Comparison of Bioorthogonal β-Lactone Activity-Based Probes for Selective Labeling of Penicillin-Binding Proteins

Abstract

Penicillin-binding proteins (PBPs) are a family of bacterial enzymes that are key components of cell-wall biosynthesis and the target of β-lactam antibiotics. Most microbial pathogens contain multiple structurally homologous PBP isoforms, making it difficult to target individual PBPs. To study the roles and regulation of specific PBP isoforms, a panel of bioorthogonal β-lactone probes was synthesized and compared. Fluorescent labeling confirmed selectivity, and PBPs were selectively enriched from Streptococcus pneumoniae lysates. Comparisons between fluorescent labeling of probes revealed that the accessibility of bioorthogonal reporter molecules to the bound probe in the native protein environment exerts a more significant effect on labeling intensity than the bioorthogonal reaction used, observations that are likely applicable beyond this class of probes or proteins. Selective, bioorthogonal activity-based probes for PBPs will facilitate the activity-based determination of the roles and regulation of specific PBP isoforms, a key gap in knowledge that has yet to be filled.

Graphical Abstract

Isoform-specific targeting: The synthesis and comparison of selective, bioorthogonal, activity-based probes for penicillin-binding proteins (PBPs) is reported. We demonstrate the expanded functional utility of bioorthogonal probes compared to fluorescent analogues and explore design considerations for the development of bioorthogonal probes that are applicable beyond the probes described in this work.

Introduction

Antimicrobial resistance is a global health crisis^[1] that has led to the evolution of myriad bacterial strains that are multidrug or pan-resistant, necessitating the discovery of new antibiotics and mechanisms for targeting resistant pathogens. Bacterial cell-wall synthesis is, and has been, an attractive target for antibiotic development, as the cell wall is essential for survival and growth. The enzymes and pathways responsible for cell wall synthesis and remodeling are orthogonal to mammalian host cell machinery, which minimizes potential off-target and toxicity concerns. Peptidoglycan (PG) is the primary component of the cell wall, consisting of alternating N-acetylmuramic acid and N-acetylglucosamine appended to a pentapeptide tail. The pentapeptide tail contains a terminal d-Ala residue, which is hydrolyzed to form a crosslink with another PG pentapeptide stem to give the cell wall its characteristic “mesh-like” structure and rigidity (Figure 1A).^[2] Penicillin-binding proteins (PBPs) are key components of PG synthesis, remodeling, and cell division; they are responsible for both the polymerization and crosslinking of PG via their transglycosylase (TG) and transpeptidase (TP) domains, respectively.^[3,4] PBPs are the target of penicillin and other β-lactam antibiotics, which act as substrate mimics to irreversibly bind the catalytic active site serine residue in the TP domain of PBPs.^[5] Despite the importance of PBPs, little is known about the precise regulatory mechanisms that govern their activity and localization.^[6,7] Most organisms contain between four and 16 PBP isoforms, which have highly homologous TP active sites.^[3] This structural similarity complicates the investigation of individual PBP isoforms, because known inhibitors such as the β-lactams bind many, if not all, PBPs within a given organism.^[8–12]

Figure 1.

A) The transpeptidase domain of PBPs crosslinks the pentapeptide tail of two peptidoglycan strands via an acyl-enzyme intermediate that is intercepted by a nucleophilic amino acid residue. B) Previously reported fluorescent β-lactone activity-based probes.^[14] The identity of fluorophore conjugated to the Phe-lactone probe scaffold can alter the binding profile and/or introduce non-PBP labeling. B = BODIPY-FL, T = TAMRA, FL = fluorescein.

Despite the structural and functional homology of the PBPs, they are differentially regulated throughout cell division and cell wall remodeling.^[13–15] To understand bacterial cell growth and division fully, the individual roles of PBPs must be assessed spatiotemporally and in an activity-dependent manner. Biochemical experiments have been conducted to study individual PBP isoform localization and protein-protein interactions with fluorescent fusion proteins or immunoaffinity tags.^[13,15–17] However, these tags do not enable the selective study of catalytically active PBPs, and larger fusion tags can perturb protein-protein interactions or localization patterns.^[18–22] While β-lactams have been utilized as activity-based probes for profiling PBP activity in different organisms,^{[11,12,23,24]} they often label several PBPs, rendering them nonselective, while also regularly exhibiting reduced (or no) labeling of some PBPs.^[8,9] As such, few probes exist that enable the study of a specific PBP,^[8,9,12] thus highlighting the need to further explore chemical space for designing PBP-selective probes.

