An efficient method for site-selective boronation of peptides and proteins applied in magnified bacterial imaging

Abstract

Borylated compounds have received tremendous attention in chemical biology and drug design. Motivated by this notion, we develop an efficient methodology for boron incorporation into peptides and proteins via chemoselective conjugation of cysteine residues under catalyst-, additive-, and metal-free conditions. By harnessing the optimum S_N2 displacement profile of bromomethyl boronate, we establish a methodology that enables the selective incorporation of methylboronic acid into various peptides, including bioactive constructs, as well as proteins with rapid kinetics (>10² M^-1s^-1) and excellent conversions at physiological pH. The method also leverages rapid access to bivalent boron conjugates, which, when applied to clinically used antimicrobial peptide UBI(29-41), augment its binding efficacy (40-fold) via covalent capture of the diols on bacterial surface glycans, mainly wall teichoic acid (and thus lipoteichoic acid). Notably, bis-boronated UBI exhibits selective staining of S. aureus over gram-negative bacteria and mammalian cells, alongside a 14-fold increase in serum half-life compared to native UBI.

A rapid and straightforward late-stage cysteine alkylation strategy is reported to synthesize boronopeptides by utilizing commercially accessible halomethyl boronates. The method installs multivalent boron to peptides and proteins. Boronopeptides were deployed to image Gram-positive bacteria.

Introduction

Chemical modifications of peptides and proteins have emerged as an indispensable tool in chemical biology to address the numerous limitations in peptide and protein research^1–5, especially towards the functional modulation of protein⁶ and the development of peptide drugs⁷. In this context, peptides and proteins edited with boronic acid (BA) have recently manifested attractive applications in drug discovery, supramolecular chemistry, and chemical biology^8–13. BA derivatives trait a dynamic interchange between sp² and sp³ hybridization with endogenous nucleophiles at physiological pH owing to the vacant p orbital of boron, and they have been shown to be excellent conjugating partners for 1,2- or 1,3-diol¹⁴, function as covalent inhibitors¹⁵, act as metal chelators¹⁶, bioisosteres of carboxylic acid¹⁷, H-bonding donor-acceptors¹⁸, etc. Its versatile reactivity can also lead to tunable de novo functional features, proteo- or thermolytic stability, and redox-responsive mutation¹⁹. These features have driven intense efforts to develop benign, biocompatible, and chemoselective methodologies for site-specific boron incorporation into large, unprotected peptides and proteins.

The early reports on the incorporation of phenylboronoalanine in proteins involved the genetic code expansion, which met with limitations due to the inhibitory effect of boron on protein translation²⁰. Contingent on the recent progress made in the ‘late-stage installation’ of BA into peptide and protein, two major strategies have come to the forefront, primarily aimed at boronoalanine construction: (1) Cu(II) catalyzed conjugate addition to preinstalled dehydroalanine (Dha) residues^19,21, and (2) desulfurization of cysteine residues and photochemical or metal-mediated boron capture of the derived β-C centered radicals^22,23 (Fig. 1a). However, the Dha-based method entails a loss in stereochemistry^19,23, whereas photochemical/metal-mediated boron incorporation renders unwanted side reactions (protodeboronation and oxidation)^23,24. Hence, the current scenario demands alternative and efficient methodologies for amalgamating boron onto peptides and proteins, which can eliminate the challenges faced by previous methods.

Fig. 1. The concept of this study.

a Two biocompatible and chemoselective strategies to install boron in peptides and proteins, reported earlier. b This work utilizes an S_N2 reaction to install boronic acids using commercially available halomethyl boronic esters by harnessing their bimolecular substitution profiles. The concept of choosing halomethyl boronic acid is presented, where the curved arrow represents a probable overlap between σ* and the vacant p (σ*C-X - pB overlap).

In this work, taking a cue from the widely used halomethyl amides across proteomics for Cys-alkylation in aqueous buffers, we hypothesize that a simple S_N2 reaction between Cys and haloalkyl boronic acids could extrapolate the installation of boronic acids in peptides (Fig. 1b) due to activation of the methylene center by the neighboring vacant p-orbital at the boron center. Guided by this rationale, we plan to utilize readily available halomethyl boronic esters, popularized in organic asymmetric synthesis by Matteson²⁵, for late-stage incorporation of methyl boronic acids via chemoselective Cys-alkylation of peptides and proteins. The developed methodology enables an operationally simple, facile boronation of several unprotected large peptides and proteins, including rapid access to multivalent borono-peptides at physiological pH and room temperature under metal- and additive-free conditions. The concept of mono- and bi-valent boronopeptides is then strategically deployed to augment the imaging efficiency of antimicrobial peptides via the dynamic covalent capture of diol-containing glycans on bacterial surface (mainly lipoteichoic acid (LTA)/wall lipoteichoic acid (WTA)), achieving selective labeling of S. aureus strains.

Results

DFT calculations were initially performed to collate and compare the LUMO orbital energies of halomethyl boronic acid pinacol esters with that of iodoacetamide. Computed outcome proposed that, relative to iodoacetamide, iodomethyl boronic acid pinacol ester (I-CH₂Bpin) could serve as a better electrophile, whereas bromomethyl boronic acid ester (Br-CH₂Bpin) may demonstrate comparable reactivity, when the HOMO orbital energy of Cys nucleophile is concerned (Fig. 2a, Supplementary Fig. 1 and Supplementary Table 1). In addition, we examined that the LUMO orbitals of halomethyl boronic esters displayed a pronounced orbital overlap between the σ*_(C-X) (antibonding orbital) and the vacant p-orbital of boron, by which halomethyl boronic esters were anticipated to manifest a large coefficient to LUMO and increased reactivity compared to general alkyl halide compounds. This computationally projected reactivity further intrigued us to proceed with the experimental interventions with peptides and proteins.

