Staphylococcus aureus AbcA transporter enhances persister formation under β-lactam exposure

ABSTRACT

We evaluated the role of Staphylococcus aureus AbcA transporter in bacterial persistence and survival following exposure to the bactericidal agents nafcillin and oxacillin at both the population and single-cell levels. We show that AbcA overexpression resulted in resistance to nafcillin but not oxacillin. Using distinct fluorescent reporters of cell viability and AbcA expression, we found that over 6–14 hours of persistence formation, the proportion of AbcA reporter-expressing cells assessed by confocal microscopy increased sixfold as cell viability reporters decreased. Similarly, single-cell analysis in a high-throughput microfluidic system found a strong correspondence between antibiotic exposure and AbcA reporter expression. Persister cells grown in the absence of antibiotics showed neither an increase in nafcillin MIC nor in abcA transcript levels, indicating that survival was not associated with stable mutational resistance or abcA overexpression. Furthermore, persister cell levels on exposure to 1×MIC and 25×MIC of nafcillin decreased in an abcA knockout mutant. Survivors of nafcillin and oxacillin treatment overexpressed transporter AbcA, contributing to an enrichment of the number of persisters during treatment with pump-substrate nafcillin but not with pump-non-substrate oxacillin, indicating that efflux pump expression can contribute selectively to the survival of a persister population.

INTRODUCTION

Staphylococcus aureus is a common human pathogen that causes diseases ranging from minor skin infections to life-threatening conditions, such as endocarditis and sepsis. The emergence of multidrug-resistant strains, such as methicillin-resistant S. aureus (MRSA), has created a healthcare challenge (1, 2). Many factors contribute to antibiotic treatment failures of staphylococcal infections, ranging from genotypic bacterial acquisition of resistance by mutation or gene acquisition to the phenotypic development of antibiotic tolerance and the occurrence of persister cells. Tolerance to antibiotics causes an increase in the minimum bactericidal concentration, without a change in the minimum inhibitory concentration (MIC) and is the result of bacterial acquisition of genetic mutations or bacterial growth restriction under certain environmental conditions. Studies have shown that stationary phase S. aureus are mostly tolerant to antibiotics (3–5). Persister cells are a cellular subpopulation that survives high levels of antibiotics without genetic changes (3, 6–9). Such cells are a phenotypic subpopulation that arises during both the exponential and stationary growth phases and in both planktonic and biofilm states. S. aureus persisters are metabolically active and reside in a non-dividing state that is reversible upon removal of the antibiotic stress without a resistance phenotype (7–9).

Treatment of methicillin-susceptible S. aureus (MSSA) in bloodstream infections commonly includes the use of β-lactams such as penicillin, oxacillin, or nafcillin, as a recommended definitive therapy. Nafcillin and oxacillin are penicillinase-resistant penicillin antibiotics that resist staphylococcal β-lactamase. Nafcillin has been recommended for empirical therapy of possible late-onset sepsis in infants without a history of MRSA colonization or infection (10). In contrast, oxacillin (together with rifampicin, oritavancin, linezolid, and moxifloxacin) is a promising treatment against intracellular S. aureus (8, 9, 11). Frequent use of antibiotics has resulted in an increase in S. aureus resistance to multiple antibiotics, including β-lactams. β-lactam drug resistance in S. aureus is caused by several mechanisms that include altered drug targets (penicillin-binding-protein PBP2A), enzymatic drug inactivation (β-lactamase), altered drug accessibility, and overproduced multidrug efflux pumps (12).

Efflux pumps and transporters play an important role in S. aureus antibiotic resistance in decreasing the intracellular concentration of antibiotics by means of the active efflux of drug compounds. S. aureus AbcA is an ATP-dependent ABC transporter, and the genetic regulation of abcA is controlled by a 420-bp overlapping promoter and regulatory region located between abcA and pbp4, which are independently and divergently transcribed (13). The transcription of abcA is controlled by at least four global regulators, MgrA, SarA, SarZ, and Rot, which bind to the 420-bp overlapping promoter region of abcA and pbp4. The involvement and the complexity of multiple regulators in the control of the abcA expression depend on the environmental conditions that trigger changes in abcA expression such as pH, oxidative stress, iron limitation, nutrient limitation, and antibiotic exposure. AbcA was previously reported to extrude β-lactam drugs, including nafcillin, and caused a low-level resistance to β-lactams when overexpressed. AbcA was also shown to secrete phenol-soluble modulins and contribute to S. aureus virulence (14–17). The pbp4 gene encodes the transpeptidase PBP4, which can also cause resistance to β-lactams when overproduced (18).

The role of S. aureus efflux pumps and transporters in persister cell formation is not well understood. A recent study demonstrated that intracellular S. aureus persisters undergo major transcriptomic reprogramming, which allows them to become less affected by antibiotics while persisting inside the host cell. This reprogramming yields a network of adaptive responses, including tolerance to multiple antibiotics acting on unrelated bacterial targets (4). Since efflux pumps and transporters often have broad substrate profiles and have regulated expression integrated into cellular physiology, we undertook to explore their involvement in the persister cell phenomenon in S. aureus exposed to β-lactam antibiotics.

Specifically, we evaluated the role of the S. aureus AbcA transporter in bacterial survival and persistence in the presence of 1×MIC and 25×MIC of nafcillin, an AbcA substrate, compared with oxacillin, a non-substrate. We find that, in the presence of either nafcillin or oxacillin, S. aureus displayed a biphasic killing curve, with a small subpopulation of S. aureus that not only survived but eventually reached a living, but non-dividing, physiological state. As expected for persisters, surviving cells then cultured in the absence of antibiotics did not exhibit an increased MIC or increased abcA expression. We then show that S. aureus lacking the AbcA transporter exhibit a lower persister population, whereas an AbcA overexpressor strain exhibited a higher persister population with nafcillin but not oxacillin exposure. When observed at the single-cell level, we demonstrated that the S. aureus survivors overexpressed AbcA on exposure to both nafcillin and oxacillin, but only exposure to nafcillin led to an enrichment of survivors over time.

These data indicated that AbcA overexpression can contribute selectively to the survival of a persister population when exposed to the pump-substrate nafcillin but not to the pump-non-substrate oxacillin.

RESULTS

The AbcA transporter reduced S. aureus susceptibility to nafcillin when overexpressed and contributed to increase the number of persister phase cells

We determined the rate of S. aureus killing when exposed to nafcillin and oxacillin at 1×MIC and 25×MIC over a period of 0–24 hours. Figure 1 shows the growth curves and kill curves of the wild-type S. aureus strain RN6390 and the RN6390 overexpressing the AbcA transporter from a plasmid [RN6390(pLI50-abcA)].

Fig 1.