The β-lactone moiety is a privileged scaffold for activity-based probe development, found in many natural products^[25–30] and acting against a diverse array of enzyme classes.^[23,31–34] Accordingly, our lab developed a suite of fluorescent β-lactone activity-based probes, which were used to label specific PBPs in the model Gram-positive organism Streptococcus pneumoniae (Spn; Figure 1B).^[14,35] Specifically, probes 7FL and 8T enabled sequential labeling of the essential Spn PBPs, PBP2x and PBP2b, in a Spn Δpbp1b strain.^[14] Across the suite of fluorophore-conjugated probes that were developed, we found that the identity of the conjugated fluorophore could impact the selectivity of the probes, as well as promote labeling of non-PBPs.^[14] Additionally, the investigation of protein-protein interactions and regulatory protein complexes involving the PBPs will require the ability to perform classical crosslinking and affinity pull-down experiments of specific PBP isoforms, for which tags other than fluorophores will be required. Herein, we report the synthesis and characterization of a versatile panel of β-lactone probes based on 7 and 8. Each probe contains a different bioorthogonal “click” chemistry handle compatible with either copper-catalyzed azide-alkyne cycloaddition chemistry (CuAAC) or copper-free tetrazine-trans-cyclooctene inverse electron-demand Diels-Alder (iEDDA) chemistry. Because fluorophore labeling occurs after PBP binding, these “clickable” probes retain the PBP selectivity of the original molecules while eliminating fluorophore-dependent off-target labeling and minimizing any potential contribution of the reporter group to PBP-probe interactions. Furthermore, we exploited the prevalence of commercially available bioorthogonal reagents for the application of synthesized probes to both fluorescent imaging and affinity enrichment of PBPs from live Spn.

Results and Discussion

Design and synthesis of bioorthogonal activity-based β-lactone probes

When designing bioorthogonal analogues of probes 7 and 8, we consulted previously reported bioorthogonal β-lactone and β-lactam activity-based probes.^{[11,31,32,36]} Given the frequency of CuAAC chemistry in the literature, we designed probes 7Ak, 7Az, 8Ak, and 8Az containing an alkyne or azide with a hexanoic acid linker (Figure 2). Hexanoic acid was chosen to mimic the six-carbon linker in the original probes 7FL and 8T. An azide-containing probe with a 4-PEG-unit linker, 7AzP, was also designed to assess the potential effect of linker length on selectivity and/or bioorthogonal labeling efficiency (Figure 2). CuAAC is one of the most widespread bioorthogonal chemistry systems in application,^[37–38] and as a result there are many commercially available reagents for affinity purification and imaging applications. Although CuAAC reactions are faster than bioorthogonal reactions such as the Staudinger ligation or strain-promoted azide-alkyne cycloadditions (SPAAC), they are still some of the slower bioorthogonal reactions reported.^[39] CuAAC possesses second-order rate constants on the order of 10–200 M⁻¹s⁻¹ and are rate-dependent on the concentration of the Cu^I catalyst.^[39] Cu^I can cause toxicity problems in live-cell applications,^[39–41] so we designed probes 7MTz and 7TCO to be fully compatible with live-cell assays (Figure 2). 7MTz and 7TCO contain iEDDA-compatible methyl tetrazine (MTz) and trans-cyclooctene (TCO) groups, respectively. iEDDA reactions are uncatalyzed and are among the fastest bioorthogonal reactions reported in the literature. With second-order rate constants nearing 10⁶ M⁻¹ s⁻¹,^[39,42–45 iEDDA chemistry is well-suited for use in low-abundance applications and live-cell imaging.^[44,46–49] While tetrazines are highly reactive, they also exhibit reduced stability. To guard against degradation, the more stable methyl tetrazine^[50] was chosen for 7MTz. 7MTz and 7TCO were both pursued to retain functional flexibility, as TCO has been shown to undergo cis-trans isomerization in the presence of thiols or copper-containing proteins.^[43,51] A 4-PEG-unit linker was used since MTz and TCO reagents based on hexanoic acid were not readily commercially available.

Figure 2.

Design of probes 7 and 8 containing various bioorthogonal groups, denoted by respective abbreviations.

Synthesis of bioorthogonal probes 7 and 8 proceeded through Boc-protected l- and d-phenylalanine-containing β-lactone intermediates 3 and 4 (Scheme 1), which were prepared according to previously published protocols (Scheme S1).^[14] Previously, deprotection of 3 and 4 was achieved in 50:50 TFA: DCM at room temperature to furnish 5 and 6, and then fluorophore coupling was conducted without further purification. This approach frequently resulted in yields less than 20%, and often required reaction time of 3 days or more.^[14] To improve the yield of the final coupling step, deprotection of 3 and 4 was performed in 95:5 TFA:DCM on ice with 1.05 molar equivalents of para-toluenesulfonic acid (pTsOH) added to form the tosylate salts 5 and 6 at 80.5% yield in both cases (Scheme 1). No further purification was required before installing the bioorthogonal groups (Scheme 1).

Scheme 1.

