Site‐Selective Modification of Proteins with Oxetanes

Abstract Oxetanes are four‐membered ring oxygen heterocycles that are advantageously used in medicinal chemistry as modulators of physicochemical properties of small molecules. Herein, we present a simple method for the incorporation of oxetanes into proteins through chemoselective alkylation of cysteine. We demonstrate a broad substrate scope by reacting proteins used as apoptotic markers and in drug formulation, and a therapeutic antibody with a series of 3‐oxetane bromides, enabling the identification of novel handles (S‐to‐S/N rigid, non‐aromatic, and soluble linker) and reactivity modes (temporary cysteine protecting group), while maintaining their intrinsic activity. The possibility to conjugate oxetane motifs into full‐length proteins has potential to identify novel drug candidates as the next‐generation of peptide/protein therapeutics with improved physicochemical and biological properties.

Abstract: Oxetanes are four-membered ring oxygen heterocycles that are advantageously used in medicinal chemistry as modulators of physicochemical properties of small molecules. Herein, we presentasimple methodf or the incorporation of oxetanes into proteins throughc hemoselective alkylation of cysteine. Wed emonstrate ab road substrate scope by reactingp roteins used as apoptotic markersa nd in drug formulation, and at herapeutic antibodyw ith as eries of 3-oxetaneb romides, enabling the identification of novel handles (S-to-S/N rigid, non-aromatic,a nd soluble linker) and reactivity modes (temporaryc ysteine protecting group),w hile maintaining their intrinsic activity.T he possibility to conjugate oxetane motifs into full-length proteins has potential to identify novel drug candidates as the next-generation of peptide/ protein therapeutics with improved physicochemical and biological properties.
Oxetanes are heterocyclic 1,3-propylene oxide moieties [1] that constitutet he core structure of many biologically active natural products (e.g. the antibiotic oxetin) and syntheticc ompounds( e.g. carbonyl bioisosters, agrochemicals, and peptidomimetics, among others). [2] They have emerged as important motifs in drug discoverydue to the modulation of the physicochemicalp roperties of the molecule to which they are attached. [3] Typically,o xetanes increasew ater solubility,m etabolic stability, and reducel ipophilicity while maintaining their parents electivity without alteringa ctivity (Figure 1). Despite the many examples reported for the introductiono fs uch motifs in small molecules, their application to modifyc omplex biomolecules, such as proteins/antibodies (beyond peptidomimetics), [4] is largely unprecedented.
The aim of the presentp roof-of-principle study is to explore methodsf or the chemoselective introduction of oxetane moieties into proteins and antibodies by alkylation of the sulfhydryl group of the side chain of cysteine (Cys) with as eries of structurally similar 3-oxetaneb romides. We reasoned that investigation of oxetane reagents (e.g. secondary vs. primaryh alides and 3-monov s. 3,3-disubstituted) and their reactivity towards an umber of different protein scaffolds would test the generality of this site-selective protein modification method for the preparation of homogeneous proteins. [5] We envisaged our methodw ould allow access to an ew familyo fp rotein conjugates with potentialt herapeutic applications, [6] for example, by enhancing ligand-protein binding since the oxetane ring servesa sah ydrogen-bond acceptor. [7] We began our investigation by exploring the site-selective incorporation of the 3-S-oxetanem otif into proteins by reacting inexpensive, commerciallya vailable 3-bromooxetane 1 with single-Cys mutants of the C2A domain of Synaptotagmin-I Cys95 (C2Am) and Annexin-V Cys315 (AnxV) as representative examples of specific apoptotic protein markers (Figure-s2aa nd 2b). [8] Incubation of C2Am with 1 in 50 mm sodium phosphate buffer (NaP i )a tp H1 1w ith up to 20 %D MF,t o ensure oxetane's solubility,a fforded expected Cys-to-oxetane alkylation product 2 in > 95 %c onversion (via an S N 2r eaction) as determined by liquid chromatography electrospray ionization mass spectrometry (LC-ESI-MS) after 30 ha t3 78C. Remarkably,a nd unlike most of the alkylating electrophiles commonly used for "on protein" Cys alkylation (typically primary C(sp 3 ) halides), the present transformationu sing as econdary C(sp 3 ) bromide represents as tereoelectronically challenging,y et successfule xample of ac onstrained