Fig. 2. Rationale of the work from theoretical studies.

a Theoretical calculation comparing energies (eV) of LUMO orbitals of XCH₂Bpin (X = Cl, Br, I) and iodoacetamide (ICH₂CONH₂). An orbital overlap between the vacant p-orbital of boron and the σ*(C-X) orbital of methylene carbon can be observed in the halomethylboron reagents, optimized using Gaussian B3LYP 6-31 G or LANL2DZ modelled in water. Hydrogen atoms have been omitted for clarity. Comparison of (b) classic boronoalanine (Bal), where NMR experiments reveal a decrease in pK_a due to the formation of a dynamic oxaborolane. c This work depicts cysteine methyl boronic acid, where theoretical studies show a moderate overlap of filled S(lp) and vacant B(p), pointing towards a similar acid-base interaction.

Unmodified alkyl BAs typically show lower binding affinity toward diols in aqueous solution compared to aromatic BAs, primarily due to their higher pK_a.^26,27 But NMR experiments by the Davis group¹⁹ revealed that when an alkyl BA (boronoalanine, Bal) is installed within a peptide, its pK_a decreases as a result of dynamic oxaborolane formation with the adjacent amide carbonyl (Fig. 2b). Motivated by this observation, we were curious to study whether the cysteine modified with methyl BA would behave in a similar fashion. Theoretical calculations using Ac-Cys(CH₂BA)-CONH₂ indicated a moderate overlap between the sulfur lone pair and the boron p-orbital, [S(lp)→B(p)], in the linear form, while the cyclic, 7-membered coordinated structure (Fig. 2c) displayed a two-fold stabilization in ground state energy as compared to the 5-membered oxaborolane in Bal. These results suggest that Bal and Cys(CH₂BA) would likely exhibit similar reaction profiles in a peptide context, which encouraged us to develop a synthetic methodology for site-selective methyl boronation on Cys.

Finding a suitable haloboronic acid reagent

We first evaluated the reactivity of readily available halo (I, Br, and Cl) derivatives of methyl boronic acid pinacol ester with Fmoc-Cys-OH by mixing them in an equimolar amount (at 50 mM) in 1:1 PBS: acetonitrile (pH 7.4). Product isolation followed by its characterization in NMR and HRMS confirmed the formation of the expected product (Supplementary Fig. 2). Briefly, the appearance of a new resonance at ~1.8 ppm for the –CH₂– group when compared with Fmoc-Cys-OH in ¹H-NMR, along with a distinct ¹¹B NMR signal at ~30 ppm, validated the product identity. Encouraged by this result, the leucine enkephalin (LeuEnk, YGGFLC, appended with a Cys residue) peptide was chosen as the model substrate to check the reaction feasibility. Halo (I, Br, and Cl) derivatives of methyl boronic acid pinacol ester (1.5 equivalent) were reacted with LeuEnk (100 µM) in PBS buffer (pH 7.4) at ambient temperature. Remarkably, the I- and Br-derivatives achieved complete consumption of the starting material within 10 min, leading to efficient formation of the desired boronated product (LeuEnk-BA), as confirmed by LC-MS analysis (LeuEnk +58 Da mass for -CH₂B(OH)₂) of the crude reaction mixture (Fig. 3a). In contrast, the Cl-derivative exhibited significantly slower reaction and incomplete conversion under identical conditions.

Fig. 3. Proof of the concept with model peptide LeuEnk.

a General scheme for the reaction of halomethyl (Cl, Br, and I) boronic acid pinacol esters (150 µM) with Cys appended LeuEnk (100 µM) in PBS (pH 7.4) at room temperature, and the table (right) represents the conversion of the reactions. b Reaction kinetics of halomethyl boronic acid pinacol esters conjugation to LeuEnk in PBS (pH 7.4). c The purity and identity of the isolated product were confirmed by HPLC and mass spec, respectively; d ¹H NMR of purified LeuEnk stacked with LeuEnk-BA in DMSO-d₆ showing a clear shift in the protons of Cys αNH, βCH₂, and appearance of new -CH₂- proton in the product. e ¹¹B NMR of LeuEnk-BA in DMSO-d₆. f The chemoselectivity studies of BrCH2Bpin with Fmoc amino acids containing reactive side chains are presented in a bar graph.

During LC-MS analyses, we noticed that the iodo-derivative produced a minor bis-alkylation product (~10%) due to its higher reactivity, which led to dimethylated side products via a probable deborylation pathway. These preliminary data prompted us to measure kinetics in PBS buffer (pH 7.4) at room temperature. HPLC monitoring revealed that both Br-CH₂Bpin and I-CH₂Bpin achieved >95% conversion within 5 min, with second-order rate constants exceeding 10² M⁻¹ s⁻¹. (Fig. 3b, Supplementary Fig. 3). Consistent with our DFT calculation, the Cl-CH₂Bpin showed the slowest kinetics (~6.5 M⁻¹ s⁻¹), reaching about 80% product conversion in 4 hours. Interestingly, when LeuEnk was treated with (bromomethyl)trifluoroborate and 4-(bromomethyl)phenylboronic acid individually under similar conditions, reactions showed ~10% conversion even after 1 h, confirmed in LC-MS (Supplementary Figs. 10, 11). Together, these experiments justify our hypothesis that halo methylboronic acids are superior electrophiles for peptide modification. Due to the optimum reactivity of the Br-CH₂Bpin, it was considered the reagent of choice for the rest of the studies.

The LeuEnk-BA product was purified by reverse-phase HPLC (Fig. 3c) and characterized through LC-MS/MS and NMR studies. Tandem MS analysis of b and y fragments in MS/MS (Supplementary Fig. 6) confirmed the site-specific incorporation of methyl boronic acid at the Cys residue, as evidenced by the characteristic loss of water molecules, a well-established mass spectrometric signature of borylated substrates^13,24. In a combined study of ¹H NMR and ¹H-¹H COSY with LeuEnk and its borylated product, we noticed changes in chemical shifts of Cys β-CH₂ and α-NH protons, indicating chemical modification on the Cys residue (Fig. 3d, Supplementary Fig. 7). Furthermore, the borylated LeuEnk showed an additional methylene peak (~ 1.8 ppm) in ¹H-NMR and the signature of a trigonal sp² boronic acid at ~30 ppm in ¹¹B NMR²⁸ (Fig. 3e), judiciously establishing the incorporation of -CH₂B(OH)₂ in the Leu-Enk peptide. Consistent with the earlier report²⁴, we also observed the loss of the pinacol group from the LeuEnk-BA product during HPLC purification, corroborated by ¹H-NMR, which is advantageous for cutting a step down to obtain free BA-inserted substrates for biological applications.

