The Next-Generation β-Lactamase Inhibitor Taniborbactam Restores the Morphological Effects of Cefepime in KPC-Producing Escherichia coli

ABSTRACT Gram-negative bacteria producing carbapenemases are resistant to a variety of β-lactam antibiotics and pose a significant health risk. Given the dearth of new antibiotics, combinations of new broad-spectrum β-lactamase inhibitors (BLIs) with approved β-lactams have provided treatment options for resistant bacterial infections. Taniborbactam (formerly VNRX-5133) is an investigational BLI that is effective against both serine- and metallo-β-lactamases, including the serine carbapenemase KPC. In the current study, we assessed the effectiveness of taniborbactam to restore antibacterial activity of cefepime against KPC-3-producing Escherichia coli by inhibiting the KPC-3-dependent hydrolysis of cefepime. Time-lapse microscopy revealed that cells treated with greater than 1× MIC of cefepime (128 μg/ml) and cefepime-taniborbactam (4 μg/ml cefepime and 4 μg/ml taniborbactam) exhibited significant elongation, whereas cells treated with taniborbactam alone did not owing to a lack of standalone antibacterial activity of the BLI. The elongated cells also had frequent cellular voids thought to be formed by attempted cell divisions and pinching of the cytoplasmic membrane. Additionally, the effect of taniborbactam continued even after its removal from the growth medium. Pretreatment with 4 μg/ml taniborbactam helped to restore the antibacterial action of cefepime by neutralizing the effect of the KPC-3 β-lactamase. IMPORTANCE β-lactam (BL) antibiotics are the most prescribed antimicrobial class. The efficacy of β-lactams is threatened by the production of β-lactamase enzymes, the predominant resistance mechanism impacting these agents in Gram-negative bacterial pathogens. This study visualizes the effects of a combination treatment of taniborbactam, a broad spectrum β-lactamase inhibitor (BLI), and the BL antibiotic cefepime on a carbapenemase-producing E. coli strain. While this treatment has been described in the context of other cephalosporin-resistant bacteria, this is the first description of a microscopic evaluation of a KPC-3-producing strain of E. coli challenged by this BL-BLI combination. Live-cell microscopy analysis of cells treated with taniborbactam and cefepime demonstrated the antimicrobial effects on cellular morphology and highlighted the long-lasting inhibition of β-lactamases by taniborbactam even after it was removed from the medium. This research speaks to the importance of taniborbactam in fighting BL-mediated antibiotic resistance.

widespread production of b-lactamases continues to compromise this first line of defense. Thus, development of b-lactamase inhibitors (BLIs) to combine with b-lactams is an essential strategy to combat MDR bacteria (2). The amoxicillin-clavulanate combination has been used clinically for decades (3). Recently, more BL-BLI combinations, such as ceftazidime-avibactam and meropenem-vaborbactam, have been approved to treat infections by MDR Gram-negative bacteria producing a wide variety of serine-b-lactamases (2)(3)(4). However, vaborbactam lacks activity against OXA carbapenemases, and neither BLI is active against metallo-b-lactamases (5).
This study aimed to evaluate the impacts and consequences of the addition of taniborbactam on cefepime-elicited morphological effects using KPC-3-producing Escherichia coli. We describe the bactericidal activity of cefepime-taniborbactam and morphological effects upon treatment. We also examined the lifetime of b-lactamase inhibition originating from pretreatment with taniborbactam using time-lapse microscopy. Our results suggest that lysis caused by cefepime while protected by taniborbactam is indistinguishable from that observed with cefepime alone and that pretreatment of cells with taniborbactam results in a prolonged cefepime protective effect to maintain antibacterial activity.

