Bicarbonate modulates delafloxacin activity against MDR Staphylococcus aureus and Pseudomonas aeruginosa

Abstract

Objectives

To investigate the utility of recently approved delafloxacin and other fluoroquinolones against leading MDR bacterial pathogens under physiologically relevant conditions.

Methods

MIC and MBC assays were conducted for MDR strains of Staphylococcus aureus, Pseudomonas aeruginosa, Acinetobacter baumannii and Klebsiella pneumoniae in the standard antibiotic susceptibility testing medium CAMHB, amended Roswell-Park Memorial Institute tissue culture medium (RPMI+) or 20% fresh human whole blood. In vivo correlation of in vitro findings was performed in a murine P. aeruginosa pneumonia model. Mechanistic bases for the findings were explored by altering media conditions and with established fluoroquinolone accumulation assays.

Results

Fluoroquinolone MICs were increased in RPMI+ compared with CAMHB for all four MDR pathogens. Specifically, delafloxacin MICs were increased 32-fold versus MDR S. aureus and 8-fold versus MDR P. aeruginosa. MBC assays in 20% human whole blood and a murine MDR P. aeruginosa pneumonia model both confirmed that delafloxacin activity was reduced under physiological conditions. Bicarbonate (HCO₃⁻), a key component of host physiology found in RPMI+ but absent from CAMHB, dictated delafloxacin susceptibility in CAMHB and RPMI+ by impairing its intracellular accumulation.

Conclusions

Standard in vitro antibiotic susceptibility testing conditions overpredicted the effectiveness of delafloxacin against MDR pathogens by failing to capture the role of the biological buffer HCO₃⁻ to impair delafloxacin accumulation. This work showcases limitations of our current antibiotic susceptibility testing paradigm and highlights the importance of understanding host microenvironmental conditions that impact true clinical efficacy.

Introduction

Antibiotic-resistant bacterial infections exert a significant disease burden worldwide including over 2 million hospitalizations annually in the USA.¹ First coined in 2008,² the ‘ESKAPE’ pathogens are a group of six highly concerning MDR bacteria³ linked to frequent clinical treatment failure and high mortality rates.⁴ The expanding burden of such MDR infections on our healthcare system necessitates a closer mechanistic appreciation of antibiotic action to optimize clinical outcomes.³

The current standardized in vitro antimicrobial susceptibility testing (AST) paradigm used by clinicians to inform therapeutic decisions is increasingly recognized to have key limitations.^5–7 AST utilizes artificially nutrient-rich bacteriological media originally designed to ensure optimal bacterial recovery and growth, but in and of itself is a poor proxy for the physiological environments found within the infected patient. Interactions between antibiotics and the host microenvironment can impact treatment outcomes; for example, endogenous peptides can synergize with antibiotics,^6,8,9 antibiotics can alter immune response and memory,^10–14 and host-derived factors can sensitize MDR bacterial pathogens to antimicrobial killing.^7,15–19

Introduced in the 1960s, multiple members of the fluoroquinolone drug class possess versatile clinical utility including norfloxacin (1980),²⁰ ciprofloxacin (1983),²¹ levofloxacin (1996)²² and moxifloxacin (2000).²³ The mechanism of fluoroquinolone action involves direct inhibition of bacterial DNA gyrase and/or topoisomerase IV to block bacterial DNA replication. Because they are one of the most commonly prescribed drug classes for treating human bacterial infections,²⁴ significant bacterial resistance to fluoroquinolones has arisen through target site mutations in the corresponding enzymes (DNA gyrase and/or topoisomerase IV) or via alterations in drug transport,²⁵ notably among ESKAPE members. However, interest in fluoroquinolone use for ESKAPE pathogens was rekindled following the introduction of delafloxacin in 2017 due to its purported broad-spectrum activity and increased potency against Staphylococcus aureus.²⁶ Bacterial resistance to delafloxacin is more difficult to elicit and less commonly reported than for other fluoroquinolone drugs.^27–29 In addition, delafloxacin exhibits increased potency (lower MIC) in standardized AST and has fewer reported side effects in patients compared with other members of this drug class.^29,30

In this study, we assessed how microenvironmental conditions affected the activity of various fluoroquinolones, including the recently introduced delafloxacin, against MDR strains of ESKAPE pathogens S. aureus, Klebsiella pneumoniae, Acinetobacter baumannii and Pseudomonas aeruginosa. Specifically, we sought to uncover how host-mimicking conditions influence fluoroquinolone susceptibility using a combination of bacteriological, physiological and ex vivo human-derived testing media coupled with in vivo murine infection models.

