ompF, osmB, CNFy, and tusB are differentially regulated in response to doxycycline

Our recent study showed there is a subpopulation of Y. pseudotuberculosis that survives doxycycline (Dox) treatment in the mouse spleen, and host-derived nitric oxide initially promotes survival of stressed cells [18]. This subpopulation persists within host tissues for 48 hours before Dox concentrations wane and the bacterial population resumes growth. We hypothesized there may be phenotypic changes that occur in response to Dox, which promote bacterial survival in the hours after treatment.

Using a fluorescent transcriptional reporter, we have approximated the concentrations of Dox in the spleen at 0.1μg/ml-1μg/ml during this period of growth inhibition following a single injection of 40mg/kg Dox [18]. To determine if there are specific pathways upregulated or downregulated during exposure to Dox, we initially chose to expose bacteria to 0.1μg/ml Dox. All bacterial cells would be exposed to at least this concentration within mouse tissues, and this concentration was sufficient for maximal de-repression of the tet operon [18], so we expected to see many transcriptional changes. We initially tested several different genes in known stress response pathways (ROS, RNS, general stress) to determine if these were differentially regulated in Dox-exposed Y. pseudotuberculosis in culture, but did not find significant changes (S1 Fig). We then took a global approach and utilized RNA-Seq to determine if there are specific pathways that would be up- or downregulated during exposure to 0.1μg/ml Dox. We exposed broth-grown bacteria to Dox and isolated bacterial RNA after 2h (hours, h) and 4h Dox exposure; treated and untreated cultures were grown in parallel. Samples were prepared for sequencing using a RNAtag-seq approach [35] and aligned to the Y. pseudotuberculosis genome. Although this concentration of Dox strongly modulates the TetR tet repressor [18], very few transcriptional changes occurred during antibiotic exposure in vitro, likely due to the low Dox concentration (Fig 1A). We identified 8 genes that were differentially regulated after 2h Dox treatment (Fig 1B), and there were no significant changes after 4h treatment, suggesting that transcriptionally, cells had returned to baseline. Of the 8 genes, only two genes were upregulated ~2 fold (osmB, ompF), and two genes were downregulated ~2 fold (tusB, cnfy), and so these 4 genes were selected for downstream analyses (highlighted in red, Fig 1B). We confirmed these 4 genes were differentially regulated after 2h Dox exposure by prepping additional bacterial samples for qRT-PCR, and also showed that gene expression changes were no longer observed after 4h Dox exposure (Fig 1C). These results were surprising; based on previous studies we had expected to see increased expression of efflux pumps and decreased expression of porins [36–38], instead we saw upregulation of a porin (ompF) and no change in efflux pump expression.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. ompF, osmB, CNFy, and tusB are differentially regulated in response to doxycycline.

Cultures of WT Y. pseudotuberculosis were incubated in the presence or absence of 0.1μg/ml Dox and bacterial transcript levels were compared by RNA-seq. Significant changes in transcript levels were detected at 2h post-treatment by DESeq2 analysis. (A) Volcano plot of aligned genes, each dot represents one gene. Horizontal dotted line: adjusted p value of 0.05, genes above this line are considered significantly altered. X-axis: log2 fold change in transcript levels. Vertical lines mark 2-fold changes, genes outside these lines are differentially regulated more than 2-fold. Hits of interest are highlighted. (B) Table of significant hits, 2h post-treatment. Hits of interest (differentially regulated ~2-fold) are highlighted in red. Base mean values represent 4 biological replicates. Log2 fold change and fold change are shown with adjusted p-value (adjusted based on gene length). (C) qRT-PCR validation of RNA-seq results. Cultures were prepared as described above, and bacterial transcripts were detected by qRT-PCR. Log2 fold change values were calculated relative to untreated cells, horizontal dotted lines depict average values for untreated cells (0log2), and 2-fold changes. Data represents four biological replicates. Statistics: (C) Two-way ANOVA with Tukey’s multiple comparison test, comparisons made relative to untreated cells. n.s.: not significant.

https://doi.org/10.1371/journal.ppat.1010556.g001

ompF, osmB, CNFy, and tusB are differentially regulated in response to doxycycline in vivo

Many studies have characterized how bacteria respond to antibiotics during growth in bacteriological media, however, few studies have identified changes that occur within the bacterial population during antibiotic treatment within host tissues. To determine if the gene expression changes we observed in culture are relevant in our mouse model of systemic infection, we infected C57BL/6 mice intravenously with WT Y. pseudotuberculosis, allowed infection to proceed 48h, and delivered a single injection of 40mg/kg Dox intraperitoneally [18]. Spleens were isolated at 48h (0h, time of Dox administration), or at 2h or 4h post-treatment. RNA was stabilized, isolated, mammalian mRNAs, tRNAs, and rRNAs were depleted to enrich for bacterial RNA, and bacterial transcript levels were detected by qRT-PCR. All 4 genes were differentially regulated in the host environment and the fold change in transcript levels was more dramatic within host tissues than seen with 0.1μg/ml Dox exposure (Fig 2A). osmB and cnfy had significant changes in transcript levels after 4h treatment, compared to the time of Dox administration (0h) (Fig 2A). For tusB, transcripts were significantly lower at 2h post-treatment, but not at 4h. Although there was high upregulation of ompF transcript levels, this was not statistically higher than baseline (0h), possibly due to some variability across samples.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. ompF, osmB, CNFy, and tusB are differentially regulated in response to doxycycline in vivo.