Further, pH-dependent kinetic studies of the reaction between LeuEnk and BrCH₂Bpin revealed a slightly reduced conversion at pH 6, reaching ~90% even after 1 hour of incubation, whereas the reaction at pH 8.5 proceeded with a rate comparable to that at pH 7.4 (Supplementary Fig. 5). It is well-studied that both highly reactive and excess haloalkyl agents could always undergo over-alkylation, as observed for ICH₂Bpin. Therefore, the reactivity of BrCH₂Bpin was inspected by incubating a range of concentrations with Leu-Enk peptide as a limiting agent, revealing that 150 µM BrCH₂Bpin offered the cleanest and most selective conversion at ambient temperature. The reaction proceeded with quantitative conversion in both HEPES and Tris buffers (pH 7.4), demonstrating its broad compatibility with standard biological buffers commonly used in bioassays. In addition, the reaction was found to be equally efficient at 5 and 40 ^oC (Supplementary Fig. 8), underscoring the versatility of this method.

Interestingly, iodoacetic acid and iodoacetamide, the frequently used Cys alkylating agents, revealed markedly slower reaction kinetics than BrCH₂Bpin under similar reaction conditions (Supplementary Fig. 9). The optimized reaction conditions were applied to individual α-N-Fmoc protected natural amino acids bearing the free side chain to test the chemoselectivity of the method. As expected, Cys showed quantitative conjugation, whereas no significant methylene-BA conjugate was detected with other amino acids tested in the HPLC and LC-MS analysis, except for methionine (~35%) (Fig. 3d, Supplementary Figs. 71–78). Notably, methionine alkylation was minimized to 3–4% when the reaction was conducted in carbonate buffer (pH 7.4) (Supplementary Fig. 75). Such a methionine methylene-BA conjugate was not observed as a byproduct in peptides and proteins (vide infra) under similar reaction conditions, either in PBS or carbonate buffer (pH 7.4), a reasonable outcome considering the superior nucleophilicity of the sulfhydryl group over a thioether. This observation highlights the differing reactivity of an amino acid in its free form versus when incorporated into peptides or proteins. The purified LeuEnk-BA conjugate displayed good chemical stability over 24 h at the physiological pH (7.4); no significant degradation was noticed as analyzed by HPLC (Supplementary Fig. 9). However, exposure to biologically relevant redox agents for 24 h induced varying degrees of degradation: H₂O₂ (32%), GSH (22%), ascorbic acid (43%), and NADH (51%) (ESI section 6.2.7). Comparable results were obtained for the SPYGRC-BA peptide (Supplementary Fig. 17). SPYGR motif represents a sialic acid-binding epitope (vide infra, Fig. 4).

Fig. 4. Peptide substrate scope study.

Peptides programmed with Cys residue(s) explored the scope of the reaction for monovalent and multivalent boronic acid incorporation. The conversion of each crude reaction was analyzed through HPLC and LC-MS (see Supplementary Information). The isolated product after HPLC purification is shown in parentheses.

Scope of the methodology for the installation of BA on peptides

We delved into the scope of the optimized reaction for peptides with diverse sequences and lengths, aiming to install monovalent and multivalent methylene-BA. Eleven bioactive peptide sequences (linear and cyclic), equipped with a single cysteine, were synthesized (Fig. 4). The peptides ranged from short (4-mer) to long (36-mer) chain lengths, with the 20 natural amino acids covering hydrophilic as well as hydrophobic sequences. An excellent LC-MS conversion (crude, > 90%) was observed for all peptide sequences, which were purified through HPLC with isolated yields ranging from 33% to 60% (Supplementary Table 2). Among these monovalent boronopeptides, we picked the borono-p53 sequence, which covers most of the natural amino acids with functional side chains, for MS/MS analysis to validate the method’s chemoselectivity. The collective b and y ions unambiguously suggested the exclusive modification of the Cys residue (Supplementary Fig. 25). We observed that the method modifies methionine to a methylated product (~5%) along with the Cys residue (100%) in the CMPep peptide when the PBS buffer (7.4) is used, confirmed through LC-MS/MS (Supplementary Figs. 28 and 30). However, the optimized method in carbonate buffer showed no methionine alkylated product with UBI and CMPep (found a single -CH₂B(OH)₂ modified mass Supplementary Figs. 29 and 31). We are unsure about the buffer effect on this favorable outcome.

Nature often employs multivalent avidity since single-site affinity is typically poor at the cellular level to attain high affinity and effective biological functions²⁹. Against this backdrop, we were encouraged to deploy this BA installation method, generating multivalent boronopeptides. Lanreotide, a somatostatin analog having a disulfide bond, was first reduced with tris(2-carboxyethyl)phosphine (TCEP) and subjected to BrCH₂Bpin in optimized conditions, but failed to demonstrate product conversion. The result did not surprise us because TCEP has been reported to react with active halo derivatives and quench alkylating reagents³⁰. Further optimization of the reaction with a higher concentration of BrCH₂Bpin (750 µM) and 1 mM TCEP (a concentration commonly used for reducing proteins)³¹ unveiled a complete bis-BA conversion with lanreotide- and BCS-somatostatin analogs, confirmed via LC-MS analysis (Fig. 4, Supplementary Figs. 19 and 22). In the next level, we designed a 9-mer model peptide sequence containing 4 Cys residues. Three variations of this sequence were deliberately designed to display two, three, and four unprotected Cys residues upon peptide cleavage from resin (Fig. 4, Supplementary Figs. 39–41). Peptides (100 μM) were first treated with TCEP (1 mM) in PBS (pH 7.4) to avoid unwanted intramolecular disulfide crosslinking. Subsequently, treatment with 750 µM, 1 mM, and 2 mM BrCH₂Bpin resulted in the installation of bis, tris, and tetrakis BA (Fig. 4). This developed method paved the way for accessing multivalent BA-modified peptides using commercially available reagents. We observed an additional mechanical loss with the isolation of multi BA-conjugated products via HPLC purification, possibly due to the increment of the stationary phase interactions with BA moieties. Overall, the designed methodology offered a highly efficient and chemoselective modification of the Cys residue, regardless of the choice of peptide sequence.