Materials and methods

Bacterial strains

MDR S. aureus strain LAC of the USA300 lineage was originally isolated from a patient in Los Angeles County.³¹ Virulent MDR A. baumannii strain AB5075, isolated from a patient with tibial osteomyelitis,³² was acquired from the Walter Reed Medical Center. MDR strains of K. pneumoniae (K1100) and P. aeruginosa (P4) are recent clinical isolates from a tertiary academic hospital in New York.^6,33

Antibiotics

Levofloxacin (Sigma–Aldrich), delafloxacin (Melinta), moxifloxacin (Matrix Scientific) and norfloxacin (TCI) were resuspended in 1× Dulbecco’s PBS (DPBS) (Corning). Ciprofloxacin (ACROS Organics) was resuspended in sterile deionized H₂O. Concentrated stocks of levofloxacin were prepared at 22 500 mg/L, ciprofloxacin at 25 000 mg/L, delafloxacin at 28 500 mg/L, moxifloxacin at 20 000 mg/L and norfloxacin at 20 000 mg/L. Fresh 10× experimental stocks of levofloxacin, ciprofloxacin, delafloxacin, moxifloxacin and norfloxacin were made in 1× DPBS at the desired concentration prior to each experiment.

Reagents and bacterial culturing conditions

Bacterial cultures were prepared for AST by first streaking onto LB agar (LA) plates from storage in 20% glycerol/80% CAMHB at −80°C then incubated overnight at 37°C. Colonies were picked from the plate and used to inoculate 5 mL of the desired liquid culture in a 14 mL Falcon polypropylene round-bottom snap cap tube (Corning #352059). Liquid culturing media were either CAMHB [MHB (Difco) supplemented with 20 mg/L Ca²⁺ and 10 mg/L Mg²⁺] or amended Roswell-Park Memorial Institute medium (RPMI+) [phenol-free RPMI (Gibco 1640) + 10% LB (Criterion)]. Liquid cultures were incubated with shaking at 100 rpm overnight at 37°C. The following day, overnight cultures were diluted 1:50, either in the 14 mL snap cap tubes or in 50 mL polypropylene conical tubes (Corning #352098) and incubated as previously described to mid-logarithmic growth phase (OD₆₀₀ = ∼0.4).

MICs (MIC₉₀)

MIC₉₀ procedures have been previously described in detail.⁷ In brief, bacterial cultures were grown in either CAMHB or RPMI+ overnight, subcultured 1:50 the following morning in the same media and grown until log phase (OD₆₀₀ = ∼0.4). Log-phase cultures were diluted to OD₆₀₀ = 0.002 (∼5 × 10⁵ cfu/mL) and 180 μL aliquoted into each well of a Costar flat-bottom 96-well plate (Corning #3370) that also contained 20 μL of the desired 10× drug treatment. Antibiotic susceptibility was assayed using a standardized 2-fold serial dilution testing scheme. The plates were incubated with shaking at 100 rpm at 37°C overnight. The following morning (∼20 h), OD₆₀₀ values of each well were measured on an EnSpire Alpha multimode plate reader (PerkinElmer) to determine growth. To calculate the MIC₉₀, defined as the drug concentration required to inhibit ≥90% of the growth of the untreated controls, OD₆₀₀ of each drug-treated well was compared with an untreated control. Investigation of bicarbonate (HCO₃⁻) buffer effect on the MIC₉₀ in RPMI was performed as above, except HCO₃⁻-free RPMI medium was prepared in advance with specified concentrations of HCO₃⁻ buffer added later.