(A) C57BL/6 mice were infected with WT Y. pseudotuberculosis, infection proceeded for 48h (hours, h), and mice were injected intraperitoneally with a single dose of 40mg/kg Dox. At the indicated timepoints spleens were harvested, and samples were enriched for bacterial RNA prior to qRT-PCR. Time: h post-treatment, 0h represents 48h post-inoculation (p.i.), the time of treatment. Log2 fold change is relative to 0h. Horizontal dotted lines depict average values for 0h (0log2), and 2-fold changes. Data represents 4 biological replicates/group. (B) Cultures of WT Y. pseudotuberculosis were incubated for 4h in the presence or absence of 1μg/ml Dox, bacterial RNA was isolated and transcripts were detected by qRT-PCR. Log2 fold change values were calculated relative to untreated cells, horizontal dotted lines depict average values for untreated cells (0log2), and 2-fold changes. Data represents four biological replicates. Statistics: (A) Two-way ANOVA with Bonferroni’s multiple comparison test, comparisons made relative to 48h p.i./time 0h timepoint. (B) Two-way ANOVA with Tukey’s multiple comparison test, comparisons made relative to untreated cells using relative expression values. ****p < .0001, ***p < .001, **p < .01, *p < .05, n.s.: not significant.

https://doi.org/10.1371/journal.ppat.1010556.g002

Based on our previous study, we believe bacteria are exposed to 0.1–1μg/ml Dox within the mouse spleen in this treatment model, and concentrations may reach as high as 4μg/ml [18,20,39]. To determine if these 4 genes are also differentially regulated in response to 1μg/ml Dox in culture, we performed additional in vitro experiments and isolated RNA following 2h and 4h exposure to 1μg/ml Dox to quantify transcript levels by qRT-PCR. All 4 genes were differentially regulated at 2h post-treatment relative to untreated cells, and in contrast with 0.1μg/ml treatment, this differential regulation was maintained 4h post-treatment (Fig 2B). The magnitude of change in transcript levels was also much higher with 1μg/ml compared to 0.1μg/ml Dox, and was between an 8-16-fold change at 4h post-treatment with 1μg/ml Dox (Fig 2B), compared to a maximum 2-fold change with 0.1μg/ml Dox (Fig 1B). The change in transcript levels within the mouse spleen fell in the middle of these two treatment condition values (Fig 2A), consistent with a range of exposure between 0.1–1μg/ml Dox in the spleen.

OmpF is a well described bacterial porin, which is localized in the outer membrane of Gram-negative bacteria, and plays a major role in antibiotic diffusion into the cell [24,36,38]. Interestingly, mutations in ompF that either reduce functionality or downregulate expression of the gene have been strongly associated with resistance to β-lactams [40,41]. Our results instead suggest that upregulation of ompF could be protective during antibiotic exposure in Yersinia. osmB encodes for an outer membrane lipoprotein, and its expression is regulated by changes in osmolarity in E. coli [42,43]. Expression of osmB is also regulated by the alternative sigma factor, RpoS, and increases in stationary phase E. coli [44,45], but osmB has not been previously associated with antibiotic susceptibility. However, increased expression of osmB in stationary phase could suggest some association of this gene product with slowed growth. TusB functions with TusC and TusD in the sulfur-relay system, which has been associated with translational fidelity, through proper incorporation of glutamic acid, glutamine and lysine into growing polypeptide chains by stabilizing the wobble position of tRNAs [33]. TusB has not been previously associated with antibiotic susceptibility or tolerance. CNFy has been previously associated with persistence in Y. pseudotuberculosis [28], and for that reason, it will not be a focus in the following figures.

ompF mutants have slightly impaired growth in the presence of doxycycline

The antibiotic susceptibility of a bacterial strain is assessed by several well-defined measurements. These include the minimum inhibitory concentration (MIC): the concentration required for complete bacterial growth inhibition during culture; the minimum bactericidal concentration (MBC): the concentration required for a certain amount of bacterial killing over a defined time period, usually 90% killing or higher; and the more recently described minimum duration for killing (MDK): the amount of time required to kill a bacterial population during exposure to inhibitory levels of drug, typically used to assess the antibiotic tolerance of a bacterial strain [4,5,46–48]. Using this framework, we set-up three sets of in vitro assays to assess the antibiotic susceptibility of our WT and mutant strains.

To determine if increased expression of osmB or ompF promotes antibiotic tolerance, we generated single deletion mutants that lack either gene product (ΔosmB, ΔompF), and a double deletion mutant strain that lacks both genes (ΔosmB ΔompF). We first wanted to test the antibiotic susceptibility of the mutant strains by performing growth curves to determine if the MIC was significantly altered by a loss of these gene products. The MIC should only be altered by the presence or absence of genes required for antibiotic resistance, while gene products contributing to tolerance should not impact the MIC of the strain [4]. Since the WT MIC is 1μg/ml [18], and we would expect mutants to show no change or increased sensitivity, the highest dose used here was 1μg/ml. Under untreated conditions, we found the ΔosmB ΔompF strain was already growing slightly slower than the WT strain (Fig 3A). This strain was also growing slower under treated conditions (0.01–1μg/ml) (Fig 3B–3D), suggesting growth rates may be slowed in this strain. Interestingly, the ΔompF strain did appear slightly more sensitive to Dox when compared to the WT strain (0.01–0.1μg/ml) (Fig 3B and 3C). During exposure to inhibitory concentrations (1μg/ml) of Dox, the absorbance of the ΔompF strain was higher than WT, albeit significantly inhibited. Given the growth inhibition of ΔompF, we would have expected the ΔosmB ΔompF strain to have similar growth inhibition, however, these results were the first indication that the osmB deletion may partially rescue the ΔompF phenotype. These slight changes in growth of the mutant strains were not sufficient to alter their MICs. To confirm these slight changes in growth were due to loss of ompF, we attempted to generate an ompF rescued strain. However, we were only able to generate a chromosomally-rescued ompF in the ΔosmB ΔompF background with osmB deleted, and it remains unclear why ΔompF could not be rescued directly. These strains (ΔosmB ompF⁺, osmB⁺ ΔompF, osmB⁺ ompF⁺) showed the expected trends in growth, where ΔosmB ompF⁺ grew slightly better and osmB⁺ ΔompF grew slightly worse than the fully rescued strain (osmB⁺ ompF⁺), but these absorbance values were not statistically different (S2A Fig).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. ompF mutants have slightly impaired growth in the presence of doxycycline.