S. aureus RN6390 time-kill curves following exposure to nafcillin and oxacillin at 1×MIC and 25×MIC. S. aureus RN6390, RN6390(pLI50), and RN6390(pLI50-abcA) (OD₆₀₀ ~ 0.6, 10⁷ CFU/mL) were exposed to nafcillin and oxacillin at 1× and 25×MIC for a period of 24 hours. The experiments were repeated three times with three biological samples for each antibiotic. The figure represents the log₁₀ of S. aureus CFU/mL with or without antibiotics (error bars = log₁₀CFU/mL ± SD for each antibiotic). RN6390(pLI50); RN6390 with empty plasmid, served as a control of AbcA expression. RN6390(AbcA), AbcA overexpressor, RN6390 with plasmid construct pLI50-abcA. (A) S. aureus RN6390 kill curves following exposure to nafcillin and oxacillin at 1×MIC (0.25 µg/mL). RN6390 had the same MIC values for nafcillin and oxacillin. (B) S. aureus RN6390 kill curves following exposure to nafcillin and oxacillin at 25×MIC (6.25 µg/mL). (C) Exposure of RN6390(pLI50) and RN6390(pLI50-abcA) to nafcillin and oxacillin at 1×MIC. RN6390(pLI50) was exposed to 0.25 µg/mL of nafcillin (1×MIC), and RN6390(pLI50-abcA) was exposed to 1 µg/mL of nafcillin (1×MIC). Both RN6390(pLI50) and RN6390(pLI50-abcA) were exposed to 0.25 µg/mL of oxacillin (1×MIC). (D) Exposure of RN6390(pLI50) and RN6390(pLI50-abcA) to nafcillin and oxacillin at 25×MIC. RN6390(pLI50) was exposed to 6.25 µg/mL of nafcillin and oxacillin (25×MIC), and RN6390(pLI50-abcA) was exposed to 25 µg/mL of nafcillin and 6.25 µg/mL of oxacillin(25×MIC).

RN6390 exhibited an approximate 3- to 4-log₁₀ reduction in viable cells over the first 6 hours of exposure to nafcillin and oxacillin at 1×MIC (Fig. 1A) and an approximate 5-log₁₀ reduction at 25×MIC (Fig. 1B), followed at 6–24 hours in both cases by a more limited or no further reduction in viable cells, a time period in which persisters are established.

RN6390(pLI50-abcA) compared to RN6390 transformed with the empty vector [RN6390(pLI50)] had the greatest effect with exposure to nafcillin at both 1× (Fig. 1C) and 25×MIC (Fig. 1D), generating a higher level of viable cells both in the initial 6 hours and in the 6–24-hour persister phase, which exhibited a continued gradual decrease in viable cells while remaining 1- to 2-log₁₀ above the similar levels of viable cells seen with nafcillin-exposed cells RN6390(pLI50) and oxacillin-exposed cells RN6390(pLI50) and RN6390(pLI50-abcA).

Thus, abcA contributed to a higher level of persister phase cells in the presence of nafcillin but not oxacillin. The differences between the number of survivors RN6390(pLI50-abcA) compared to that of RN6390(pLI50) following exposure to nafcillin at 1× and 25×MIC were significant as determined by a one-way analysis of variance (ANOVA) with a pairwise t-test and Bonferroni adjustment (P < 0.05).

To assess whether the native expression of abcA also affected persister cell subpopulations in the presence of nafcillin and oxacillin, we attempted to create an abcA deletion mutant in RN6390 but were unsuccessful. Instead, we used the reference strain Newman (NM) and its previously reported abcA knock-out mutant (Newman ∆abcA::pSF151) that was named MN0599, kindly provided by the Kaito laboratory (17). S. aureus Newman was transformed with plasmids pLI50 and pLI50-abcA to serve as a control [NM(pLI50)] and an AbcA overexpressor [NM(pLI50-abcA)], respectively. The MICs of Newman mirrored those of RN6390, and MN0599 had the same MICs for both nafcillin and oxacillin at 0.25 µg/mL. Likewise, NM(pLI50-abcA) showed an MIC of 1 µg/mL for nafcillin and 0.25 µg/mL for oxacillin.

Figure 2 shows the growth curves and kill curves of the wild-type S. aureus strain Newman, mutant MN0599, and NM(pLI50-abcA).

Fig 2.

S. aureus Newman time-kill curves following exposure to nafcillin and oxacillin at 1×MIC and 25×MIC. S. aureus Newman, abcA mutant MN0599, NM(pLI50), and NM(pLI50-abcA) (OD₆₀₀ ~ 0.6, 10⁷ CFU/mL) were exposed to nafcillin and oxacillin at 1× and 25×MIC for a period of 24 hours. The figure represents the log₁₀ of S. aureus CFU/mL with or without antibiotics (error bars = log₁₀CFU/mL ± SD for each antibiotic). The experiments were done in triplicate with three independent biological samples. Chloramphenicol at 10 µg/mL (Cm10) was added to maintain the plasmid. (A) S. aureus Newman and abcA knockout mutant MN0599 kill curves following exposure to nafcillin and oxacillin at 1×MIC (0.25 µg/mL). (B) S. aureus Newman and abcA knockout mutant MN0599 kill curves following exposure to nafcillin and oxacillin at 25×MIC (6.25 µg/mL). (C) S. aureus Newman NM(pLI50) kill curves following exposure to 0.25 µg/mL (1×MIC) of nafcillin and oxacillin, and NM(pLI50-abcA) kill curves following exposure to nafcillin (1µg/mL, 1×MIC) and oxacillin (0.25 µg/mL, 1×MIC). (D) S. aureus Newman NM(pLI50) kill curves following exposure to 6.25 µg/mL (25×MIC) of nafcillin and oxacillin, and NM(pLI50-abcA) kill curves following exposure to nafcillin (25 µg/mL, 25×MIC) and oxacillin (6.25 µg/mL, 25×MIC). NM(pLI50), Newman strain with plasmid pLI50 and NM(AbcA), Newman strain with plasmid pLI50-abcA.

MN0599 exposed to nafcillin at 1×MIC exhibited an initial more rapid loss of viable cells and an approximate 0.5- to 1-log₁₀ reduction in viable cells in the 6–24 hours persister phase relative to the Newman parental strain on nafcillin exposure and both strains on oxacillin exposure (Fig. 2A). The differences between the number of survivors of MN0599 compared to that of Newman following exposure to nafcillin at 1×MIC were significant as determined by a one-way ANOVA with a pairwise t-test and Bonferroni adjustment (P < 0.05).

At 25×MIC, the initial loss of viability was similar for all strains in both nafcillin and oxacillin, and in the persister phase (6–24 hours), a slight reduction (non-significant) in viable cell numbers for MN0599 was observed (Fig. 2B). NM(pLI50-abcA) generated results similar to those of RN6390(pLI50-abcA) (Fig. 2C and D) (P < 0.05). Thus, the absence of abcA reduces the level of surviving persister cells in the presence of nafcillin but has little effect in the presence of oxacillin, supporting the role of abcA as a contributor to the persister population in the presence of a pump substrate when expressed from wild-type cells.

MIC and abcA expression of S. aureus persisters remained unchanged upon removal of nafcillin at 1×MIC

Persister cells are defined by the absence of stable resistance to the exposed antimicrobials. We thus plated RN6390 cells surviving at 8 hours (3.3 × 10⁴ CFU/mL) and 20 hours (1.2 × 10³ CFU/mL) of 1×MIC of nafcillin exposure on drug-free TSB agar and tested 100 separate colonies by replica plating on agar with or without 1×MIC nafcillin (0.25 µg/mL) for nafcillin MIC testing and abcA transcript levels by RT-PCR. We found 16 replicates from 100 colonies at 8-hour exposure and 3 replicates from 100 colonies at 20-hour exposure that grew on nafcillin-supplemented plates but only at 48 hours. All colonies (8- and 20-hour exposure) grew within 24 hours on antibiotic-free plates. Testing of these 19 replicates from the drug-free agar plates found unchanged nafcillin MICs (0.25 µg/mL) and no increase in transcript levels of abcA relative to those for RN6390 (0.38–0.42) (Table 1). Thus, no stable resistance phenotypes among 100 sampled cells contributed to the survival of cells in the persister phase of exposure to nafcillin.