MBCs (MBC₉₉)

To determine the MBC₉₉, defined as the minimum amount of drug required to kill ≥99% of the bacteria in the untreated controls, cultures were prepared in CAMHB as for the above MIC₉₀ determination. Log-phase cultures were concentrated, washed and inoculated into 15 mL of either CAMHB, RPMI+ or 80% RPMI + 20% fresh human whole blood using approximately 5 × 10⁵ cfu. Each culture was aliquoted into a 96-well plate, as described for the MIC₉₀ determination, with 180 μL of culture and 20 μL of the desired 10× drug treatment. The plates were incubated overnight with shaking at 100 rpm at 37°C, each well serially diluted in 1× DPBS at ∼20 h, plated onto LA and incubated at 37°C overnight for cfu enumeration.

Delafloxacin accumulation

Intracellular accumulation of delafloxacin was measured as previously described for fluoroquinolones.^34–36 Briefly, cultures were grown to logarithmic phase from overnight cultures as described for the MIC₉₀ experiments. Aliquots of 1.5 mL were removed from the cultures when they reached OD₆₀₀ = 0.5 (∼1.25 × 10⁸ cfu/mL) and placed into 1.7 mL microcentrifugation tubes (Corning #CLS3620). Delafloxacin was added at 200 mg/L for MDR S. aureus and 400 mg/L for MDR P. aeruginosa. Cultures were then incubated with shaking at 100 rpm at 37°C. Every 30 min for MDR S. aureus until 2 h, and every 1 h for MDR P. aeruginosa for 6 h, aliquots were removed and centrifuged at 13 200 rpm for 1 min (Eppendorf 5418 R). Supernatants were removed for each sample and the cell pellet was washed in 1.5 mL of 1× DPBS. Washes were repeated three times. After the final centrifugation, pellets were resuspended in 120 μl of 1× DPBS for MDR S. aureus, or 120 μl of molecular grade H₂O for P. aeruginosa (H₂O was found to aid in the lysis of P. aeruginosa) and placed into a heating block at 95°C for 30 min. After incubation, samples were cooled at 4°C and the supernatants were transferred to a Costar flat-bottom 96-well plate (Corning #3370) and delafloxacin was measured with a 350 nm excitation and 480 nm emission on a EnSpire Alpha multimode plate reader (PerkinElmer).

Murine P. aeruginosa pneumonia model

All animal experiments were conducted under veterinary supervision and approved by the UC San Diego IACUC. P. aeruginosa strain P4 was cultured in CAMHB as previously described for the MIC₉₀ experiments. Mid-logarithmic P4 cultures (OD₆₀₀ = ∼0.4) were pelleted via centrifugation at 3203 × g, the supernatant removed, and the pellet resuspended and washed three times in an equal volume of 1× DPBS (Corning). After the final centrifugation, the pellet was resuspended in a low volume of 1× DPBS, 10 μL diluted into 990 μL of 1× DPBS, and the OD₆₀₀ measured. The culture was then diluted to 2.5 × 10⁸ cfu/mL using a previously determined OD to concentration ratio for P. aeruginosa P4 (OD₆₀₀ of 0.4 = ∼1 × 10⁸ cfu/mL).

All mice were housed in an SPF barrier facility on a 12/12 light/dark cycle in Innovive Innocage pre-bedded corn cob disposable cages on a 2020X diet from Envigo. Mice were randomized into groups of five mice by the UCSD vivarium staff (blind to the researcher) at least 48 h prior to infection. To reach N = 7 for the antibiotic treatment studies, groups of five mice were further randomized and split into smaller groups of two mice per cage. No cage contained more than five mice and no more than one group was in any cage.

Survival experiments were conducted using 8-week-old juvenile female C57Bl/6J mice (Jackson Labs). Prior to infection, each mouse was anaesthetized with 100 mg/kg ketamine (Zoetis) + 10 mg/kg xylazine (VetOne), then administered a 40 μL intratracheal dose of the prepared 2.5 × 10⁸ cfu/mL culture using an operating otoscope (Welch Allyn), giving an infectious dose of ∼1 × 10⁷ cfu/mouse. Post-infection, mice were recovered on a sloped heated pad before being returned to their respective cages. At 1 h post-infection, mice received an intraperitoneal injection of the desired dose of each fluoroquinolone via a 1 mL U-100 syringe (Becton Dickinson). Mouse survival was monitored for 6 days.