Protein modification and its integrity studies

Being one of the least abundant amino acids in proteins, Cys plays a multifaceted role in proteins, largely acting as a structural component and a redox-active residue, contributing to biogenesis, protein stability, folding, signal transduction, and catalysis³². Some of these functions could be well complemented, as shown by the modification of BA in proteins^19,33. Having established the reaction strategy with peptides, we sought to test the efficacy of the method for selective protein modifications (Fig. 5). We synthesized a short protein sequence (51-mer) by arbitrarily combining two peptide fragments (sequence number Ala1-Ala34 and truncated Cys106-Pro116-(Arg)₄, namely PDL1-A and PDL1-D, respectively³⁴) of programmed death ligand (PDL1) by native chemical ligation³⁵. The protein (~6 kDa, 10 µM) was selectively modified at the Cys residue using the optimized method that did not use TCEP. The LC chromatogram of the conjugated product did not appear to show a clear peak shift from the starting material. However, the MS spectra revealed the full conversion of the ~6 kDa protein with the addition of +40 Da mass [+58 Da – loss of water molecule] (Fig. 5a, Supplementary Fig. 34). Further, lysozyme (~14 kDa, 10 μM, PDB ID: 1DPX) was partially reduced with TCEP ( ~ 1 mM); consequently, the treatment of BrCH₂Bpin (~750 μM) delivered the methylene-BA conjugates of 2-Cys residues as the major product (91%), and 1-Cys residues as a minor product (9%) (Fig. 5b, Supplementary Fig. 36). We performed an in-depth MS/MS sequencing with the major product of lysozyme conjugates to confirm the BA conjugation sites and compared the results with native lysozyme. After a quick desalting step, a bottom-up approach was followed for protein sequencing, where peptide mapping and MS/MS studies ensured the modification of Cys6 and Cys115 residues (Fig. 5b, Supplementary Figs. 42, 43). We believed that the TCEP treatment reduced >1 disulfide bond, and using the optimized stoichiometry reagent for bis-BA modification enabled conjugations to the two most reactive free thiols. The combined data demonstrate the method’s ability to control the incorporation of the number of boron units into a protein.

Fig. 5. The scope of the reaction in protein substrates.

a PDL1 (A-D) and its mass, (b) installation of bivalent boronic acid on lysozyme was a major product, and BA-modified peptide mapping through protein digestion and subsequent MS/MS studies ensured the modification of Cys6 and Cys115 residues. c The deconvoluted mass confirmed the installation of BA on BSA. d Subsequent labeling of BSA-BA by a fluorophore SHA-FITC and SDS-PAGE analysis showing the visible protein bands (left) and fluorescent band (right), where BSA unmodified BSA protein, BSA-Fl unmodified BSA incubated with SHA-FITC, BSA-BA protein modified with BA handle, BSA-BA-Fl modified BSA-BA protein incubated with SHA-FITC. e The circular dichroism signature of BSA and BA-modified BSA unveiled a comparable signature.

The method would expect a single site modification of the free thiol residue in the native BSA protein (~63 kDa, 10 µM, PDB ID: 3V03) with the optimized method when the reagent concentration is 150 μM. The accompanying deconvoluted MS spectra of the reaction mixture (conversion >80%, Fig. 5c) showed the addition of methylene-BA at a single site, proving the developed strategy’s suitability for installing boron in a native protein. To further validate the existence of the BA handle in BSA, we employed a bioorthogonal fluorescence-tagging approach using salicylhydroxamic acid (SHA), a known BA-binding motif^36,37. An SHA derivative tagged with a fluorophore (SHA-FITC) was synthesized to enable its visualization. Following a 10-minute incubation of SHA-FITC with borono-BSA (BSA-BA), an SDS-PAGE analysis was performed where the modified and unmodified proteins were run with and without SHA-FITC. In line with our expectations, only the SHA-FITC incubated boronic acid-modified protein (BSA-BA-Fl) showed a bright green fluorescence, whereas all control lanes did not reveal any noticeable fluorescence (Fig. 5d, Supplementary Method 5). The results suggested that pinacol hydrolysis occurs rapidly under physiological conditions and is immediately followed by efficient SHA-FITC conjugation. We found an exciting result in the circular dichroism (CD) study, where the CD signature of the BSA-BA replicated that of native BSA (Fig. 5e, Supplementary Fig. 38), proving the biocompatibility and superiority of the method for protein modification without perturbing the native fold. Purified lysozyme-BA and BSA-BA proteins were found to be stable for over a week as a lyophilized powder, confirmed by the deconvoluted mass presented.

Design of boronopeptides for augmenting bacterial imaging

Staphylococcus aureus (S. aureus) is a leading cause of a wide range of infections in clinics³⁸ and millions of deaths globally due to antimicrobial resistance³⁹. Such clinical emergencies demand the expansion of innovative biochemical agents to circumvent these challenges. Chemical modulation of antimicrobial peptides (AMP) that improve bacterial membrane recognition and target bacterial lipids has been shown to engender improved imaging methods for bacteria and better solutions to the issue of antibiotic resistance^40–42. Building on our ongoing efforts to develop next-generation bacterial imaging agents^43–45, we led to engineering of antimicrobial peptides by applying the developed BA installation method. We conceptualized that the addition of Cys-modified methylene-BA moiety to the cationic AMP backbone would preserve its functional activity, which mainly relies on electrostatic interactions with the negatively charged bacterial membrane. Importantly, this chemical translation of cationic AMP with BA may anticipate a dynamic covalent anchoring mechanism with 1,2-diol or 1,3-diol functional groups present on bacterial cell surfaces, mainly from WTA/LTA. Such a dual binding paradigm, covalent conjugation uniting with the electrostatic interaction of AMP, must provide synergistic, tight membrane binding (Fig. 6a). Synergistic interaction, or synergism, refers to the combined effect/interaction of two or more substances or actions that is greater than the sum (additive) of their individual effects⁴⁶. The mentioned binding synergism could significantly enhance bacterial targeting and imaging efficiency. This approach is particularly relevant for S. aureus strains, which employ an intriguing lipid mutation strategy by lysylation of phosphatidylglycerol (Lys-PG) in its plasma membrane⁴⁷, resulting in a reduced net negative charge on the membrane surface, thereby compromising electrostatic interactions by cationic AMP (Fig. 6a).

