The clinical zoliflodacin-susceptible N. gonorrhoeae strain with GyrB S467N (SE600/18) was cultured in Sweden in 2018. Additionally, the zoliflodacin-resistant mutant of SE600/18 with an additional GyrB D429N substitution selected in the HFIM (SE600/18-D429N) was examined.

For determination of zoliflodacin MICs (mg/L), agar dilution on GCVIT agar plates (Foerster et al., 2019) and microbroth dilution [in triplicates in the HFIM medium, i.e., modified Fastidious Broth (mFB)] were performed, as previously described (Jacobsson et al., 2021). Etest was used to determine MICs (mg/L) of ceftriaxone, cefixime, ciprofloxacin, and azithromycin, in accordance with the manufacturer’s instructions (bioMérieux, Marcy-l’Etoile, France).

For simulation of a gonococcal infection and the PK/PD of current and new antimicrobials, such as zoliflodacin, against N. gonorrhoeae, we recently developed and optimized a dynamic HFIM using cellulosic cartridges (FiberCell Systems Inc., Frederick, MD, United States) (Jacobsson et al., 2021). In brief, our HFIM is a two-compartment model system, in which N. gonorrhoeae cells grow in the extracapillary space of a cellulosic cartridge containing a bundle of microfibers (FiberCell Systems Inc., Frederick, MD, United States). A syringe pump administered zoliflodacin into the HFIM and peristaltic pumps isovolumetrically replaced zoliflodacin-containing liquid growth medium with zoliflodacin-free liquid growth medium to simulate the half-life of zoliflodacin and non-protein bound (free) zoliflodacin concentration-time profiles reported in human plasma throughout 7 days. N. gonorrhoeae quantitative cultures (colony forming units (CFUs)/mL) for total N. gonorrhoeae burden and zoliflodacin-resistant N. gonorrhoeae population and determination of zoliflodacin concentrations in the HFIM were performed over 7 days (Drusano, 2017).

Briefly, on the first day 0.5 ml of N. gonorrhoeae cultures (18–24 h) from GCAGP agar plates (Foerster et al., 2019) were inoculated in 49.5 ml of mFB and incubated at 36°C in a humidified 5% CO₂-enriched atmosphere to mid-log phase. 10 ml (∼10⁶ CFU/ml) of the N. gonorrhoeae suspension were then inoculated into each HFIM cartridge to mimic a clinically relevant N. gonorrhoeae cell concentration (Bissessor et al., 2011; Chow et al., 2016; Priest et al., 2017; Van Der Veer et al., 2020). Zoliflodacin was administrated to mimic an adult human PK concentration-time profile following a single oral dose of zoliflodacin [PK parameters for zoliflodacin 3 g oral dose were used (linearly adjusted for other doses): 17% fraction of zoliflodacin free (protein-unbound) in plasma, 6.47 h half-life (t_1/2), and a 3 h infusion time] (O’Donnell et al., 2019), as previously described (Jacobsson et al., 2021). One HFIM cartridge per examined strain and experiment was used as an untreated growth control.

Dose-range experiments simulated zoliflodacin single dose oral regimens of 0.5, 1, 2, 3, and 4 g against the clinical SE600/18 strain, and single dose oral regimens of 2, 3, 4, and 6 g against SE600/18-D429N. Dose-fractionation experiments simulated zoliflodacin oral dose therapy with 4, 6, and 8 g administered as one half of the total dose given at 0 h and at 12 h (q12 h) against SE600/18-D429N. All experiments were followed for 7 days.

To determine the N. gonorrhoeae total population and zoliflodacin-resistant subpopulations, bacterial solution (1 ml) was sampled from each HFIM cartridge at time points 3, 6.5, 24, 48, 72, 96, 120, 144, and 168 h for the dose-range experiments, and at 3, 6.5, 12, 15, 18.5, 24, 48, 72, 96, 120, 144, and 168 h for the q12 h dose-fractionation experiments. Samples were serially diluted in mFB and quantitatively plated on GCAGP agar plates (Foerster et al., 2019) and GCAGP agar plates (Foerster et al., 2019) containing 2 × MIC of zoliflodacin, resulting in a detection limit of ≥100 CFUs per HFIM cartridge, as previously described (Jacobsson et al., 2021). Colony counts (log10 CFU/ml) were quantified after incubation for up to 72 h at 36°C in a humid 5% CO₂-enriched atmosphere using an automated colony counter (Scan 4000, Interscience, Saint-Nom-la-Bretèche, France).

