Expanded profiling of β-lactam selectivity for penicillin-binding proteins in Streptococcus pneumoniae D39

Abstract

Penicillin-binding proteins (PBPs) are integral to bacterial cell division as they mediate the final steps of cell wall maturation. Selective fluorescent probes are useful for understanding the role of individual PBPs, including their localization and activity during growth and division of bacteria. For the development of new selective probes for PBP imaging, several β-lactam antibiotics were screened, as they are known to covalently bind PBP in vivo. The PBP inhibition profiles of 16 commercially available β-lactam antibiotics were evaluated in an unencapsulated derivative of the D39 strain of Streptococcus pneumoniae, IU1945. These β-lactams have not previously been characterized for their PBP inhibition profiles in S. pneumoniae and these data augment those obtained from a library of 20 compounds that we previously reported. We investigated seven penicillins, three carbapenems, and six cephalosporins. Most of these β-lactams were found to be co-selective for PBP2x and PBP3, as was noted in our previous studies. Six out of 16 antibiotics were selective for PBP3 and one molecule was co-selective for PBP1a and PBP3. Overall, this work expands the chemical space available for development of future β-lactam-based probes for specific pneumococcal PBP labeling and these methods can be used for the development of probes for PBP labelling in other bacterial species.

Introduction

Antimicrobial resistance is one of the major challenges to global health due to the emergence of multidrug resistant bacteria (World Health Organization 2020). Bacterial strains have been found to be resistant to all classes of clinically-utilized antibiotics, including those that target cell wall synthesis (Nikaido 2009; Nikolaidis et al. 2014). Cell wall synthesis is an essential process for bacterial growth and division, making it an attractive target for antibiotics. The bacterial cell wall is a complex structure primarily composed of a three-dimensional heteropolymeric peptidoglycan (PG) backbone, which is responsible for maintaining its structure and rigidity (Pasquina-Lemonche et al. 2020). PG is composed of repeating units of N-acetylglucosamine (GlcNAc) and N-acetyl muramic acid (MurNAc) that are joined by β−1–4 glycosidic linkages. These polysaccharide chains are crosslinked through a transpeptidation reaction of small branched peptide units called stem peptides whose composition varies among different bacterial species (Lovering et al. 2012; Vollmer et al. 2008; Vollmer and Seligman 2010). It is the proteins that perform this transpeptidation reaction, the penicillin-binding proteins (PBPs), that serve as popular targets for antibiotic development as they are susceptible to the β-lactams, which form a covalent complex with the active-site serine residue (Macheboeuf et al. 2006; Sauvage et al. 2008; Tipper 1985; Waxman and Strominger 1983). This covalent interaction can be exploited to develop fluorescent probes for the PBPs that will enhance our understanding of their activity and localization.

Bacterial cells possess multiple PBPs that are classified into three classes based on their molecular weight, function, and various conserved domains. Class A PBPs are high molecular weight bifunctional enzymes, which act as both a transglycosylase to polymerize GlcNAc and MurNAc and a transpeptidase to crosslink adjacent stem peptides. Class B PBPs are high molecular weight monofunctional enzymes with transpeptidase activity, while Class C are low-molecular weight PBPs with carboxypeptidase activity (Macheboeuf et al. 2006; Zapun et al. 2008). Carboxypeptidases are responsible for cleaving the pentapeptide chain to a tetrapeptide to prevent crosslinking and aid in the regulation of PG levels (Macheboeuf et al. 2006). PBP transpeptidase domains are highly structurally homologous, and different bacterial species can contain as few as four or as many as 16 PBPs that perform similar functions (Sauvage et al. 2008). Despite the structural and enzymatic redundancy of the many PBPs, their specific roles and regulation throughout cell growth and division are known to differ (Popham and Young 2003). However, it has been difficult to fully characterize the catalytic activity of each PBP isoform as the development of selective tools for an individual PBP remains challenging. To date, tools that enable the selective monitoring of the activity of an individual PBP are only available for a small subset of PBPs under specific genetic backgrounds.