Fig. 6. Engineering borono-UBI peptides and their recognition to S. aureus and serum stability.

a Chemical structure of major lipid compositions of S. aureus. A cartoon representation proposing our hypothesis of augmenting the bacterial cell wall recognition through synergistic interaction. b The sequence of UBI (29-41) peptide (P0) and its mutation site with BA led to the design of boronopeptides (P1-2). A short non-antimicrobial peptide having Cys(methylene-BA), P3, was added as a control. c S. aureus staining ability of P0-P3 was quantitatively assessed via flow cytometer at different concentrations (Data were presented as mean values  ±  SEM; n  =  3 independent replicates). Median fluorescence intensities with error bars were plotted for better comparison. d The staining efficiency of P0-P2 (2 µM) followed a similar pattern under a confocal microscope. P2 showed bacterial labeling even at 50 nM (right lane) without any wash condition. The scale bar represented in the images is 20 μm. Representative images are from 2 independent experiments. e Half-life studies of P0 and P2 in human serum monitored in HPLC (Data were presented as mean values  ±  SEM; n  =  3 independent replicates).

To assess our hypothesis, a clinically used cationic AMP UBI (29-41), derived from Ubiquicidine protein⁴⁸, was strategically programmed with mono- and bis-Cys(CH₂BA) (namely P1 and P2) by replacing Glu(9) and Asn(11) in the UBI (29-41) sequence. UBI (29-41) is a cationic peptide, carrying +6 charges with side chains at neutral pH, faces poor salt tolerance, resulting in inferior bacterial membrane binding⁴³. Incorporation and incrementation of BA moieties in the UBI peptide are anticipated to result in better membrane binding. In our design, we restricted UBI (29-41) modulation to mono- and bis-BA incorporation by replacing the native Gln and Asn residues (Fig. 6b, Supplementary Fig. 44). A limited number of BA modifications minimizes bacterial clustering, as observed by Hayashita et al.⁴⁹ through BA-modified dendrimers, which avoids non-specific interactions of multipoint BA moieties with glycans in mammalian biosystems. To detect and visualize bacterial staining, tailor-made boronopeptides (P1 and P2) were labeled with fluorescein at the N-terminus by adding β-alanine as a spacer. The fluorescent-labeled native UBI (29-41) (P0) was considered for comparison with boronated-UBI peptides. We included a short non-antimicrobial peptide attached to Cys-(CH₂BA) (P3) in the bacterial staining experiment as a control to investigate the affinity of only a BA moiety without UBI and evaluate whether the boronated-UBI illustrates synergism (Fig. 6b, Supplementary Fig. 45). The peptides were then incubated with S. aureus in a series of concentrations (from 0.05 to 2 µm) in PBS (pH 7.4) for 30 min and subjected to stringent two washes in PBS (pH 7.4). Stained samples and non-stained bacteria (blank) were subjected to flow cytometer analysis to quantitatively assess and compare their staining efficacy. Remarkably, P2 exhibited ~40-fold higher staining efficiency than P0 and P3 at 2 µM, while P1 showed ~8-fold improvement of labeling (Fig. 6c, see flow cytometry median value, Supplementary Figs. 46, 47), demonstrating the importance of additional BA-moieties for better membrane recognition. By contrast, P0 and P3 displayed minimal labeling, even at the highest concentration (2 µM), suggesting that their association relies solely on electrostatic interactions or weak diol binding, which are insufficient to withstand post-incubation washes. However, P1 and P2 are anchored tightly to the membrane through synergistic interactions. The finding was made unambiguous when the combined staining effect of P0 and P3 was found to be lower than the individual P1 outcome (see Supplementary Fig. 48; the median of (P0 + P3) is shown in a dotted line). Precisely, P1 showed >4 times higher staining efficacy than the combined median fluorescence outcome of P0 and P3 at 2 μM treatment. Similar trends were observed when we studied P2 staining efficacy under a confocal microscope (Fig. 6d). We speculated that S. aureus might uptake P1 and P2 in the cytosol because of their tight binding ability to the plasma membrane, which perhaps be the reason for the bright staining. The additional examination of optical sectioning on stained bacterial cells under a confocal microscope validated the uptake of P2 in bacterial cytosol (Supplementary Movie 1). Although P1 and P2 showed a slight increase (8–15%) in bactericidal activity compared to P0 at 2 µM reagent, even after 12 h of incubation (Supplementary Fig. 49). Notably, P2 was proven as a highly effective staining agent that S. aureus could visualize with as low as 50 nM concentration without any wash (Fig. 6d lane 4). To the best of our knowledge, this represents the first demonstration of an alkylboronic acid-modified AMP that gains remarkable imaging efficiency through the synergy of electrostatic interactions and enhanced complexation ability with bacterial surface diols. Due to the presence of a neighboring sulfur atom within the BA moiety, the diol association ability of BA might be facilitated via the moderate interaction between S(lp)→B(p), as supported computationally. In a serum stability study with P0 and P2 (fluorophore unlabeled), P2 elicited around 14 times higher half-life than the parent UBI peptide (P0), monitored by HPLC (Fig. 6e), consistent with the earlier observation on borono-cyclic peptide⁵⁰. Collectively, our design and investigation demonstrate the preeminence of chemical modification with Cys(CH₂BA) on AMP peptides in the development of advanced bacterial infection imaging.