To evaluate the biofitness of the zoliflodacin-resistant mutant selected in the HFIM (SE600/18-D429N) compared to the zoliflodacin-susceptible clinical SE600/18 parent strain, competition experiments using coculture were performed in the HFIM. Briefly, bacteria were harvested from GCAGP agar plates (Foerster et al., 2019) and suspended in mFB to a quantity of ∼10⁶ CFU/ml. Equal volumes (5 ml/strain) of the suspensions of each strain were inoculated into the same HFIM cartridge. Aliquots (1 ml) were sampled at 24, 48, 72, 96, 120, 144, and 168 h, serially diluted in mFB and quantitatively plated on GCAGP agar plates (Foerster et al., 2019), as previously described (Jacobsson et al., 2021). Colony counts (log10 CFU/ml) were quantified after incubation for up to 72 h at 36°C in a humid 5% CO₂-enriched atmosphere using an automated colony counter (Scan 4000, Interscience, Saint-Nom-la-Bretèche, France). The competitive index (CI) was determined by dividing the ratio of the SE600/18-D429N mutant to wild-type SE600/18 at each time point with the ratio of the SE600/18-D429N mutant to wild-type SE600/18 in the initial inoculum (Vincent et al., 2018).

To confirm that the predicted zoliflodacin PK profiles were observed in the HFIM, broth samples (500 µl) were collected at time points 1, 2, 3, 6.5, 18.5, 24, 48, 72, 96, 120, 144, and 168 h for the dose-range experiments, and at 1, 2, 3, 6.5, 12, 15, 18.5, 24, 48, 72, 96, 120, 144, and 168 h for the q12 h dose-fractionation experiments. All zoliflodacin concentrations were determined from 100 μl sample aliquots using liquid chromatography-tandem mass spectrometry (LC-MS/MS), as previously described (Jacobsson et al., 2021).

We simultaneously modeled 3 system outputs for the analysis of the experimental data. The system outputs were: concentration of zoliflodacin, total N. gonorrhoeae burden, and burden of N. gonorrhoeae with lower susceptibility/resistance to zoliflodacin (containing MIC-increasing gyrB mutation selected during treatment). Population modeling was performed employing the Non-Parametric Adaptive Grid (NPAG) program of Leary et al. (2001) and Neely et al. (2012). Modeling choices (weighting, etc.) and goodness of fit evaluations were as previously published (Brown et al., 2015). Simulation was performed with the ADAPT V Program of D’Argenio et al. (2009) using Bayesian posterior parameter estimates.

Whole genome sequencing (WGS) was performed, as previously described (Jacobsson et al., 2016; Golparian et al., 2020a), on selected colonies that grew on the zoliflodacin-containing plates and that also had increased MICs of zoliflodacin by agar dilution. The WGS was primarily performed to identify zoliflodacin resistance-associated gyrB mutations, i.e., the previously identified gyrB mutations that were verified to cause the increased MICs of zoliflodacin in in vitro selected zoliflodacin-resistant mutants (Alm et al., 2015; Foerster et al., 2015; Foerster et al., 2019; Jacobsson et al., 2021) or novel gyrB mutations. However, the whole genome sequences of the zoliflodacin-resistant mutants were examined to identify also any other zoliflodacin resistance-associated mutations selected in the HFIM. All reads were quality controlled and trimmed accordingly using our previously described CLC Genomics Workbench v20.0.4 workflow (Golparian et al., 2020b), and all quality-controlled reads were mapped to the gyrB reference obtained from Genbank (Genbank: AE004969.1) using local alignment with CLC Genomics Workbench with match score 1, mismatch cost of 2 and linear gap cost of 3, the variants across the gene were called with a minimum coverage of 10x and a minimum frequency of 35%. WGS reads of SE600/18 with a pre-existing GyrB S467N mutation and the zoliflodacin-resistant mutant of SE600/18 with an additional GyrB D429N substitution (SE600/18-D429N) are available through the European Nucleotide Archive (ENA) accession number PRJEB50904.

The main experiments of the zoliflodacin-susceptible N. gonorrhoeae SE600/18 parent strain (with a GyrB S467N mutation) and the zoliflodacin-resistant N. gonorrhoeae SE600/18-D429N mutant [with GyrB S467N plus GyrB D429N selected in the Hollow Fiber Infection Model (HFIM)] have been summarized in Supplementary Figure 1.

The MICs of zoliflodacin determined using agar dilution and microbroth dilution methods, GyrB substitutions, and additional relevant characteristics of the two examined strains are summarised in Table 1.

Briefly, the clinical zoliflodacin-susceptible N. gonorrhoeae SE600/18 strain, containing a GyrB S467N substitution, and the zoliflodacin-resistant SE600/18-D429N mutant (GyrB S467N plus an in vitro selected GyrB D429N substitution) were examined in the HFIM. The zoliflodacin MICs of both strains were one MIC dilution higher using microbroth dilution compared with agar dilution. Notably, SE600/18 belonged to MLST ST7363, which has been a common gonococcal ST internationally during latest decades and it has also been associated with multi-drug resistance, including decreased susceptibility and resistance to extended-spectrum cephalosporins such as ceftriaxone and particularly cefixime (Shimuta et al., 2015; Unemo, 2015; Harris et al., 2018; Sánchez-Busó et al., 2021) (Table 1).