Synthesis of biorthogonal probes 7 and 8, with respective yields. Specific bioorthogonal groups are indicated by abbreviations defined in Figure 2.

As the synthesis was designed to utilize commercially available carboxylic acids, the final step was generalized across all seven probes. The TCO carboxylic acid was supplied as a racemic mixture of stereoisomers, resulting in an unresolvable mixture of diastereomeric products. Given the long PEG linker, however, it was deemed unlikely that the stereochemistry of the TCO moiety would have any effect on the labeling efficiency of 7TCO. Probe yields of 23–50% were achieved with reaction times of 18–24 hours (Scheme 1), representing a significant improvement from the final fluorophore coupling for 7FL and 8T. Following the synthesis and characterization of the suite of 7 and 8 bioorthogonal analogues, the PBP-binding profiles were assessed in live Spn cells.

PBP labeling profile of bioorthogonal probes in live S. pneumoniae

To assess the PBP binding profiles of the newly synthesized analogues of 7FL and 8T, cell suspensions of S. pneumoniae IU1945^[52] were incubated with each probe. Incubation was followed by a click reaction with the appropriate fluorescent reporter molecule, then cell lysis and gel-based fluorescent imaging (Figure 3A). Five μM Bocillin–FL (Boc–FL), a fluorescent analogue of penicillin V which labels all PBPs in Spn,^[9–10,14] was used as a positive control. The conditions for whole-cell PBP labeling and analysis were analogous to those reported previously,^[14] with the addition of a bioorthogonal click chemistry reaction (see the Experimental Section). For direct comparison with 7FL, fluorescein (FL or FAM) was the preferred bioorthogonal fluorophore, while the nearly spectroscopically identical AF488 was used as a replacement when needed. For comparison with 8T, TAMRA was selected, with Cy3 serving as an alternative where TAMRA was not available (Table S1). Following click chemistry coupling to the appropriate fluorophores, samples were separated via gel electrophoresis and analyzed with in-gel fluorescence imaging (see the Supporting Information). At 6.5 μM, 7FL primarily labeled PBP1b and PBP2x in Spn, with minor PBP2b labeling, and 8T labeled PBP1b, 2x and 2b.^[14] To assess concentration-dependent labeling, Spn cells were incubated with increasing probe concentrations, followed by click chemistry coupling of the fluorescent reporter (FAM/FL/AF488 for 7 analogues, TAMRA/Cy3 for 8 analogues). For all probes the expected profile was retained (Figure S1A–G). PBP labeling was concentration-dependent, with higher concentrations of probe resulting in increased off-target labeling. The labeling intensity relative to the Boc–FL control at a given concentration differed between probes, but the expected labeling profile was evident for all probes at 10 μM (Figure S1A–G). 7Ak and 7Az labeled PBP2b more strongly at higher probe concentrations than 7AzP, 7MTz and 7TCO. This could be due to the difference in linkers, as the 7Ak and 7Az contain a short, six-carbon chain compared to a four-unit PEG linker for the 7AzP, 7MTz and 7TCO. 7MTz labeling with AF488-TCO resulted in some concentration-independent non-PBP labeling. This is likely a result of AF488-TCO nonspecific labeling, which did not interfere with interpretation or analysis of PBP labeling data. Given the apparent variation in probe labeling intensities among the bioorthogonal probes, each was directly compared to its fluorescent analogue at a fixed probe concentration of 10 μM, and the average relative labeling plotted for each PBP labeled (Figure 3B, C). Overall, 7Ak, 7Az, 8Ak and 8Az exhibited reduced labeling compared to 7FL or 8T, though 7Az and 8Az label PBP2x more strongly than the alkyne-containing analogues. 8Ak and 8Az label PBP2b to a similar degree, and in general, exhibit lower PBP labeling compared to 8T. 7MTz and 7TCO labeled PBP2x more strongly than the CuAAC probes, with the exception of 7AzP, which contains the same four-unit PEG linker as 7MTz and 7TCO. iEDDA reactions are known to be more efficient than CuAAC reactions,^{[37,39,42,46]} but iEDDA probes labeled PBP1b at a similar or lower level compared to the CuAAC probes (Figure 3B), in particular 7AzP, despite containing the same four-unit PEG linker. These results indicate that a faster bioorthogonal reaction does not necessarily result in greater intensity of labeling. Accessibility of the bioorthogonal group may have a greater impact on labeling for some PBP isoforms, as illustrated by the stronger labeling of PBP1b and/or PBP2x by 7AzP and 7Az compared to 7Ak. Although 7Ak, 8Ak, 7Az and 8Az are all CuAAC probes with a hexanoic acid linker, the alkyne in 7Ak and 8Ak comprises carbons 5 and 6 of the hexanoic acid linker, while the azide group of 7Az and 8Az is attached to carbon 6 of the same moiety (Figure S2). The shortened alkyne linker could restrict access to the fluorescent reporter molecule during the click reaction, and account for the reduced signal of the alkyne-containing probes for PBP2x. This effect seems to be dependent on the specific protein environment since PBP1b labeling is consistent between Ak and Az probes (Figure 3B, C). The difference between PBP1b and PBP2x labeling intensity for 7MTz and 7TCO also indicate that protein environment can have a strong effect on bioorthogonal labeling efficiencies. PBP TP active sites are highly structurally homologous, but differences in antibiotic binding between PBPs, along with limited available structural comparisons, suggest that there are subtle differences in active site architecture between PBP isoforms.^{[8,9,12,53–56]} Further structural characterization of PBPs in complex with β-lactone probes will be necessary to achieve a structure-based understanding of differences in labeling efficiency.