We sought to identify scaffolds that could be used for selective PBP labeling in the Gram-positive bacteria Streptococcus pneumoniae (Spn), a well-characterized model organism and known opportunistic pathogen which causes pneumonia, meningitis, otitis media, septicemia- and bacteremia-like infections (Brooks and Mias 2018; Donkor 2013; Henriques-Normark and Tuomanen 2013; Sham et al. 2012). Spn possesses six PBPs, out of which three are class A PBPs (PBP1a, PBP1b, PBP2a), two are class B (PBP2x and PBP2b), and one is class C PBP (PBP3) (Land et al. 2013; Massidda et al. 2013; Vollmer et al. 2019). We have previously reported the identification of both β-lactam- and β-lactone-containing compounds that are selective or co-selective for several PBPs, for example four β-lactams are selective for PBP3, one for PBP2x, and most of the other structures are co-selective for PBP2x and PBP3 in Spn; similarly β-lactones are selective for PBP2x in a Δpbp1b mutant strain (this deletion has no observable phenotype) (Kocaoglu et al. 2015; Sharifzadeh et al. 2017). To expand the understanding of lactam selectivity for the PBPs and identify more candidates for probe development, we investigated 16 additional β-lactam antibiotics in an unencapsulated derivative of the D39 strain of S. pneumoniae, IU1945 (Lanie et al. 2007) using a previously established fluorescent gel-based method (Gee et al. 2001; Kocaoglu and Carlson 2013; Zhao et al. 1999). In this assay, a fluorescent penicillin V analogue, Bocillin-FL (Boc-FL), is used as a readout probe for PBP activity due to its ability to label all six isoforms in Spn (Kocaoglu et al. 2015). This fluorescence-based assay has been previously used to visualize the inhibition of various compounds against the PBPs (Kocaoglu and Carlson 2013, 2015; Kocaoglu et al. 2015; Sharifzadeh et al. 2017; Sharifzadeh et al. 2018, 2020). Additionally, such penicillin-based probes have been useful to monitor transpeptidation activity of PBPs in live cells (Sharifzadeh et al. 2018; Zhao et al. 1999).

In the present study, we evaluated the PBP selectivity of 16 commercially available β-lactam antibiotics that have not been tested in Spn. Among these molecules, we found six β-lactams to be selective for PBP3, one for PBP2x, eight to be co-selective for PBP2x and PBP3, and a single compound that targets PBP3 and PBP1a. These results are comparable to the earlier study where eight of the β-lactams were found to be co-selective for PBP2x and PBP3, four were selective for PBP3, and one was selective for PBP2x (Kocaoglu et al. 2015). This work expands the library of molecules for the design of next-generation probes for selective activity labeling of pneumococcal PBPs. Further development of chemical probes for PBPs will enable the visualization of these proteins in live cells and ultimately, expand our understanding of bacterial growth and division.

Results and discussion

Experimental rationale

β-lactam antibiotics are known suicide inhibitors, as they acylate the active site serine residue of the transpeptidase domain of the PBPs, resulting in a stable covalent acyl-enzyme complex with a very slow rate of hydrolysis (Waxman and Strominger 1983). The β-lactam ring found in this class of antibiotics functions as a substrate mimic because it is structurally similar to the terminal portion of the stem peptide substrate, D-Ala-D-Ala (D-alanyl-D-alanine), and thereby covalently labels the PBPs in an activity-based manner (Blumberg et al. 1974; Tipper and Strominger 1965). As such, β-lactam labeling reports on PBP activity. To augment our previous studies to identify selective PBP inhibitors in Spn (Kocaoglu et al. 2015), we evaluated the PBP inhibition profiles of 16 additional β-lactam antibiotics from multiple subclasses (Supplementary Figure S1). For this purpose, live Spn cells are treated with a β-lactam antibiotic at various concentrations, followed by Boc-FL treatment to label the unbound PBPs in a gel-based assay. The β-lactam antibiotic, once bound to a PBP, will not allow binding of Boc-FL due to formation of an acylated complex and hence a fluorescent band will not be visible (or will have decreased signal compared to control) on the SDS-PAGE gel (Figure 1).

Figure 1:

Schematic diagram of the profiling of β-lactam selectivity for penicillin-binding proteins.

S. pneumoniae (Spn) cells possess six PBPs, which are targets of various β-lactam antibiotics. The Spn cells are incubated with varying concentrations of a β-lactam antibiotic followed by incubation with Bocillin-FL, which will bind to the uninhibited PBPs and SDS-PAGE is performed on the cell lysates. The disappearance of bands on the gel corresponds to binding of the antibiotic.

We evaluated molecules from multiple β-lactam subclasses, including seven penicillins (dicloxacillin, cloxacillin, ticarcillin, azlocillin, carbenicillin, flucloxacillin and nafcillin), six cephalosporins (cefixime, cefpirome, cefepime, cefotetan, cefpodoxime and cefaclor), and three carbapenems (biapenem, imipenem and ertapenem). Inhibition profiles were assessed from the gel band Boc-FL labeling intensities, which are weaker when inhibition with a given antibiotic is stronger, and the IC₅₀ values were calculated. A compound was considered selective for a PBP if its IC₅₀ value was at least 4-fold lower than that of the next-most inhibited PBP. However, the compound was deemed co-selective if the IC₅₀ value of the next lowest PBP was less than 4-fold of that of the most inhibited PBP. This 4-fold method was used as this range is much larger than the standard error of the mean for most of the IC₅₀ values and is similar to the specifications used in previous reports (Kocaoglu and Carlson 2015; Kocaoglu et al. 2015; Sharifzadeh et al. 2020).

