [Cys30-Trp(CHO)64]-␣thioalky- applicable to other c-type lysozyme family members, for example, lester was performed under the same conditions, except that hen egg-white lysozyme. MPAA and TCEP were omitted from the reaction mixture. Convergent synthesis is made possible by KCL, which has proven to be a practical way of controlling the reactivity of peptide Folding and Disulfide Formation. Purified, reduced, and fully side- thioesters (24). The 130-residue lysozyme protein molecule is the chain deprotected polypeptide corresponding to [Lys1-Val130] (6.3 second and thus far the largest example prepared in a convergent mg) was dissolved in 6 M guanidine hydrochloride, 0.2 M sodium fashion by using a combination of NCL and KCL. The convergent phosphate, and 20 mM 1,4-DTT (pH 8.0) at a peptide concentra- synthesis of lysozyme by KCL comes at a price, namely, the tion of 7 mg/ml. Folding was achieved by 1:40 dilution into refolding requirement for additional steps in the synthetic scheme: transthio- buffer [0.8 M guanidine hydrochloride, 0.1 M Tris⅐HCl, 5 mM esterification reactions and the need for additional Cys protecting oxidized glutathione, 2 mM DTT, and 1 mM EDTA (pH 8.0)]. groups and their removal. Although the transthioesterification may After gentle stirring overnight at room temperature, the reaction be obviated by direct on-resin synthesis of the peptide ␣thio(4- mixture was centrifuged and decanted, and the supernatant was carboxymethyl)phenyl esters (37), we believe that the use of Cys loaded onto a semipreparative Vydac C4 column (10 ϫ 250 mm). protecting groups needs further examination. The ability to effi- Correctly folded material was eluted by using a linear gradient of ciently assemble proteins from four synthetic peptide segments, and 10–40% buffer B in buffer A over 60 min at a flow rate of 10 the increased flexibility in the design of a synthetic route, will ml/min. Fractions were pooled based on LC and LC-MS analysis BIOCHEMISTRY and lyophilized. [Yield: 2.5 mg (40%). ESI-MS: oxidized synthetic provide a further boost to the field of chemical protein synthesis, Ϯ which we expect to translate into a more detailed understanding of human lysozyme, obsd., 14,693.4 1.0 Da; calcd., 14,692.7 Da. MALDI-MS: obsd., 14,690 Ϯ 8 Da.] the chemical basis of protein function. Materials and Methods Lysozyme Activity Assay. Lysozyme enzymatic activity was determined as described in ref. 36 (see SI Text). Peptide Synthesis and Purification. Peptides were synthesized as described in refs. 26 and 27 (see SI Text). The identity of the Crystal-Structure Determination and NMR Spectroscopy. Details for CHEMISTRY peptides was confirmed by LC-ESI-MS on an Agilent 1100 series 1 6 crystallization of synthetic human lysozyme, x-ray data acquisition, instrument equipped with ion-trap MS: [Lys -Cys(Acm) - model building and refinement, and NMR spectroscopic experi- 28 29 ␣ Ϯ Trp(CHO) -Met ]- thioalkylester obsd. mass, 4,070.6 0.4 ments are described in SI Text. Da, calcd. mass (average isotope composition), 4,070.7 Da; [Cys30-Trp(CHO)34-Trp(CHO)64]-␣thioalkylester obsd., 4,734.7 Ϯ 0.5 Da, calcd., 4,735.1 Da; [Thz65-Cys(Acm)77- ¶Transthioesteriﬁcation instead can be carried out on the crude peptide thioalkyl ester His(DNP)78-Cys(Acm)81-Ala94]-␣thioalkylester obsd., 3,437.0 Ϯ before a single puriﬁcation (data not shown). 0.5 Da, calcd., 3,436.5 Da; and [Cys95-Trp(CHO)109-Trp- (CHO)112-Cys(Acm)116-Cys(Acm)128-Val130] obsd., 4,444.0 Ϯ 0.4 Da, calcd., 4,443.9 Da. We thank Brad L. Pentelute, Zachary P. Gates, and Valentina Tereshko for help with x-ray data collection and structure refinement and Michael Weiss

␣ ␣ and Qingxin Hua at Case Western Reserve University (Cleveland, OH) for Conversion of a Peptide Thioalkylester to a Thio(4-carboxymethyl) performing the NMR measurements. This research was supported by U.S. phenyl Ester. Purified peptide was dissolved in 6 M guanidine Department of Energy Genomes to Life Genomics Program Grant DE- hydrochloride/0.2 M sodium phosphate buffer at a peptide FG02–04ER63786 (to S.B.H.K.).