RESULTS AND DISCUSSION
Susceptibility of KPC-3-producing E. coli to cefepime and taniborbactam. To examine potentiation of cefepime antibacterial activity by taniborbactam, we used E. coli CDC-0001 producing the KPC-3 carbapenemase. Broth microdilution assays were performed to determine MICs of antibiotics against E. coli CDC-0001 (Table 1). The modal MIC for cefepime against CDC-0001 from 3 biological replicates was 128 mg/ml, consistent with Antibiotic Resistance (AR) Isolate Bank reporting (.32 mg/ml). Taniborbactam did not inhibit growth of CDC-0001 even at 128 mg/ml when tested alone, but it did potentiate the antibacterial activity of cefepime to an MIC of 4 mg/ml when tested at a fixed BLI concentration of 4 mg/ml. Attenuated growth of E. coli CDC-0001 was also observed at sub-MICs of taniborbactam-protected cefepime (Fig. S1 in the supplemental material). Taniborbactam restores the bactericidal activity of cefepime. Time-kill assays were performed with E. coli CDC-0001 ( Fig. 1) to examine whether the addition of taniborbactam affects the rate of killing by cefepime. The results of these assays showed a concentration-dependent killing for both treatments at 37°C ( Fig. 1A and B) and at room temperature ( Fig. 1C and D). Cell killing by cefepime occurred quickly at 37°C, with CFU falling below the limit of detection (10 2 CFU/ml) after 4 h of treatment with 16Â MIC of cefepime with or without 4 mg/ml taniborbactam. Given that all subsequent experiments were performed at room temperature, the following discussion highlights the differences seen between Fig. 1C and D. When comparing the rate of killing at 16Â MIC by cefepime alone (2,048 mg/ml) versus cefepime plus taniborbactam (64 mg/ml cefepime plus 4 mg/ml taniborbactam), cefepime alone resulted in a greater extent of cell killing at 4 h. There are at least two related explanations for this observation. First, initial periplasmic cefepime concentrations may be higher in the cefepime alone condition simply due to the different dosing concentration (2,048 mg/ ml versus 64 mg/ml). Second, initial periplasmic cefepime concentrations may be higher in the cefepime alone conditions due to parameters of kinetic turnover of cefepime by KPC-3. While the K m and k cat of KPC-3 are not known, KPC-2 differs from KPC-3 by only one amino acid substitution (H272Y) and has a K m of .1,000 mM and a k cat of .6 s 21 (14,15). These values suggest that in the first 2 h, cefepime may be slowly sequestered by binding to KPC-3 without hydrolysis. If the initial cefepime concentration is higher, it is more likely that cefepime would saturate the binding sites of KPC-3 proteins and that excess cefepime would irreversibly inhibit both PBP3 and other PBP targets, which would induce rapid lysis.
Bacterial morphology with antibiotic treatments. PBP3 is integral to bacterial cell division, and, therefore, inhibition results in a septation defect with continuous elongation inducing cell filamentation (16)(17)(18). To monitor the effect of these PBP inhibitors on PBP3, we measured resulting changes in cell length by time-lapse microscopy. Because the trend of the time-kill curves at room temperature was similar to that at 37°C (Fig. 1), we observed growth and filamentation of single cells at room temperature. The average length of cells (30 # N # 100) was determined from images taken every hour for 4 h at 0Â and 1Â MIC for cefepime in the presence of 4 mg/ml taniborbactam (Fig. 2). To monitor the phenotype of E. coli CDC-0001 cells treated with cefepime alone, cells were imaged following treatment with 4 mg/ml and 128 mg/ml cefepime (concentrations equivalent to 1Â MIC cefepime plus taniborbactam and 1Â MIC of cefepime alone, respectively). After exposure to cefepime-taniborbactam (4 mg/ml each) and cefepime alone (128 mg/ml), cells exhibited significant elongation at 1 h posttreatment and by 4 h reached a maximum length of 32 6 5 mm and 29 6 7 mm, respectively. Cells treated with 4 mg/ml cefepime alone or 4 mg/ml taniborbactam alone did not elongate and reached maximum average lengths of 4.3 6 2.6 mm and 3.7 6 1.6 mm, respectively. Overall, these results provide visual confirmation that taniborbactam restores antibacterial activity of cefepime at 4 mg/ml against KPC-3-producing E. coli CDC-0001.
