Phage-antibiotic synergy against daptomycin-nonsusceptible MRSA in an ex vivo simulated endocardial pharmacokinetic/pharmacodynamic model

ABSTRACT

Phage-antibiotic combinations (PAC) offer a potential solution for treating refractory daptomycin-nonsusceptible (DNS) methicillin-resistant Staphylococcus aureus (MRSA) infections. We examined PAC activity against two well-characterized DNS MRSA strains (C4 and C37) in vitro and ex vivo. PACs comprising daptomycin (DAP) ± ceftaroline (CPT) and a two-phage cocktail (Intesti13 + Sb-1) were evaluated for phage-antibiotic synergy (PAS) against high MRSA inoculum (10⁹ CFU/mL) using (i) modified checkerboards (CB), (ii) 24-h time-kill assays (TKA), and (iii) 168-h ex vivo simulated endocardial vegetation (SEV) models. PAS was defined as a fractional inhibitory concentration ≤0.5 in CB minimum inhibitory concentration (MIC) or a ≥2 log₁₀ CFU/mL reduction compared to the next best regimen in time-kill assays and SEV models. Significant differences between regimens were assessed by analysis of variance with Tukey’s post hoc modification (α = 0.05). CB assays revealed PAS with Intesti13 + Sb-1 + DAP ± CPT. In 24-h time-kill assays against C4, Intesti13 + Sb-1 + DAP ± CPT demonstrated synergistic activity (−Δ7.21 and −Δ7.39 log₁₀ CFU/mL, respectively) (P < 0.05 each). Against C37, Intesti13 + Sb-1 + CPT ± DAP was equally effective (−Δ7.14 log₁₀ CFU/mL each) and not significantly different from DAP + Intesti13 + Sb-1 (−Δ6.65 log₁₀ CFU/mL). In 168-h SEV models against C4 and C37, DAP ± CPT + the phage cocktail exerted synergistic activities, significantly reducing bio-burdens to the detection limit [2 log₁₀ CFU/g (−Δ7.07 and −Δ7.11 log₁₀ CFU/g, respectively)] (P < 0.001). At 168 h, both models maintained stable MICs, and no treatment-emergent phage resistance occurred with DAP or DAP + CPT regimens. The two-phage cocktail demonstrated synergistic activity against two DNS MRSA isolates in combination with DAP + CPT in vitro and ex vivo. Further in vivo PAC investigations are needed.

INTRODUCTION

Methicillin-resistant Staphylococcus aureus (MRSA) infections present a significant medical challenge, especially in cases of daptomycin (DAP)-nonsusceptible (DNS) MRSA strains. Due to an approximate treatment failure rate of ~30% with vancomycin and DAP monotherapies, sometimes attributed to emergent vancomycin-intermediate S. aureus strains and/or DNS phenotypes, combination therapies are often used (1–3). The National Action Plan (2020–2025) emphasizes two primary objectives in battling antibiotic-resistant microbes: the acceleration of research for new treatments and measures to retard the emergence and spread of antibiotic-resistant bacteria (4). Specifically, for MRSA, both these goals necessitate innovative therapeutic strategies.

High-inoculum bacterial infections, such as infective endocarditis, pose additional challenges in treatment (5). The ex vivo simulated endocardial vegetation (SEV) two-compartment pharmacokinetic/pharmacodynamic (PK/PD) model, extensively validated versus in vivo animal models, has proven pivotal for such scenarios (6–9). It mimics human PK and dynamic antibiotic exposures over extended treatment periods against high bacterial inocula commonly associated with deep-seated infections including infective endocarditis (IE) (6–9). These ex vivo model characteristics present an opportunity to study the dynamics of treatment under high-inoculum conditions, which is especially crucial given MRSA’s frequent failure during standard of care antibiotic therapy (1–3).