Selectivity profiles of carbapenems

The carbapenems tested for PBP selectivity included biapenem, imipenem, and ertapenem. The representative images are shown in Figure 2 and IC₅₀ values are given in Table 1. The three carbapenems had some differences in their binding profiles, but overall were highly specific for PBP3 (Supplementary Figure S2–S4). Biapenem and imipenem were selective for PBP3 with an IC₅₀ value of 0.006 and 0.007 μM, respectively (Figure 2, Supplementary Figures S2, S3 and Table 1). The inhibition profile of imipenem also showed that it binds to other PBPs, but at higher concentrations than it does for PBP3, indicating dose-dependent selectivity. Ertapenem, on the other hand, binds PBP3, as well as PBP1a and/or PBP1b (Figure 2, Supplementary Figure S4 and Table 1). PBP1a and PBP1b migrate closely on the gel making accurate gel band quantification extremely challenging. When it was clear that one or both of these PBPs may be a main target of a given antibiotic, such as the case for ertapenem, we also performed the assay with knockout mutants of PBP1a or PBP1b as further confirmation (Δpbp1a or Δpbp1b; Table 1). The Δpbp1a strain data showed that ertapenem was selective for PBP3 and not PBP1b as the IC₅₀ value for PBP1b was more than 4-fold higher as compared to that of PBP3 (Supplementary Figure S5 and Table 1). However, using the Δpbp1b mutant with ertapenem confirmed co-selectivity towards PBP3 and PBP1a (Supplementary Figure S6 and Table 1). The variation in the selectivity of ertapenem as compared to the other two carbapenems in this work may be due to the presence of a substantially larger group off of the five membered ring conferring its binding to PBP1a (Supplementary Figure S1). Indeed, previous assessment of doripenem and meropenem, which possess similar large groups at this position, also showed co-selectivity towards PBP3 along with PBP1a, but these molecules were additionally co-selective for PBP2x.

Figure 2:

Representative gel images and graphs for relative Boc-FL labelling of carbapenems to PBPs in a dose-dependent manner.

Structure, gel-based profile and relative percentage Bocillin-FL labelling of ((A)–(C)) Biapenem and ((D)–(F)) Ertapenem. ((A) and (D)) Structure of β-lactam antibiotics; ((B) and (E)) fluorescent gel images; and ((C) and (F)) relative percentage labelling of all the PBPs as compared to control (without antibiotic) and average ± SD from two biological replicates have been plotted.

Table 1:

MIC, IC₅₀ values and PBP selectivity of β-lactam antibiotics in S. pneumoniae IU1945, Δpbp1a and Δpbp1b strains.

