In Vitro Activity of Tebipenem (SPR859) against Penicillin-Binding Proteins of Gram-Negative and Gram-Positive Bacteria

Tebipenem (SPR859) is the microbiologically active form of SPR994 (tebipenem-pivoxil), an orally available carbapenem with activity against extended-spectrum β-lactamase (ESBL)-producing Enterobacteriaceae. Measurement of the relative binding of SPR859 to the bacterial cell targets revealed that it is a potent inhibitor of multiple penicillin-binding proteins (PBPs) but primarily a Gram-negative PBP 2 inhibitor, similar to other compounds in this class.

ABSTRACT

Tebipenem (SPR859) is the microbiologically active form of SPR994 (tebipenem-pivoxil), an orally available carbapenem with activity against extended-spectrum β-lactamase (ESBL)-producing Enterobacteriaceae. Measurement of the relative binding of SPR859 to the bacterial cell targets revealed that it is a potent inhibitor of multiple penicillin-binding proteins (PBPs) but primarily a Gram-negative PBP 2 inhibitor, similar to other compounds in this class. These data support further clinical development of SPR994.

TEXT

The β-lactam class of antibiotics has enjoyed a long track record of clinical efficacy and safety against many important pathogens, making these antibiotics a mainstay for treating patients suffering from many different types of bacterial infections (1). The carbapenem subclass possesses the broadest spectrum and stability for many clinically relevant β-lactamases (2). The broad-spectrum antibacterial activity of the carbapenems originates from low nanomolar potency against multiple penicillin-binding proteins (PBPs) (3). Tebipenem (SPR859) is the microbiologically active form of SPR994 (tebipenem-pivoxil), an orally available carbapenem with activity against extended-spectrum β-lactamase (ESBL)-producing Enterobacteriaceae (4). Tebipenem-pivoxil is currently marketed only in Japan by Meiji Seika Pharma Co. as a fine granule formulation (Orapenem) with indications for pediatric otitis media, sinusitis, and pneumonia. A new oral formulation of tebipenem-pivoxil (SPR994) is currently under development to optimize the pharmacokinetic-pharmacodynamic (PK/PD) profile for adult complicated urinary tract infection (cUTI) because of the recent increase in ESBL-producing Enterobacteriaceae, for which SPR859 is active. As expected of a carbapenem, tebipenem was found to be a potent inhibitor of multiple PBPs in pathogens like Streptococcus pneumoniae and Haemophilus influenzae, which translated to antimicrobial activity through cell wall synthesis inhibition (5, 6). In this work, we sought to characterize SPR859 activity against PBPs from pathogens typically found in cUTI, including Escherichia coli and Klebsiella pneumoniae.

Tebipenem (SPR859), provided by Spero Therapeutics, was characterized in vitro by assessing its potency against putative target PBPs. The PBP binding and competition assay with the fluorescent reporter molecule Bocillin FL (Invitrogen Molecular Probes, Eugene, OR) revealed the relative binding of test compounds for the different PBPs of each bacterial species investigated (7). Meropenem, avibactam, clavulanic acid, and ceftazidime were purchased from USP (Rockville, MD), and amdinocillin and oxacillin were purchased from Sigma-Aldrich (Oakville, ON, Canada). The strains used were Escherichia coli K-12, Klebsiella pneumoniae ATCC 13883, and Pseudomonas aeruginosa ATCC 27853 as the Gram-negative test strains. Staphylococcus aureus ATCC 29213 and 67-0 (8) were used as reference strains for methicillin-susceptible S. aureus (MSSA) and methicillin-resistant S. aureus (MRSA) strains, respectively.