Exponential phase cultures of the WT and mutant strains (ΔosmB, ΔompF, ΔosmB ΔompF) were treated with the indicated concentrations of Dox to assess (A-D) growth inhibition, (E-H) changes in viable counts, and (I-L) competitive survival. (A-D) Strains were incubated with the indicated concentrations of Dox and growth inhibition was assessed based on absorbance (A_600nm) at the indicated timepoints (time: hours, h). Data represents the mean and standard deviation of three biological replicates. (E-H) Strains were incubated with the indicated concentrations of Dox and viable counts were determined by quantifying CFUs. Fold change in CFUs is shown relative to time 0h. Dotted line: a value of 1, no change relative to 0h. Data represents the mean and standard deviation of three biological replicates. (I-L) Competitive survival in the presence of the indicated concentrations of Dox; WT, single and double mutants were tested alongside the WT GFP⁺ strain. Competitive index (CI): CFUs of the test strain/WT GFP⁺ divided by the ratio of test strain /WT GFP⁺ in the culture at time 0h. Values above 1 indicate the mutant preferentially survives, values less than 1 indicate the WT GFP⁺ preferentially survives. Dotted line: value of 1, equal fitness. Dots: biological replicates, lines connect biological replicates sampled across the timepoints, 3–4 biological replicates shown. Statistics: Two-way ANOVA with Tukey’s multiple comparison test, (A-H): comparisons made relative to the WT strain; (I, K): comparisons made between 4h treatment CI values. ****p < .0001, ***p < .001, **p < .01, *p < .05, n.s.: not significant.

https://doi.org/10.1371/journal.ppat.1010556.g003

We then tested the viability of the WT and mutant strains during exposure to heightened, inhibitory levels of Dox (0.1–4μg/ml). The highest concentration used here was 4μg/ml, within the range of peak concentrations in serum in the mouse model (2–10μg/ml) [20,49,50]. We exposed exponential phase cells to Dox for 4h, and plated cells at the point of adding Dox (0h), 2h after exposure, and 4h after exposure to determine whether bacteria continued to grow, were growth inhibited, or lost viability during the exposure. Each strain was grown in a separate test tube with rotation, and we calculated the fold change in CFUs relative to the 0h timepoint. Strains continued to grow under untreated conditions and during exposure to 0.1μg/ml Dox (Fig 3E and 3F), while all strains were growth-inhibited during exposure to 1–4μg/ml Dox (Fig 3G and 3H). There were not significant differences in the viability of the strains, with the exception of the ΔosmB strain, which had significantly more CFUs than WT at 2h treatment with 4μg/ml. These data also suggest there is not a detectable difference in the MIC of the mutant strains relative to the WT, and again indicate the MIC for all strains is 1μg/ml. Because bacterial killing was not observed, the MBC or MDK cannot be determined based on these results.

We then tested in vitro Dox sensitivity in one additional assay, designed to test competitive survival of two co-cultured strains. The WT strain was transformed with a plasmid that contains a chloramphenicol resistance cassette and also constitutive GFP signal [51,52]. Mutants lacked an antibiotic resistance cassette, allowing us to identify the abundance of each strain by plating on selective and non-selective media. Strains were grown to exponential phase then added to individual wells of a 96-well plate. Mutant strains were tested for competitive survival during co-culture with the WT GFP⁺ strain, and single strain controls were also included to assess growth and survival of individual strains during Dox exposure in the plate, to ensure this mirrored the viability results. Competitive index values were calculated based on the ratio of mutant/WT GFP⁺ CFUs at the indicated timepoints after treatment relative to the ratio of mutant/WT GFP⁺ CFUs in the well at 0h, when Dox was added. Competition between an unmarked WT and WT GFP⁺ was included as a control, and showed a WT unmarked strain outcompetes WT GFP⁺ at high Dox concentrations, although this was not statistically significant (Fig 3I). Data comparing survival of the WT and ΔosmB strain suggested that the ΔosmB may have a slight fitness advantage relative to the WT strain, and showed no increased Dox sensitivity (Fig 3J). Consistent with the slightly impaired growth of the ΔompF strain (Fig 3B and 3C), ΔompF showed decreased relative survival during exposure to 4μg/ml Dox, especially when compared the WT unmarked strain (Fig 3I), suggesting this strain does have increased Dox sensitivity (Fig 3K). Interestingly, the ΔosmB ΔompF strain had very similar survival compared to the WT strain (Fig 3L), suggesting this strain does not have Dox sensitivity and that the osmB deletion may partially rescue the ΔompF phenotype. To confirm these changes in competitive survival were due to loss of ompF, we again utilized the chromosomally-rescued mutants in the ΔosmB ΔompF strain background. The osmB⁺ ΔompF strain showed slightly more sensitivity to 4μg/ml than 1μg/ml Dox, similar to ΔompF, although this was not statistically significant (S2B Fig). Without a significant phenotype here, and with high levels of variability in the assay, it was difficult to assess rescue with the osmB⁺ ompF⁺ strain; however, the average CI value of 1 suggests osmB⁺ ompF⁺ competes similarly to WT GFP⁺. Collectively, these data suggest the ΔompF strain has increased Dox sensitivity, however the osmB deletion has no detectable impact on Dox sensitivity, and actually seems to improve the fitness of the strain.