Results

Media-dependent MIC activities of fluoroquinolones versus ESKAPE pathogens

As a first step in exploring the impact of host physiological conditions on bacterial fluoroquinolone susceptibility, we measured the activity of ciprofloxacin against MDR S. aureus in standardized AST medium, CAMHB and in RPMI+, a more physiologically relevant medium. A 4-fold increase in the S. aureus ciprofloxacin MIC₉₀ was found in RPMI+ compared with CAMHB (Figure 1). This approach to identify potential media-dependent differential activity was then expanded to other fluoroquinolones (levofloxacin, norfloxacin, moxifloxacin and the recently approved delafloxacin) and three additional MDR ESKAPE pathogens (P. aeruginosa, K. pneumoniae and A. baumannii). Across these fluoroquinolones, we saw an average loss in the activity of ciprofloxacin (1.7-fold), levofloxacin (1.2-fold), moxifloxacin (2.1-fold), norfloxacin (4.0-fold) and delafloxacin (12-fold) (Figure 1 and Table S1, available as Supplementary data at JAC Online). Among the bacterial pathogens, we saw an average fluoroquinolone activity loss of 1.2-fold for K. pneumoniae, 1.6-fold for A. baumannii, 3.9-fold for P. aeruginosa and 10.2-fold for S. aureus (Figure 1 and Table S1). We next assayed a panel of S. aureus and P. aeruginosa strains to assess whether the observed loss of delafloxacin activity in RPMI+ was a strain-specific phenomenon (Table S2). Delafloxacin activity was reduced by 32-fold in RPMI+ compared with CAMHB for all four additional S. aureus strains examined (Newman, ATCC 29213, ATCC 25923 and ATCC 33591), mirroring the results seen with S. aureus LAC (Table S2). Similarly, delafloxacin activity was reduced in RPMI+ for the panel of P. aeruginosa strains by either 8-fold (PAO1, ATCC 10145) or 3.3-fold (ATCC 2785) (Table S2). These results indicate that the reduced delafloxacin activity in RPMI+ in S. aureus and P. aeruginosa is not a strain-specific phenomenon.

Figure 1.

Differential fluoroquinolone activity was observed between amended tissue culture-based medium (RPMI+) and standard bacteriological medium (CAMHB) for MDR ESKAPE pathogens. The activities of the fluoroquinolones ciprofloxacin (CIP), levofloxacin (LVX), moxifloxacin (MXF), norfloxacin (NOR) and delafloxacin (DLX) were compared in CAMHB versus RPMI+ against MDR S aureus strain LAC, MDR A. baumannii AB5075, MDR K. pneumoniae K1100 and MDR P. aeruginosa P4. Data are plotted as a heat map comparing the log₂-fold change in activity of each fluoroquinolone, as determined by MIC₉₀, in CAMHB versus RPMI+. An increased susceptibility in RPMI+ versus CAMHB is indicated as green, while reduced susceptibility is red. All experiments were performed in triplicate. This figure appears in colour in the online version of JAC and in black and white in the print version of JAC.

Media-dependent bactericidal activity of fluoroquinolones versus S. aureus and P. aeruginosa