The results of the zoliflodacin dose-range studies for the zoliflodacin-susceptible clinical SE600/18 strain are summarised in Figure 1.

Briefly, the SE600/18 strain grew well in the untreated growth control arms and reached a bacterial density of 10¹⁰–10¹¹ CFU/ml at 24 h (Figure 1). All untreated controls also maintained growth at approximately 10⁹–10¹¹ CFU/ml throughout the 7-days experiments (Figure 1). A rapid bacterial kill was documented during the first 6.5 h for all doses. However, after zoliflodacin 0.5 g single dose therapy the SE600/18 strain had recovered growth at 24 h (approximately 10⁷ CFU/ml) and at 48 h the bacterial density was as high as in the untreated control (>10¹⁰ CFU/ml) (Figure 1A). Using both the zoliflodacin 1 and 2 g treatment (Figures 1B,C, respectively), SE600/18 was rapidly killed and at 24 h no growth was detected. However, at 48 h SE600/18 had recovered growth at bacterial density of approximately 10⁹ CFU/ml (Figure 1B) and 10⁷ CFU/ml (Figure 1C), respectively. Using the zoliflodacin 3 and 4 g treatments, SE600/18 was eradicated at 6.5 h time point and the strain did not recover any growth during the 7-days experiments (Figure 1D).

Zoliflodacin-resistant mutants grew on the zoliflodacin-containing plates for all treatments arms where growth was recovered, i.e., in the 0.5, 1 and 2 g treatment arms (Figures 1A–C). These zoliflodacin-resistant populations emerged after 6.5 h (0.5 g treatment; Figure 1A) or 48 h (1 and 2 g treatments; Figures 1B,C). Notably, these zoliflodacin-resistant populations were maintained at low concentrations of approximately <10²–10⁴ CFU/ml during the 7-days experiment. Accordingly, further amplification of the zoliflodacin-resistant populations was not observed, and these zoliflodacin-resistant mutant populations appeared to grow slower and in smaller colonies compared to the SE600/18 strain on agar plates, which all may indicate a suboptimal biofitness. These selected zoliflodacin-resistant mutants had zoliflodacin MICs of 2 mg/L (agar dilution) and contained the GyrB D429N substitution in addition to the pre-existing GyrB S467N substitution (referred to as SE600/18-D429N hereafter). No other selected mutations that appeared to be associated with zoliflodacin susceptibility were found.

The results of the zoliflodacin dose-range studies for the zoliflodacin-resistant SE600/18-D429N mutant are summarised in Figure 2.

In brief, the SE600/18-D429N mutant grew well in the untreated growth control arms and reached a bacterial density of 10¹¹ CFU/ml at 24 h (Figure 2). All untreated controls maintained growth at around 10¹⁰–10¹¹ CFU/ml throughout the 7-day experiments (Figure 2). However, the zoliflodacin 2, 3, and 4 g single dose oral therapy did not result in any kill of SE600/18-D429N (Figures 2A–C) and the zoliflodacin 6 g single dose regimen only caused bacterial kill the first 6.5 h (Figure 2D). Growth was recovered in all treatment arms and ranged from approximately 10⁷ CFU/ml (6 g arm) to 10¹¹ CFU/ml (2 g arm) at 24 h and at the bacterial density of the untreated control at 48 h (10¹¹ CFU/ml), which was maintained during the remaining 5 days of the experiments (Figure 2).

Furthermore, zoliflodacin-resistant gonococcal populations grew on the zoliflodacin-containing plates for all treatments arms (Figures 2A–D). These zoliflodacin-resistant populations grew at approximately 10⁵ CFU/ml after 24 h in all treatment arms. Notably, these zoliflodacin-resistant populations fluctuated at 10⁵–10⁷ CFU/ml during the whole 7-days experiment (Figure 2). Accordingly, further amplification of these zoliflodacin-resistant populations was not observed, and these mutant populations appeared to grow slower and in smaller colonies on agar plates, which all may indicate a suboptimal biofitness. These zoliflodacin-resistant populations had zoliflodacin MICs of 2–8 mg/L (agar dilution) and contained the GyrB D429N substitution in addition to the pre-existing GyrB S467N substitution.

The results of the zoliflodacin dose-fractionation experiments for the zoliflodacin-resistant SE600/18-D429N mutant are summarised in Figure 3.