ΔosmB has increased in vivo fitness relative to the WT strain, ΔompF does not appear significantly more susceptible to doxycycline in vivo

To determine if the ΔosmB or ΔompF strains had increased doxycycline (Dox) sensitivity in vivo, we tested their survival in our mouse model of infection during exposure to a single dose of Dox. Because the in vitro phenotypes were subtle, we decided to test the survival of mutant strains relative to the WT in a co-infection experiment, essentially mirroring the competitive survival experiments we had set-up in culture (Fig 3J and 3K). Mutant strains used for infection contained a yopE::mCherry construct that would allow us to detect mutant cells with mCherry fluorescence [52,53]. The ΔosmB ΔompF strain was not tested here since it showed no difference in antibiotic susceptibility relative to the WT strain (Fig 3L).

Equal amounts of the WT and mutant strain were mixed, mice were inoculated intravenously, and 48h post-inoculation mice received a single dose of Dox or were left untreated. 24h after Dox injection (+Dox 24h), spleens were harvested to quantify CFUs and examine microcolony areas by fluorescence microscopy. The WT strain was marked with a constitutive GFP plasmid and chloramphenicol resistance, and the mutants contained yopE::mCherry to mark all mutant cells with mCherry signal, as described above. Total CFU/spleen were first quantified to ensure Dox treatment significantly decreased bacterial CFUs, then the numbers and microcolony areas of each strain were quantified. Although the Dox treatment significantly decreased CFUs (Fig 4A), the ΔosmB strain outcompeted the WT strain in the presence and absence of Dox (Fig 4B), again showing the ΔosmB strain does not have increased Dox sensitivity, and actually has a fitness advantage relative to WT. Consistent with this, the microcolony areas of the ΔosmB strain appeared slightly larger than WT, although this was not statistically significant (Fig 4C). The number of microcolonies decreased with treatment for both the WT and ΔosmB strains, but microcolonies did not significantly decrease in size after treatment for either strain (Fig 4C). We were expecting to see decreased microcolony areas after treatment, but that may have been difficult to assess without comparing to the areas at the time of injection (48h) [18]. We then quantified the CFUs of each strain before and after treatment within the same tissues. There were more ΔosmB CFUs prior to and after treatment, and both WT and ΔosmB CFUs significantly dropped in response to treatment (Fig 4D).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. ΔosmB has increased in vivo fitness relative to the WT strain, ΔompF does not appear significantly more susceptible to doxycycline in vivo.

C57BL/6 mice were infected with equal numbers of WT and mutant Y. pseudotuberculosis, infection proceeded for 48h (hours, h), and mice were injected intraperitoneally with a single dose of Dox or left untreated. Spleens were harvested 24h later to quantify CFUs and microcolony areas. (A) Co-infection with ΔosmB yopE::mCherry + WT GFP⁺. Total CFU/spleen for treated mice (+Dox 24h) compared to untreated mice at the same timepoint. Dots: individual mice. (B) Competitive index: CFUs of the mutant/WT in the spleen divided by the ratio of mutant/WT in the inoculum. Values above 1: mutant is more fit. Dotted line: value of 1, equal fitness. Dots: individual mice. (C) Microcolony areas were quantified for individual strains within the harvested spleens (6 untreated mice and 6 treated mice). 20–62 microcolonies were quantified in one section for untreated mice, and 1–13 microcolonies were quantified in one section for treated mice. (D) CFU/spleen of each strain in the ΔosmB + WT competition experiments. (E) Fold change in bacterial growth after 4h culture in the presence (+NO) or absence (untreated) of the NO donor compound, DETA-NONOate (2.5mM). WT, ΔosmB, and osmB⁺ (osmB rescued strain) are compared. Fold change relative to time 0, dotted line represents time 0 values. (F) Co-infection with ΔompF yopE::mCherry + WT GFP⁺. Total CFU/spleen for treated mice (+Dox 24h) compared to untreated mice at the same timepoint. Dots: individual mice. (G) Competitive index: CFUs of the mutant/WT in the spleen divided by the ratio of mutant/WT in the inoculum. Values above 1: mutant is more fit. Dotted line: value of 1, equal fitness. Dots: individual mice. (H) Microcolony areas were quantified for individual strains within the harvested spleens (4 untreated mice and 4 treated mice). 14–72 microcolonies were quantified in one section for untreated mice, and 1–7 microcolonies were quantified in one section for treated mice. (I) CFU/spleen of each strain in the ΔompF + WT competition experiments. All mouse experiments were carried out on two independent days. Statistics: (A, B, C, F, G, H) Mann-Whitney; (E) One-way ANOVA with Tukey’s post-test; (D, I) Kruskal-Wallis one-way ANOVA with Dunn’s post-test. ****p < .0001, **p < .01, *p < .05, n.s.: not significant.

https://doi.org/10.1371/journal.ppat.1010556.g004

To determine if the increased growth of the ΔosmB strain in the mouse model could be due to reduced sensitivity to nitric oxide (NO), we completed in vitro experiments with the NO donor compound, DETA-NONOate. WT, ΔosmB, and the osmB⁺ (osmB rescue) strains were grown in the absence and presence of 2.5mM DETA-NONOate for 4h, and sensitivity was assessed based on CFUs. As shown in previous studies [18,51,54], this concentration inhibits growth of the WT strain, and we observed similar levels of growth inhibition with the ΔosmB strain, suggesting increased growth in vivo is not due to altered NO sensitivity (Fig 4E). We also observed a clear increase in growth of the ΔosmB strain relative to WT under untreated conditions, which was rescued in the osmB⁺ strain (Fig 4E). These data together show that the loss of osmB improves growth of Y. pseudotuberculosis, both in vitro and in vivo.