IC50(μM)
β-lactam class and drug	MIC (μM)	PBPla	PBP1b	PBP2x	PBP2a	PBP2b	PBP3	PBP selectivity
Carbapenems
Biapenem	0.01	0.060 ± 0.021	6.824 ±4.147	N.D.	0.146 ± 0.446	0.111 ± 0.106	0.006 ± 0.001	3
Imipenem	0.10	0.027 ± 0.005	0.052 ±0.022	0.048 ± 0.013	0.062 ± 0.021	0.052 ± 0.011	0.007 ± 0.001	3
Ertapenem	0.01	0.118 ±0.064	0.137 ±0.375	0.752 ± 0.817	26.77 ± 4.077	13.79 ± 3.495	0.061 ± 0.043	3, la
Ertapenem (Δpbp1a)	–	–	0.326 ±0.082	0.341 ± 0.251	N.D.	3.861 ± 7.401	0.012 ± 0.006
Ertapenem(Δpbp1b)	–	0.038 ± 0.012*	–	0.729 ± 0.122	6.694 ± 0.401	>1000	0.031 ± 0.011
Penicillins
Cloxacillin	0.01	0.561 ± 0.219	0.977 ±0.372	0.119 ± 0.062*	4.468 ± 2.236	2.362 ± 1.765	0.102 ± 0.092	2x, 3
Dicloxacillin	0.01	0.385 ± 0.261	1.143 ± 1.371	0.042 ± 0.055	4.842 ± 4.737	8.698 ± 34.31	0.065 ± 0.051*	2x, 3
Flucloxacillin	1.00	12.43 ± 4.201	17.67 ±4.161	0.832 ± 0.323	35.29 ± 25.98	18.54 ± 11.07	0.911 ± 0.252*	2x, 3
Flucloxacillin(Δpbp1a)		–	50.46 ±2.031	0.889 ± 0.389*	16.04 ± 2.012	16.83 ± 9.301	0.783 ± 0.581
Flucloxacillin(Δpbp1b)		11.04 ± 2.475	–	2.008 ± 0.878*	51.52 ± 5.161	33.47 ± 15.96	1.954 ±0.884
Azlocillin	0.01	1.525 ± 0.532	0.963 ±0.477	0.081 ± 0.018*	1.401 ± 0.693	0.294 ± 0.131	0.065 ± 0.013	2x, 3
Ticarcillin	0.10	0.827 ± 0.365	1.008 ±0.331	0.251 ± 0.109*	7.743 ± 2.604	3.338 ± 1.485	0.146 ± 0.087	2x, 3
Carbenicillin	0.10	1.112 ± 0.367	0.859 ±0.285	0.962 ± 0.506	5.637 ± 4.466	2.125 ± 1.154	0.018 ±0.011	3
Nafcillin	0.01	94.47 ± 88.61	32.41 ±3.781	0.014 ± 0.006	4.537 ± 1.752	1.275 ± 0.306	0.125 ± 0.061	2x
Cephalosporins
Cefaclor	10.0	3.601 ± 1.333	8.588 ±3.634	3.673 ± 2.522	N.D.	55.36 ± 26.33	0.034 ± 0.009	3
Cefotetan	1.00	14.61 ± 6.695	0.538 ±0.201	14.37 ± 5.769	54.34 ± 21.47	N.D.	0.069 ±0.015	3
Cefixime	0.10	32.34 ± 34.45	8.962 ±4.598	0.336 ± 0.224	4.847 ± 2.056	>1000	0.026 ± 0.007	3
Cefepime	0.01	0.995 ± 0.099	1.061 ± 0.413	0.009 ± 0.002	0.119 ± 0.048	>1000	0.012 ± 0.006*	2x, 3
Cefepime(Δpbp1a)		–	1.167 ±0.401	0.013 ± 0.002	0.245 ± 0.121	125.9 ± 16.12	0.013 ± 0.002	2x, 3
Cefepime(Δpbp1b)		0.839 ± 0.119	–	0.014 ± 0.001	0.221 ± 0.123	>1000	0.015 ± 0.001*
Cefpodoxime	0.01	1.917 ±0.581	0.907 ±0.406	0.041 ± 0.015*	0.341 ± 0.125	0.522 ± 0.231	0.032 ± 0.005
Cefpodoxime(Δpbp1a)		–	0.301 ±0.071	0.134 ± 0.082*	0.621 ± 0.161	>1000	0.032 ± 0.004
Cefpodoxime(Δpbp1b)		2.777 ± 0.886	–	0.035 ± 0.018*	0.179 ± 0.081	>1000	0.018 ± 0.004
Cefpirome	0.01	0.051 ± 0.011	0.024 ±0.008	0.002 ± 0.001*	0.016 ± 0.004	5.527 ± 2.991	0.0015 ± 0.097	2x, 3

IC₅₀ values have been calculated from the gel band intensities using GraphPad Prism as described in the material and methods section using [inhibitor] versus response (four parameters) with nonlinear regression fitting. Data has been obtained from two experiments and the average with standard error of the mean has been reported in the Table. Values in bold are the lowest IC₅₀ values for each compound. Asterisk indicates the PBPs that had IC₅₀ values that are within 4-fold of the lowest reported IC₅₀ value and hence show co-selectivity. Italics formatting indicates values that were re-tested using Δpbp1a and Δpbp1b mutant strains to confirm the selectivity profiles. An IC₅₀ of >1000 was assigned when GraphPad Prism reported an ambiguous number and significant inhibition was not seen at 1000 μM. ND, not determined (GraphPad Prism reported an ambiguous number and/or an extremely wide confidence interval for IC₅₀ determination [>10⁶] for both data sets).