Given the above conundrums, the scientific community is exploring the synergistic potential of bacteriophages (phages) plus traditional antibiotics (10–12). Unlike standard antibiotics, lytic phages offer unique potential benefits such as in situ replication, avoiding microbiome disruption and enhanced localization at the site of infection (11, 13). Additionally, phage use in humans has been demonstrated to be, in general, safe. The only concern is the removal of endotoxin levels during the purification process adjusted to 5 EU/mL or lower (14). However, despite isolated in vitro studies and clinical case reports of phage-antibiotic combinations (PACs) (9, 15–21), current research often lacks depth in studying PAC interactions. Thus, in terms of S. aureus, such case reports have utilized non-humanized antibiotic dose strategies or focused on MRSA strains other than DNS (10).

The paradigm of PAS presents an innovative approach in addressing MRSA treatment challenges. The experimental PAS studies herein are anchored in addressing the following research gaps pertaining to three fundamental readout metrics of the PAS paradigm: (i) bacterial eradication, (ii) circumvention of phage-resistant bacteria, and (iii) potential stabilization and/or reversion of pre-existing antimicrobial resistance (13). This study aims to investigate strategic PACs in high-inoculum in vitro and ex vivo DNS MRSA microenvironments representative of infective endocarditis, using well-characterized clinical strains.

RESULTS

DNS MRSA isolate selection

Two clinical DNS MRSA strains isolated from bacteremic patients and belonging to the two clonal complex genotypes currently in most widespread United States circulation were studied in this work: C4, a USA300/ST8 clonotype with the mprF single nucleotide polymorphism P314L, and C37, a USA100/ST5 clonotype with the mprF single nucleotide polymorphism V351E. These strains were a gift from the Cubist Biorepository (Cambridge, MA).

Phage “cocktail” selection

Selection of the lytic S. aureus myophages used for the experiments herein was based on our previous phage screening evaluations that included (i) host range, (ii) mechanism of action, (iii) genetic similarity, (iv) isolation of spontaneous bacteriophage-insensitive mutants, and (v) bacterial growth suppression (15, 16). Based on those metrics, which included demonstrated in vitro susceptibility to our target DNS MRSA strains (C4 and C37) via phage plaque assay (22), Kayvirus phages Sb-1 and Intesti13 were selected for in vitro and ex vivo assays outlined below.

Modified checkerboard (CB) synergy assays

Using CB assays, we evaluated PAS and identified an effective range of phage theoretical multiplicity of infection (tMOI) in PAC prior to 24-h time-kill analyses (time-kill assays) and 168-h SEV models. Thus, bacterial growth was assessed against 2-fold and 10-fold dilutions of antibiotic and phage titer, respectively, using high bacterial inoculum (10⁹ CFU/mL) in a modified CB minimum inhibitory concentration (MIC) microtiter plate assay; the optical density of the culture in each well was measured following 24 h of incubation (Fig. 1). A high inoculum of 10⁹ CFU/mL was selected to simulate deep-seated, difficult-to-treat high-inoculum infections such as infective endocarditis that have a high risk of mortality and often fail antibiotic therapy, including combination therapy. The two-phage cocktail containing Intesti13 + Sb-1 was combined with DAP and/or ceftaroline (CPT) in CB MIC assays of PACs. In CB MIC assays that contained DAP + CPT, a standard MOI of 0.1 [2 × 10¹⁰ plaque-forming units (PFU)/mL] for each phage was added to appropriate wells. Spectrophotometric data were converted to a heat map of percent growth compared to growth control (GC) (23).

Fig 1.

High-inoculum (10⁹ CFU/mL) modified checkerboard MIC assays of daptomycin-nonsusceptible MRSA isolates (A) C4 and (B) C37 against (A1 and B1) daptomycin plus Intesti13 + Sb-1, (A2 and B2) ceftaroline plus Intesti13 + Sb-1, and (A3 and B3) daptomycin plus ceftaroline plus Intesti13 + Sb-1. Phage was added to each non-growth control (GC) and non-media control (MC) well in A3 and B3 at a phage MOI of 0.1. Values were normalized to the MC and converted to percent growth versus the untreated GC. MOI, multiplicity of infection; DAP, daptomycin; CPT, ceftaroline. CB squares within the orange outline represent synergistic bacterial killing (fractional inhibitory concentration ≤0.5).