OsmB contributes to membrane repair after exposure of stationary phase E. coli to high pressure conditions [55]. Because osmB is also expressed in E. coli under high osmolarity conditions [43–45], we hypothesized OsmB would be required for Y. pseudotuberculosis survival under high osmolarity conditions. To better understand the role of OsmB, we cultured exponential and stationary phase WT, ΔosmB, and the osmB⁺ (osmB rescue) strains under high osmolarity conditions (1M sorbitol, 0.5M and 1M NaCl), which have been shown to induce osmB expression [44]. We found all strains were similarly sensitive to high osmolarity conditions in both stationary and exponential phase suggesting osmB is not required (S3A and S3B Fig). These results are consistent with redundancy in the pathways that protect cells from outer membrane stress, and suggest additional lipoproteins are sufficient to protect cells when osmB is deleted.

During co-infection with the WT and ΔompF strains, we confirmed Dox treatment significantly decreased CFUs (Fig 4F). However, the WT and ΔompF strains were equally fit in the absence and presence of Dox, hence while the ΔompF strain may have slight increased susceptibility to Dox in vitro, this wasn’t sufficient to impact survival of the strain within mouse tissues (Fig 4G). The microcolony areas of the WT and ΔompF strains were similar, although the ΔompF microcolonies were slightly larger than WT in untreated mice, but the areas were not significantly altered by treatment (Fig 4H). There were few ΔompF microcolonies remaining within host tissues after treatment, which may suggest increased susceptibility of the strain. However, the number of WT microcolonies in the same spleens also decreased (Fig 4H). The CFUs of the WT and ΔompF strain were similar before and after treatment, however the decrease in ΔompF CFUs after treatment was more pronounced, which could indicate an increase in ΔompF Dox sensitivity in vivo (Fig 4I).

ΔompF has altered doxycycline permeability

Increased expression of the OmpF porin in response to Dox was unexpected, since OmpF expression could increase the diffusion of antibiotic into bacterial cells [36,40,41]. We hypothesized that Dox may use a different outer membrane protein for entry into Y. pseudotuberculosis, and that increased expression of ompF may promote antibiotic diffusion back out of the cell. This would lead to increased levels of Dox accumulation in the ΔompF strain, where Dox could enter the cell, but not diffuse out. Similarly, we hypothesized that OsmB lipoprotein could impact membrane permeability, and sought to determine if there was also differential Dox accumulation in ΔosmB cells. However, the limited impact of OsmB on Dox sensitivity meant that any change in Dox accumulation in the ΔosmB strain would likely be minimal.

To determine if Dox differentially accumulates in mutant strains, we utilized a TetON reporter system to detect intracellular Dox concentrations. In this reporter system, increasing concentrations of Dox results in heightened levels of mCherry signal, until concentrations are sufficient to inhibit mCherry translation [18]. This results in a bell curve, or normal distribution, of mCherry expression. The WT and deletion strains (ΔosmB, ΔompF, ΔosmB ΔompF) were first transformed with the constitutive GFP⁺ then transformed with the TetON reporter plasmid. The GFP signal was used here to select individual cells. Strains were exposed to increasing concentrations of Dox in culture, and we quantified mCherry fluorescence within individual cells by fluorescence microscopy. In the absence of treatment, background mCherry fluorescence was similar in all strains, but slightly higher in ΔompF (Fig 5A). With low concentrations of Dox (0.01μg/ml), we detected high levels of fluorescence in the ΔompF strain, consistent with increased antibiotic accumulation (Fig 5B). Interestingly, the ΔosmB ΔompF had slightly increased fluorescence compared to the WT strain, but much lower fluorescence than ΔompF, again indicating deletion of osmB may have partially rescued the ompF phenotype (Fig 5B). Consistent with increased antibiotic accumulation in ΔompF, exposure to 0.1μg/ml Dox caused fluorescence to drop below WT levels, suggesting this relatively low concentration was accumulating to sufficient levels to inhibit mCherry translation (Fig 5C). Exposure to 1μg/ml Dox resulted in decreased mCherry fluorescence in all mutants relative to WT, suggesting they may all have some level of increased antibiotic accumulation (Fig 5D). Throughout these in vitro experiments we saw the most dramatic changes with ΔompF, providing evidence this gene product impacts antibiotic accumulation within the cell. The impact of osmB was subtle, while the ΔosmB ΔompF strain showed an intermediate phenotype; this will be discussed in more detail in the Discussion.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. ΔompF has altered doxycycline permeability.

(A-D) WT and deletion strains (ΔosmB, ΔompF, ΔosmB ΔompF) transformed with the constitutive GFP plasmid and TetON reporter plasmid were incubated with the indicated concentrations of Dox for 4h. mCherry fluorescence was detected within individual cells by fluorescence microscopy. Relative mCherry was calculated by normalizing to the WT untreated average cell value. Data represents the mean and standard deviation of three biological replicates. (E) C57BL/6 mice were infected with WT or mutant strains (GFP⁺ TetON). Infections proceeded 48h, and mice were injected intraperitoneally with a single dose of Dox. Spleens were harvested 24h after treatment to (E) quantify CFUs, (F) quantify microcolony areas, and (G) quantify reporter signal by fluorescence microscopy. Relative reporter signal was quantified by dividing the mean mCherry (TetON) signal by the mean GFP signal from each microcolony. 5 mice were quantified/strain, and 1–15 microcolonies were quantified in one section for each mouse. (H) Representative images, scale bars: 20μm. All mouse experiments were carried out on two independent days. Statistics: Kruskal-Wallis one-way ANOVA with Dunn’s post-test. ****p < .0001, *p < .05, n.s.: not significant.