Time-lapse videos of individual cells treated with antibiotics were generated to evaluate morphological changes elicited from antibiotic exposure. E. coli CDC-0001 cells were treated with 1Â, 2Â, and 4Â MIC of cefepime combined with 4 mg/ml taniborbactam (Movie S1). As controls, time-lapse microscopy experiments were also performed with cells treated with 4, 8, or 16 mg/ml cefepime alone or 4 mg/ml taniborbactam alone (Movies S2 and S3). Treatment of cells with 1Â, 2Â, and 4Â MIC cefepime combined with 4 mg/ml taniborbactam resulted in elongated cells, subsequent cessation of growth, and bulge formation at the midpoint after 2 to 3 h of treatment (Fig. 3B). These cells exhibited midcell bulges leading to lysis from disruption of the cell envelope. Notably, treatment with 1Â MIC cefepime combined with 4 mg/ml taniborbactam induced a phenotype exhibiting an apparent cellular void (Fig. 3B), which consistently appeared and disappeared within 12 6 10 min. Comparable images of cells treated with aztreonam were obtained as a control (Fig. 3A). Aztreonam, a PBP3-specific b-lactam antibiotic, was used as a control in order to establish a baseline morphological effect of an antibiotic acting on PBP3. The morphological changes observed for both treatments are consistent with a disruption in cell division caused by inhibition of PBP3. For cells treated with cefepime-taniborbactam, voids occurred in 62% 6 9% of cells and could be the result of inner membrane pinching at initiated division sites, where the constrictive force is not enough to contract the cell wall and outer membrane layers. Similar cellular voids have been described in relation to a nonfunctional cell division protein variant, FtsK (19).
Cells treated with various concentrations of cefepime alone exhibited differing extents of elongation and cell morphologies. The treatment with 4 mg/ml cefepime (1/32Â MIC) initially resulted in slightly elongated cells, which then returned to a regular pattern of cellular division (Movie S2). Cells treated with 8 mg/ml and 16 mg/ml (1/16Â MIC and 1/8Â MIC) cefepime elongated significantly without cell lysis or any cessation of growth (Movie S2). These treatments also resulted in the same apparent cellular voids as seen with the combined cefepime plus taniborbactam treatment (16 mg/ml, 4Â MIC). As expected, an increasing propensity for elongation was observed at higher concentrations of cefepime. As bacteria produce a finite number of b-lactamases, their effect can be overwhelmed at elevated b-lactam concentrations. As anticipated from its lack of standalone antibacterial activity, taniborbactamtreated cells grew and divided normally even at 16 mg/ml (Movie S3).
Treatment with either 16 mg/ml cefepime combined with 4 mg/ml taniborbactam (4Â MIC) or 128 mg/ml cefepime alone (1Â MIC) commonly resulted in large midcell bulges, indicating a weakness in the cell wall at the initiated division site (Fig. 3). These same cells, when viewed in time-lapse images by phase microscopy, lysed within minutes after the midcell bulges formed (Fig. 3B). Scanning electron microscopy (SEM) analysis of these samples revealed that after 4 h of treatment, cells showed similar midcell bulging with 4Â MIC cefepime combined with 4 mg/ml taniborbactam (Fig. 4A) and with 1Â MIC cefepime alone (128 mg/ml) (Fig. 4B) compared to what was observed in untreated cells (Fig. 4C). Because cefepime has high affinity to PBP3 but also binds PBP2 and PBP1s (9), cefepime at high concentrations inhibits multiple PBPs, leading to cell lysis, while cefepime induces elongation by inhibiting PBP3 predominantly at low concentrations. This further supports initial findings from time-kill assays where cell death occurred more rapidly in the presence of cefepime at higher multiples of MIC, allowing it to inhibit multiple PBPs simultaneously.