Evaluations of the two DNS MRSA strains, C4 and C37, revealed synergistic activity for both strains against DAP ± CPT with Intesti13 + Sb-1 with phage MOI ranging from 0.1 to 10 [fractional inhibitory concentration (FIC) ≤0.5 for each combination] (Fig. 1 A1, A3, B1 and B3). Additionally, in strain C37, the combination of CPT plus the phage cocktail also demonstrated synergistic activity at a phage MOI ranging from 0.1 to 10 (FIC ≤0.5) (Fig. 1 B2).

Time-kill analyses

To further prioritize the treatment regimen selection for PK/PD SEV models, we performed 24-h high-inoculum (10⁹ CFU/mL) time-kill assays with C4 and C37 to assess for synergistic bactericidal impacts with our various phage-antibiotic combinations (Fig. 2). Based on the CB MIC data, we selected the following concentrations: DAP and CPT, each at 0.5× MIC, and each phage in the cocktail at a tMOI of 0.01, 0.1, 1, and 10 (all data not shown). Our target bactericidal and synergistic activity thresholds were defined as a ≥3 log₁₀ CFU/mL reduction from baseline and a ≥2 log₁₀ CFU/mL kill compared to the next most effective regimen, each at 24 h.

Fig 2.

Bacterial quantification in high-inoculum (10⁹ CFU/mL) 24-h time-kill experiments of GC, DAP, CPT, or DAP + CPT (each at 0.5× MIC) combined with phage cocktail Intesti13 + Sb-1 (MOI ranging from 0.01 to 1 each) against DAP-nonsusceptible MRSA isolates (A) C4 and (B) C37. The error bars indicate standard deviation of two replicate experiments. P values were determined using one-way ANOVA and Tukey’s post hoc test. Significant differences in CFU/mL between regimens is notated with a bracket and asterisk. Abbreviations: CFU, colony-forming units; GC, growth control; DAP, daptomycin; CPT, ceftaroline; MIC, minimum inhibitory concentration; MOI, multiplicity of infection; MRSA, methicillin-resistant Staphylococcus aureus; ANOVA, analysis of variance.

In the 24-h time-kill assays vs C4, Intesti13 + Sb-1 (MOI 1 and 0.1) plus either DAP or DAP + CPT demonstrated robust bactericidal synergy (−Δ7.21 and −7.39 log₁₀ CFU/mL, respectively) vs the next most effective regimen of CPT + Intesti13 + Sb-1 (P < 0.05 each) (Fig. 2A). Against C37, Intesti13 + Sb-1 (MOI of 1 and 0.1) with CPT or DAP + CPT were equally potent and effective regimens (−Δ7.14 log₁₀ CFU/mL each), although neither was significantly better at 24 h than the synergistic regimen of DAP alone + Intesti13 + Sb-1 (−Δ6.65 log₁₀ CFU/mL) (Fig. 2B).

Ex vivo SEV PK/PD model

We next determined the impact of humanized DAP and CPT exposures and the two-phage cocktail on high-inoculum bioburdens of DNS MRSA isolates C4 and C37. We conducted a series of 168-h SEV PK/PD model experiments, prioritizing the “top performing” regimens as identified in time-kill assays. In total, 12 distinct ex vivo PK/PD SEV models were run in duplicate, with five treatment regimens and a GC for each strain (Fig. 3 and 4). Humanized doses of DAP (10 mg/kg q24h) and CPT (600 mg q12h) were administered alone and in various combinations with the two-phage cocktail. CPT alone and CPT + the two-phage cocktail were not tested in SEV models due to underwhelming bacterial killing in preliminary CB and time-kill assays analyses. Each phage was administered at a tMOI of 1 (4 × 10¹⁰ PFU/mL) dosed q24h based on preliminary CB MIC and 24-h time-kill assays results under high-inoculum bioburdens (Fig. 1 and 2). Table 1 summarizes the observed PK parameters of human-simulated regimens. Targets for bactericidal and synergistic activity were the same as that used for 24-h time-kill assays.