https://doi.org/10.1371/journal.ppat.1010556.g005

To determine if differential antibiotic accumulation occurs in mutant strains during infection, we performed single strain infections with WT, ΔosmB, ΔompF, or ΔosmB ΔompF GFP⁺ TetON strains, allowed infection to proceed 48h, injected intraperitoneally with Dox, and harvested spleens 24h after treatment to quantify CFUs, microcolony areas, and detect reporter signal by fluorescence microscopy. The constitutive GFP signal was used to assess differential TetON (mCherry) reporter expression across microcolonies [18]. The CFUs and microcolony areas for each strain were similar, consistent with the previous findings in Fig 4B and 4E, indicating these strains are not significantly more sensitive to Dox than WT (Fig 5E and 5F). The relative Dox exposure was then quantified by dividing the mean mCherry signal by the mean GFP signal for each microcolony. ΔompF microcolonies had significantly higher reporter signal relative to the other strains, suggesting that heightened Dox accumulation also occurs within this mutant during infection (Fig 5G and 5H). It is important to note that while ΔompF may accumulate more Dox, this is not sufficient to significantly impact bacterial survival (Fig 5E) or microcolony area (Fig 5F) when comparing across single strain infections, indicating the impact of OmpF on antibiotic susceptibility remains subtle. We again observed fewer microcolonies within ΔompF infected tissues; it required sectioning deeper into tissues to find bacterial centers, indicating the mutant strain does not retain as many microcolonies after treatment.

To show that altered antibiotic accumulation is due to loss of ompF, we utilized the chromosomally-rescued ompF strains in the ΔosmB ΔompF background. We performed in vitro antibiotic accumulation experiments in strains that contained the TetON plasmid (ΔosmB ΔompF, ΔosmB ompF⁺, osmB⁺ ΔompF, osmB⁺ ompF⁺, WT), and quantified mCherry fluorescence within individual cells by microscopy after 4h exposure to 0.01 or 0.1 μg/ml Dox. We expected to see a clear peak of mCherry signal in the mutant strains at 0.01μg/ml, which we did not observe (S4 Fig). At 0.01μg/ml Dox, the ΔosmB ΔompF strain had lower fluorescence than the other strains, which was not consistent with the GFP⁺ ΔosmB ΔompF strain in Fig 5, and rescues of this strain suggested both osmB and ompF promote antibiotic accumulation based on heightened signal (S4 Fig). All strains increased in mCherry fluorescence with 0.1μg/ml Dox, and all mutant strains had lower signal than WT. This suggested both ompF and osmB mutants had altered Dox uptake, but this was not restored to WT levels in the fully rescued strain (osmB⁺ ompF⁺) (S4 Fig). It is important to note these strains were independently-derived from unmarked ΔosmB ΔompF, and these results suggest the strains may have acquired additional genetic modifications that both impacted our ability to rescue ompF, and altered the phenotypes of rescued strains.

tusB overexpression promotes Y. pseudotuberculosis doxycycline sensitivity in vitro

tRNA 2-thiouridine synthesizing protein B (tusB) is a protein that functions as a heterohexamer with TusC and TusD in the sulfur-relay system, responsible for the post-transcriptional 2-thiolation of the uracil 34 at the wobble position [33,34]. This modification is necessary for the identification of purines, specifically promoting proper translation and incorporation of glutamic acid, glutamine and lysine into growing polypeptide chains [33]. Our RNA-seq results showed that tusB is significantly downregulated in response to Dox, thus, we chose to generate a tet-inducible (TetON) tusB overexpression strain (P_tetA::tusB), to specifically induce tusB expression in the presence of Dox and prevent downregulation [18] (Fig 6A). We used qRTPCR to detect tusB transcript levels in WT and P_tetA::tusB strains under untreated conditions and during 4h exposure to 1μg/ml Dox, and confirmed P_tetA::tusB results in tusB overexpression in the presence of Dox (Fig 6B). It is important to note the P_tetA::tusB strain has heightened levels of tusB prior to Dox exposure as well, although these levels are not statistically different from WT cells (Fig 6B). To determine if the Dox MIC was altered by increased expression of tusB, we performed growth curves with increasing levels of Dox. Overexpression of tusB did not alter bacterial growth kinetics or the MIC of the strain, and we observed similar absorbance levels for the WT and P_tetA::tusB strains under untreated and treated conditions (Fig 6C–6F).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. tusB overexpression promotes Y. pseudotuberculosis doxycycline sensitivity in vitro.

(A) Schematic of tusB overexpression strain. tusB expression is repressed by TetR in the absence of tetracyclines (-Tet). In the presence of antibiotics (+Tet), TetR repression is relieved, and tusB is expressed. (B) qRTPCR was used to detect tusB transcript levels in WT and P_tetA::tusB strains in the absence and presence of 1μg/ml Dox (4h treatment). Fold change shown relative to tusB transcript levels in untreated WT cells (normalized to a value of one, dotted line), median values depicted, three biological replicates are shown. Exponential phase cultures of the WT and tusB overexpression strain (P_tetA::tusB) were treated with the indicated concentrations of Dox to assess (C-F) growth inhibition, (G-J) changes in viable counts, and (K) competitive survival. (C-F) Strains were incubated with the indicated concentrations of Dox and growth inhibition was assessed based on absorbance (A_600nm) at the indicated timepoints (time: hours, h). Data represents the mean and standard deviation of five biological replicates. (G-J) Strains were incubated with the indicated concentrations of Dox and viable counts were determined by quantifying CFUs. Fold change in CFUs is shown relative to time 0h. Dotted line: a value of 1, no change relative to 0h. Data represents the mean and standard deviation of ten biological replicates. (K) Competitive survival in the presence of the indicated concentrations of Dox. Competitive index: CFUs of the P_tetA::tusB /WT divided by the ratio of P_tetA::tusB /WT in the culture at time 0h. Values above 1 indicate the tusB strain preferentially survives, values less than 1 indicate the WT preferentially survives. Dotted line: value of 1, equal fitness. Dots: biological replicates, lines connect biological replicates sampled across the timepoints, five biological replicates shown. Statistics: (B) Friedman paired one-way ANOVA with Dunn’s post-test, (C-K) Two-way ANOVA with Bonferroni’s multiple comparison test, (C-J): comparisons made relative to the WT strain; (K): comparisons made between timepoints at the same Dox concentration. ****p < .0001, ***p < .001, **p < .01, *p < .05, n.s.: not significant.