TABLE 1.

MICs of nafcillin and oxacillin and the abcA relative transcripts of S. aureus RN6390, abcA overexpressor, and persisters

	MIC (μg/mL)		Real-time qRT-PCRs
	Nafcillin	Oxacillin	Relative transcription of abcA (mean FC ± SEM)a
S. aureus
RN6390 (pLI50)	0.25	0.25	1.0
RN6390 (pLI50-abcA)	1 (4×)	0.25	6.0 ± 0.1
S. aureus persistersb
Nafcillin-8-hour (16 colonies)	0.25	0.25	0.39–0.42 ± 0.03
Nafcillin-20-hour (3 colonies)	0.25	0.25	0.38–0.40 ± 0.01

^a Relative expressions of abcA of S. aureus RN6390 were represented as fold change (FC) values ± SEM. The real-time assays were done in triplicate with three separate biological samples. The FC of the persisters represents a range of mean values of three biological samples.

^b Nafcillin-8-hour and nafcillin-20-hour represent the number of colonies grown on TSA plates and on replicate TSA plates with nafcillin at 0.25 µg/mL. These colonies originated from the 100 colonies selected from the CFU/mL of RN6390 grown on TSA plates after growth in TSB supplemented with nafcillin at 1×MIC (0.25 µg/mL) for 8 hours and 20 hours, respectively.

Nafcillin at 1×MIC increased the number of AbcA overexpressor cells as observed by confocal imaging

Initial exposure of exponentially growing S. aureus RN6390 to nafcillin or oxacillin at 1× or 25×MIC for 1 hour resulted in modest increases in abcA transcript levels of 1.5- to 2.5-fold at the population level (Fig. 3). In order to evaluate AbcA expression over longer time periods of exposure to nafcillin or oxacillin and at the single-cell level, we constructed abcA translational fluorescent reporters for use with confocal fluorescence microscopy and a high-throughput single-cell microfluidics platform. As indirect markers of cell viability for comparison in confocal imaging, we used separate constitutively expressed fluorescent reporter plasmid pAH16 transformed into RN6390 (19, 20).

Fig 3.

Induction of abcA transcript levels by nafcillin and oxacillin at 1× and 25×MIC. RN6390 was exposed to nafcillin and oxacillin at 1× and 25×MIC for 1 hour. Then, quantitative real-time RT-PCR assays were performed to assess the level of abcA transcription. The relative transcript level of abcA was expressed as the fold change (FC) in abcA transcripts of bacteria exposed versus non-exposed to antibiotics. The assays were repeated three times with three different biological samples. The error bars represent the means of FC ± SEM for each assay. At 1×MIC after 1-hour induction, the differences (*) in the FC of abcA of RN6390 exposed to nafcillin (solid red) versus RN6390 non-exposed or exposed to oxacillin (blue) at 1×MIC for 1 hour were statistically significant as determined by a one-way ANOVA test (P < 0.05). At 25×MIC, the increase in the FC of abcA transcript of RN6390 exposed to nafcillin and oxacillin was statistically significant compared to the FC of abcA transcript of RN6390 non-exposed to antibiotic, as determined by an ANOVA test (P < 0.05). The differences in the FC of abcA transcripts between nafcillin and oxacillin at 25×MIC were not significant.

To analyze confocal imaging data, we counted the number of YFP-expressing cells (green) in six contiguous confocally imaged fields of view per well, each field of view had an average initial number of 1.6 × 10⁴ bacteria. The number of bacterial cells reported in the section below represented the average number of cells per field of view for the six chosen fields at each indicated time point and condition, starting at time 0 with 1.6 × 10⁴ bacteria (average number of cells = sum of all viable cells/six fields of view). The number of bacterial cells in the six fields was similar.

At time 0, all RN6390 transformed with the YFP-plasmid construct pAH16 [RN6390(pAH16)] expressed YFP as shown by the fluorescence image and by the differential interference contrast (DIC) image (Fig. 4, nafcillin A and B, time 0, too many to count). When 1×MIC nafcillin was added (0.25 µg/mL), RN6390(pAH16) decreased rapidly and approximately 100 RN6390(pAH16) remained at 8 hours (Fig. 4, -nafcillin A and B, time 8 hours, average number of cells = 100 per field of view for the six chosen fields).

Fig 4.

Confocal imaging of S. aureus exposed to nafcillin and oxacillin at 1×MIC. S. aureus RN6390 (10⁵ bacteria/well) carrying plasmid construct pAH16 (A and B, expressing YFP constitutively) or pRN-abcAp-yfp (C and D, expressing YFP by induction of abcA or spontaneous expression) were grown in TSB media supplemented with erythromycin or chloramphenicol at 10 µg/mL, respectively. Bacteria were exposed to nafcillin at 1×MIC (0.25 µg/mL) and then imaged at time points 0 and 8 hours. Six fields of view were selected per well and all fields of view showed similar DIC (A and C) and fluorescent observations (B and D). Shown are representative fields of view for each transformant at each time point. The experiments were done in triplicate with three biological samples. The number of viable bacterial cells was counted for each of the six fields of views and reported as the average number of each condition (sum of all counts/total six views). Top panel: (A) DIC image of S. aureus RN6390(pAH16). We observed a decrease in the numbers of S. aureus from 0 to 8 hours in the presence of nafcillin. (B) Fluorescent image of RN6390(pAH16). We observed a decrease of S. aureus expressing YFP (average 100 cells at 8 hours) that was in correlation with the DIC image in panel A during the same time frame. (C) DIC image of S. aureus RN6390(pRN-abcAp-yfp). We observed a decrease in the numbers of S. aureus from 0 to 8 hours in the presence of nafcillin. (D) Fluorescent image of S. aureus RN6390(pRN-abcAp-yfp). We observed an increase of S. aureus expressing YFP between 0 (20 cells at 0 hour) and 8 hours (120 cells at 8 hours) of nafcillin exposure. *Time = 0, nafcillin at MIC (0.25 µg/mL) was added to the S. aureus culture. Bottom panel: (A) DIC image of S. aureus RN6390(pAH16). We observed a decrease in the numbers of S. aureus from 0 to 8 hours in the presence of oxacillin. (B) Fluorescent image of RN6390(pAH16). We observed a decrease of S. aureus expressing YFP that was in correlation with the DIC image in panel A during the same time frame (50 cells at 8 hours). (C) DIC image of S. aureus RN6390(pRN-abcAp-yfp). We observed a decrease in the numbers of S. aureus from 0 to 8 hours in the presence of oxacillin. (D) Fluorescent image of S. aureus RN6390(pRN-abcAp-yfp). No visible YFP-expressing RN6390 could be detected at 8 hours of exposure. The same DIC and fluorescent images at time zero (T = 0) were shown for RN6390 (pAH16)–nafcillin 1×MIC and RN6390 (pAH16)–oxacillin 1×MIC (top and bottom panels, A and B, Time = 0). Both images (DIC and fluorescent) were used as internal experimental controls for both conditions. RN6390 (pAH16) constitutively expressed YFP and was used as a bacterial viability control of the experiments (DIC, number of bacteria; fluorescence, constitutive expression of YFP). The same DIC image at time zero (T = 0) was shown for RN6390 (pRN-abcAp-yfp)–nafcillin 1×MIC and RN6390 (pRN-abcAp-yfp)–oxacillin 1×MIC (top and bottom panels, C, Time = 0). The identical DIC images were used as internal experimental controls for the viability and the number of bacteria of the assay at time zero. The fluorescent image at time zero (Time = 0, D) of RN6390 (pRN-abcAp-yfp)–nafcillin 1×MIC was different from the fluorescent image at time zero (Time = 0, D) of RN6390 (pRN-abcAp-yfp)–oxacillin 1×MIC. RN6390 (pRN-abcAp-yfp) expressed YFP under induction by nafcillin or oxacillin at 1×MIC. The fluorescent bacteria (green dots) in the images (D, top and bottom panels) represented the bacteria that spontaneously expressed YFP in the S. aureus population represented in the DIC image (C, top and bottom panels). Two different fluorescent views were chosen to illustrate this observation.