Figure 3.

Live Spn cell bioorthogonal probe treatment and labeling. A) Spn cells treated with the bioorthogonal probe of interest, washed, resuspended, and reacted with the respective bioorthogonal fluorophore reporter prior to lysis, SDS-PAGE separation, and BODIPY-FL (green) and/or TAMRA (red) fluorescent imaging. B) Comparison of 10 μM 7FL to 10 μM bioorthogonal analogues; 5 μM Boc–FL is included as positive control and PBP marker. C) Comparison of 10 μM 8T to 10 μM bioorthogonal analogues; 5 μM Boc–FL is overlaid as positive control and PBP marker. Bar graphs represent relative % intensity labeling compared to the indicated fluorescent probe (error bars represent standard deviations from biological duplicates).

The ability to exchange reporter molecules is a key advantage of using bioorthogonal probes. Spn cultures were incubated with each probe then reacted with both a “red” and “green” fluorescent reporter to ensure that changing the reporter did not affect the probes’ selectivity. The PBP selectivity profile of each probe did not change appreciably when the fluorophore was changed, with the exception of 7Ak PBP2x labeling, which increased significantly for TAMRA–Az compared to FL–Az (Figure S3). The difference between FL–Az labeling and TAMRA–Az labeling of 7Ak samples is likely due to a longer linker between the azide and fluorophore for TAMRA–Az (six carbons in TAMRA–Az vs. three carbons for FL–Az). As evidenced earlier, in the context of native protein environments, the probe linker length might have a pronounced effect on labeling reactions, and this may extend to the fluorescent reporter molecule. Visible non-PBP labeling was observed for TAMRA-alkyne-treated samples and TCO-containing samples (both probe-treated and negative controls), but it did not interfere with the gel-based analysis of PBP labeling at 10 μM β-lactone probe concentration. In general, alkyne-containing fluorophores exhibited greater background labeling than azide-containing fluorophores, an important observation for future consideration CUAAC probe design and experiments. The differences in background labeling between different fluorophores highlights the need to carefully select the probe, fluorescent reporter molecule, and reaction conditions when designing experiments. While the level of background observed is acceptable for gel-based analysis, for other applications such as microscopy, fluorophore and probe concentrations and incubations must be optimized to ensure that there is not any interference from nonspecific labeling. We next set out to investigate the differences in relative labeling intensities between identical probes with different bioorthogonal groups. Variations in probe labeling intensities could be due to differences in either (or both) of two distinct factors: labeling efficiency of PBPs by the probe of interest, or click chemistry reaction efficiency, which is independent of probe binding. These factors are indistinguishable in the above experiments, so to separate the individual contributions of PBP labeling and bioorthogonal reaction efficiencies we sought to investigate probe labeling independent of bioorthogonal reaction efficiency.