1. Blake CC, Koenig DF, Mair GA, North AC, Phillips DC, Sarma VR (1965) Nature 206:757–761. 20. Lu WY, Qasim MA, Laskowski M, Kent SBH (1997) Biochemistry 36:673–679. 2. Dobson CM, Evans PA, Radford SE (1994) Trends Biochem Sci 19:31–37. 21. Smith R, Brereton IM, Chai RY, Kent SBH (1996) Nat Struct Biol 3:946–950. 3. Matagne A, Dobson CM (1998) Cell Mol Life Sci 54:363–371. 22. Bang D, Gribenko AV, Tereshko V, Kossiakoff AA, Kent SB, Makhatadze GI (2006) Nat Chem 4. Kirby AJ (2001) Nat Struct Biol 8:737–739. Biol 2:139–143. 5. Vocadlo DJ, Davies GJ, Laine R, Withers SG (2001) Nature 412:835–838. 23. Bang D, Kent SB (2004) Angew Chem Int Ed Engl 43:2534–2538. 6. Sauter C, Otalora F, Gavira JA, Vidal O, Giege R, Garcia-Ruiz JM (2001) Acta Crystallogr D 24. Bang D, Pentelute BL, Kent SB (2006) Angew Chem Int Ed Engl 45:3985–3988. 57:1119–1126. 25. Hendrickson JB (1977) J Am Chem Soc 99:5439–5450. 7. Jolles P, Jolles J (1984) Mol Cell Biochem 63:165–189. 26. Schno¨lzer M, Alewood P, Jones A, Alewood D, Kent SB (1992) Int J Pept Protein Res 40: 8. Kuroki R, Weaver LH, Matthews BW (1995) Nat Struct Biol 2:1007–1011. 180–193. 9. Ibrahim HR, Aoki T, Pellegrini A (2002) Curr Pharm Des 8:671–693. 27. Hackeng TM, Griffin JH, Dawson PE (1999) Proc Natl Acad Sci USA 96:10068–10073. 10. Booth DR, Sunde M, Bellotti V, Robinson CV, Hutchinson WL, Fraser PE, Hawkins PN, 28. Johnson EC, Kent SB (2006) J Am Chem Soc 128:6640–6646. Dobson CM, Radford SE, Blake CC, Pepys MB (1997) Nature 385:787–793. 29. Villain M, Vizzavona J, Rose K (2001) Chem Biol 8:673–679. 11. Pepys MB, Hawkins PN, Booth DR, Vigushin DM, Tennent GA, Soutar AK, Totty N, Nguyen 30. Dawson PE, Churchill MJ, Ghadiri MR, Kent SBH (1997) J Am Chem Soc 119:4325–4329. O, Blake CC, Terry CJ, et al. (1993) Nature 362:553–557. 12. Kent S (2003) J Pept Sci 9:574–593. 31. Najib J, Letailleur T, Gesquiere JC, Tartar A (1996) JPeptSci2:309–317. 13. Dawson PE, Kent SBH (2000) Annu Rev Biochem 69:923–960. 32. Fujii N, Otaka A, Watanabe T, Okamachi A, Tamamura H, Yajima H, Inagaki Y, Nomizu M, 14. Kochendoerfer GG, Kent SB (1999) Curr Opin Chem Biol 3:665–671. Asano K (1989) J Chem Soc Chem Commun 283–284. 15. Kenner GW (1977) Proc R Soc London B 197:237–253. 33. Redfield C, Dobson CM (1990) Biochemistry 29:7201–7214. 16. Sharp JJ, Robinson AB, Kamen MD (1973) J Am Chem Soc 95:6097–6108. 34. Kumeta H, Kobashigawa Y, Miura K, Nishimiya Y, Oka C, Nemoto N, Miura A, Nitta K, Tsuda 17. Schno¨lzer M, Kent SBH (1992) Science 256:221–225. S (2002) J Biomol NMR 22:183–184. 18. Dawson PE, Muir TW, Clark-Lewis I, Kent SB (1994) Science 266:776–779. 35. Harata K, Abe Y, Muraki M (1998) Proteins Struct Funct Bioinform 30:232–243. 19. Kochendoerfer GG, Chen SY, Mao F, Cressman S, Traviglia S, Shao H, Hunter CL, Low DW, 36. Morsky P (1983) Anal Biochem 128:77–85. Cagle EN, Carnevali M, et al. (2003) Science 299:884–887. 37. Bang D, Pentelute BL, Gates ZP, Kent SB (2006) Org Lett 8:1049–1052.

Durek et al. PNAS ͉ March 20, 2007 ͉ vol. 104 ͉ no. 12 ͉ 4851 Downloaded by guest on September 25, 2021