Selectivity and synergistic binding studies with P2

The development of precision imaging agents highly relies on target selectivity^51,52. UBI (29-41) peptide has been shown to be more selective towards gram-positive bacteria⁵³, primarily used in monitoring S. aureus infection in vivo^44,54,55. Motivated by this, we further sought to investigate the selectivity of the adequate reagent P2 (bis-BA modified UBI) in our hands for gram-positive bacteria over gram-negative bacteria and mammalian cells. The initial staining of E. coli with P2 (5 μM) resulted in no significant staining (Fig. 7a), which was ambiguous to us, because Pacheco et al. have demonstrated that Gram-negative bacteria (Salmonella enterica) could be stained with BA-fabricated quantum dots⁵⁶. To better comprehend this result, a co-culture of E. coli and S. aureus was incubated with P2 (5 µM), and stringent-washed cells were analyzed under a confocal microscope, which revealed only strong fluorescence staining with S. aureus (shown in blue square, Fig. 7a), demonstrating the S. aureus selectivity. The observation was supported by the flow cytometry outcome, where no significant staining for the E. coli was noticed (Supplementary Fig. 47), which we believe is due to the inability to uptake boronate-UBI peptide through the thick lipopolysaccharide cell wall. Furthermore, the incubation of P2 (5 µM) with a normal cell (HEK 293) and a cancer cell (A549) showed minimal staining compared to the brightness detected on S. aureus. The result is not surprising because the UBI (29-41) AMP peptide exhibits selective binding to bacterial membranes over mammalian cells. However, the insertion of BA moieties in P0 could non-specifically bind to mammalian cell surface glycans and glycoproteins. However, it could be unlikely to happen with alkyl BA due to the poor association with membrane diols at physiological pH, compared to most aromatic BA reagents having relatively lower pK_a²⁷. Additionally, the conjugates of diol boronate ester are reversible, so only BA moieties will be less likely to help associate strongly with mammalian cells, as we demonstrated with P3, where BA individually does not stain S. aureus. Later, P0-P2 (2 µM) were incubated with varied human serum concentrations to extend the selectivity studies, and staining efficiency was measured in a flow cytometer. The result suggested that all peptides showed a similar trend of inhibition properties even in 20% human serum due to unwanted non-specific interactions (Fig. 7b, Supplementary Fig. 50). Altogether, these boronopeptides are promising for the development of bacterial infection imaging agents, as manifested by the in vitro experiments.

Fig. 7. Binding selectivity and probable mode of binding studies.

a Selectivity studies of P2 (5 µM) imaging to Gram (+), S. aureus, over Gram (-) and mammalian cells under a confocal microscope. (1^st row) E. coli did not show significant staining; (2^nd row) A co-culture of S. aureus and E. coli, where E. coli (white circle) did not show fluorescent labeling. The labeled S. aureus is shown in a blue square. (3^rd row) HEK293 and (4^th row) A549 were minimally stained. The white scale bar inside the image represents 20 μm. Representative images are from 2 independent experiments. b Human serum inhibition studies with P0-P2 in S. aureus (ATCC 6538) labeling. Normalized data is presented to visualize and compare data. Data were presented as mean values  ±  SEM; n  =  3 independent replicates. Two-way ANOVA was used, ****P  <  0.0001. Source data are provided as a source data file. c The concept of inhibiting synergistic binding of P2: masking BA via complexation with mannitol (left) and selective blocking of lipoteichoic and teichoic acid on S. aureus membrane (right). d The presented flow cytometer histograms compare the S. aureus labeling, which clearly proves the inhibition of P2 binding. All flow cytometer experiments were performed twice with consistent outcomes. e The bar graph represents the inhibition of covalent binding between BA-moieties and diols present on the S. aureus cell surface in a dose-dependent manner. Scrambled hemoglobin subunit β (HSβ) peptide exhibited a minimal inhibition. Data were presented as mean values  ±  SEM; n  =  3 independent replicates. One-way ANOVA was used, ****P  <  0.0001, ***P  =  0.0007 for P2 + HSβ 20 µM, **P  = 0.001 for P2+Mannitol 100 mM. Source Data are provided as a Source Data file.

It is crucial to prove whether the complexation between BA and diols present on the bacterial cell surface, particularly within WTA/LTA, indeed enhances the labeling efficiency of boronated-UBI. In order to validate our hypothesis, the P2 peptide (2 µM) was first complexed with mannitol at three concentrations (20, 50, and 100 mM), which would mask the BA moieties (Fig. 7c). Mannitol can form a diol complex with BA, as established earlier²⁷. Once the complexation between mannitol and P2 is built, the binding aptitude of P2 on the bacterial cell membrane is anticipated to diminish. The flow cytometer data favorably revealed a reduced staining outcome of P2 in a dose-dependent manner when complexed with mannitol (Fig. 7d, e, Supplementary Fig. 51). For instance, 50 mM mannitol complexation with P2 reduced its staining ability to ~60%. We also speculated that WTA and LTA represent the major cell surface components to which P2 might significantly form diol complexes, resulting in better staining. In such contexts, pre-blocking of LTA and WTA should diminish the binding efficacy of P2 (Fig. 7c). A peptide that binds to LTA, outsourced from Hemoglobin subunit β (VLGAFSDGLAHLDNLK, namely HSβ, Supplementary Fig. 52)⁵⁷, was incubated with S. aureus for 30 min with a range of concentrations (1, 5, and 20 μM) and subsequently treated with P2. These stained samples, including only P2-treated S. aureus as a control, were analyzed using flow cytometry. As expected, pre-treatment of S. aureus with 20 µM HSβ resulted in an approximately 80% decrease in P2 staining efficiency, consistent with competitive blocking of WTA/LTA binding sites. (Fig. 6d, e, Supplementary Figs. 53, 54). However, it is challenging to fully ascertain the synergistic effect of P2 owing to the dynamic nature of HSβ interactions with LTA/WTA. Additionally, glycoproteins on the bacterial cell surface could associate with BA moieties. To validate the selectivity of HSβ to LTA/WTA, we showcased a control experiment, where the incubation of a scrambled HSβ (20 μM) (Supplementary Fig. 52-54) demonstrated a slight binding inhibition (~18%) of P2, while the observed effect perhaps stems from transient kinetic trapping of the scrambled HSβ with the membrane (Fig. 7e). These experiments and previously shown flow cytometry data (Fig. 6c) judiciously prove our hypothesis of synergistic interaction and the objective of staining enhancement. Based on the HSβ inhibition results, we hypothesize that the BA moiety likely binds to the 4,6-diol of GlcNAc present in LTA/WTA (Fig. 6a); nonetheless, it cannot be ruled out that the possibility of BA covalent complexation with other bacterial surface-exposed sugar moieties containing 1,2-diol or 1,3-diol groups. Additional biophysical and biochemical analyses are necessary to elucidate the underlying mechanisms more comprehensively in the future.