Selectivity profiles of penicillins

Another class of β-lactam antibiotics is the penicillins and to understand the specificity of these molecules to the PBPs in Spn, we assessed seven molecules that were not examined in our previous study. These include azlocillin, carbenicillin, cloxacillin, dicloxacillin, flucloxacillin, nafcillin and ticarcillin (Supplementary Figure S1). Azlocillin, cloxacillin, dicloxacillin, flucloxacillin and ticarcillin were found to be co-selective for PBP2x and PBP3, while carbenicillin was designated as selective for PBP3 and nafcillin was selective for PBP2x. Cloxacillin and dicloxacillin exhibited very similar PBP binding, with specificity for PBP3 and PBP2x, which is not surprising as they only differ by the addition of a second chlorine atom at the meta position on the benzyl group (Figure 3A–C, Table 1 and Supplementary Figures S1, S7, S8). Both molecules have lower labelling efficiency for PBP2a, 2b and PBP1a/1b. Compared to dicloxacillin, flucloxacillin differs by replacement of a chlorine with a fluorine atom (Supplementary Figure S1). This replacement does not lead to a significant change in the binding profile of flucloxacillin, which was co-selective for PBP2x and PBP3 with IC₅₀ values of 0.83 μM for PBP2x and 0.91 μM for PBP3 (Table 1 and Supplementary Figure S9). Selectivity of flucloxacillin towards PBP1a and PBP1b was also assessed using the Δpbp1a and Δpbp1a mutant strains to determine if the PBP1b band was partially masked due to the higher intensity of PBP1a on the gel. We found that flucloxacillin still exhibited co-selectivity for only PBP2x and PBP3 as was seen in the parent strain (Supplementary Figures S10, S11 and Table 1).

Figure 3:

Representative gel images and graphs for relative Boc-FL labelling of dicloxacillin and cefaclor to PBPs in a dose-dependent manner.

Structure, gel-based profile and relative percentage Bocillin-FL labelling of ((A)–(C)) Dicloxacillin and ((D)–(F)) Cefaclor. ((A) and (D)) Structure of β-lactam antibiotics; ((B) and (E)): fluorescent gel images; and ((C) and (F)) Relative percentage labelling of all the PBPs as compared to control (without antibiotic) and average ± SD from two biological replicates have been plotted.

Azlocillin and ticarcillin have very different groups off of the “left side” of the β-lactam ring (as drawn in Supplementary Figure S1), but both exhibit high labelling efficiency towards PBP2x and PBP3 (Supplementary Figures S12 and S13). Carbenicillin appeared to be selective for PBP3 on the basis of gel band quantitation and this was supported by the IC₅₀ value of 0.018 μM for PBP3 while the rest of the PBPs showed higher IC₅₀ values (Supplementary Figure S14 and Table 1). Nafcillin, on the other hand, exhibited selectivity for PBP2x (Supplementary Figure S15) unlike all of the other tested penicillins in this study, as well as those which were evaluated earlier, perhaps due to the bulk of the napthyl substituent (Kocaoglu et al. 2015).

Selectivity profiles of cephalosporins

The PBP selectivity of cephalosporins was examined for six molecules including cefaclor, cefotetan, cefixime, cefpodoxime, cefepime and cefpirome (Supplementary Figure S1). Cefaclor was found to be selective for PBP3 with an IC₅₀ value of 0.034 μM (Figure 3D–F and Supplementary Figure S16), and similarly, cefotetan and cefixime were found to be selective for PBP3 with IC₅₀ values of 0.069 and 0.026 μM, respectively (Table 1 and Supplementary Figures S17, S18). The other three cephalosporins were co-selective for PBP3 along with other PBPs. Amongst these three, cefepime was selective for PBP3 but also bound PBP2x making it co-selective for both proteins with IC₅₀ values of 0.012 μM and 0.009, respectively (Supplementary Figure S19). Likewise, cefpodoxime and cefpirome also exhibited co-selectivity towards PBP3 and PBP2x, which was further confirmed by assessment in the Δpbp1a and Δpbp1a mutant strains (Supplementary Figures S20–25 and Table 1).

All of the cephalosporins inhibited PBP3 strongly (lowest IC₅₀ value of the PBPs), and none of them potently inhibited other PBPs except PBP2x, consistent with the analogs tested in earlier work (Kocaoglu et al. 2015). In fact, PBP3 selectivity was seen for most of the β-lactam antibiotics examined in the present study, with six out of the 16 antibiotics labeling PBP3 specifically. The PBP3 active site possesses conserved penicilloyl serine transferase superfamily motifs similar to all other PBPs in Spn (Morlot et al. 2005), but is not as selective as these proteins due perhaps to yet unknown interaction(s) or flexibility within the active site.

While most of the antibiotics showed strong labeling of PBP3, PBP2x was the next most potently inhibited protein. Out of 16 antibiotics tested, eight antibiotics among the penicillin and cephalosporin subclasses showed co-selectivity towards PBP2x along with PBP3. It is interesting to note that the only β-lactam that was selective for PBP2x alone was nafcillin. PBP2b was not selectively inhibited by any of the antibiotics used in the present study, again consistent with the previously reported work (Kocaoglu et al. 2015). While most of the molecules lacked potency for this essential PBP, several exhibited greater inhibitory activity, biapenem, imipenem and azlocillin, as well as other carbapenems, penicillins and a penem in the previous report.