Taniborbactam elicits a lasting effect of cefepime. Taniborbactam has been previously reported to inhibit KPC-2 b-lactamase with a half-life of 105 min (6). Here, experiments were designed to wash away excess taniborbactam after a pretreatment of the cells. This allowed us to specifically monitor the time course of morphological recovery to rod shape by release of taniborbactam from the KPC-3 active site and subsequent restoration of cefepime hydrolysis by KPC-3. We first treated E. coli CDC-0001 cells with 1Â MIC cefepime (4 mg/ml) combined with 4 mg/ml taniborbactam for 4 h, washed the cells, and then captured time-lapse images of cells every 3 min for 12 h. Cells that were observed at time zero of imaging elongated further, and cellular division resumed after ;3.5 h (Movie S4). Almost all first divisions (89%) were at the poles of the cells (Table 2), in good agreement with previous reports (20). Subsequent divisions occurred predominantly at the poles of cells (67%) but with an increased proportion of cells exhibiting midcell divisions. This return to midcell division is likely a result of shorter cell length and proper placement of the cell division machinery. A similar phenomenon was also observed for cells treated with 128 mg/ml cefepime (i.e., 1Â MIC) alone (Table 2).
In the second recovery assay, cells were pretreated with 4 mg/ml taniborbactam for 4 h, after which the taniborbactam was washed away and cefepime was added at 4 mg/ml (Fig. 5A, bottom; Movie S4). As a control, cells were not pretreated with taniborbactam prior to treatment with 4 mg/ml cefepime (Fig. 5A, top). Thus, normal rod cells were observed at time zero of imaging in the experiments. While cells without pretreatment with taniborbactam began to divide after 21 min, cells pretreated with taniborbactam exhibited a significant (P , 0.0005) delay in division and began to divide after ;4 h (Fig. 5B). This delayed return to normal cell division likely reflects a combination of new synthesis of KPC-3 and PBPs and slow release of taniborbactam from the KPC-3-taniborbactam complex. The results from both of these experiments are consistent with a long residency time of taniborbactam within the active site of KPC-3 (6) in E. coli cells and provides prolonged protection from KPC-3 for cefepime to exhibit its antibacterial activity.

CONCLUSION
In this study, we investigated the morphological effects of cefepime-taniborbactam on KPC-3-producing E. coli CDC-0001 and the post-BLI effects on cefepime after pretreatment with taniborbactam. Taniborbactam was successful in inhibiting the KPC-3 b-lactamase and shifting the MIC of cefepime from 128 mg/ml alone to 4 mg/ml in the presence of taniborbactam. Treating E. coli CDC-0001 with the cefepime-taniborbactam combination resulted in cell death, morphological defects, and physical lysis, consistent with the controls of aztreonam and cefepime at a much higher concentration. When taniborbactam was used as a pretreatment, it exhibited a lasting protective effect, enabling the antibacterial activity of the cefepime in E. coli. This result confirms the long residency time of taniborbactam in KPC-3, consistent with the previously studied KPC-2 (6).

MATERIALS AND METHODS
Strains, compounds, and bacterial growth. E. coli CDC-0001 (BioSample accession number SAMN04014842) was used in this study. Unless otherwise stated, bacterial cultures were grown at 37°C in cation-adjusted Mueller-Hinton broth (CAMHB) (Fisher Scientific) in the presence or absence (negative controls) of taniborbactam (Venatorx Pharmaceuticals, Inc.) or cefepime hydrochloride (USP).