Discussion

In view of the multifarious applications of boronopeptides in chemical biology and medicinal chemistry, we have developed an efficient method for the facile and highly chemoselective installation of an alkyl boronic acid handle on peptides and proteins using readily available halomethylboronic esters in aqueous buffer at physiological pH. This easily operable strategy renders excellent conversions with rapid kinetics (~10² M⁻¹s⁻¹), eliminating complex technical steps. Our studies revealed that the method may lead to undesired methionine alkylation, a side reaction that can be effectively eliminated by performing the reaction in carbonate or Tris buffer. The method showcased the ability to graft bis-, tris-, and tetravalent BA onto both peptides and proteins with high fidelity via Cys-selective conjugation. Moreover, this methodology was applied to modulate the membrane binding efficiency of an AMP (UBI 29-41) to S. aureus through the synergistic mechanism, resulting in 8- and 40-fold enhanced imaging ability with mono- and bivalently labeled borono-UBI, respectively. We believe that bisborono-UBI (P2) exhibits enhanced binding efficacy due to the increased valency of the BA moiety. However, P2 does not compromise its binding selectivity to S. aureus over gram-negative bacteria (E. coli) and mammalian cells (HEK293 and A549). In a wash-free condition, P2 was found to label S. aureus cells at 50 nM concentrations and in the presence of human serum. Although our experiments suggest that the BA moieties likely recognize the 4,6-diol component of GlcNAc present in LTA/WTA, comprehensive biochemical and biophysical investigations are necessary to elucidate the precise recognition mechanism of boronic acid toward membrane-associated sugars. Currently, we are investigating the efficacy of P2 compared to UBI for imaging of intramuscularly S. aureus infected mouse models. Additionally, the biodistribution in the mouse model organs and the toxicity of P2 will be studied. We foresee numerous potential applications for this alkyl BA installation strategy.

Methods

Theoretical calculations

The energy calculation and small molecule optimization were performed using the Gaussian 09 program at the B3LYP level of theory with the 6–31 G or LANL2DZ basis set. The conductor-like polarizable continuum model (CPCM) solvation model was employed to simulate systems in aqueous solution. The GaussView 6.0 software program was used for the visualization and analysis of the orbital shape and overlap (isovalue = 0.02).

Synthesis of peptides

The peptides were synthesized on Rink amide AM or CTC resin (loading capacity 0.67 mmol/g for Rink amide and 1.2 mmol/g for CTC) using a standard Fmoc-deprotection strategy. The resin was used with the specified loading or downloaded to a lower loading as required. Three equivalents of the commercially available amino acids, HBTU as a coupling agent, and 4.5 equivalents of DiPEA as a base, were used for the coupling reaction for 15–20 min, followed by 3 times wash with DMF. Fmoc deprotection was achieved using 3 mL 20% piperidine/DMF twice for 3 min each, followed by washing with DMF six times. The peptides were cleaved off the resin and globally deprotected with reagent B (88% TFA, 5% phenol, 5% H₂O, and 2% TIPS) for shorter 6 to 8-mer peptides or with reagent K (82.5% TFA, 5% H₂O, 2.5% EDT, 5% thioanisole, and 5% phenol) for longer sequences. A typical time of 2 h was provided for cleavage. Precipitation with chilled diethyl ether yielded the crude peptides, which were subsequently purified by RP-HPLC. The purity of the peptides was determined to be greater than 90% using LC-MS (Waters) and HPLC (Shimadzu). The characterization data are presented in Supplementary Table 2. All the peptides were purified by semi-preparative RP-HPLC and lyophilized to obtain them as a white powder with an isolated yield of 25–35%.

Installation of boronic acid in peptides

Stock solutions of peptides and bromomethyl boronate pinacol esters (BrCH₂Bpin) were freshly prepared in DMSO and acetonitrile, respectively. 50–100 µM of peptides were reacted with 1.5 equivalents of BrCH₂Bpin in PBS buffer (pH 7.4) for 10 min at 25 °C. Subsequently, the reaction mixtures were quenched with 10% TFA/H₂O and analyzed by RP-HPLC and LC-MS. For methionine-containing peptides, using carbonate buffer instead of PBS resulted in a marked suppression of the alkylation reaction of methionine residues. Site-selective modification to cysteine residues was confirmed by ¹H NMR and MS/MS studies.

Installation of multivalent boronic acid

For 50–100 µM peptides containing more than one cysteine residue or disulfide bonds, peptides were incubated for 1 h at pH 7 to reduce the probability of forming disulfide linkages. Taking into account the reactivity of phosphine reagent with halomethyl reagents, 750 µM of BrCH₂Bpin was added to the reaction mixture to achieve clean quantitative conversion to bis boronopeptides, 1 mM and 2 mM of boronic acid reagent to obtain tris and tetra boronopeptides, respectively, with reaction times extending to 30 min.