https://doi.org/10.1371/journal.ppat.1010556.g006

We then tested the viability of the WT and P_tetA::tusB strains during exposure to heightened, inhibitory levels of Dox (0.1–4μg/ml). As described above, we exposed exponential phase cells to Dox for 4h, and plated cells at the point of adding Dox (0h), 2h, and 4h after exposure to quantify the number of viable cells during the exposure. Strains were grown in separate test tubes with rotation, and we calculated the fold change in CFUs relative to the 0h timepoint. As seen with the osmB and ompF mutants, strains continued to grow under untreated conditions and during exposure to 0.1μg/ml Dox (Fig 6G and 6H). However, there was a significant drop in the number of P_tetA::tusB cells at 4h of treatment with 1μg/ml Dox compared to WT cells (Fig 6I). This phenotype was more pronounced with 4μg/ml Dox, where there was a significant drop in P_tetA::tusB cells relative to WT after only 2h exposure, and greater loss in viability at 4h post-treatment (Fig 6J). These data strongly support a role for tusB downregulation in Dox tolerance, and show that Dox can have bactericidal activity against the P_tetA::tusB strain. This defines the P_tetA::tusB strain as susceptible, and the WT strain as tolerant under these conditions.

We also tested the competitive survival of the P_tetA::tusB strain relative to WT during co-culture. The P_tetA::tusB construct contains a chloramphenicol resistance cassette, which was used to quantify the CFUs of this strain. An unmarked WT strain was used alongside P_tetA::tusB. Exponential phase cells were added to individual wells of a 96 well plate, and competitive index values were calculated based on the ratio of P_tetA::tusB /WT CFUs at the indicated timepoints after treatment relative to the ratio when Dox was added. These data were consistent with the viability count data, and showed the P_tetA::tusB strain was significantly outcompeted by WT during Dox exposure (Fig 6K). The P_tetA::tusB strain lost significantly more viable cells than WT during 4h exposure to 1μg/ml Dox, and at both 2h and 4h after exposure to 4μg/ml Dox. Both methods of detecting antibiotic tolerance showed that tusB downregulation significantly contributes to antibiotic tolerance, and that reversing this phenotype by expressing tusB during Dox treatment results in heightened drug susceptibility. Bacterial killing was observed in the P_tetA::tusB strain, although we would need to extend experiments longer to accurately determine MBC and MDK values with 90% killing or more stringent cut-offs. However, based on these experiments, approximately 75% of cells were killed with 4μg/ml Dox at 4h, which would mean the MBC₇₅ is ~4μg/ml and the MDK₇₅ would be ~4h.

tusB expressing cells have increased susceptibility to doxycycline in vivo

To determine if the increased Dox sensitivity of the P_tetA::tusB strain resulted in impaired survival during Dox treatment in the mouse model of infection, we compared the survival of the WT and P_tetA::tusB strains during co-infection. Equal amounts of the WT and P_tetA::tusB strain were mixed, mice were inoculated intravenously, then at 48h post-inoculation mice either received a single dose of doxycycline or were left untreated. At 24h post-treatment, spleens were harvested to quantify CFUs. Enumeration of total CFUs indicated that the Dox was sufficient to significantly decrease CFUs (Fig 7A), however the competitive index values were similar with and without treatment (Fig 7B). Notably, the competitive index values were below one, indicating the P_tetA::tusB strain was slightly outcompeted by WT in mouse tissues even under untreated conditions, despite very similar fitness in culture. We then quantified the CFUs of each strain before and after treatment within the same tissues. We saw a pronounced reduction in the CFUs of the P_tetA::tusB strain after treatment and also a reduction in WT CFUs, although this was not statistically significant (p = 0.0502) (Fig 7C). While the WT and P_tetA::tusB CFUs were similar prior to treatment, there was a significant drop in P_tetA::tusB CFUs relative to WT +Dox treatment, indicating the P_tetA::tusB strain does have heightened sensitivity to Dox within mouse tissues (Fig 7C). To determine if this heightened sensitivity also results in smaller microcolony areas after treatment, we transformed WT Y. pseudotuberculosis containing yopE::mCherry with the P_tetA::tusB plasmid and repeated infections with this strain. Mice were inoculated intravenously with the yopE::mCherry P_tetA::tusB strain, at 48h post-inoculation mice received a single dose of Dox or were left untreated, and spleens were harvested 24h post-treatment. The CFUs of the yopE::mCherry P_tetA::tusB strain significantly decreased with treatment (Fig 7D), however this was not reflected by a change in microcolony size (Fig 7E). We did observe there were fewer total microcolonies in treated tissues, as we had also seen with the WT, ΔosmB, and ΔompF infections. These results could suggest that larger and smaller microcolonies are equally susceptible to treatment, and may be eliminated or survive at a similar rate.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. tusB expressing cells have increased susceptibility to doxycycline in vivo.