small oxygen heterocycle alkylation, which may account for the "harsh"c onditions employed. The modified protein was identified by LC-ESI-MS, which showed as ingle peak (16 279 Da) with am ass shift corresponding to the addition of as ingle 3-oxetane unit (+ 57 Da). Negative Ellman's test and peptide mapping/MS 2 analysis revealed the alkylation proceed with exquisite chemoselectivity at Cys95 (Figure 2c and the Supporting Information, Figures S1-S5). Experiments to furthere valuate putative lysine (Lys) cross-reactivity were conducted with single Lys-peptide S1 and 3-bromooxetane 1 using conditions similart ot hose employed for "on protein" Cys alkylation (50 mm NaP i at pH 11 with 20 %D MF). No reactivityw as noticed after 36 ha t3 78C as monitored by 1 HNMR (Supporting Information, Figure S77). Importantly,t he 3-S-oxetane motif proveds table upon incubation with human plasma (37 8C, 24 h) and in the presence of endogenous thiols such as reduced glutathione (GSH) (20 mm, 37 8C, 24 h), which are prerequisite conditions for in vivo applications.I dentical stability was also noticed upon incubation with other S-or N-nucleophiles (b-mercaptoethanol (BME), PhSH, bGalSNa, and BnNH 2 )u nder forcingc onditions (50mm NaP i at pH 4.5-11w ith/without 20 %C H 3 CN, 37 8C, up to 24 h). Additionally,a dditives known to promote oxetane ring-opening polymerization [9] were also tested without success. These included oxophilic metallicc omplexes (Y(OTf) 3 and MgCl 2 ·6 H 2 O) and organocatalytic promoters (urea and thiourea) (Supporting Information, Figures S6-S21). The inertness of the oxetane ring towards nucleophiles is in line with previous studies that described their lacko fg eno/cytotoxicity and mutagenicity,u nlike that of epoxidesa nd b-lactones, because they do not act as alkylating agents at physiological pH. [10] Likewise, other proteins such as AnxV reacteds imilarly at pH 11 with 10 %D MF and expected product 3 was obtained in > 95 %c onversion, thus demonstrating the robustness of this alkylation protocol to modify this widely used apoptotic protein marker( Figure 2b). Finally,t he impact of such am odification and reactionc onditions in protein's structure and function was evaluated using circular dichroism (CD) and surfacep lasmon resonance (SPR), respectively. [8] The chemoselective incorporation of the 3-S-oxetanem otif in 2 is mild as determined by CD analysis, which showedn os tructure alteration between native and modified proteins (Supporting Information, Figure S80). Additionally,d espite the SPR functional assay for C2Am vs. C2Am-3-S-Ox 2 shows that its intrinsic binding activity against phosphatidylserine( PS), an internal membrane lipid externalized during apoptosis, is only partially eroded (60 %a ctivity is maintained), differences in its binding mechanismi s noticedf rom the analysiso ft he binding and dissociation curves ( Figure 2d). Having established conditions for efficient site-selectivei ncorporation of the 3-oxetane motif into Cys-tagged proteins, we explored the extension of this approacht oo ther oxetanes. We continued by evaluating the homobifunctional electrophile 3,3-bis(bromomethyl)oxetane 4 (Figure 3). We anticipate the facile alkylation of Cys due to the ap riori more accessible, reactive primary C(sp 3 )b romide will enablee fficient desymmetrization of 4,t hus providing ap latform to chemoselectively incorporate an electrophilic handle (BrCH 2 -S-oxetane) into ap redefined site of proteins. This handle is amenablet oasecond round of chemical modification (via an S N 2r eaction) after incu-   (Figure 3a). Indeed, this is ad ifficult task and very few chemicalm ethods allow for the precise installation of electrophilic reactingp oints on as ingle residue, [11] which is typically achievedb yg enetic encoding protocols. [12] Thus,r eaction of 4 with C2Am in NaP i (50 mm,p H8)w ith 10 %D MF afforded 5 in > 95 %c onversion as determined by LC-ESI-MS after 5h at 37 8C. Fine-tuning of the reaction conditions and using pH 11 reduced the reactiont ime to only 2h,w hile maintaining chemoselectivity as confirmed by negative Ellman's test and 1 HNMR experiments with Lys-peptide S1 (Supporting Information, Figures S22-S26 and S78). The protein scope was further expandedt oA nxV that reacted similarly at pH 11 with 10 % DMF to obtain 6 in > 95 %c onversion (Figure 3b and Supporting Information, Figures S54 and S56). We next set up to evaluate the incorporation of several S-and N-nucleophiles taking advantage of the flexible introduction of the privileged 