Fig 3.

Efficacy of phage-antibiotic combinations in 168-h ex vivo PK/PD SEV models against daptomycin-nonsusceptible MRSA strain C4. Abbreviations: CFU, colony-forming units; GC, growth control; DAP, daptomycin; CPT, ceftaroline.

Fig 4.

Efficacy of phage-antibiotic combinations in 168-h ex vivo PK/PD SEV models against daptomycin-nonsusceptible MRSA strain C37. Abbreviations: CFU, colony-forming units; GC, growth control; DAP, daptomycin; CPT, ceftaroline.

TABLE 1.

Pharmacokinetic parameters for antibiotics used in ex vivo SEV models^a

	Target		Achieved
Antibiotic (dose)	Cmax (µg/mL)	t1/2 (h)	Cmax (µg/mL)	t1/2 (h)	AUC0–24 (µg*h/mL)
DAP (10 mg/kg)	141.1	8	144.7 ± 0.5	8.03 ± 0.1	1,772.3 ± 19.4
CPT (600 mg)	21.3	2.66	23.9	2.81 ± 0.09	82.1 ± 1.5

^a SEV, simulated endocardial vegetation; DAP, daptomycin; CPT, ceftaroline; Cmax, maximum concentration; t_1/2, half-life; AUC_0–24, area under the concentration curve over the 24-h dosing interval.

In the SEV models against DNS MRSA isolate C4, DAP 10 + CPT + the two-phage cocktail exerted both bactericidal and synergistic activities, with significant bioburden reductions down to the limit of detection of 2 log₁₀ CFU/g (−Δ 7.07 log₁₀ CFU/g) vs both DAP and phage cocktail monotherapy (P < 0.001). The addition of CPT offered no statistically significant improvement in bacterial killing at 168 h when compared to DAP + the two-phage cocktail (P = 0.263); however, in practical terms, DAP + CPT + the two-phage cocktail maintained bacterial counts below the limit of detection, whereas regrowth was observed with DAP + the two-phage cocktail (Fig. 3).

Against strain C37, DAP + CPT + the two-phage cocktail again demonstrated both bactericidal and synergistic activities, with significant bioburden reductions down to the detection limit of 2 log₁₀ CFU/g (−Δ 7.11 log₁₀ CFU/g) compared to antibiotic and phage monotherapy regimens (both P < 0.001) and DAP plus the two-phage cocktail (P = 0.018) (Fig. 4). No significant difference in bacterial killing at 168 h was identified between DAP monotherapy and DAP plus the two-phage cocktail (P = 0.296).

Post-SEV susceptibility testing

Antibiotic MIC

To evaluate possible changes in antibiotic MICs during the SEV exposures, we assessed the antibiotic MICs of the DNS MRSA C4 and C37 isolates at baseline and at the end of therapy in triplicate. First, we plated 100 μL of the SEV sample onto tryptic soy agar (TSA) plates containing 3× the baseline MIC of each drug of interest (DAP or CPT) tested in the model. Plates were examined for growth after 24 and 48 h of incubation at 37°C. We then followed up with formal broth microdilution MICs for each drug, following Clinical and Laboratory Standards Institute (CLSI) guidelines (24). If the end-of-therapy sample demonstrated an MIC change of ≥2 dilutions from the baseline (either MIC elevation or reduction), then samples were assessed for emergence of resistance in a backward stepwise manner from 144 h to earlier time points until a ≤1 dilution change in MIC was identified for the sample. At 168 h for both models, all SEV samples demonstrated stable DAP and CPT MICs compared to baseline (Table 2).

TABLE 2.