In contrast, in separate wells and fields of view, few RN6390 that expressed YFP from an abcA promoter [RN6390(pRN-abcAp-yfp)] were observed at time 0 without antibiotic (average 20 live bacteria per field of view, six total fields, Fig. 4, nafcillin C and D, time 0). These YFP bacteria represented those with spontaneous increased expression of AbcA, representing a ratio of 20 cells per 1.6 × 10⁴ viable cells (ratio ~ 1.25 × 10⁻³) within the population of RN6390. When 1×MIC nafcillin (0.25 µg/mL) was added, RN6390(pRN-abcAp-yfp) also decreased with a rate similar to that of RN6390(pAH16), and ~120 YFP bacteria remained at 8 hours (Fig. 4, nafcillin C and D, time 8 hours, average number of cells = 120 per field of view, six total fields). The ratio of RN6390(pRN-abcAp-yfp) to RN6390(pAH16) was 120 induced-YFP to 100 expressed-YFP ~1.2, suggesting that survivors RN6390(pAH16) expressed AbcA (represented by YFP) were strongly enriched for AbcA-expressing cells.

Confocal imaging carried out at 6 hours provided similar changes in the number of survivors RN6390(pAH16) (average number of cells = 120 per field of view, six total fields) and RN6390(pRN-abcAp-yfp) (average number of cells = 110 per field of view, six total fields) which yielded a ratio of 0.96. The experiments were repeated three times with three separate biological samples.

In similar experiments using oxacillin at 1×MIC final concentration (0.25 µg/mL), RN6390(pAH16) was rapidly killed by oxacillin at 8 hours (~50 per field of view, six total fields) (Fig. 4, oxacillin A and B). At time 0, spontaneous RN6390(pRN-abcAp-yfp) were found (10 per field of view, six total fields), and none was found at 8 hours of oxacillin exposure (Fig. 4, oxacillin C and D). These experiments were done in triplicate using three independent biological samples. The differences in the numbers of bacteria RN6390(pAH16) and RN6390(pRN-abcAp-yfp) exposed to nafcillin at 0 hour versus 8 hours were statistically significant as determined by a one-way ANOVA with pairwise t-test and Bonferroni adjustment (P < 0.05).

Although there was enrichment of AbcA-expressing cells among persisters in the confocal experiments with nafcillin, it was not possible to identify a one-to-one correspondence of AbcA-expressing cells and persisters in the confocal imaging, since RN6390(pRN-abcAp-yfp) and RN6390(pAH16) were evaluated in parallel but different culture wells. To strengthen and complement the confocal imaging data, we next used a single-cell high-throughput microfluidic platform to study abcA overexpression in nafcillin- and oxacillin-surviving persisters through time. A combination of the two methods provided a more complete view of the persister phenomenon of this study with the single-cell technique allowing continuous observation of the evolution of a single S. aureus over a period of 20 hours.

Nafcillin and oxacillin at 1×MIC and nafcillin at 25×MIC increased the number of AbcA overexpressor cells as observed by a single-cell high-throughput microfluidic platform

For these experiments, we constructed a second fluorescent reporter of abcA, pTH-abcAp-cfp, which we introduced into RN6390 together with pAH16, enabling reporters of both abcA expression and cell viability in the same cell. Figure 5 followed cell growth, the number of viable bacteria, and the AbcA overexpression from the CFP reporter following the addition of nafcillin or oxacillin at 1×MIC (0.25 µg/mL) or 25×MIC (6.25 µg/mL).

Fig 5.

AbcA expression in a single-cell high-throughput microfluidic device. S. aureus RN6390 transformed with two plasmids (pAH16) and (pTH-abcAp-cfp) was grown in TSB media supplemented with erythromycin (10 µg/mL), chloramphenicol (10 µg/mL), and 800 µg/mL of F108 for 4 hours. Nafcillin or oxacillin at 1×MIC or 25×MIC was added at time 0. Single-cell high-throughput microfluidic experiments were carried out to follow the division of thousands of individual S. aureus cells over a period of 27 hours. (A) No antibiotic. Kymograph showing an untreated single lineage of RN6390(pAH16 + pTH-abcAp-cfp) cells through time. YFP fluorescence shows the viability of the cells, and the abcA-promoted CFP fluorescent protein shows the induction of the AbcA transporter. These bacteria represented the spontaneous AbcA expression. (B) With antibiotics. Percentage of CFP-overexpressing cells through time. The graphs showed the percentage of RN6390(pAH16 + pTH-abcAp-cfp) cells that overexpressed AbcA/CFP in the presence of nafcillin or oxacillin at 1×MIC or 25×MIC. In antibiotic-free media, an average of 1.24% cells of the RN6390(pAH16 + pTH-abcAp-cfp) showed CFP-overexpressing bacteria (blue fluorescence). This represented the spontaneous AbcA expression. Six hours after the addition of either nafcillin or oxacillin at 1×MIC (0.25 µg/mL) or 25×MIC (6.25 µg/mL), the number of AbcA/CFP cells increased from 0.81% (untreated control) to 15.45% (1×MIC nafcillin), 37.71% (1×MIC oxacillin), and 69.38% (25×MIC nafcillin). No increase in CFP-overexpressing cells was observed with 25×MIC oxacillin-treated cells. (C) With antibiotics. Number of YFP expressing RN6390(pAH16 + pTH-abcAp-cfp) through time. The viability of RN6390 was followed through the number of YFP cells over time in the absence or presence of antibiotics. Oxacillin was more efficient in bacterial killing than nafcillin at both 1×MIC (0.25 µg/mL) and 25×MIC (6.25 µg/mL) with the latter showing the highest efficiency in bacterial killing of all the antibiotic regimens. Cells treated with 1×MIC nafcillin resumed growth after the removal of the antibiotic at time 1,050 min.

We observed the growth of thousands of RN6390(pAH16 + pTH-abcAp-cfp) single cells through time with two different optical configurations, YFP from plasmid pAH16 and CFP from plasmid pTH-abcAp-cfp. The total CFP intensity of the cells varied among the different testing conditions. The mean total CFP intensity was 9.8 × 10⁴ ± 4.5 × 10⁴, 1.6 × 10⁵ ± 7.1 × 10⁴, 2.2 × 10⁵ ± 2.0 × 10⁵, 1.8 × 10⁵ ± 9.2 × 10⁴, or 1.0 × 10⁵ ± 4.7 × 10⁴ arbitrary units for the untreated control, the 1×MIC nafcillin, 25×MIC nafcillin, 1×MIC oxacillin, or 25×MIC oxacillin treatments, respectively. The highest CFP expression level was obtained when the cells were exposed to 25×MIC nafcillin (2.2 × 10⁵ ± 2.0 × 10⁵), which differed from the transcript levels trend found with early exposure to nafcillin and oxacillin (Fig. 3).