Bacterial membranes for the PBP binding and competition assay with Bocillin FL were isolated and prepared according to previously published protocols (7, 9). Briefly, bacteria were cultured in a 1.5-liter brain heart infusion (BHI) broth (Becton Dickinson Canada, Inc., Mississauga, ON, Canada) at 35°C to an A₆₀₀ of ∼1.0. For the heterogeneous MRSA strain 67-0, the culture was supplemented with 2% NaCl and 5 μg/ml of nafcillin to induce a strong expression of PBP 2a as previously described (8, 10). The bacterial cell pellets were collected and suspended in a 20 mM potassium phosphate-140 mM NaCl (KPN) buffer (pH 7.5). E. coli, K. pneumoniae, and P. aeruginosa cells were treated with lysozyme (400 μg/ml) for 1 h at 37°C with gentle agitation, while lysostaphin (400 μg/ml) was used for S. aureus strains. DNase (10 μg/ml), RNase (10 μg/ml), and a protease inhibitors cocktail (Sigma-Aldrich) were added for another hour of incubation at 37°C. Cells were then disrupted with a French pressure cell, and membranes were collected by centrifugation at 12,000 × g for 10 min to remove the remaining unbroken cells. The supernatant was centrifuged at 150,000 × g for 40 min at 4°C. The membrane pellet was suspended in a minimal volume (typically 300 to 500 μl of KPN buffer) and stored at −80°C. The protein concentration in the membrane preparations was estimated by the Micro BCA protein assay kit (Thermo Fisher Scientific, Rockford, IL) using bovine serum albumin as a standard. The amount of membrane preparation used for each species and the final concentration of Bocillin FL used in the assay were selected so that the PBP banding patterns obtained on gels reflected those previously published in the literature. In each assay mixture, 25 to 40 μg of membrane proteins was used for E. coli, K. pneumoniae, and P. aeruginosa, and 5 to 10 μg was used for S. aureus ATCC 29213 and MRSA 67-0. Before the competition assay, membranes of the MRSA strain 67-0 were incubated with 100 μg/ml of clavulanic acid to saturate high-affinity PBPs and specifically reveal PBP 2a (8, 10). For S. aureus ATCC 29213, labeling of PBP 4 by Bocillin FL was possible by adding 10 μg/ml of avibactam for 10 min prior to the addition of the test antibiotic and the reporter molecule as previously described (9). For the competition, increasing concentrations of each test compound were first added for 10 min at 30°C before Bocillin FL was added at a final concentration of 12.5 μM for an additional 20 min of incubation. Thereafter, reactions were stopped by adding the Laemmli-loading buffer (Bio-Rad Laboratories, Richmond, CA) with freshly added β-mercaptoethanol (Sigma-Aldrich). Prior to loading on gel, the reaction mixtures were heated for 3 min at 100°C.

For PBP visualization, membrane proteins from the competition assay mixtures were separated by electrophoresis using an SDS-polyacrylamide discontinuous gel system as previously described (9). To obtain the fluorescent PBP images, gels were scanned by the Typhoon FLA 9500 instrument (GE Healthcare Bio-Sciences Inc., Baie D’Urfe, QC, CA) using a 488-nm excitation wavelength (emission, 520 to 530 nm) corresponding to the peak absorption of Bocillin FL. The volume (intensity × constant area) of each PBP band was then measured using Quantity One 1-D analysis software (version 4.6.6; Bio-Rad Laboratories, Richmond, CA). The 50% inhibitory concentration (IC₅₀) (or relative binding of test compound) for each PBP was defined as the antibiotic concentration (μg/ml) needed to reduce by 50% the binding of Bocillin FL. IC₅₀ data were obtained from at least three independent PBP binding assays using three different sets of antibiotic concentrations. Unlabeled membranes (no Bocillin FL and antibiotic) were also included in each gel electrophoresis to reveal nonspecific autofluorescent bands that are not PBPs.