Media-dependent alterations in the MIC of multiple fluoroquinolones were probed further using bactericidal assays in CAMHB, RPMI+ and 20% freshly isolated human whole blood. MDR S. aureus and MDR P. aeruginosa were selected for this analysis based on the most dynamic media-dependent alterations in fluoroquinolone MIC. Likewise, three fluoroquinolones representing a range of responses were studied: levofloxacin (largely media-independent activity changes), delafloxacin (highly media-dependent activity changes) and ciprofloxacin (modest media-dependent activity changes). Bactericidal action of levofloxacin, determined as MBC₉₉ in each condition, was slightly reduced in RPMI+ compared with CAMHB for S. aureus (1.58 log₂-fold change, P = 0.0117) and MDR P. aeruginosa (1.25 log₂-fold change, P = 0.0310), equivalent for CAMHB and 20% whole blood for S. aureus, and slightly reduced (1.58 log₂-fold change, P = 0.0117) in 20% whole blood versus CAMHB for P. aeruginosa (Figures 2 and 3). The MBC₉₉ for ciprofloxacin determined in CAMHB was decreased in RPMI+ compared with CAMHB for S. aureus (1.58 log₂-fold change, P = 0.0117) but not for MDR P. aeruginosa (0.58 log₂-fold change, P = 0.1089) and unchanged for both pathogens in 20% whole blood (Figures 2 and 3). Consistent with the inhibition results, the MBC₉₉ of delafloxacin was abolished against S. aureus in RPMI+ (3.58 log₂-fold change, P = 0.0002) and 20% whole blood (3.25 log₂-fold change, P = 0.0003) compared with CAMHB (Figure 2). Similarly, delafloxacin bactericidal activity was largely ablated in MDR P. aeruginosa in RPMI+ (3.00 log₂-fold change, P = not applicable) and 20% whole blood (3.00 log₂-fold change, P = not applicable) compared with CAMHB (Figure 3).

Figure 2.

Loss of bactericidal activity of fluoroquinolones in differential media is not a class-wide phenomenon against MDR S. aureus. Bactericidal activity of levofloxacin (LVX), ciprofloxacin (CIP) and delafloxacin (DLX) against S. aureus LAC. LVX (a, d and g), CIP (b, e and h) and DLX (c, f and i) activities were assessed in the standard bacteriological medium CAMHB (a–c), in the physiologically relevant amended tissue culture medium RPMI+ (d–f) and in 20% fresh human whole blood (g–i). Media-dependent log₂-fold change in the MBC₉₉ of LVX (j), CIP (k) and DLX (l). Statistical significance was calculated using one-way ANOVA with P ≤ 0.05 (*) and P ≤ 0.01 (**). All experiments were performed in triplicate. NS, not significant. This figure appears in colour in the online version of JAC and in black and white in the print version of JAC.

Figure 3.

Loss of bactericidal activity of fluoroquinolones in differential media is not a class-wide phenomenon against MDR P. aeruginosa. Bactericidal activity of levofloxacin (LVX), ciprofloxacin (CIP) and delafloxacin (DLX) against MDR P. aeruginosa P4. LVX (a, d and g), CIP (b, e and h) and DLX (c, f and i) activities were assessed in the standard bacteriological medium CAMHB (a–c), in the physiologically relevant amended tissue culture medium RPMI+ (d–f) and in 20% fresh human whole blood (g–i). Media-dependent log₂ fold-change in the MBC₉₉ of LVX (j), CIP (k) and DLX (l). Statistical significance was calculated using one-way ANOVA with P ≤ 0.05 (*) and P ≤ 0.01 (**). All experiments were performed in triplicate. NS, not significant. This figure appears in colour in the online version of JAC and in black and white in the print version of JAC.

Correlation of fluoroquinolone activities in testing media versus 20% human blood

We next asked whether fluoroquinolone activity assessed in CAMHB or RPMI+ was a more accurate predictor of the corresponding drug’s activity in 20% human whole blood. To accomplish this, we used the calculated MBC₉₉ values for each fluoroquinolone in each media type, and then created a ratio of the MBC₉₉ for CAMHB and RPMI+ compared with whole blood, log₂[(medium MBC₉₉)/(whole blood MBC₉₉)], so that a perfect agreement would yield a value of 0 (Figure 4 and Table S3). For levofloxacin against MDR S. aureus, CAMHB was slightly more predictive of the potency in blood then RPMI+ (−0.41 versus 1.25) while for MDR P. aeruginosa, CAMHB was less predictive then RPMI+ (−1.67 versus −0.33) (Figure 4 and Table S3). CAMHB was more predictive for ciprofloxacin bactericidal activity in blood than RPMI+ for both S. aureus (−0.08 versus 1.58) and MDR P. aeruginosa (0.33 versus 1.00) (Figure 4 and Table S3). CAMHB was not as predictive as RPMI+ for delafloxacin activity in whole blood for either S. aureus (−3.40 versus 0.26) or MDR P. aeruginosa (−3.00 versus 0.00) (Figure 4 and Table S3).