MIC assays. MIC values (Fig. S1 in the supplemental material) were determined using the method described by Wiegand and colleagues and produced MIC results equivalent to the CLSI M07 edition 11 (2018) broth microdilution method (21,22). The bacterial strain used also produced results consistent  with the quality control ranges stated in the CLSI M100 edition 30 document (23). Briefly, the conversion factor to obtain a CFU/ml based on an optical density at 600 nm (OD 600 ) was determined experimentally (an OD 600 of 1 was equal to 7.2 Â 10 8 CFU/ml). An overnight culture was diluted to a final inoculum of 5 Â 10 5 CFU/ml in each test well along with the appropriate serially diluted antibiotic to be tested. Each sample was run in technical quadruplicate and biological triplicate, and MICs were determined using the mode of the concentration at which there was a 90% reduction in cell turbidity compared to the growth control, as determined using a spectrophotometer. These experiments were performed in flat-bottom polystyrene 96-well plates.
Time-kill assays. The procedure for determining the time it takes for an antibiotic to kill bacterial cells is described in detail in the CLSI-approved guideline document M26-A (22). Briefly, cells were inoculated in liquid medium to 5 Â 10 5 CFU/ml with various concentrations of antibiotic, based on previously determined MICs (0.25Â, 1Â, 4Â, and 16Â MIC) along with an untreated control (0Â MIC). These were then incubated for set amounts of time (0, 2, 4, 6, and 24 h) and assessed for viability by diluting serially 10-fold and spot plating each dilution of the cultures on antibiotic-free agar plates. These experiments were conducted in technical and biological triplicates.
Time-lapse microscopy. Time-lapse images were taken by using NIS-Elements and an Eclipse Ti2 microscope (Nikon) equipped with a Â40 long working distance objective and an Orca Flash 4.0 LT camera (Hamamatsu). An overnight culture of E. coli CDC-0001 was diluted 100-fold in fresh CAMHB and grown at 30°C to an OD 600 of 0.4. Cells were then diluted 2-fold in medium containing antibiotics. These treated cells were then placed in an 8-chamber slide (Thermo Scientific Nunc Lab-Tek II chambered coverglass) and covered with a nutrient agarose pad (1%) containing the same desired concentration of antibiotic (0Â, 1Â, 2Â, or 4Â MIC). The nutrient agarose pad was topped with a small piece of coverglass and sealed at the top with Breathe-Easy polyurethane moisture barrier (Diversified Biotech). The 8-chamber slides were then imaged every 3 min for each time-lapse experiment. Images were then compiled into videos at 7 frames per second in ImageJ (24).
Cell length measurements. Snapshots taken from time-lapse experiments were analyzed to track cell lengths under different treatment conditions over time. For the time-lapse experiments, cells were monitored at room temperature during treatment. All cells in a field of view were measured using ImageJ software to a maximum N of 100 every hour for 4 h. Data were compiled in Prism, and two-way analysis of variance (ANOVA) was performed comparing each test data point to the control data point at the corresponding time point.
Scanning electron microscopy. Samples of cells for scanning electron microscopy (SEM) were grown up to an OD 600 of 0.4 from the diluted overnight culture. Cells were then diluted 2-fold into medium containing antibiotics and incubated for 4 h. Samples were then adhered to carbon planchets and fixed with glutaraldehyde (2%) and osmium tetroxide (1%) followed by an ethanol dehydration series. Samples were then fully dried using a Denton DCP-1 critical point dryer and were sputter-coated in gold-palladium using a Denton Desk V TSC. Samples were then imaged using a Quanta FEG 250 SEM operated at 10 kV.
Recovery assays. Recovery assays were set up similar to the time-lapse experiments with a few modifications. After dilution and the addition of antibiotics, cells at log phase were incubated in a 1.5-ml microcentrifuge tube for 4 h in the absence or presence of antibiotic. After this treatment, cells were pelleted and resuspended in fresh CAMHB and imaged in the same 8-chamber slide setup as for the timelapse experiments. Samples were imaged every 3 min for 12 h.

ACKNOWLEDGMENTS
We thank Greg Moeck for initiation of the collaborative work and Steve Boyd for editorial support.