Bioorthogonal reactions in Spn lysates

To further explore fluorescent labeling intensities of PBPs independent of bioorthogonal group accessibility, we conducted bioorthogonal labeling reactions in denatured cell lysates rather than intact cells as previously described. We chose to test the CuAAC probes as the comparison of similar probes with different labeling intensities gave us a unique opportunity to tease apart underlying factors contributing to bioorthogonal labeling intensity. Spn cell suspensions were incubated with probes 7Ak, 7Az, and 7AzP and immediately lysed. The bioorthogonal reactions were then performed on the cell lysates under protein-denaturing conditions using a 0.2% SDS in 1 × PBS solution. The covalently bound β-lactone probes remain attached to denatured proteins, but all bioorthogonal groups should be fully exposed and equally accessible to the fluorescent reporter molecules and other click chemistry reagents. As hypothesized, 7Ak labeled PBP2x less strongly than 7Az and 7AzP in whole cell labeling experiments. In addition, 7AzP labeled PBP1b more strongly than 7Ak and 7Az (Figure 3), but in denaturing cell lysates the labeling intensity was the same within error across all three probes (Figures 4 and S4). The same experiment was then conducted with 8Ak and 8Az (Figure S5), revealing distinctly different labeling patterns between bioorthogonal labeling in whole cells and reactions conducted in cell lysates. PBP1b is much more strongly labeled in lysates than in whole cells, while PBP2a labeling is also observed, which was not the case in all previous experiments involving bioorthogonal reactions on whole cells (Figures 3 and S1F, G). We also attempted to determine apparent IC₅₀ values for each bioorthogonal probe against Spn PBPs, however the probes proved to be poor inhibitors, and reliable data could not be generated (Supplementary Protocol 1). The β-lactones are intended as selective probes of PBP activity rather than PBP inhibitors, so the lack of potency was of little concern. Interestingly, similarly to Figure S4, inhibition of PBP2a could be seen visually for 8Ak and 8Az in the ${\mathrm{IC}}_{50}^{\mathrm{app}}$ experiments (Figure S6F, G), suggesting that 8Ak and 8Az label PBP2a, but the folded, native protein environment of PBP2a prevents the installation of a fluorescent reporter by CuAAC. Overall, these results support the hypothesis that the accessibility of the probe bioorthogonal group may account for the difference in whole-cell PBP labeling between probes with similar structures, and again emphasizes the need to carefully consider multiple factors when designing bioorthogonal activity-based probes.

Figure 4.

Relative percent labeling compared to Boc–FL of 7Ak, 7Az, and 7AzP in Spn lysates. Whole cells were labeled with respective probes (10 μM for bioorthogonal probes, 5 μM for Boc–FL), followed by washing and lysis. Membrane fractions were collected, and CuAAC reactions were carried out under denaturing conditions using 0.2% SDS in 1 × PBS. The bar graph represents average relative Boc–FL labeling intensities for respective probes from biological duplicates (error bars represent standard deviations). The labeling intensity of all three probes is the same, within error, thus indicating that the accessibility of the bioorthogonal handle to respective fluorophore partners is potentially a confounding factor in whole-cell analysis. A representative fluorescent gel can be found in Figure S4.

Ligand comparison for bioorthogonal reactions in cell lysates

Unexpectedly, CuAAC reactions in whole cells differed compared to reactions in lysates depending on the Cu^I ligand chosen. In whole cells, we used BTTAA, a water-soluble ligand that was developed by Besanceney-Webler et al.^[57] BTTAA easily dissolves in aqueous solutions, making it more appealing for whole cell applications, as opposed to organic solutions that are required for dissolving the more commonly used ligand, TBTA. While the whole cell CuAAC reactions using BTTAA yielded acceptable PBP-labeling profiles, this was not the case when reactions with BTTAA were carried out in lysates. Significant labeling was not detected under denaturing or native conditions. When TBTA dissolved in 1:4 DMSO/tert-butanol was used as the ligand, however, strong fluorescent labeling of the desired PBPs was observed (Figure S7; see the Supporting Information for full experimental details). Although the specific reason for this discrepancy is unknown and will require further investigation, one explanation could be that in lysed samples the PBPs are surrounded by residual membrane patches. This could create a more hydrophobic protein environment, necessitating the use of a more hydrophobic ligand such as TBTA to deliver Cu^I to the reaction site. This is a unique discovery that has implications for experiments using not only the probes described herein, but other bioorthogonal probes that utilize CuAAC reactions. As such, all CuAAC reactions conducted on lysates were conducted with TBTA as the ligand.

Selective pull-down of PBP2b from Spn cell lysates by 8Ak

In addition to reducing the steric footprint of the molecules, the bioorthogonal probes are functionally diverse, as they can be clicked onto not only fluorophores containing the cognate bioorthogonal moiety, but also bioorthogonal group-containing affinity reagents (i.e., biotin) to conduct affinity-based pull-down experiments. The use of affinity reagents permits the proteins that interact with the PBPs to be “pulled-down” with the bioorthogonal probe-bound PBPs and analyzed by western blot or tandem mass spectrometry (MS/MS). Following an adapted protocol for activity-based protein profiling in bacterial cells,^[58] Spn cultures were treated with 8Ak, followed by cell lysis and “clicking” the labeled PBPs onto a tri-functional desthiobiotin-TAMRA-azide reagent (desthiobiotin-TMR–Az). Following the click reaction, the labeled proteins were immobilized on neutravidin-conjugated beads, the beads washed to remove nonspecifically bound proteins, then boiled in a mixture of SDS and biotin to elute the bound PBPs for SDS-PAGE separation and fluorescent gel analysis (Figure 5). PBP1b, PBP2x, and PBP2b were eluted in the presence of 8Ak, with PBP2b eluted preferentially. The SYPRO total protein scan also shows no nonspecific elution from the neutravidin beads (Figure S8, elution lanes), confirming the selective enrichment of labeled proteins. The fluorescent gel indicated minimal PBPs remained in the bead supernatant, suggesting that all labeled PBPs were immobilized onto the neutravidin beads. Further optimization will be required for the application of the bioorthogonal probes to proteomics pull-down experiments that will help elucidate the isoform-specific regulation of PBPs in Spn throughout cell division.