Figure 4.

Ex vivo predictivity of AST in standard bacteriological or tissue culture medium is fluoroquinolone dependent. The bactericidal activities of the fluoroquinolones levofloxacin (LVX), ciprofloxacin (CIP) and delafloxacin (DLX) in 20% fresh human whole blood were compared with standard bacteriological (CAMHB) and amended tissue culture medium (RPMI+) against MDR S. aureus LAC and MDR P. aeruginosa P4. Data are plotted as a heat map comparing the log₂-fold change in activity of each fluoroquinolone, as determined by MBC₉₉, in CAMHB and RPMI+ versus in the presence of 20% human whole blood. Accurate predictivity of activity is indicated as white, while underprediction is green and overprediction is red. All experiments were performed in triplicate. This figure appears in colour in the online version of JAC and in black and white in the print version of JAC.

Examination of fluoroquinolone potency in a murine pneumonia model of infection

Despite delafloxacin having a lower MIC₉₀ in CAMHB, our data indicated that this drug’s activity may be negatively impacted by conditions present in the host, leading to an overestimation of its potency in standardized AST. To explore whether our findings translated to an in vivo context, we compared the efficacy of levofloxacin, ciprofloxacin and delafloxacin in a murine pneumonia model of MDR P. aeruginosa infection. Mice were treated with either 250, 125 or 62.5 mg/kg of each drug 1 h post-infection and survival monitored for 6 days. The selected doses promoted different degrees of survival in a dose-dependent fashion for each fluoroquinolone (Figure 5a–c). When the data were grouped by dose instead of by drug, no statistically significant differences in survival were detected between levofloxacin, ciprofloxacin or delafloxacin at the three doses examined (Figure 5d–e), despite the clear differences in potencies observed in standard MIC.

Figure 5.

Survival in MDR P. aeruginosa murine pneumonia infection model is not correlated with specific fluoroquinolone treatment. Therapeutic assessment of (a) levofloxacin (LVX), (b) ciprofloxacin (CIP) and (d) delafloxacin (DLX) in a murine pneumonia model of P. aeruginosa. Survival curves compared by individual drug dose response (a–c) or by dose comparison between all three drugs (d–e). C57BL/6J mice were infected intratracheally with 1 × 10⁷ cfu of MDR P. aeruginosa strain P4. Each drug was administered through an intraperitoneal injection 1 h post-infection at the indicated dosages (mg/kg). Mouse survival was monitored for 6 days with n = 10 for the untreated controls and n = 7 animals per treated group. Statistical significance was calculated with a Mantel–Cox survival curve analysis with P ≤ 0.05 (*). NS, not significant.