Correlation of growth inhibition and PBP binding

Given that the β-lactam antibiotics are the major class of antibiotics used clinically to treat pneumococcal infections (Henriques-Normark and Tuomanen 2013; Sham et al. 2012), we sought to further investigate the roles of individual PBPs in the lethality of the organism upon treatment with various β-lactams. PBP2b and PBP2x are essential proteins for Spn growth and mutations in these proteins can result in resistance development. To understand the correlation of our PBP inhibition profiles with cell death, we determined the minimum inhibitory concentration (MIC) for each of the antibiotics used in the present study and compared them to the corresponding IC₅₀ values found with each antibiotic for every individual PBP. Scatter plots were obtained using GraphPad Prism with the MIC of the antibiotics and IC₅₀ values calculated from the gel images. A linear fitting was performed to assess the correlation between IC₅₀ and MIC values (Figure 4). It was found that most of the MIC values for the β-lactam antibiotics were near-linearly correlated with the PBP2x IC₅₀ values. This signifies that inhibition of PBP2x is highly correlated with lethality due to β-lactam binding and is consistent with results from our previous work, reinforcing the essential role of PBP2x inhibition in the functionality of these antibiotics. As before, inhibition of other PBPs does not show a linear relationship with the MIC values of the β-lactam antibiotics suggesting a weak correlation between PBP binding and cell lethality.

Figure 4:

Correlation of MIC of β-lactam antibiotics with IC₅₀ values of PBP inhibitors.

Scatter plot of IC₅₀ values for individual PBP versus MICs. All the data points have been plotted in log scale on both axes and a linear regression performed in GraphPad Prism.

Our titration data indicates that PBP3 is inhibited by most of the β-lactams tested in this study, but the strength of PBP3 inhibition is not correlated with the MIC values of the antibiotics. This suggests that even though β-lactam antibiotics more effectively label PBP3, this binding does not necessarily translate to inhibition of growth or loss of cell viability. Given that PBP3 is non-essential for the growth of Spn, this is not a surprising result. It can be noted, however, that while some antibiotics bind most strongly to PBP3, they also bind to other PBPs at a higher concentration, and inhibition of these PBPs may result in cell death.

Data summary and comparison

In sum, we conclude that most of the penicillins including dicloxacillin, cloxacillin, flucloxacillin, ticarcillin, and azlocillin displayed co-selectivity towards PBP3 and PBP2x while carbenicillin was only selective for PBP3 and nafcillin was selective for PBP2x. On the other hand, cephalosporins like cefaclor, cefotetan, cefixime were selective for PBP3, while other cephalosporins exhibited inhibition of PBP3 along with PBP2x. Carbapenems were found to be primarily selective towards PBP3 and unlike any of the other molecules, ertapenem inhibited PBP1a along with PBP3. The PBP selectivity results for the 16 profiled β-lactams has been summarized in Figure 5 and shows that six β-lactams including three cephalosporins, one penicillin, and two carbapenems are selective for PBP3, one β-lactam is selective for PBP2x, and the rest of the β-lactams are co-selective. In addition, we have compiled the data from this study and our prior work to provide an overall story of β-lactam-PBP selectivity in S. pneumoniae with an array of molecules (Supplementary Figure S26) (Kocaoglu et al. 2015). We found that most of the β-lactams are co-selective for PBP2x and PBP3, which includes 11 penicillins (six from a previous report and five from this study) and five cephalosporins (two from a previous report and three in this study). We also determined that PBP3 is the most commonly inhibited protein, with 10 selective molecules identified including six cephalosporins (three from a previous report and three from current study), one monobactam (previous report), one penicillin, and two carbapenems (this study). Other β-lactam compounds such as cefuroxime (previous report) and nafcillin (this study) were selective for PBP2x. However, given the structural diversity of these molecules both in terms of core scaffold (e.g., cephalosporin versus penicillin) and side chains, it is difficult to pinpoint the structural components required for inhibition of a specific PBP. One of the carbapenems, ertapenem showed co-selectivity towards PBP3 and PBP1a which was not found with any other molecule class in this study, but a related profile, co-selectivity for PBP3, PBP1a, and PBP1b was previously seen for cephalothin.

Figure 5:

Overall summary of PBP binding specificity based on the IC₅₀ values obtained from the gel-based assay and calculated using GraphPad Prism.