Antibiotic MIC for daptomycin-nonsusceptible MRSA isolates C4 and C37 at baseline and in 168-h SEV samples^a

	Antibiotic MIC (μg/mL)
	C4		C37
	DAP	CPT	DAP	CPT
Baseline	4	0.5	4	0.25
	Post-SEV model MIC (168-h sample)
Growth control	4	0.5	4	0.25
Intesti13 + Sb-1	4	0.5	4	0.25
DAP 10 mg/kg	4	0.5	4	0.25
DAP 10 mg/kg plus Intesti13 + Sb-1	4	0.5	4	0.25
DAP 10 mg/kg plus CPT 600 mg Intesti13 + Sb-1	4	0.5	4	0.25

^a SEV, simulated endocardial vegetation; DAP, daptomycin; CPT, ceftaroline.

Phage susceptibility

We next evaluated whether there was a change in phage susceptibility at the end of therapy compared to baseline by defining the apparent frequency of resistance (Table 3). Since bacterial colonies may arise due to either fixed mutations or transient changes in gene expression, this initial screening was also followed by confirmatory testing of subcultured colonies in triplicate, as previously described for the two phages used in our SEV models (25). Phage resistance was identified in the end-of-therapy SEV samples for both phages when the model regimen consisted of phage cocktail monotherapy. In contrast, neither C4 nor C37 end-of-therapy isolates demonstrated reduced susceptibility to either phage following regimens that included DAP or DAP + CPT, suggesting that these PACs reduced the frequency of phage resistance in the SEV model.

TABLE 3.

Frequency of apparent resistance in daptomycin-nonsusceptible MRSA strains C4 and C37 sensitive to each phage^a,b,c

Strain	Treatment	Timepoint (h)	Colony count	Mean CFU/g in sample	Apparent FoR	LoD (1/CFU on plate)
C4
	No antibiotic
	Intesti13 + Sb-1	0	0	2.56 × 108	<LoD	4.15 × 108
		168	Est. 1,000	7.42 × 108	5.84 × 10−5	7.89 × 108
	DAP	0	0	2.89 × 108	<LoD	2.53 × 108
		168	523	5.21 × 108	5.84 × 10−7	6.47 × 108
	Intesti13 + Sb-1	0	0	4.78 × 108	<LoD	9.03 × 108
		168	167	6.34 × 108	<LoD	5.72 × 108
	DAP + CPT	0	0	9.15 × 108	<LoD	8.26 × 108
	Intesti13 + Sb-1	168	0	8.47 × 108	<LoD	3.68 × 108
C37
	No antibiotic
	Intesti13 + Sb-1	0	0	1.29 × 108	<LoD	7.49 × 108
		168	Est. 1,000	2.53 × 108	3.74 × 10−5	1.34 × 108
	DAP	0	0	4.62 × 108	<LoD	8.52 × 108
		168	449	5.93 × 108	4.15 × 10−7	3.74 × 108
	Intesti13 + Sb-1	0	0	3.47 × 108	<LoD	2.68 × 108
		168	307	6.28 × 108	4.15 × 10−7	4.93 × 108
	DAP + CPT	0	0	7.54 × 108	<LoD	2.65 × 108
	Intesti13 + Sb-1	168	0	8.39 × 108	<LoD	9.17 × 108

^a Abbreviations: CFU, colony-forming unit; FoR, frequency of resistance; LoD, limit of detection; DAP, daptomycin; CPT, ceftaroline.

^b FoR: [colonies that arise on the resistance plate]/[CFU per milliliter used to create the plate].

^c LoD: If <LoD is indicated, then the apparent FoR equals 0.

DISCUSSION

MRSA remains a formidable clinical challenge, given the limitations and decreasing efficacy of standard-of-care anti-MRSA antibiotics. The worrisome prevalence of DNS MRSA strains (26–28), combined with the fact that nearly one-third of MRSA infections treated with vancomycin fail, clearly indicates an urgency to develop novel, advanced therapeutic approaches. This is especially relevant in the context of high-inoculum infections such as IE.