Figure 5.

Affinity purification of PBPs from Spn lysates using trifunctional biorthogonal reagent (for full structure, see Figure S9). “Elution” = proteins eluted from neutravidin beads. “Load” = proteins remaining in the supernatant after neutravidin immobilization. BODIPY-FL (green) and TAMRA (red) fluorescent composite gel image of eluted proteins indicates PBP1b (faint), PBP2x and PBP2b are labeled and eluted by 8Ak, with little labeling in the bead immobilization supernatant and none in the negative control elution.

Conclusion

We have synthesized and characterized a panel of selective, bioorthogonal, activity-based β-lactone probes for S. pneumoniae PBPs based on previously synthesized fluorescent probes. Replacing directly conjugated fluorophores with bioorthogonal groups has resulted in activity-based probes that retain the initial selectivity observed for the fluorescent probes, while allowing the use of different reporter molecules. Both copper-catalyzed (CuAAC) and copper-free (iEDDA) bioorthogonal modalities were exploited. CuAAC probes permit the utilization of a vast array of CuAAC-compatible reagents while iEDDA probes support live cell imaging applications where the toxicity of copper is an impediment, or low-abundance applications where a more efficient reaction than CuAAC is required. In addition to fluorescent imaging, we demonstrated the utility of bioorthogonal probes for activity-based pull-down experiments. 8Ak selectively pulled down PBPs from Spn cell lysates, which is an important proof-of-concept for the future application of these bioorthogonal probes to activity-based chemical proteomics.

Synthesis of an array of bioorthogonal probes also revealed differences in labeling intensity between bioorthogonal moieties, highlighting the importance of carefully designing bioorthogonal probes, observations which are applicable beyond the design of the probes reported herein. The choice of bioorthogonal reagents must be made based on the specific application required, as the observed levels of bioorthogonal fluorophore-dependent background labeling are acceptable for some applications, while further optimization of experimental conditions will be required for more sensitive applications like microscopy. In particular, alkyne-containing and TCO-containing fluorophores consistently gave more background labeling compared to azide- and tetrazine-containing fluorophores. iEDDA reactions are known to be faster and more efficient than CuAAC probes, but the primary determinant of labeling intensity across all probes appears to be accessibility of the bioorthogonal handle in the native protein environment during the click reaction, as all probes containing a multi-PEG-unit linker exhibited similar strong labeling compared to probes containing shorter hexanoic-acid-based linkers, regardless of the bioorthogonal moiety present. Even the difference in length of two atoms can have a marked effect, as demonstrated by the severely reduced labeling of PBP2x by 7Ak vs. 7Az, which was ameliorated upon denaturing of labeled PBPs prior to click chemistry labeling. Additionally, the difference in labeling intensities between PBPs with the same probe, as well as the emergence of PBP2a labeling by 8Ak and 8Az when bioorthogonal labeling was conducted in a denaturing environment, emphasizes the need for greater structural knowledge of PBP TP active sites and their interactions with β-lactone probes, experiments which are currently underway.

The synthesis and application of the suite of bioorthogonal probes described herein represents the first instance of selective bioorthogonal activity-based probes for penicillin-binding proteins. Given the importance of PBPs in cell survival throughout all bacterial species and the continued relevance of bacterial cell-wall synthesis as an antibiotic target, selective, activity-based bioorthogonal probes are widely applicable. Newly selective probes are currently in development for S. pneumoniae as well as other Gram-positive and Gram-negative organisms. The observation that bioorthogonal group accessibility to reporter molecules is the primary determinant of labeling intensity has implications beyond the selective labeling of PBPs with β-lactones, as design of bioorthogonal probes must take into account not only the bioorthogonal reaction used, but also whether reporter tag conjugation may be different in native protein environments or under protein-denaturing conditions. For CuAAC, this also impacts the specific bioorthogonal reaction conditions that should be used, as evidenced by the difference between TBTA and BTTAA ligands in denatured cell lysates. In sum, the ability to target active PBPs in an isoform-specific manner will provide insight into potential new bacterial vulnerabilities and represents an important step on the pathway to sustainability in the fight against antimicrobial resistance.

Experimental Section

General materials, further detailed protocols, and synthetic procedures and characterization available in the Supporting Information.