Conjugation of boronic acid to protein

To a solution of 10 µM protein containing one cysteine residue prepared in PBS (pH 7.4), 150 µM of BrCH₂Bpin was added and stirred for 30 min at 25 °C. In the case of lysozyme, it was first treated with an optimal amount of TCEP to reduce a vulnerable disulfide linkage, followed by treatment with the BrCH₂Bpin reagent, a similar method to that used for multivalent peptides. The reactions were acidified after completion, and the product was confirmed by deconvoluted MS spectra. PAGE analysis also revealed successful installation of the boronic acid residue on the BSA protein. Proteomics was also carried out to confirm the site-selective installation of bis-boronic acid in lysozyme.

Flow cytometry for bacterial staining measurement

Stock solutions of all fluorophore-labeled peptides were prepared in biological grade DMSO (2 mM). 1 mL of bacterial (S. aureus (MTCC 7443) or E. coli (MTCC 1687), CSIR- IMTECH, India) solution was subsequently incubated with the fluorophore-labeled peptides (P0-P2) at six different concentrations, ranging from 0.05 to 2 μM, for 30 minutes at 37 °C, along with a negative control without peptide. Post incubation, all samples were thoroughly washed 3 times with PBS and then resuspended in 1 mL of PBS. Flow cytometry analysis was performed using a Sysmex Cube 6 system, with the following settings: laser wavelength, 488 nm; detector channel, FL1; and bandpass filter wavelength, 533/30 nm. Sample analyses were conducted with FSC voltage set to 250, SSC voltage set to 350, and FL1 voltage set to 550. Fifty thousand events were collected in all samples. Fluorescence readout was measured based on a forward scatter threshold. Data analysis was performed using FlowJo software (Tree Star, Inc.), with constant gating applied to non-treated bacteria for the experiments. The median fluorescence intensities of the stained bacteria were then extracted and plotted. The gating strategy was kept conserved across all samples. Each data point was taken in triplicate. The entire experiment was repeated (i.e., n = 2) to assess the consistency of the results.

Confocal imaging

For confocal imaging, samples were prepared in a similar manner to the flow cytometry measurement. In this measurement, a single concentration of 2 M was used for incubation at 37 ^oC for 30 min. 5 μL of bacterial cells were placed on a glass slide, and a 22 × 22 × 0.15 mm³ Fisherbrand microscope cover glass was placed on top. The coverslip was pressed down on the cell droplet to give a single layer of cells on the glass slides. In cases of mammalian cells (A549 and HEK293, from NCCS repository, India), 70% confluent cells were detached from 25 cm² tissue culture flasks using Trypsin-EDTA (2 min at 37 °C, 5% CO₂), pelleted, and seeded into confocal dishes with glass-bottom (20 mm diameter) at a density of 1 × 10⁵ cells/well. After 24 h of incubation, the media was removed and replaced with 2 mL of PBS containing fluorophore-labeled P2 at a concentration of ~2 μM. Cells were incubated for 30 min at 37 °C, then gently washed three times with PBS. Finally, 2 mL of fresh media was added before imaging under the confocal microscope. Images were taken on a Carl Zeiss LSM 880 (laser scanning confocal microscopy). Acquired images were processed using ZEISS ZEN 3.10 software. A 40× oil objective was used with an Argon laser at 10% laser power. The gain was adjusted to a range of 900 HV to 1100 HV with an offset of −0.5%. Acquired images were processed using ZEISS ZEN 3.10 software.

Bacteria co-culture preparation

For the co-culture staining experiment, E. coli (500 μL) and S. aureus (500 μL) were separately diluted (1:100) from the secondary culture and allowed to grow till OD₆₀₀ = 0.4. The bacteria were mixed in a separate microcentrifuge tube, combining 500 µL of each resuspended cell, then spun down at 5000 × g. The combined cell pellet was treated with 1 mL fluorophore-labeled P2 (2 μM) in PBS buffer (1X, pH 7.4) and was incubated at 37 °C for 30 min. Cells were then spun down and washed twice with PBS before being diluted in 100 μL of PBS buffer (1X, pH 7.4) for microscopic imaging analysis.

Serum inhibition study

Serum concentrations ranging from 0% to 20% were prepared from fresh serum isolated from a healthy human. Bacterial cell pellets were resuspended in this solution, and to it the fluorophore-labeled peptide P2 was added at a final concentration of 0.5 μM. The mixture was then incubated at 37 °C for 30 minutes. The cells were then washed three times with PBS and analyzed by flow cytometry according to the protocol described above. Each measurement was performed in triplicate, and the entire experiment was repeated (n = 2) to confirm reproducibility of the results.

Statistics, reproducibility, and figures

All the experiments were independently repeated two to three times. Data are presented as mean values ± SEM (Standard Error of the Mean) calculated using OriginPro 8.5. Figures were generated using GraphPad Prism 8, PyMol 2.4.1, and Microsoft PowerPoint 2021.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

A.B. conceived the project, supervised the entire project, analyzed data, and wrote the manuscript; S.C. performed all reaction optimizations, peptide and protein substrate scopes, DFT calculations, and assisted with manuscript preparation; NMT synthesized the PDL1-related peptide substrates. A.C., N.V., and V.G. synthesized AMP-borono peptides, performed all biological assays, and helped with data analysis. S.B. performed native and BA-modified protein sequencing and data analysis under the supervision of V.R. All authors contributed to the preparation of the manuscript and the details of the supplementary information.

Peer review

Peer review information

Nature Communications thanks the anonymous reviewers for their contribution to the peer review of this work. A peer review file is available.

Data availability

All data supporting the findings of this study are available within the Article and its accompanying Supplementary Information file. The source data for Figs. 3b, 3c, 3f, 5d, 5e, 6c, 6e, 7b, 7e and Supplementary Figs. 48, 49 are provided in the Source Data file. The data that support the findings of this study are available within the main text, its Supplementary Information, and from the corresponding author(s) upon request. The raw data generated in this study have been deposited in the Figshare public repository [10.6084/m9.figshare.30102463]. Source Data are provided with this manuscript. 

Competing interests

The authors declare the following competing financial interest(s): A.B. and S.C. have filed an Indian patent application (202411017017) on this invention. All other authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-025-67141-5.