Figure 5B shows that in the absence of antibiotic, an average of 1.24% cells of the RN6390(pAH16 + pTH-abcAp-cfp) strain showed CFP-overexpression (blue fluorescence), representing AbcA spontaneous overexpression as seen in the kymograph of Fig. 5A.

The expression of AbcA as measured by the percentage of cells overexpressing the CFP reporter over the total number of YFP-expressing cells at each time point was increased by exposure to nafcillin (both 1× and 25×MIC) and oxacillin (1× but not 25×MIC) (Fig. 5B).

Specifically, at 8–16 hours (480–960 min) exposure to nafcillin (1×MIC) in the persister phase, we saw a ratio of 0.24, which meant ~24% of cells overexpressing AbcA/CFP out of ~24,600 viable cells (Fig. 5B and C, red line), which was less than the ratio of ~1 as estimated in the confocal experiments (Fig. 4). Furthermore, we observed that 70% of cells overexpressed AbcA/CFP out of ~9,000 and decreasing viable cells following exposure to 25×MIC nafcillin (ratio of 0.7) (Fig. 5B and C, dotted red line), and 50% of cells overexpressed AbcA/CFP out of ~8,000 and decreasing viable cells following exposure to 1×MIC oxacillin (ratio of 0.5) (Fig. 5B and C, blue line).

These findings differed from the confocal imaging observation, which used two different RN6390 transformants, RN6390(pAH16) for viability and RN6390(pRN-abcAp-yfp) for abcA expression by proxy. The observations done with confocal imaging gave a general view of two different populations of RN6390 cells that were separately exposed to nafcillin and oxacillin at 1×MIC for 8 hours. The ratio RN6390(pRN-abcAp-yfp)/RN6390(pAH16) from confocal imaging was higher than the ratio found with the single-cell method (0.96 versus 0.24) following exposure to nafcillin at 1×MIC, while no clear signal of RN6390(pRN-abcAp-yfp) was detected in confocal imaging following exposure to oxacillin at 1×MIC for 8 hours.

These data emphasized the advantage of having two plasmid constructs in the same RN6390 for comparison accuracy. The single-cell technique complemented the confocal imaging while providing a more detailed view of the phenomenon of persisters when S. aureus was exposed to different antibiotics.

Notably, at 1×MIC, despite a greater percentage of CFP-expressing cells on exposure to oxacillin than nafcillin, surviving cells as measured by the YFP reporter were lower with oxacillin (Fig. 5C), a finding in keeping with nafcillin but not oxacillin being a substrate for AbcA. Additionally, AbcA overexpression was sustained for the whole duration of the 1×MIC nafcillin treatment but decreased during both, 25×MIC nafcillin and 1×MIC oxacillin treatments before the antibiotic was removed at time 1,050 min. In accordance with this observation, 1×MIC nafcillin was the only antibiotic condition that resumed growth after the antibiotic was removed. These findings suggested that at 25×MIC nafcillin and 1×MIC oxacillin, the AbcA transporter was overwhelmed by the antibiotics.

These data highlighted that extended exposure to nafcillin (more than exposure to oxacillin) after the establishment of a persister population enriched surviving cells that overexpressed AbcA. These findings supported the role of functional drug efflux in enhancing a persister population.

DISCUSSION

The mechanisms behind the phenomenon of S. aureus persisters have been previously investigated but are still not fully understood with studies suggesting a link between antibiotic tolerance and persistence (3, 21). Several S. aureus transporters were known to be implicated in antibiotic resistance, bacterial virulence, and adaptation responses. The presence of a phenotypic subpopulation of S. aureus variants, which survive a high level of antibiotic exposure strongly suggested a possible involvement of S. aureus transporters in the phenomenon of persistence (22, 23). S. aureus AbcA transporter was found to confer β-lactam resistance and also contributed to starvation response (14, 17). We previously demonstrated that overexpression of AbcA caused a fourfold increase in the MIC of nafcillin but no change in the MIC of oxacillin (14). Oxacillin has been used in a study involving time-kill curves of intracellular S. aureus infecting macrophages (4).

In this study, we show that time-kill curves performed on S. aureus RN6390 and Newman strains, using both low and high concentrations of nafcillin and oxacillin (1×MIC and 25×MIC), yield conventional biphasic killing curves, with a rapid phase that eliminated a substantial majority of antibiotic-susceptible S. aureus over a relatively short period of time, followed by a slower killing or stabilization phase that lasted for an 18-hour period populated by persisters; these persisters did not have a resistance phenotype after the removal of the antibiotics.

Overexpression of AbcA from a plasmid allowed a higher number of cells to survive after exposure to nafcillin at 1× and 25×MIC when compared with wild-type S. aureus. However, AbcA overexpression did not improve the survival of the bacteria when exposed to oxacillin at 1×MIC and 25×MIC. We repeated the experiments with the MSSA Newman strain and the Newman AbcA overexpressor and obtained similar results. A lack of AbcA transporter increased the susceptibility of the abcA mutant MN0599 to nafcillin killing at concentrations equal to 1×MIC after 6 hours of exposure, but no additional susceptibility to oxacillin at 1×MIC was detected for the mutant MN0599. At 25×MIC, the effects of nafcillin and oxacillin on Newman and MN0599 followed the same patterns with a smaller magnitude. Thus, increased AbcA expression in the presence of its substrate nafcillin enhances persister numbers, whereas its increased expression in the presence of oxacillin, a non-substrate, does not, indicating that it is the efflux activity of AbcA and not other functions that enhance persister levels.

Since short-term exposure to nafcillin and oxacillin resulted in increased transcript levels of abcA at the population level, we assessed the relationship between increased AbcA expression and cell survival at the single-cell level using fluorescent reporters by confocal microscopy and a high-throughput microfluidic platform. As shown in the data obtained through the two techniques, we observed a more detailed view of the phenomenon of persisters when S. aureus was exposed to different antibiotics. Further attempts to perform confocal imaging using RN6390 with the two plasmids and to identify the cause of the differences in oxacillin data, as well as testing other β-lactams using the single-cell microfluidic platform are underway.

Notably, in the absence of drug exposure, a subset of cells (1.24% of the population) was seen to spontaneously overexpress AbcA, in keeping with findings of spontaneous expression of other efflux pumps (24, 25). Due to the cell division modality of S. aureus, mother cells at the end of the microfluidic trenches could lose their position; therefore, it was not possible to determine if the AbcA/CFP overexpression was transient using the microfluidics platform that enabled nearly continuous monitoring of single-cell fluorescence reporters. On exposure to nafcillin or oxacillin, single-cell reporter of cell viability decreased over time as expected. Remarkably, AbcA-expressing cells increased initially, albeit still within a subset of the population, indicating that induction related to drug exposure was not uniform across all cells. Strikingly, in the case of nafcillin but not oxacillin exposure at extended persister-phase time periods, the proportion of AbcA-expressing cells to viable cells increased substantially (from 0.81% up to 24% following 1×MIC nafcillin exposure for 17 hours). This difference likely reflects the substrate profile of AbcA, which includes nafcillin but not oxacillin, and thus differences in selective enrichment of AbcA-expressing cells between the otherwise similar β-lactams. In future studies, we will test other β-lactams (cefazolin, cefotaxime, and ceftaroline) to determine if AbcA-expressing cells would also affect S. aureus viability when exposed to these antibiotics.