All 16 antibiotics tested have been placed according to their PBP inhibitions, here x-axis and y-axis are different PBPs and the antibiotics that are placed on the diagonal represent specific binding, e.g., cefixime was found to be selective for PBP3 but ertapenem was found to be co-selective for PBP3 and PBP1a. Placements have been made as per the selectivity profiles from Table 1.

Conclusions

This study builds on a previous report describing the PBP selectivity profiles of 20 β-lactam antibiotics in S. pneumoniae (Kocaoglu et al. 2015) and includes an additional 16 β-lactams. Out of the 16 compounds tested, seven selectively inhibited a single PBP as determined by relative IC₅₀ values, while the remainder were co-selective for multiple PBPs. Selectivity (and thus low IC₅₀ values) was most commonly achieved for PBP3, while the least targeted protein was PBP2b. These data further indicate that it is difficult to correlate the structure of β-lactams to their PBP selectivity, and additional investigations will be necessary to determine potential structure-activity relationships. The IC₅₀ values generated from this study can nonetheless be an important component for designing activity-based probes to study the localization and temporal activity of specific PBPs in live cells (Sharifzadeh et al. 2017, 2018, 2020). Such probes can be designed on the basis of desired selectivity profiles as determined herein, and can be synthesized by installing a reporter moiety onto the chosen β-lactam scaffold through incorporation of a fluorescent or bioorthogonal tag (Brown Jr et al. 2021; Lang and Chin 2014; Sharifzadeh et al. 2017, 2020). Given the variety of PBP labeling by the β-lactams included in this study, a number of different probes are possible based on the IC₅₀ data and selectivity profiles. In addition, the molecules utilized in this work can be used to selectively inhibit the activity of a given PBP to further evaluate the role of that protein in cell growth, division, and morphology.

Materials and methods

Bacterial strains and growth conditions

S. pneumoniae strain IU1945, an unencapsulated derivative of D39 (Δcps) (Lanie et al. 2007), E177 (Δpbp1a), and E193 (Δpbp1b) (gifts from Malcolm E. Winkler, Department of Biology, Indiana University–Bloomington, USA), was grown in Becton-Dickinson brain heart infusion (BHI) media at 37 °C with 5% CO₂ atmosphere and kept under static conditions (without shaking). The growth was monitored by measuring the optical density of the cultures at 620 nm (OD₆₂₀) (Land et al. 2013). For all assays, cultures were prepared from frozen glycerol stocks by inoculating 5 ml of BHI in a culture tube and performing several 10-fold serial dilutions that were made by inoculating 4.5 mL of BHI media with 0.5 ml of the previous tube. All cultures were grown for 12–14 h and the growth monitored by OD₆₂₀. Following incubation, the cultures in exponential phase (within the range of 0.2–0.4 OD₆₂₀) were used to prepare secondary cultures by 10-fold dilution (i.e., 0.5 ml primary culture in 4.5 ml of fresh BHI media). These cultures were grown until they reached early exponential phase of 0.11–0.15 OD₆₂₀, at which point the cells were harvested by centrifugation for titration assays (see below).

Antibiotics stocks

Cloxacillin, dicloxacillin, ticarcillin, flucloxacillin, azlocillin, nafcillin, carbenicillin, cefixime, cefepime, cefpodoxime, cefotetan, cefpirome, cefaclor, biapenem, imipenem, ertapenem were purchased from MilliporeSigma. All stocks were freshly prepared from powdered solids immediately before use. Stocks for cloxacillin, dicloxacillin, ticarcillin, flucloxacillin, azlocillin, nafcillin, carbenicillin, cefotetan, biapenem, imipenem, ertapenem were prepared in ultrapure water to obtain a final concentration of 5 mM. Similarly, 5 mM stocks were made for cefixime, cefepime, cefpodoxime, cefpirome, cefaclor in DMSO. All the stock solutions were diluted 5-fold in 1X PBS (phosphate-buffered saline, pH 7.4) to reach a concentration of 1 mM and then serially diluted 10-fold in 1 × PBS for titration assays (see below). Boc-FL was purchased from Life Technologies (Grand Island, NY).