Contribution of HCO₃⁻ to media-dependent fluoroquinolone activities

Testing conditions that more closely mimicked the host environment impacted the calculated potency of each of the examined fluoroquinolones in a different manner, and the inaccuracy of the susceptibility predictions in the standard CLSI medium were confirmed in an in vivo model. Thus, we next sought to identify the specific media component that is absent from standard CAMHB used for AST, but present within RPMI+ and in vivo, that dictates bacterial fluroquinolone susceptibility. HCO₃⁻ buffer, present within RPMI+ and human blood but absent from CAMHB, represents a key difference in composition among the media that have previously been linked to alterations in antibiotic activities.^7,15,37 We investigated whether the concentration of HCO₃⁻ buffer in media impacted the bactericidal activity of fluoroquinolones. To that end, we titrated pH 8.4 NaHCO₃ into HCO₃⁻-free RPMI+ and CAMHB media, increasing the concentration in 2-fold increments from 0.38 mM to the physiological concentration of 24 mM.³⁸ Increasing concentrations of HCO₃⁻ did not affect levofloxacin activity versus S. aureus, and caused a small activity loss at high concentrations for MDR P. aeruginosa (Figure 6). Ciprofloxacin activity was also largely found to be independent of HCO₃⁻ (Figure 6). In contrast, delafloxacin activity was directly linked to HCO₃⁻, with higher HCO₃⁻ concentrations directly correlating with a loss in its activity in both CAMHB and RPMI+ (Figure 6). As maintaining pH homeostasis is one of the physiological functions of HCO₃⁻, and fluoroquinolone antibiotics can have pH-dependent activity profiles,^26,39 we explored whether their differential media-based activities were correlated with pH. Examining the fluoroquinolone activity in pH 6.0-adjusted CAMHB and RPMI+ revealed that while they are more effective in acidic environments, the differential activity of delafloxacin in response to HCO₃⁻ was independent of pH (Table S4). These data demonstrate that HCO₃⁻ directly modulates the activity of delafloxacin and is a major contributor to its media-dependent activity profile.

Figure 6.

HCO₃⁻ is the basis for media-dependent differential fluoroquinolone activity in both MDR S. aureus and MDR P. aeruginosa. Levofloxacin (LVX), ciprofloxacin (CIP) and delafloxacin (DLX) sensitivity was measured in CAMHB (a and b) and HCO₃⁻-free RPMI+ (c and d) after the titration of 2-fold increases of exogenous HCO₃⁻ for both MDR S. aureus (a and c) and MDR P. aeruginosa (b and d). The MIC₉₀ was measured at each condition and the fold change in MIC based on 0 mM HCO₃⁻ was calculated. Statistical significance was calculated with two-way ANOVA, comparing the activity of each respective drug at the indicated HCO₃⁻ concentration with the control, 0 mM HCO₃⁻, per panel. **P ≤ 0.01, ***P ≤ 0.001 and ****P ≤ 0.0001. All experiments were performed in triplicate. This figure appears in colour in the online version of JAC and in black and white in the print version of JAC.

Basis for HCO₃⁻-mediated antagonism of delafloxacin

We next pursued a mechanistic correlation for the observed loss of delafloxacin activity in the presence of HCO₃⁻. We therefore examined the intracellular accumulation of delafloxacin in either the presence or absence of 24 mM HCO₃⁻ in CAMHB or RPMI+ for both MDR S. aureus and MDR P. aeruginosa. Cultures were treated for the indicated times, washed and lysed, and intracellular delafloxacin was measured using its inherent fluorescence as previously demonstrated.^34–36 In all conditions examined, significantly less delafloxacin accumulated intracellularly in the presence of HCO₃⁻ (Figure 7).

Figure 7.

HCO₃⁻ reduces intracellular accumulation of delafloxacin (DLX). DLX accumulation was examined in MDR S. aureus (a and c) and MDR P. aeruginosa (b and d) in CAMHB (a and b) or RPMI+ (c and d) either in the presence (squares) or absence (circles) of 24 mM HCO₃⁻. DLX was added at t = 0 at 200 mg/L for MDR S. aureus and 400 mg/L for MDR P. aeruginosa. Intracellular levels of DLX were detected from bacterial lysates using the drug’s inherent fluorescence profile. Statistical significance was calculated with two-way ANOVA, comparing the accumulation of DLX in either the presence or absence of HCO₃⁻. *P < 0.05, **P ≤ 0.01, ***P ≤ 0.001 and ****P ≤ 0.0001. All experiments were performed in triplicate. This figure appears in colour in the online version of JAC and in black and white in the print version of JAC.

Discussion

In this study we contrasted the activity of multiple fluoroquinolones against selected MDR bacterial pathogens in nutrient-rich bacteriological conditions (CAMHB) used in standardized AST with those in a more physiologically relevant nutrient-deficient tissue culture medium (RPMI+). Drug activities were impacted by the selection of the growth medium; however, this was not a homogeneous class-wide phenomenon as the magnitude of activity alterations differed significantly among fluoroquinolones. Based on the varying response, neither CAMHB nor RPMI+ were found to be entirely predictive of fluoroquinolone potency in 20% freshly isolated human whole blood. While antibiotic activity in CAMHB may not correlate with in vivo potency, as has previously been recognized,^6,7,15,40 a simple change to tissue culture-based RPMI+ is in and of itself not a reliable gauge of drug activity in blood.