Our data show that the AbcA transporter exhibits spontaneous expression, as has been shown for other transporters, and that exposure to active β-lactams increases the subpopulation of expressing cells. Its effects on β-lactam activity differ for nafcillin and oxacillin with an effect on nafcillin MIC correlating with an enrichment of single AbcA-expressing cells in the persister phase of drug exposure greater than seen with oxacillin. This effect can also be seen as a partial effect in kill-curve experiments, indicating that although AbcA enhances persister levels, other important factors such as ATP depletion, regulation of energy metabolism via the activity of the TCA cycle, nutrient starvation, as well as deprivation of extracellular magnesium may also contribute to formation and survival of persisters on extended drug exposure (26–28). Furthermore, the abcA gene can also be induced by environmental factors such as a nutrient limitation condition, which could contribute to antibiotic tolerance in a subpopulation of S. aureus persister cells. Future studies are underway to investigate these persister-inducing conditions and their impact on the antibiotic tolerance of persister cells.

In conclusion, this study showed that survivors of nafcillin and oxacillin treatment overexpressed transporter AbcA and contributed to an enrichment of the number of persisters during treatment with pump-substrate nafcillin but not with pump-non-substrate oxacillin. These findings highlighted that efflux pump expression can contribute selectively to the survival of a persister population.

MATERIALS AND METHODS

Bacterial strains and growth conditions

The bacterial strains, plasmids, and primers used in this study are listed in Table 2.

TABLE 2.

Bacterial strains, plasmids, and primers used in this study

Strains, plasmids, and primers	Genotypes or relevant characteristic(s)	Reference or source
S. aureus
RN6390	Wild type	(29)
Newman	Laboratory strain, high level of clumping factor	(17)
MN0599	Newman ΔabcA::pSF151	(17)
RN6390 (pLI50-abcA)	abcA-overexpressor, CmR	This study
RN6390 (pAH16)	sarA P1-dependent YFP with sod RBS, ErmR	(19, 20)
RN6390 (pRN-abcA/sodp-yfp)	abcApromoter-dependent YFP with sod RBS, CmR	This study
RN6390 (pTH-abcA/sodp-cfp)	abcApromoter-dependent CFP with sod RBS, CmR	This study
NM (pLI50-abcA)	abcA-overexpressor, CmR	This study
E. coli
DH5α	F- Φ80lacZΔM15 Δ(lacZYA-argF) U169 recA1 endA1	Life Technologies
	hsdR17(rk-, mk+) phoA supE44 thi-1 gyrA96 relA1 λ-
Plasmid
pLI50	Shuttle plasmid E. coli-S. aureus, CmR	(29)
pAH16	Plasmid sarA P1-dependent YFP with sod RBS, ErmR	(19)
pRN12	Shuttle vector E. coli-S. aureus, CmR	(30)
	Expression of sGFP from the sarA-P1 promoter
pTH2	Shuttle vector E. coli-S. aureus, CmR	(30)
	Expression of CFP from the sarA-P1 promoter
Primers for real-time RT-PCR assays
gmk	Forward 5′TCAGGACCATCTGGAGTAGGTAAAG 3′
	Reverse 3′CAAATGCGTGAAGGTGAAGTTGATG 5′
abcA	Forward 5′CAAGAACCTATTGAACCGACAGAA 3′
	Reverse 3′TGTGTCGTTCCAAATCCCAC 5′
yfp	Forward 5′GTCAGTGGAGAGGGTGAAGG 3′
	Reverse 3′CCATGCCCGAAGGTTATGTA 5′
cfp	Forward 5′AGTGGAGAGGGTGAAGGTGA 3′
	Reverse 3′ATGCTTTGCGAGATACCCAG 5′
Primers for abcA overexpressor
abcA-F (SphI)	Forward 5′ATATGCATGCATGAAACGAGAAAATC 3′
Primers for plasmid constructs pRN-abcAp-yfp and pTH-abcAp-cfp
abcA-pF (NheI)	Forward 5′ ATATGCTAGC AAAGCGTTAATCTT 3′
abcA-sRa	Reverse 3′ CTATAAATTTCAAAAGGAGGATGATTATTT 5′
yfp-sFa	Forward 5′ AGGAGGATGATTATTTATGAGTAAAGGAGAAGAACTTTTC 3′
yfp-R (EcoRI)	Reverse 3′GTTGTTGAATTCGCATGGATGAACTATACAAATAA 5′
abcA-pR (KpnI)	Reverse 3′ ATATGGTACC ACTATAAATTTCAAAC 5′

^a The underlined and bold sequences in primers (abcA-sR) and yfp-sF indicated the overlapping sequences necessary for the plasmid construction.

Kanamycin, chloramphenicol, anhydrotetracycline, nafcillin, oxacillin, and lysostaphin were purchased from Sigma-Aldrich (St. Louis, MO, USA).

S. aureus containing plasmids pLI50, pRN12, pTH2, and their various derived constructs were grown at 37°C in trypticase soy broth media (TSB) supplemented with chloramphenicol at 10 µg/mL. S. aureus containing plasmid pAH16 was grown at 37°C in TSB media supplemented with erythromycin at 10 µg/mL. S. aureus strain MN0599 carrying the knocked-out abcA gene was grown in TSB media supplemented with kanamycin at 50 µg/mL (17). All other S. aureus were grown in TSB media unless otherwise stated.

Antibiotic susceptibility assay

The MIC was determined by broth microdilution at 37°C for 24 hours, as previously described (31). A log-phase culture of S. aureus (OD₆₀₀ = 0.5) grown in TSB media was diluted 100-fold and inoculated into microtiter plates (Fisher Scientific, Pittsburgh, PA, USA) containing twofold serial dilutions of nafcillin or oxacillin. MIC was the lowest drug concentration that produced no visible turbidity after incubation at 37°C for 24 hours.

Construction of plasmids carrying yfp or cfp genes fused to the abcA promoter

We constructed the plasmid pRN-abcAp-yfp using a technique that was previously described with some modifications (32). All primers used in this study were synthesized by Eton Bioscience Inc. (Eton Bioscience, Boston, MA, USA) and are listed in Table 2. The 420-bp abcA promoter sequence was amplified using primers designed from the upstream region of the abcA gene that was reported in our previous study (14). The restriction enzyme sequence of NheI (GCTAGC) was inserted in the forward primer abcA-pF (NheI), and the reverse primer abcA-sR carried a 16-nucleotide sequence of the sod RBS at the 5′ end (AGGAGGATGATTATTT). The sod RBS-yfp gene was amplified from plasmid pAH16 (19, 20, 32) using the forward primer yfp-sF that carried the same 16-nucleotide sequence of the sod RBS (AGGAGGATGATTATTT) and the reverse primer yfp-R (EcoRI) with the sequence of EcoRI inserted (GAATTC). The abcA promoter and yfp were amplified and then the two PCR products served as templates for another PCR amplification using the forward primers abcA-pF (NheI) and the reverse primer yfp-R (EcoRI) (Table 2). Plasmid pRN12 was purchased from Addgene (30) (Addgene, Watertown, MA, USA) and was digested with enzymes NheI and EcoRI to remove the sarA P1 promoter-mAmetrine reporter gene of pRN12 and replaced with abcA/sod RBS promoter–yfp to yield plasmid construct pRN-abcAp-yfp that was used in confocal experiments to visualize and evaluate the expression of YFP when the abcA promoter was induced.