β-lactam titration for detection of PBP binding

Secondary cultures of S. pneumoniae IU1945 or mutant strains growing in early exponential phase with 0.11–0.15 OD₆₂₀ (see above) were used for the β-lactam titrations. For each sample, 1.5 ml of culture was harvested by centrifugation at 16,000 × g for 2 min at room temperature. Cell pellets were prepared for each β-lactam titration experiment. Pellets were washed by resuspension in 1 ml of 1 × PBS, pH 7.4 and harvested again by centrifugation. The cell pellets were then resuspended in 50 μl of 1 × PBS containing different concentrations of a β-lactam antibiotic, which were prepared by serial dilutions as mentioned above with a range of 1000–0.0001 μM. A control sample without antibiotic (1 × PBS only) was also prepared for each experiment. These samples were incubated for 20 min at room temperature followed by centrifugation at 16,000 × g for 2 min. The cell pellets were washed again with 1 ml of 1 × PBS and then resuspended in 50 μl of 1 × PBS containing Boc-FL with the final concentration of 15 μM. Following Boc-FL treatment, samples were kept away from light whenever possible. All the samples were incubated for 10 min at room temperature in the dark followed by centrifugation at 16,000 × g for 2 min. Cells were washed again with 1 ml of 1 × PBS and pelleted by centrifugation. The cells pellets were resuspended in 100 μl of 1 × PBS containing 10 mg/ml lysozyme (Chicken egg white, Research Products International) and incubated at 37 °C for 20 min in the dark. The samples were lysed by sonication using a Hielscher vial tweeter UP200St (90% C, 95% A, 10 cycles of 30 s on, 30 s off on ice in the dark) for a total of 10 min. The lysed samples were centrifuged at 21,000 × g for 15 min at 4 °C to pellet the membrane fraction. The supernatant was discarded and the pelleted membrane fractions were resuspended in 100 μl of 1 × PBS. The concentration of protein was measured using the NanoPhotometer P330 (IMPLEN). The concentration of all the samples was adjusted to 2 μg/μl using 1 × PBS. From these samples, 30 μl was aliquoted and added to 10 μl of 4X SDS Laemmli buffer. These samples were boiled for 5 min at 95 °C, briefly vortexed, and 13 μl loaded onto a 10% SDS-PAGE gel (acrylamide:bis-acrylamide = 29:1). Proteins were separated by running the gel in 1 × Trisglycine SDS running buffer for 1.5 h at 180 V and 40 mA on ice and in the dark. The gel was imaged for fluorescence using a GE Typhoon 9500 gel scanner with the BODIPY-FL setting of 473 nm laser excitation with LPB filter (≥510 nm) and imaged at 50 μm resolution. The gel was stained with Brilliant Blue Coomassie stain to visualize the protein bands for 30 min and then destained in buffer containing 40% methanol and 10% glacial acetic acid for 12 h. This gel was imaged for total protein using a GE Typhoon 9500 gel scanner with the Coomassie setting of 635 nm laser excitation with LPR filter (≥665 nm) at a 50 μm resolution. All the results have been analyzed from biological duplicate samples.

Image processing

All of the gel images were analyzed using ImageJ software (National Institutes of Health, Bethesda, MD, USA). First, the background signal was subtracted followed by adjusting the brightness and contrast to optimize the signal-to-noise ratio. Both of these processes were performed on the entire gel image in a uniform manner. Each band on the gel was analyzed and integrated density values were measured for gel band quantitation in ImageJ. A background integrated density for each lane was subtracted from the integrated density of each Boc-FL labelled PBP in a single lane, and the values obtained were normalized by protein band intensities in the same lane from the Coomassie-stained gel image. Relative percentage of Boc-FL labelling was calculated for each PBP band in antibiotic-treated samples in comparison to the PBP band intensity of the untreated sample (1 × PBS only treatment) from the same gel (usually the first band of the gel). IC₅₀ values were calculated using GraphPad Prism (GraphPad Software, La Jolla, CA, USA) with [Inhibitor] versus response (four parameters) and the values were plotted on the log scale. The gels were analyzed from biological duplicate experiments and the IC₅₀ values were calculated and standard error of mean was obtained for the duplicate samples using GraphPad Prism.

Antimicrobial susceptibility assay for S. pneumoniae IU1945

Antimicrobial susceptibility assays were performed as described previously (Wiegand et al. 2008). Briefly, the primary cultures at an OD₆₂₀ of 0.1 were diluted 10-fold with BHI in a culture tube, 500 μl of the culture was added to 4.5 ml of BHI media. Fresh stocks of β-lactam antibiotics were made and filter sterilized using 0.2 μm pore size sterile syringe filters. For water soluble antibiotics, PES membrane filters were used and for DMSO soluble antibiotics, nylon membrane filters were used. The stocks were serially diluted 2-fold in BHI. In 96-well clear plates, 50 μl of diluted culture were added and 50 μl of serially diluted antibiotics were added in each well. The plate was kept for incubation at 37 °C in an atmosphere of 5% CO₂ for 18–20 h. Duplicates of each antibiotic concentration were performed on a different plate. After incubation, the plates were examined by eye and absorbance at 620 nm was measured using a plate reader. The lowest concentration that inhibited cell growth was noted as the MIC.