The identification and designation of each PBP band revealed by the binding of Bocillin FL matched PBP profiles from the scientific literature and were confirmed notably by competition assays with control antibiotics having well-known PBP binding preferences. For example, amdinocillin has a very selective binding for E. coli PBP 2 (11), whereas ceftazidime binds primarily PBP 3 and PBP 1a and 1b (12). The effective concentrations of the test β-lactams that prevent binding of Bocillin FL to PBPs by 50% under assay conditions (IC₅₀) are reported in Table 1. Carbapenems like meropenem are known to display a broad-spectrum affinity for PBPs (3). As SPR859 is also a carbapenem, we used meropenem as a comparator for the E. coli PBPs binding patterns observed in competition with Bocillin FL (Fig. 1). As expected, SPR859 bound all E. coli PBPs, and IC₅₀ measurements were similar to those obtained for meropenem (Table 1). Meropenem PBP binding was in turn consistent with published data (13). PBP 2 was the main target of SPR859 for E. coli and K. pneumoniae (Table 1; see also Fig. S1 in the supplemental material for K. pneumoniae PBP profiles). As for amdinocillin and meropenem, the MIC of SPR859 correlated well with the IC₅₀ for E. coli PBP 2 (Table 1). The exception was K. pneumoniae showing a very low amdinocillin IC₅₀ for PBP 2 but an MIC of 512 μg/ml for that antibiotic (Table 1). MICs were measured by using a broth microdilution technique following the Clinical and Laboratory Standards Institute guidelines (14). The mechanism for amdinocillin resistance in K. pneumoniae is not well understood. Resistance is partly explained by the fact that K. pneumoniae ATCC 13883 produces an SHV-1 β-lactamase (15). In the presence of 2 μg/ml clavulanic acid, the MIC of ampicillin for K. pneumoniae dropped from 256 to 4 μg/ml and that of amdinocillin dropped from 512 to 32 μg/ml (data not shown). The MIC of clavulanate was 64 μg/ml. Thus, SHV-1 inhibition clearly affects K. pneumoniae susceptibility to amdinocillin, and the enzyme may not be the only factor contributing to resistance. In E. coli, amdinocillin resistance also involves the stringent response (16) and mutations that counteract the toxic effect of the futile cycle of peptidoglycan synthesis induced by amdinocillin (17). Nevertheless, microscopy confirmed inhibition of PBP 2 by amdinocillin, meropenem, and SPR859 in bacterial cells (Fig. 2). Indeed, exposure of E. coli cells to PBP 2 inhibitors like amdinocillin causes the rounding of cells (11), as this PBP is involved in cell elongation and rod shape maintenance (18). Besides, SPR859 and meropenem also had good binding to PBP 3. Binding to PBP 1a and 1b was equivalent for SPR859, meropenem, and ceftazidime, but ceftazidime had no significant binding to PBP 4 and PBP 5 and 6, as previously reported (19), compared to that of SPR859 and meropenem. Preferential binding to PBP 3 by ceftazidime provoked cell filamentation in both E. coli and P. aeruginosa (Fig. 2), as this PBP is involved in septation in E. coli (18) and is closely related in function in P. aeruginosa (20). While both E. coli (Fig. 2) and K. pneumoniae (data not shown) present rounding of cells when exposed to amdinocillin, meropenem, and SPR859, P. aeruginosa morphological changes included filament bulging and the presence of cells having a conical shape in the presence of the carbapenems (Fig. 2). This may represent the dual action of the carbapenems on both PBP 2 and PBP 3 in P. aeruginosa (19). Indeed, SPR859 showed high binding to both P. aeruginosa PBP 2 and PBP 3 (Table 1; see also Fig. S2 in the supplemental material for P. aeruginosa PBP profiles). Recently, the essentiality of P. aeruginosa PBP 3 was demonstrated (21), while the essential role of PBP 2 for this species is questioned since some mutants lacking this PBP are viable (22). Note that PBPs are designated according to their relative electrophoretic motility and that the same PBP designation across phylogenetically distant species does not necessarily imply that they are orthologs (18, 20).

TABLE 1.

Binding of antibiotics to PBPs of Gram-negative bacteria

Strain	Compounda	MIC(s) (μg/ml)	Relative binding (IC50 [μg/ml]) for PBP:b
			1a	1b	1a/1b	2	3	4	5	5/6
E. coli K-12	AMD	0.12–0.25			>20	0.040 ± 0.0090	>20	>20		>20
	CAZ	0.25			2.56 ± 1.90	>20	0.11 ± 0.018	>20		>20
	MEM	0.03			2.59 ± 1.66	0.015 ± 0.0039	0.45 ± 0.42	0.31 ± 0.0088		2.10 ± 1.21
	SPR859	0.015–0.03			3.94 ± 0.96	0.022 ± 0.0066	0.56 ± 0.28	0.65 ± 0.25		2.07 ± 1.02
K. pneumoniae ATCC 13883	AMD	512			>20	0.055 ± 0.017	>20	>20		>20
	CAZ	0.25			1.38 ± 0.24	>20	0.065 ± 0.026	>20		>20
	MEM	0.03–0.06			1.50 ± 0.77	0.010 ± 0.0035	0.16 ± 0.040	0.011 ± 0.0064		3.31 ± 1.12
	SPR859	0.06–0.12			1.34 ± 0.84	0.013 ± 0.0083	0.92 ± 0.31	0.019 ± 0.013		3.31 ± 0.95
P. aeruginosac ATCC 27853	AMD	256	>20	>20		0.34 ± 0.11	>20	>20	>20
	CAZ	2	0.043 ± 0.021	0.97 ± 0.12		>20	0.0096 ± 0.0043	2.61 ± 1.89	>20
	MEM	0.25–1	0.69 ± 0.29	0.15 ± 0.026		0.026 ± 0.018	0.0094 ± 0.0029	0.049 ± 0.031	4.77 ± 0.43
	SPR859	2	0.36 ± 0.18	1.94 ± 1.17		0.015 ± 0.012	0.031 ± 0.019	0.040 ± 0.022	3.84 ± 2.02

^aCAZ, ceftazidime; AMD, amdinocillin; MEM, meropenem.

^bA greater than symbol (>) preceding a value indicates that the IC₅₀ was greater than the highest dose tested.

^cThe IC₅₀ for PBP 2 of P. aeruginosa was measured by including the SDS-PAGE comigrating autofluorescent protein in the background.

FIG 1.