The plasmid construct pTH-abcAp-cfp was created using the same technique as described above with the purchased plasmid pTH2, which carried the sod RBS linked to the cerulean cfp gene under the control of the sarA promoter (Addgene, Watertown, MA, USA). To replace the sarA promoter of plasmid pTH2, the abcA promoter was amplified using the forward primer abcA-pF (NheI) and the reverse primer abcA-pR (KpnI). The PCR product and plasmid pTH2 were digested with NheI and KpnI and then ligated together. RN6390 was transformed with two plasmid constructs pAH16 and pTH-abcAp-cfp, and the double transformants were selected on TSB supplemented with Erm10 and Cm10.

Growth curves of S. aureus RN6390 and Newman strains with and without plasmid constructs

S. aureus RN6390, RN6390(pLI50), and RN6390(pLI50-abcA) were cultured overnight at 37°C in TSB broth supplemented with 10 µg/mL of chloramphenicol (Cm 10) when carrying plasmids. Then, 0.1 mL of each culture was transferred into 10 mL of fresh TSB liquid media supplemented or not with Cm10 and allowed to grow at 37°C under shaking for a period of 24 hours. Bacterial samples were collected every 2 hours, diluted, and plated on TSB or TSB + Cm10 for colony counts (CFU/mL) (29).

S. aureus Newman, the abcA knockout mutant MN0599, NM(pLI50), and the abcA overexpressor NM(pLI50-abcA) were grown overnight and started fresh as was described above for S. aureus RN6390. TSB media of the cultures were supplemented with Cm10 if the bacterial strains carried a plasmid. Bacteria were grown for 24 hours, and samples were taken every 2 hours for colony counts (CFU/mL) (17, 29).

All experiments were repeated using three independent biological samples. Statistical analyses were performed based on a one-way ANOVA with a pairwise t-test as post hoc test and a Bonferroni adjustment to determine the significance of differences in the growth of different S. aureus strains wild type, mutant, and overexpressors.

Kill curves of S. aureus strains RN6390 and Newman with and without plasmids

Nafcillin and oxacillin at two different concentrations 1×MIC and 25×MIC were used to perform S. aureus killing curves. The concentrations of antibiotics at 1× and 25× were adjusted depending on the MICs of the bacterial strain wild types, knockout mutant, or overexpressors. S. aureus RN6390 and Newman were cultivated overnight, and 0.5 mL of the overnight cultures was added to 50 mL of fresh TSB media supplemented or not with Cm10 to start the growth. At OD₆₀₀ ~ 0.6 (10⁷ CFU/mL), nafcillin or oxacillin at 1×MIC or 25×MIC were added to a 10 mL subculture. All S. aureus continued to grow for 24 hours with samples collected every 2 hours for colony counts (CFU/mL).

All experiments were repeated using three independent biological samples. Statistical analyses were performed using a one-way ANOVA with a t-test as post hoc test and a Bonferroni adjustment to determine the significance of differences in the growth of different S. aureus strains wild type, knockout mutant, and overexpressors in the presence of nafcillin or oxacillin at 1×MIC and 25×MIC.

Determination of S. aureus persister assay

The experimental design was based on a previous study by Conlon et al. with some modifications (26). S. aureus RN6390 was grown overnight in TSB under shaking at 37°C. A volume of 0.5 mL of the overnight culture was added to 50 mL of fresh TSB to initiate growth at 37°C under shaking (OD₆₀₀ ~ 0.05). At OD₆₀₀ ~ 0.6 (10⁷ CFU/mL), the culture was divided into subcultures of 10 mL each then challenged with nafcillin or oxacillin at a concentration equal to 1×MIC (0.25 µg/mL). A subculture of 10 mL was grown in parallel as a control of S. aureus growth without antibiotics. At time points 0, 8, and 20 hours, aliquots of bacterial cells were collected, washed with phosphate buffer saline (PBS 1× ), plated on TSB agar plates without antibiotics, and incubated at 37°C for 24 hours to count the surviving cells. From TSB agar plates of time points 8 and 20 hours, 100 colonies were selected from each plate, and then the colonies were replicated in parallel on fresh TSB plates and TSB plates supplemented with nafcillin or oxacillin at 1×MIC (0.25 µg/mL). The antibiotic in this selection step should be the same as the one used in the TSB broth subculture at the start of this experiment. TSB and TSB plus antibiotic plates were incubated overnight at 37°C, and then the colonies that grew on both types of plates were counted. S. aureus that grew on antibiotic plates were selected and re-grown on TSB agar plates and then subjected to MIC testing and quantitative RT-PCR assays to determine their phenotypic susceptibility to nafcillin and oxacillin, as well as the transcript levels of the abcA gene. All experiments were repeated using three different biological samples. Statistical analyses were performed using a one-way ANOVA with a t-test as post hoc test and a Bonferroni adjustment to determine the significance of differences in S. aureus survivors.

Quantitative real-time RT-PCR assay

The real-time RT-PCR assays were done as previously described (14). Total S. aureus RNA was extracted from lysostaphin-treated cells using the RNeasy midi kit (Qiagen, Valencia, CA, USA). cDNAs were synthesized using the Verso cDNA synthesis kit (Thermo Scientific, ABgene, Epsom, Surrey, UK), followed by real-time qRT-PCR assays using EvaGreen dye and the CFX96 real-time system (Bio-Rad, Hercules, CA, USA). Primers designed for the qRT-PCR assays were synthesized at Eton Bioscience Inc. (Eton Bioscience, Boston, MA, USA) and are listed in Table 2. The housekeeping gene gmk was used as an internal control. All samples were analyzed in triplicate, and expression levels were normalized against gmk gene expression, which remained unchanged following exposure to 1×MIC and 25×MIC of nafcillin or oxacillin. The assays were repeated with three independent biological samples. Statistical analyses were performed using a one-way ANOVA with a t-test as post hoc test and a Bonferroni adjustment to determine the significance of differences in gene expression values.

Confocal imaging assay

S. aureus RN6390 transformed with pAH16 and pRN-abcAp-yfp were grown in TSB supplemented with 10 µg/mL of erythromycin (plasmid pAH16) or 10 µg/mL of chloramphenicol (plasmid pRN-abcAp-yfp) at 37°C. At OD₆₀₀ ~ 0.6 (time = 0), nafcillin or oxacillin was added to the cultures at a final concentration equal to 1×MIC (0.25 µg/mL), and the bacteria were exposed to antibiotics for an additional 8 hours (time = 8). Then, the samples were submitted to confocal imaging as previously described (20, 32). The S. aureus cultures were adjusted to have 10⁵ bacterial cells per well in an 8-well chambered cover glass. The chambered cover glass was mounted onto a Nikon Ti-E inverted microscope fitted with a spinning disc confocal detection head (Yokogawa, Sugar Land, TX, USA). Solid-state lasers were used to produce excitation wavelengths of 405, 488, 514, or 647 nm. Bacteria were imaged using a Nikon 100× objective (1.49 NA, oil immersion objective, Nikon). Images were captured using an electron-multiplying charge-coupled device camera (C9100-13; Hamamatsu, Bridgewater, NJ, USA) and analyzed using MetaMorph software (Molecular Devices, Downingtown, PA, USA) (33). The confocal assays were repeated using three independent biological samples. Statistical analyses were performed using a one-way ANOVA with a t-test as post hoc test and a Bonferroni adjustment to determine the significance of the differences in the number of S. aureus RN6390 cells exposed to antibiotics over time and the significance of the differences in the number of S. aureus RN6390 cells exposed to nafcillin versus oxacillin at 1×MIC and over time.