S. pneumoniae cell culture.

S. pneumoniae IU1945, an unencapsulated derivative of D39,^[52] or El77 (Δpbp1a)^[59] were obtained as a generous gift from the Malcolm Winkler lab at Indiana University, and cultured in Becton-Dickinson brain heart infusion (BHI) broth (cat# 237500) at 37 °C with 5% CO₂ and without shaking. Cultures were prepared by inoculating a tube containing 5 mL BHI with a sterile 200 μL pipette tip-full of frozen glycerol stock. Several serial dilutions (3–5) were prepared by inoculating 4.5 mL BHI with 0.5 mL of the previous dilution. Cultures were grown overnight (12–14 h) and OD₆₂₀ measurements were taken. An overnight culture that was still in exponential phase (OD₆₂₀ = 0.2–0.4) was chosen to prepare fresh cultures by inoculating 4 mL fresh BHI with 1 mL overnight culture (other serial dilutions were discarded). Fresh cultures were then grown to an OD₆₂₀ = 0.2–0.3 for various assays. The number of fresh cultures required depends on the number of samples needed for assays (generally 1 mL fresh culture/sample required unless otherwise stated).

General procedure for whole-cell probe treatment and bioorthogonal labeling.

Fresh cultures of S. pneumoniae were cultured as described above. Once cultures reached an OD₆₂₀ = 0.2–0.3, 1 mL culture per sample was harvested by centrifugation (16 000×g, 2 min, RT). Cell pellets were washed with 1 mL 1 × PBS (pH 7.4) and collected by centrifugation. Pellets were resuspended in 50 μL of PBS containing respective probe concentrations (0.5–2000 μM) and incubated at RT for 30 min. A negative control for each bioorthogonal probe was included that was resuspended in 50 μL PBS only and a Boc–FL control was included that was resuspended in 50 μL containing 5 μM Boc–FL. Following the 30 min incubation, cells were pelleted by centrifugation, the supernatant was removed, and pellets were washed with 1 mL PBS. Pellets treated with the fluorescent probes 7FL, 8T, and the Boc–FL control were resuspended and directly lysed for SDS-PAGE analysis (below). Pellets for the negative controls and probe-treated samples for CuAAC reactions (7Ak, 7Az, 7AzP, 8Ak, 8Az) were resuspended in 47 μL PBS and the following reagents were added in order: 1) 0.5 μL 5 mM fluorophore (final concentration = 50 μM) 2) 1 μL 125 mM Na ascorbate (final concentration = 2.5 mM) 3) 0.5 μL 20 mM BTTAA (final concentration = 200 μM) 4) 1 μL 5 mM CuSO₄ (final concentration = 100 μM). Alkyne-containing probes were reacted with azide-containing fluorophores, and vice versa. Probe-treated samples for Cu-free reactions (7MTz, 7TCO) were resuspended in 49 μL PBS, followed by the addition of 1 μL 500 μM respective fluorophore conjugates (final concentration = 10 μM). 7MTz was reacted with TCO-containing fluorophores, while 7TCO was reacted with tetrazine-containing fluorophores. All “click” samples were vortexed and incubated for 30 min in the dark at RT. Following incubation, cells were pelleted, the supernatant was removed, and cells were washed with 1 mL PBS and pelleted by centrifugation for lysis.

General procedure for cell lysis.

Pellets were resuspended in 100 μL PBS containing 10 mg/mL lysozyme and incubated at 37 °C for 20 min, followed by sonication using a Hielscher vial tweeter UP200St (90% C, 95% A, 5% adjustment snap, 7 cycles of 30 s on, 30 s off on ice in the dark). The membrane fraction was collected by centrifugation (21 000×g for 15 min, 4 °C) and the supernatant was removed.

Bioorthogonal labeling in lysates.

For click reactions conducted in cell lysates, the standard labeling procedure was carried out with slight modifications. Whole cells were labeled with respective probes, as described above. Following labeling and washing, cells were lysed using the standard lysis protocol. Following lysis, bioorthogonal reactions were conducted on the membrane pellet. For CuAAC reactions, membrane pellets were resuspended in 47 μL PBS + 0.2% SDS and the following reagents were added in the following order: 1) 0.5 μL 5 mM fluorophore (final concentration = 50 μM) 2) 1 μL 50 mM TCEP (final concentration = 1 mM) 3) 0.5 μL 10 mM TBTA dissolved in 1:4 DMSO/tert-butanol (final concentration = 100 μM) 4) 1 μL 50 mM CuSO₄ (final concentration = 1 mM). For Cu-free reactions, membrane pellets were resuspended in 49 μL PBPS, followed by the addition of 1 μL 500 μM respective fluorophore conjugate (final concentration = 10 μM). For the control Boc–FL sample, the membrane pellet was resuspended in 50 μL PBS. Samples were briefly vortexed and incubated in the dark at RT for 1 h. Following incubation, samples were centrifuged (21 000×g, 15 min, 4 °C) and the supernatant was removed. Membrane pellets were washed with 0.5 mL PBS and centrifuged again. The supernatant was removed, and pellets were resuspended in 30 μL PBS + 0.1% Triton-X-100. Protein concentration was adjusted to 5 mg/mL as described in the Supporting Information, 10 μL 4X SDS LB were added to 30 μL of protein solution, and samples were boiled for 5 min at 95 °C. Samples were briefly vortexed and 10 μL sample were loaded into a 10% SDS-PAGE gel and gels were ran and analyzed as described in the Supporting Information.