Our investigation into PACs for treating DNS MRSA strains revealed several interesting findings. Using the ex vivo SEV model, evaluations of two clinically relevant DNS MRSA strains demonstrated significant synergistic activity when treated with selected phage cocktails in conjunction with anti-MRSA antibiotics. Specifically, DAP combined with CPT and the two-phage cocktail exhibited potent bactericidal activities against both MRSA strains, reducing bioburden significantly. Importantly, these experiments used humanized antibiotic PK, and subsequent tests indicated both stable antibiotic MICs and no emergent phage resistance when PAC therapies were employed. It is also noteworthy that we observed PAS using two Kayvirus phages. Other work has shown that Kayvirus phages, including Intesti13 and Sb-1, are less active in vitro against USA300 strains such as C4 than against other S. aureus lineages unless they carry a particular loss-of-function mutation (29) In our current study, the phages alone did not substantial reduce bacterial populations but still exhibited synergy with DAP and CPT.

A notable strength of this study was the use of the SEV ex vivo model. It addressed the critical need for translatability from experimental in vitro findings to mimic human microenvironments in vivo. Unlike in vitro experiments, which often neglect the intricate dynamics of the host microenvironment, the SEV model simulates the complexity of high bioburden contexts in combination with vital elements like human platelets, fibrin, fibrinogen-fibronectin, and host defense peptides. This results in a setting where MRSA, antibiotics, and phage interact in a milieu that more closely mimics conditions of actual human infection. Hence, the results obtained provide deeper insights into phage-antibiotic-microbial interactions within a realistic host setting. The anticipated progression of our work into an animal IE model will further strengthen similar research by incorporating even more diverse environmental factors, such as distinct tissue-specific dynamics, immune interactions, organ-specific penetration, and human endovascular mechanics (e.g., shear forces).

While our results are encouraging, it is imperative to acknowledge several limitations. The study was primarily focused on two DNS MRSA clinical strains, which might not encompass the broader spectrum of MRSA genetic diversity and resistance mechanisms. Additionally, while the SEV model offers a sophisticated representation of a high-inoculum bioburden that mimics IE, it still cannot fully capture the multifaceted interactions seen in living organisms, underscoring the need for subsequent in vivo evaluations.

Building upon these results, future research endeavors should look into expanding the range of MRSA strains studied to ensure the broader applicability of the findings. Furthermore, in-depth investigations into the molecular mechanisms driving the observed synergism will provide valuable insights that can guide the optimization of PACs. Additionally, as the National Action Plan emphasizes, efforts should also concentrate on ways to prevent the emergence and spread of antibiotic-resistant bacteria, thus augmenting the dual approach of innovative treatments and preventative measures.

MATERIALS AND METHODS

Bacterial strains

The DNS MRSA strains selected for evaluation in the studies herein were initially isolated from bacteremic patients and obtained from the Cubist Pharmaceuticals Isolate Collection (30). These strains are archived in the Rybak Anti-Infective Research Laboratory and have been periodically confirmed for viability and retention of their DNS phenotypes by MIC testing. Of note, these strains have maintained their DNS phenotype on prolonged storage at −80°C over many years. Importantly, this strain set encompasses the clonal complex genotypes currently in most widespread US circulation (USA100 and 300 clonotypes).

Bacteriophages

The selection of S. aureus bacteriophages Intesti13 and Sb-1 for this study followed methods outlined in prior research (31, 32). In summary, the choice of the two-phage cocktail for PAC was based on previously published data from our group demonstrating distinct phage host ranges, evidence that spontaneous resistance to one phage did not necessarily lead to cross-resistance with other phages in the cocktails, and demonstrating bacterial growth inhibition, both with and without antibiotics present (15, 16).

Intesti13 and Sb-1 originated from bacteriophage solutions procured from the Georgia Eliava Institute (Tbilisi, Georgia) and were cultivated on S. aureus ATCC 19685 and D712, respectively.

Antimicrobials

Daptomycin was commercially obtained from Xellia Pharmaceuticals (Buffalo Grove, IL, USA), and CPT analytical powder was sourced from AbbVie, Inc. (North Chicago, IL, USA). Mueller-Hinton broth (MHB) (Difco, Detroit, MI, USA) supplemented with calcium and magnesium at concentration of 50 µg/mL and 12.5 µg/mL, respectively, was utilized for the modified CB MIC and 24-h time-kill assays, while 75 µg/mL and 12.5 µg/mL, respectively, were utilized for 168-h SEV models as previously described (33). Colony counts were performed on TSA plates (Difco, Detroit, MI, USA).