PBP binding competition assay between Bocillin FL and SPR859 (a) or meropenem (b) using E. coli K-12 cell membranes. The fluorescent Bocillin FL signal decreases with increasing antibiotic concentrations. The calculated IC₅₀ values are reported in Table 1. The controls (CTRL) are the unlabeled membranes (no Bocillin FL and antibiotic), showing nonspecific (ns) autofluorescent bands that are not PBPs.

FIG 2.

Microscopy of E. coli K-12 (a) and P. aeruginosa ATCC 27853 (b) incubated with 1 × MIC of the indicated antibiotics for 4 h. Paraformaldehyde 1% was used to fix the cells before phase-contrast microscopy. CAZ mainly targets PBP 3 (Table 1), leading to cell filamentation. AMD, MEM, and SPR859 target PBP 2 causing the rounding of the cells for E. coli K-12. AMD caused a similar morphological change in P. aeruginosa ATCC 27853, but conical shaped cells and filament bulging are observed with MEM and SPR859.

The tested antibiotics showed different relative binding affinities for the PBPs of Gram-positive S. aureus. Figure S3 in the supplemental material shows that the carbapenems meropenem and SPR859 bind to all PBPs. For SPR859, the IC₅₀ values for PBP 1 and PBP 4 were as low as 0.027 ± 0.018 μg/ml and 0.017 ± 0.001 μg/ml, respectively, and those values correlated with the antibiotic MIC for S. aureus ATCC 29213 (Table 2). Note that, as described before (8), better visualization of PBP 4 on S. aureus PBP banding patterns was enhanced using the β-lactamase inhibitor avibactam during the competition assay with Bocillin FL (Fig. S3). To a lesser degree, S. aureus PBP 2 and PBP 3 were also targets of SPR859. This was similar to that found for meropenem (Table 2). Yang et al. (23) showed the same relative binding of meropenem to PBP 2 and PBP 3, although it was reported that formation of the meropenem-PBP 3 acyl-enzyme may have a short half-life, thus, increasing its relative IC₅₀ value in PBP binding assays (24). Oxacillin on the contrary displays good binding for PBP 1, 2, and 3 but shows very little binding to PBP 4 (Table 2). The binding profile of oxacillin to S. aureus PBPs was consistent with other published results (10). Using light microscopy, the morphological changes observed for S. aureus in the presence of all tested antibiotics were equivalent (small enlargement of cells) (see Fig. S4 in the supplemental material), showing that the binding of carbapenems to PBP 4 (as opposed to that seen with oxacillin) did not contribute to this effect.

TABLE 2.

Binding of antibiotics to PBPs of S. aureus ATCC 29213 and MRSA 67-0

Strain	Compounda	MIC(s) (μg/ml)	Relative binding (IC50 [μg/ml]) for PBP:b
			1	2	2a	3	4
S. aureus ATCC 29213	OXA	0.25–0.5	0.058 ± 0.015	0.81 ± 0.25	NA	0.086 ± 0.050	>20
	MEM	0.12–0.5	0.047 ± 0.032	0.54 ± 0.24	NA	1.43 ± 0.45	0.0057 ± 0.0026
	SPR859	0.03–0.12	0.027 ± 0.018	0.53 ± 0.32	NA	1.01 ± 0.46	0.017 ± 0.00058
MRSA 67-0	OXA	256	ND	ND	440 ± 79	ND	ND
	MEM	32–64	ND	ND	167 ± 8	ND	ND
	SPR859	8–16	ND	ND	45 ± 15	ND	ND

^aOXA, oxacillin; MEM, meropenem; NA, not available; ND, not determined.

^bA greater than symbol (>) preceding a value indicates that the IC₅₀ was greater than the highest dose tested.

In S. aureus, the presence of the low-affinity PBP 2a provides resistance to β-lactam antibiotics and is present in clinical isolates of MRSA but not in those of MSSA. PBP 2a transpeptidase activity supports cell growth and viability even in the presence of β-lactam concentrations that inhibit PBPs 1 to 4 (25), and the MICs of these antibiotics, thus, correlate with their relative binding to PBP 2a (Table 2; see also Fig. S5 in the supplemental material for MRSA PBP 2a profiles). Against MRSA, SPR859 was superior to meropenem based on its MIC and IC₅₀ for PBP 2a (Table 2 and Fig. S5), but both induced cell enlargement like oxacillin when used at their MICs (see Fig. S4). Still, SPR859 cannot be compared to anti-MRSA β-lactams like ceftobiprole (12) or ceftaroline (10) in terms of relative efficacy.

In conclusion, SPR859 was found to be a potent inhibitor of multiple PBPs among clinically important bacteria but was primarily a PBP 2 inhibitor in Enterobacteriaceae, similar to other compounds in the carbapenem class. These data support further clinical development of SPR994 to become the first oral carbapenem for treatment of serious Gram-negative infections.