Statistical analysis

The experiments were performed in triplicate with three biological samples. The data were expressed as a mean ± SD. The data were analyzed using one-way analysis of variance. The pairwise comparison was done with a t-test to compare sample groups combined with a Bonferroni adjustment. The threshold for significance was set at a P-value < 0.05.

Single-cell high-throughput microfluidic platform experiments

A microfluidic silicon wafer master mold was designed, nanofabricated, and used to prepare microfluidic chips as described before with some modifications (34). The silicon wafer consisted of six chips, each with eight lanes of approximately 5,000 trenches per lane. Trenches were dead-end 2 µm wide, 30 µm long, and 1.9 µm high features that originated at both sides of a feeding channel. The microfluidic design was printed by exposing each thin layer of SU8 photoresist deposited on the silicon wafer to 375 nm UV light in a Maskless aligner (MLA150).

An overnight culture of S. aureus RN6390(pAH16 + pTH-abcA-cfp) was grown in 2 mL of TSB with 10 µg/mL of erythromycin, 10 µg/mL of chloramphenicol, and 800 µg/mL Pluronic F108 (Sigma-Aldrich) at 37°C under shaking (New Brunswick Scientific Excella E24 shaker). Five lanes of the microfluidic device were pretreated with 0.1 M EDTA (Sigma Aldrich) for 10 min, Biofloat for 30 min (faCellitate BIOFLOAT FLEX coating solution), and rinsed with MOPS buffer to prevent biofilm formation during imaging. One milliliter of the resuspended pellet of the overnight culture was deposited into each of the microfluidic device lanes (Fisherbrand Gel-Loading Tips), then the chip was attached to a 3D-printed custom-built holder and spun in a centrifuge at 500 g for 5 min in one direction and 500 g for 15 seconds in the other direction to load both sides of the device. The remaining culture was washed with a new growth medium from the device’s feeding channel before the image acquisition.

The microfluidic device was connected to several syringe pumps (New Era-300) containing the experimental media through 0.02-inch internal diameter and 0.06-inch outer diameter Tygon tubing (ND 100-80 Microbore Tubing, US plastic) and gauge 20 luer-lock, stainless steel, 1-inch and 0.5-inch, blunt needles (McMaster). The cells in the device were exposed to different media combinations at a flow rate of 20 µL/min. Initially, all lanes were administered TSB with 10 µg/mL of erythromycin, 10 µg/mL of chloramphenicol, and 800 µg/mL Pluronic F108 for the first 4 hours of the experiment, then replaced by different antibiotic conditions, i.e., untreated, 1×MIC nafcillin, 25×MIC nafcillin, 1×MIC oxacillin, or 25×MIC oxacillin for 17 hours 30 min, and finally the antibiotic was removed by TSB with 10 µg/mL of erythromycin, 10 µg/mL of chloramphenicol, and 800 µg/mL Pluronic F108 for 5 hours 30 min. To mimic the conditions of the confocal imaging bulk assay (OD₆₀₀ ~ 0.6), the TSB media used in the microfluidic experiment consisted of filter-sterilized spent TSB media after a bulk culture reached OD₆₀₀ ~ 0.8. Twenty fields of view for each one of the experimental conditions, each with approximately 60 trenches, were imaged every 10 min in a Nikon Eclipse Ti2 fluorescence microscope. The microscope was equipped with a Perfect Focus System (PFS, Nikon), an Iris 9 camera (Teledyne Photometrics), a 40× objective with an additional 1.5× intermediate magnification integrated in the microscope base, an automated XY stage (Nikon), a Spectra III light engine (Lumencor), an incubator with temperature and humidity control that was kept at 37°C (Okolab), high-speed emission filter wheels (Finger Lakes Instrumentation HS-625), and a set of quad-band and triad-band dichroics for fast fluorescence image acquisition. The microfluidic device with S. aureus RN6390 (pAH16 + pTH-abcA-cfp) was imaged with phase contrast, YFP, and CFP filters at 100 ms exposure time and 100% LED intensity.

Micrographs were analyzed using a custom data analysis pipeline written in Python (version 3.10.8), called TrenchCoat. A Nikon ND2 file was converted to HDF5 using the nd2reader (version 2.1.3) and PyTables (version 3.7.0) libraries. At each step, micrographs and their features were visualized using Napari (version 0.4.17) (35). Microfluidic trenches were detected using a custom algorithm with two main steps: determination of the vertical location for each row of trenches and determination of the horizontal shift to optimally center the trenches. To determine the vertical offset(s) for each potential row of trenches, a set of uniform rectangles with user-defined lengths (27.5 µm), widths (3.4 µm), and spacing (3.46 µm) was generated and then exhaustively compared to all possible vertical pixel positions. At each position, an optimization metric was used to score the presence of trenches. The metric compared all horizontal positions and computed the maximum difference in summed pixel intensity; vertical positions without trenches had approximately zero difference, regardless of the horizontal shift, whereas positions with trenches had a large difference, reflecting how the pixel intensities of the space between trenches were different from the pixel intensities within the trenches. A peak detection algorithm was used to find locally maximal score(s), with one peak per trench row. After detecting the trench rows, for each row, the same set of rectangular regions was used to determine the optimal horizontal offset, either by maximizing the sum of pixel intensities (useful for fluorescence data), or the sum of pixel intensities at the edges of each trench (useful for phase contrast data). The trench detection procedure was executed for each field of view and time frame in parallel, and the four trench corner coordinates were written to an HDF5 table. Automated cell detection was performed using Omnipose (version 0.3.6) (36) with YFP as the cell detection channel. For model.eval, the following parameters were chosen: channels: [0,0], normalize true, flow threshold: 1.0, min size: 5, mask threshold: 2.0, cluster: true, rescale: 1.25, transparency: False, and omni: False. For CellposeModel, the bact_fluor_omni model was used and the following parameters were chosen: omni: true, concatenation: true, gpu: false, and padding: null. For each field of view, time frame, and optical configuration, the trench coordinates found previously were read from the disk and used to extract image slices for each trench. Each image slice was then fed into Omnipose, and the resulting labeled mask was used to determine the locations of each cell. Cell features and properties were then measured from the labeled mask and from the intensity images using the scikit-image (version 0.19.3) (10.7717/peerj.453) regionprops function in Python, as well as the total cellular pixel intensity in each channel. Finally, these measurements were written to an HDF5 table. After manual parameter optimization, for each of the five data sets, trench detection and cell detection were computed.

To compare a given field of view over time, it was sometimes necessary to account for small (micrometer) amounts of stage drift and renumber the trenches in the event that the leftmost trench ever fell out of bounds. The leftmost pixel coordinate of the zeroth trench was compared across all time points, and a drift event was flagged if the horizontal difference (in pixels) between the zeroth and nth trench coordinate exceeded an empirically determined threshold value. Image analysis was carried out in a computer with a Windows 10 version 22H2 operating system and an Intel i7-9700 processor. CFP-overexpressing cells were defined as the number of cells with total CFP intensity higher than the mean total CFP intensity of the untreated control plus three times its standard deviation.