3-bromomethyl-3-S-oxetaneh andle.T his represents au nique example of spiro, heterobifunctional S-to-S/N linker [13] that enables the diversity-oriented introduction of severaln ucleophilesf rom as ingle Cys mutant. [14] To this end, 5 wasi ncubated with the S-nucleophiles bGalSNa, BME, and PhSH as representative ex-ampleso fp rotein glycosylation, aliphatic, and aromaticm oieties, respectively as well as BnNH 2 as an N-nucleophile present in several drug fragments to afford 7a-d in > 95 %c onversion (Figure 3c). Extending the reactiont ime once the alkylation is completed (e.g. up to 15 hw ith PhSH) did not show any noticeable degradation (Supporting Information, Figures S27-S33). Moreover,s imilart ot hat observedw ith the 3-S-oxetane motif, S-to-S/N oxetane linker in 7c also provedstable upon incubation with human plasma (37 8C, 24 h) and GSH (20 mm, 37 8C, 24 h) (Supporting Information, Figures S34 and S35), which reinforces its application as an ovel linker for stable, covalent protein modification (e.g. including disulfide stapling). [5] The displayofbGalS as arepresentative example of carbohydrate epitope wasf urther studied by molecular dynamics (MD) simulations and the result compared to ag eneral aliphatic linker with the same number of carbons, obtainedb yt hiol-ene chemistry at S-allyl cysteine( Sac) (Figure 3d). [15] Interestingly, the oxetane linker in 7a provedt ob eam ore rigid scaffold than that in 8 for the presentation of bGalS as determined by the analysiso ft he angle between Ca-Cb-C 1 where am ajor population is observed. Ta ken together,w eb elieve this rigid, non-aromatic, and soluble linker will be advantageous, for example,i nt he development of homogeneous carbohydrate- based vaccine conjugates,h opefully,w ith reduced anti-linker response. [16] Encouraged by these results, we next evaluated the use of 3-(bromomethyl)oxetane 9 (Figure 4). Experiments with as eries of proteins (C2Am,A nxV, and recombinant human serum albumin (rHSA))a nd one antibody( Trastuzumab) revealed the application of this moiety as anovel temporary protecting (PG) group for Cys-containing biomolecules [17] (via the following full sequence Cys-protection!SCH 2 -Ox!Cys-deprotection) (Figures 4a and 4b). Unlike previousr eactions with 3oxetaneb romides 1 and 4 that necessitatedu pt opH 11 or prolonged reactiont imes to reach completion, the alkylation with 9 proceed smoothly at pH 8( for C2Am) and pH 9( for AnxV), respectively to afford 10 and 11 in uniformly complete conversionsw ith am ass difference corresponding to the incorporation of as ingle CH 2 -oxetane moiety (+ 71 Da). Negative Ellman's test confirmedt he chemoselectivity at Cys95 in C2Am and Cys315i nA nxV (Supporting Information, Figures S36, S37, S57, and S58). rHSA (Albumedix Recombumin), [18] an approved ingredient for the manufacture of human therapeuticst hat possess multiple Cys residues, 17 structurally relevantdisulfides and as ingle Cys at position34, gave the expected monoalkylation product 12 (25 %c onv.) upon incubation with 50 equiv of 9 in NaP i (50 mm,p H8)w ith 10 %D MF as determined by LC-ESI-MS after 2h at room temperature. However,t he addition of several 3-methyloxetane units (+ 71 Da) was observed after prolonged reactiont imes by increasing the equivalents of 9 (> 50 equiv) and warming the reactionu pt o3 78C. We reasoned this might be attributed to the reaction of 9 with partially reduced disulfides (Supporting Information, Figures S64-S67). [19] Similar results are also obtained with the monoclonal antibody (mAb) Trastuzumab containing severals tructurald isulfides (Figure 5a). [20] We then examined the stabilityo f3 -S-monomethyl oxetane proteins.I ncubation of 10 with 100 equiv of BnNH 2 in NaP i (50 mm,p H8)u nexpectedly resulted in the full recovery of the originalC ys moiety after 2h at 37 8C( Figure4c). Interestingly, ac onceptually similar N-dealkylation reactionb yo xidative cleavage of the CH 2 bridge was found to be the main metabolization product in 3-monomethyl oxetanes. [21] This study also suggests, as we found in our work with the 3,3-disubstituted analogue, the introductiono fb ulky gem-dimethyl substituents increases the stability of the oxetane. Ap utative S-dealkylation scenarioi nw hich the nucleophile attacks the alpha carbon (Ca)o ft he oxetane,r eleasing Cys as al eaving group perhaps with the participation/anchimeric assistance of the non-bonding electron pair of the oxygen might account for the reactivity observed, although alternative mechanisms( e.g. oxidative cleavage