Briefly, the SE600/18-D429N mutant grew well in the untreated growth control arms and reached a bacterial density of 10¹¹ CFU/ml at 24 h (Figure 3). All untreated controls maintained growth at around 10¹⁰–10¹¹ CFU/ml throughout the 7-day experiments (Figure 3). The equivalent zoliflodacin oral therapy of 4 and 6 g administered as equally divided doses at q12 h did not result in any substantial kill of SE600/18-D429N (Figures 3A,B) and the zoliflodacin 8 g q12 h regimen only caused bacterial kill the first 24 h (Figure 3C). Accordingly, growth was recovered in all treatment arms and ranged from approximately 10³ CFU/ml (8 g q12 h arm) to 10⁸ CFU/ml (4 g q12 h arm) at 24 h and the bacterial density at 48 h was at approximately 10¹⁰–10¹¹ CFU/ml, which was maintained during the remaining 5 days of the experiments (Figure 3).

Zoliflodacin-resistant populations grew on the zoliflodacin-containing plates for all treatments arms (Figures 3A–C). These zoliflodacin-resistant populations grew at approximately 10³–10⁴ CFU/ml after 24 h and at >10⁶–10⁹ CFU/ml at 48 h in all treatment arms (Figure 3). The zoliflodacin-resistant population selected in the zoliflodacin 8 g q12 h treatment arm appeared to recover growth at nearly the same bacterial density as the untreated control (Figure 3C), however, also this mutant population appeared to grow slower and in smaller colonies compared to the SE600/18 strain on agar plates, which may indicate a suboptimal biofitness. These zoliflodacin-resistant populations had zoliflodacin MICs of 2–8 mg/L (agar dilution) and contained the GyrB D429N substitution in addition to the pre-existing GyrB S467N substitution.

The three output PK/PD model was fit to all the data for SE600/18. The mean and median values for SE600/18 are displayed in Table 2.

The fit of the model to the data was acceptable, with exception of the pre-Bayesian (population) analysis for the zoliflodacin-resistant mutants (SE600/18-D429N). The predicted-observed regressions for the analysis are displayed in Supplementary Figure 2. The reason for the poor model fit for the zoliflodacin-resistant mutants is likely because of poor biofitness of these zoliflodacin-resistant mutants. This hypothesis is also supported by their very small K_g-r (Table 2) and, even when zoliflodacin was essentially gone, the zoliflodacin-resistant mutants did not amplify (Supplementary Figure 2; Figure 1).

Regarding the zoliflodacin PK profiles, the agreement between observed and predicted zoliflodacin concentrations in the HFIM during the experiments were high (Supplementary Figures 2A,D).

For the HFIM dose-ranging study for the clinical N. gonorrhoeae strain SE600/18, we performed simulation with the identified parameter values to obtain an approximate exposure of zoliflodacin that would suppress amplification of selected mutants with increased zoliflodacin MIC and gyrB resistance mutations. That exposure was a zoliflodacin 2.7 g single dose oral treatment in the HFIM.

The growth rate constant for the susceptible SE600/18 population (Table 2) was 58–60% of the growth rate constants previously measured for WHO F and WHO X (Jacobsson et al., 2021). However, the growth rate constant for the zoliflodacin-resistant SE600/18-D429N population (Table 2) was only 7–16% of the corresponding growth rate constants for zoliflodacin-resistant populations of WHO F and WHO X (Jacobsson et al., 2021). This suboptimal growth rate of the zoliflodacin-resistant SE600/18-D429N population shows that its growth is impaired and accordingly biofitness decreased, which was further supported by the kill rate constant for the zoliflodacin-resistant SE600/18-D429N population that was more than 20 times higher than the growth rate constant for the mutant (Table 2).

To confirm if the in vitro-selected zoliflodacin resistance in the clinical zoliflodacin-susceptible SE600/18 strain impaired bacterial growth and accordingly decreased the biofitness, the zoliflodacin-susceptible parent SE600/18 and the zoliflodacin-resistant SE600/18-D429N mutant were cocultured in the same HFIM cartridge for 7 days (Figure 4A) and the competitive index calculated (Figure 4B). The growth of the zoliflodacin-susceptible clinical SE600/18 strain was maintained at ∼10¹⁰–10¹² CFU/ml during the 7 days experiments, which was at a similar level as in the SE600/18 monocultures. However, the growth of the zoliflodacin-resistant SE600/18-D429N mutant was substantially lower particularly during the first 24 h and it peaked at ∼10⁸ CFU/ml at 24 h and then decreased for each day resulting in a bacterial density of ∼10⁶ CFU/ml after 7 days (Figure 4A). The clinical zoliflodacin-susceptible SE600/18 strain appeared to outcompete the zoliflodacin-resistant SE600/18-D429N mutant, which was even more clearly shown when plotting the competitive index over the 7 days experiments (Figure 4B). Accordingly, the in vitro-selected zoliflodacin-resistant mutant SE600/18-D429N was substantially less fit compared to the zoliflodacin-susceptible clinical SE600/18 parent strain (Figure 4).