In vivo environmental components present during infection impact the efficacy of antibiotic therapies. Here we identify a role of HCO₃⁻ in impacting the activity of some, but not all, of the members of the fluoroquinolone class of antibiotics. As the major physiological buffer present in human blood, one of the many roles of HCO₃⁻ is in maintaining pH homeostasis that is critical for preserving immune functions and control of bacterial infections.^41–45 Due to its zwitterionic character, delafloxacin is known to have increased activity in acidic conditions;^26,46 however, we observed a consistent reduction in delafloxacin activity independent of the experimental pH, decoupling the role of HCO₃⁻ in regulating pH from its impact on the observed drug activity. When contrasted with comparator fluoroquinolones, delafloxacin has higher serum protein-binding affinities, which could influence its pharmacodynamics in vivo, although in the current study we observed similar delafloxacin MBC₉₉s in both RPMI+ and 20% fresh human whole blood. HCO₃⁻ also enhances host responses in acidotic patients,^47,48 alters bacterial susceptibility to host factors,¹⁶ inhibits biofilm formation¹⁷ and disrupts bacterial proton motive force.¹⁵ Here we demonstrate that HCO₃⁻ antagonizes the activity of delafloxacin by impairing its intracellular accumulation in both MDR S. aureus and MDR P. aeruginosa. Further studies are required to identify the exact mechanistic basis for how HCO₃⁻ modulates bacterial cell uptake or influx mechanisms or the drug itself to impact its accumulation.

Innate immune factors in patients, such as HCO₃⁻, provide a critical non-specific defence against pathogens. In cases where HCO₃⁻ homeostasis is disturbed, patients can suffer from metabolic or respiratory acidosis, increasing the risk of fatal bacterial infections.^49–51 Our data, and prior studies,^52,53 increasingly indicate that some of these factors could also influence antibiotic activities in patients with MDR bacterial infections. More precise assessment of an individual patient’s physiological status (blood pH and HCO₃⁻ levels, for instance) may be predictive of therapeutic outcomes and could be incorporated into selection of treatment regimens. Although these data showed reduced fluoroquinolone activity in the presence of HCO₃⁻, prior studies have observed enhanced activities of other antibiotic classes with added HCO₃⁻,^5,7,15,18 suggesting that different agents may be advantageous based on the patient’s current conditions and electrolyte status.

Incorrect antibiotic choice or dosing based on an overestimation of antibiotic activity in vitro can lead to treatment failure and/or promote the evolution of antibiotic resistance. Given the broad usage of fluoroquinolones, and the recent introduction of delafloxacin into the clinic, our study underscores the potential value of examining antibiotic activities in conditions that recapitulate microenvironments found within the host to help ascertain their full clinical potential. By gaining a thorough understanding of the immune conditions modulating antibiotic efficacies, we may be able to both develop a more predictive AST scheme and optimize our current therapies to be more effective in patients.

Funding

This research was supported by a National Institute of Health (NIH) through the National Institute of Allergy and Infectious Disease (NIAID) grant (1-U01-AI124316). N.D. was additionally supported by an NIH (5T32HD087978-05).

Transparency declarations

V.N. has consulted for Cellics Therapeutics, Vaxcyte, Clarametyx Biosciences, SNIPR Biome, Boehringer Ingelheim and Iogen, all unrelated to the current study. All other authors: none to declare. The authors do not hold patents related to the current study.

Author contributions

M.H. and N.D. carried out the laboratory experiments, compiled the data, analysed results and composed the manuscript. E.B. assisted with the in vivo experiments. N.D. and V.N. conceptualized the project. All authors edited the final manuscript.

Supplementary data

Table S1 to S4 are available as Supplementary data at JAC Online.