with atmospheric O 2 ) [22] cannot be discarded. With these preliminaryc onditions in hand, the scope of the reaction was evaluated with ar ange of S-, N-, and P-nucleophiles (BME, PhSH, bGalSNa, BnNH 2 ,a nd tris(2-carboxyethyl)phosphine (TCEP)) and proteins (10 and 11). Despite startingC ys-proteins (C2Am and AnxV) were recovered in up to > 95 %c onversion, the rate of this transformation seems to be dependent on the combination of protein/nucleophile/conditions( Supporting Information,F igures S38-S47 and S59-S61). For example, deprotection of 10 was slower with both TCEP (9 h) and PhSH (22 h), necessitating al arge excess of nucleophile (10,000 equiv) at 37 8Ct or each completion, whereas bGalSNa required only 5h (at pH 8) or 2h (at pH 11). Similarly,A nxV was obtained from 11 after incubation with 100 equivalents bGalSNai nN aP i (50 mm,p H9)a fter only 1h at 37 8C. Controls to demonstrate how oxetanes modulates the electrophilic character of the Ca enablingt he S-dealkylation when treated with excess S-, N-, and P-nucleophiles werec arried out using the Sac handle [23] as as table 3C-surrogateo ft he S-methyloxetane unit (SCH 2 -Ox). Sac-containingp roteins C2Am S2 and AnxV S3 were subjected to the same S-dealkylation conditions and starting materials were recovered unaltered. Thus, our preliminary findings suggest ar ole of the methyloxetane moietyi nt he nucleophile-induced deprotection step, at least on full-length proteins (Fig-Figure 4. Selective incorporation of the SCH 2 -3-oxetane motif into proteins as aC ys temporary protectingg roup. a) Generalstrategy with 9.b)Proteinscope withC 2Am,A nxV, and rHSA. c) Schematicrepresentationo ft he deprotection step (SCH 2 -Ox!Cys) with severalP -, N-, and S-nucleophiles and unreactive controls using Sac (see the SupportingInformation for details). § Multiple additionso bservedu pon incubation with > 50 equiv of 9.
ure 4c andS upporting Information, FiguresS48, S49, S62, and S63). [24] Finally,t he utility of our protocol was extended to the reversible modification of Trastuzumab; am Ab used in the clinic for the treatment of HER2 + metastatic breast and gastric cancers. We found the Cys-engineered Trastuzumab analog (Thiomab 4D5 LC-V205C) [20] reacted with 9 in NaP i (50mm,p H8)w ith 10 %D MF to afford modified 13 containing several (3-methyloxetane) n units (+ 71 n Da) in the light chain as determined by LC-ESI-MSafter 5hat room temperature (Figure 5a and SI, Figures S68 and S69). [25] Similarly to 3-oxetaneb romides 1 and 4, 1 HNMR experiments with 9 and single Lys-and single Cys-peptides S1 and S4,r espectively demonstrated the introduction of these multiple 3-methyloxetane units is due to alkylationo f the free engineered Cys together with those resultingf rom inefficient re-oxidation of structural disulfides [20] rather than for Lyscross-reactivity (Supporting Information, Figure S79).
The (SCH 2 -Ox)-to-Cys deprotection in 13 wast riggeredb y TCEP,B nNH 2 ,o rB ME and represents as uccessful metal/lightfree example of ap rotection-deprotection sequence on an intact IgG mAb (Supporting Information, Figures S70-S73). We then investigated the influence of incorporation/removal of SCH 2 -Ox on mAb's function by bio-layer interferometry( BLI) analysisa nd comparedi ts activity before and after modification. We found, the binding affinity is maintained along the entire protection-deprotection cycle with their binding constants remaining within the same nanomolar range (Figure 5b and Supporting Information, Figures S83 and S84). We expect this protocol will find use as at ool for transient, site-selective manipulation of Cys on proteins and antibodies. [26] In summary,w ed isclosed an operationally simple and advantageousm ethod for the chemoselective introduction of oxetane moietiesi nto proteins through alkylation of Cys residues. We validated this mild transformationo naseries of proteins and antibodies, which maintained their inherent activities. Screeningo fo xetane variantse nabledt he discoveryo fn ovel handles( spiro S-to-S/N linker)a nd reactivity modes (temporary Cys-PG). This work providest he basis for the development of novel oxetane reagents and complementsc urrent methods for site-selective chemical protein/peptide modification. [5] We anticipate the knowledge derived from this thorough proof-ofprinciple study will help in the selection of reaction conditions suitable for introducing other strained heterocyclic motifs openingn ew horizons in the field of protein engineering and biological therapeutics. [6]