Antibiotic susceptibility testing

Minimum inhibitory concentrations of study antimicrobials were determined in duplicate by manual broth microdilution at approximately 10⁶ CFU/mL according to the CLSI (24).

Modified checkerboard MIC assay

Antibiotic and phage synergy against DNS MRSA isolates C4 and C37 was initially evaluated using modified CB MIC analyzing the activity of antibiotics and phage in combination compared to their individual activities, as previously described (25). In brief, 100 µL of MHB was added to each well of a 96-well round-bottom microtiter plate. Next, 100 µL of DAP or CPT at twofold the MIC was added to column 1 and serially diluted to column 10, like MIC testing. For checkerboards analyzing synergy between two antibiotics, 100 µL of the other antibiotic was added to the first row in columns 1 to 10 and diluted twofold through row 7. For PAS screening, 10 µL of phage (MOI, 0.1) instead of antibiotic was added to the first row in columns 1 to 10 with phage dilutions performed 10-fold rather than 2-fold to achieve the desired range of MOI. Next, 100 µL of MHB inoculated with 10⁹ starting inoculum of test organism was then added to each well of the 96-well plate except for the last column. The plate was then incubated at 37°C for 24 h followed by spectrophotometric reading at an optical density of 570 nm. In triple-therapy checkerboards assessing the synergy of DAP plus CPT in the presence of constant phage, phage was added to each well, excluding controls, at a subinhibitory MOI determined in a single phage-antibiotic CB. Synergy, additive activity, and antagonism were defined as a calculated FIC index of ≤0.5, 1 to 4, and >4, respectively (25, 34)

Time-kill analyses

In vitro high-inoculum (10⁹ log₁₀ CFU/mL) time-kill assays experiments were carried out in microwell plates over 24 h as previously described (25) against DNS MRSA strains C4 and C37. DAP ± CPT was used, each at 0.5× MIC to recreate clinical treatment failure concentrations and the PAC at MOIs ranging from 0.01 to 1 based on CB MIC results. Calcium- and magnesium-supplemented MHB inoculated with 10⁹ log₁₀ CFU/mL (2 mL) was added to each well. Antibiotics DAP and CPT were introduced to their designated wells at 0.5× MIC. In wells containing PAC regimens, antibiotic was added first, followed directly by phage. At 0, 4, 8, and 24 h, 100-µL samples were ascetically removed from each well. Subsequent steps involved eliminating antibiotic and phage carryover through two rounds of centrifugation, supernatant removal, and appropriate dilutions in 0.9% saline. Samples were then plated on TSA and incubated at 37°C for 24 h before bacterial colony counting (Scan 1200, Interscience for Microbiology, Saint Nom la Breteche, France; detection limit of 10² CFU/mL). Synergy and bactericidal activity were defined as a ≥2 log₁₀ CFU/mL kill compared to the next most effective regimen and a ≥3 log₁₀ CFU/mL reduction from baseline at 24 h. SPSS version 29.0 (IBM Corp., Armonk, NY, USA) software was utilized for statistical analysis. Assessment of significant differences between phage-antibiotic regimens in terms of bacterial killing metrics (i.e., log₁₀ CFU/mL reductions from time 0 to 24 h) was accomplished through analysis of variance (ANOVA) with Tukey’s post hoc test (P < 0.05).

Ex vivo PK/PD model

Ex vivo SEV PK/PD models were conducted, and SEV clots were prepared as previously documented (7, 8). The models were incubated at 37°C for the duration of each 168-h model. MHB was cycled into and out of each model at a rate to simulate antibiotic half-life, as appropriate, with humanized doses of DAP 10 mg/kg and CPT 600 mg injected into each model every 24 and 8 h, respectively. When included in the regimens, each phage was injected into the models every 24 h at an MOI of 1. Once SEV samples were removed in duplicate from each model (total of four) at hour time points 0, 4, 8, 24, 32, 48, 72, 96, 120, 144, and 168, each SEV clot was homogenized and then centrifuged twice, with supernatant replaced with normal saline following each centrifuge cycle to eliminate antibiotic and phage carryover.