Affinity pull-downs of PBPs from S. pneumoniae cell lysates.

The affinity pull-down protocol was adapted from a previously published method by Sieber and co-workers.^[60] Briefly, a 100 mL culture of S. pneumoniae IU1945 was grown overnight as previously described. Following overnight growth, 10 mL of starter culture was diluted into 100 mL of BHI medium and allowed to grow until O.D.₆₀₀ = 0.2–0.3. Cells were then harvested by centrifugation of 50 mL of culture medium at 4000×g for 10 min at 4 °C. Cells were washed with 5 mL of 1 × PBS, centrifuged at 4000×g for 10 min at 4 °C, and the supernatant discarded. Next, cells were resuspended 500 μL of 1× PBS containing either 2 μL of 3 mM 8Ak or DMSO control and incubated at room temperature for 1 h with rotation. Probe-treated cells were then pelleted at 6000×g for 10 min at RT and the supernatant discarded. Cell pellets were washed 2x with 1 mL 1 × PBS, with centrifugation at 6000×g for 5 min at RT. Cells were then suspended in 500 μL of 1 mg/mL lysozyme and rotated for 30 min at RT. Cell lysis was then performed using Hielscher vial tweeter UP200St for 10 min (30 s on, 30 s off at 80% amplitude, 90% C, 5% adjustment snap) under ice cooling, followed by centrifugation at 21 000×g at 4 °C for 30 min. The supernatants were removed, and the lysis pellets were resuspended in 200 μL 1 × PBS with water bath sonication, followed by centrifugation at 21 000×g at 4 °C for 10 min. The supernatants were removed, and the pellets resuspended in 500 μL of 1 × PBS by water bath sonication and vortexing. Following resuspension, click reagents were added in the following order followed by brief vortex to mix: 5 μL of 5 mM trifunctional TAMRA-desthiobiotin-N₃ in DMSO, 10 μL of 50 mM TCEP in water, 30 μL of 1.7 mM TBTA in DMSO/tBuOH (1:4, v/v), and 10 μL of 50 mM CuSO₄·5H₂O in water. The reaction was then incubated in darkness for 1 h. Following incubation, reaction solutions were added to 4 mL of pre-chilled acetone and left at −20 °C for 1 h to precipitate proteins. Protein precipitate was collected via centrifugation at 4000×g for 20 min at 4 °C. The protein pellet was washed 2 × with 200 μL methanol, then left for 15 min to air dry. Dried pellets were dissolved in 1 mL 0.2% SDS by water bath sonication and vortexing, added to 50 μL of pre-equilibrated neutravidin-agarose beads (according to manufacturer’s instructions) and rotated for 1 h at RT in the dark. Bead suspensions were centrifuged for 2 min at 400×g at RT, then the supernatant was removed and 50 μL set aside to assess total loaded protein. The beads were then transferred to a fritted centrifuge column (Pierce cat# 89868) and washed 3x with 500 μL 0.2% SDS, 3× with 500 μL 6 M urea, and 3× with 500 μL PBS. Following final PBS washing, beads were transferred to a microcentrifuge tube and centrifuged at 400×g for 2 min at RT. The bead supernatant was removed, then 25 μL of 4X SDS laemmli buffer and 25 μL of 1.22 mg/mL biotin was added to the beads and 50 μL protein loading sample. Samples were then incubated for 15 min at 95 °C, and the bead-containing sample centrifuged at 21 000×g for 2 min at RT. The bead supernatant was then removed, and both bead sample and protein loading sample were analyzed by SDS-PAGE. For every gel, a Bocillin–FL positive control lane was run. This Bocillin–FL positive control sample was prepared according to above protocols for whole cell Bocillin–FL labeling for SDS-PAGE analysis. Proteins were resolved by SDS-PAGE on 15% polyacrylamide gels with 50 μL protein sample loading using 180 V for 1.5 h. Gels were subsequently scanned using fluorescent gel scanner with BODIPY–FL and TAMRA filters. Gels were then stained with SYPRO Orange protocol to visualized total proteins. To 100 mL of 200 mM Glycine, 2.5 mM Tris in 80%/20% water/MeOH, 20 μL of SYPRO Orange in DMSO was added. 30 mL of solutions was then added to a gel for 30 min. Gels were then washed briefly with water and imaged on a fluorescent gel scanner with TAMRA filter.