Pharmacokinetic analysis

Pharmacokinetic samples were obtained in duplicate through the injection port of each infection model at 0, 4, 8, 24, 32, 48, 72, 96, 120, 144, and 168 h for verification of target antibiotic concentrations as previously described (35). All samples were stored at −80°C until ready for analysis. CPT concentrations were determined by bioassay using Bacillus subtilis ATCC 6633 (36). Blank 1/4-in disks were spotted with 10 mL of standard concentrations or samples. Each standard was tested in duplicate by placing the disk on agar plates (antibiotic medium 11) inoculated with a 0.5 McFarland suspension of the test organism. This assay demonstrated an intraday coefficient of variance of less than 4.7% for high, medium, and low broth standards. The plates were incubated for 24 h at 37°C, at which time the zone sizes were measured using a ProtoCOL plate reader (Microbiology International, Frederick, MD, USA). DAP concentrations were determined using a validated high-performance liquid chromatography assay that conforms to the guidelines set forth by the College of American Pathologists and demonstrated an intraday coefficient of variance of less than 2% for high, medium, and low standards. The half-lives, peak concentrations, and area under the curve (AUC) (by trapezoidal method) were determined as appropriate for all antimicrobials utilizing PK Analyst software (version 1.10; MicroMath Scientific Software, Salt Lake City, UT, USA).

Post-SEV antibiotic and phage susceptibility testing

Antibiotic susceptibility

Changes in antibiotic susceptibility compared to baseline were evaluated in each 168-h SEV sample using antibiotic-embedded agar and broth microdilution. For antibiotic-embedded agar, we plated 100 μL of the 168-h SEV samples onto individual TSA plates containing threefold the baseline MIC of the drug (DAP or CPT) tested in the model. Plates were examined for growth after 24 and 48 h of incubation at 37°C. For broth microdilution, we evaluated 168-h SEV sample antibiotic MICs according to CLSI guidelines (24). For either assay, if samples demonstrated MIC changes of >2 dilutions from baseline (elevation or reduction in MIC), then they were passed for a 3-day consecutive period with MIC testing each day. For samples maintaining the >2 dilution change in MIC from baseline following the 3-day pass, additional SEV samples were assessed for resistance in a backward stepwise manner from 144 h to earlier time points until a ≤1 dilution change in MIC was identified for the sample. Antibiotics tested included those used in the SEV models (e.g., daptomycin and ceftaroline).

Phage susceptibility

The development of phage resistance for phages used in the SEV models (e.g., Intesti13 and Sb-1) was tested in triplicate populations of the post-SEV C4 and C37 strains in a manner similar to that previously described (25). Prior to SEV exposure, the frequency of spontaneous phage resistance in C4 and C37 populations was estimated to be approximately 10⁻⁹ (22). First, surviving colonies were streak purified on TSA plates followed by an evaluation of phage sensitivity using the double-drop method on TSA plates as previously described (22). The apparent frequency of resistance was calculated as the number of colonies on each test plate divided by the bacterial replicate number.

Statistical analysis

Statistical analyses were conducted with SPSS version 29 (IBM Corp., Armonk, NY, USA) software and Prism software version 9.2.0 (GraphPad, La Jolla, CA, USA). Bacterial burden reduction differences between phage and/or antibiotic regimens were compared by one-way ANOVA with Tukey’s post hoc test. A P value of <0.05 was considered significant. PK Analyst software version 1.10 (MicroMath, Salt Lake City, UT, USA) was used for the pharmacokinetic analysis of antibiotics used in the SEV models (e.g., daptomycin and ceftaroline) including t_1/2, total C_max concentrations, and AUC_0–24.