C57BL/6 mice were infected with equal numbers of the WT and P_tetA::tusB strains, infection proceeded for 48h, and mice were either injected intraperitoneally with a single dose of Dox or left untreated. 24h later spleens were harvested to quantify CFUs. (A) Total CFU/spleen, each dot represents an individual mouse. (B) Competitive index values. Dotted line: value of 1; equal fitness between the two strains. Values less than 1 indicate the tusB strain is less fit than WT. (C) CFU/spleen of each strain in the WT + P_tetA::tusB competition experiments. (D) C57BL/6 mice were infected with the yopE::mCherry P_tetA::tusB strain, infection proceeded for 48h, and mice were either injected intraperitoneally with a single dose of Dox or left untreated. 24h later spleens were harvested to quantify CFUs. (E) Microcolony areas of the yopE::mCherry P_tetA::tusB strain were quantified from 4 untreated mice and 4 treated mice, each dot represents an individual microcolony. 6–38 microcolonies were quantified in one section for untreated mice, and 1–19 microcolonies were quantified in one section for treated mice. All mouse experiments were carried out on two independent days, horizonal lines depict median values. Statistics: (A, B, D, E) Mann-Whitney; (C) Kruskal-Wallis one-way ANOVA with Dunn’s post-test. **p < .01, *p < .05, n.s.: not significant.

https://doi.org/10.1371/journal.ppat.1010556.g007

ompF, tusB, and cnfy are differentially regulated in response to inhibitory concentrations of chloramphenicol, while osmB expression was only altered by doxycycline

Many antibiotics target the ribosome to inhibit translation and arrest bacterial growth. To begin to determine if the differential regulation of osmB, ompF, tusB and cnfy was a response to doxycycline, or a more general response to ribosomal inhibition, we also performed experiments with chloramphenicol (Cm), which targets the 50S ribosomal subunit by binding to 23S rRNA [56]. Cm was also chosen for these experiments because it can be used to treat Yersinia infections [21,23]. We performed growth curves with increasing doses of Cm (0.0025μg/ml-50μg/ml) to determine the sensitivity of WT Y. pseudotuberculosis to Cm, and define the minimum inhibitory concentration (MIC). All concentrations tested resulted in significant growth inhibition after 6 hours of growth in bacteriological media (Fig 8A). However, the growth curve with 0.0025μg/ml chloramphenicol mirrored untreated growth through 4 hours of growth, suggesting this low dose minimally impacted growth, and was sub-inhibitory. Treatment with 0.025–0.25μg/ml resulted in similar levels of growth inhibition, and treatment with 2.5–50μg/ml resulted in severe growth inhibition, to the point that no growth was detected (Fig 8A). Based on these data, the MIC of our WT Y. pseudotuberculosis strain to Cm was 2.5μg/ml.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 8. ompF, tusB, and cnfy are differentially regulated in response to inhibitory concentrations of chloramphenicol, while osmB expression was only altered by doxycycline.

(A) WT Y. pseudotuberculosis was grown in the presence of the indicated concentrations of chloramphenicol (Cm). Growth was measured by absorbance (A_600nm) at the indicated timepoints (hours, h). Bacteria were grown in the presence or absence of (B) 0.0025μg/ml Cm or (C) 0.025μg/ml Cm. RNA was isolated after 2h and 4h exposure, and the indicated transcripts were detected by qRT-PCR. Fold change is relative to untreated cells (represented by a dotted line at a value of 1). Statistics: (A-C) Two-way ANOVA with Tukey’s multiple comparison test, comparisons made relative to untreated cells, comparisons between treatment conditions where indicated by brackets. In C, statistics represent treated vs. untreated comparisons for cnfy, and ompF compared to osmB, which did not change with treatment. ****p < .0001, ***p < .001, n.s.: not significant.

https://doi.org/10.1371/journal.ppat.1010556.g008

We then compared the levels of osmB, ompF, tusB and cnfy transcripts during exposure to the subinhibitory 0.0025μg/ml Cm dose and the inhibitory 0.025μg/ml dose. The subinhibitory 0.0025μg/ml Cm did not significantly change any transcript levels when comparing untreated and treated cells (Fig 8B). Interestingly, the inhibitory 0.025μg/ml Cm dose did not change osmB transcript levels, which may suggest that the change in osmB transcripts in response to Dox is antibiotic-specific, and not a general response to ribosomal inhibition (Fig 8C). ompF was significantly upregulated in response to this concentration of Cm, and cnfy was significantly downregulated, suggesting these could be general responses to ribosomal inhibition. tusB transcript levels also decreased in response to Cm, although this was not statistically significant. Collectively, these data suggest that transcriptional regulation of ompF, cnfy, and tusB may be a response to ribosomal targeting by antibiotics, while heightened osmB expression may be a specific response to doxycycline.

Ethics statement

The animal experiments within this manuscript were approved by the Johns Hopkins Institutional Animal Care and Use Committee under protocol number: MO19H418. Experiments with Y. pseudotuberculosis were approved by the Johns Hopkins Institutional Biosafety Committee under protocol numbers P1611010206 and DN1611010106.

Bacterial strains & growth conditions

The WT Y. pseudotuberculosis strain, IP2666, was used throughout this study. For all in vitro experiments, overnight cultures of bacteria were grown for 16–18 h in lysogeny broth (LB) at 26°C with rotation. Bacteria were then sub-cultured 1:100 into fresh LB and incubated at 37°C with rotation for experiments. Overnight cultures were considered stationary phase, exponential phase cells were prepared as described below. Doxycycline was added to the samples at the indicated concentrations. For in vivo experiments, bacteria were cultured overnight as described above, in 2xYT broth (LB with 2x yeast extract and tryptone) [18,52]. Bacteria were then diluted in sterile phosphate-buffered saline solution (PBS) to a final concentration of 10³ CFU/100μl.

qRT-PCR to detect bacterial transcripts in broth-grown cultures

Bacterial cells were grown in the presence or absence of antibiotics for the indicated time points, pelleted, resuspended in Buffer RLT (QIAGEN) + ß-mercaptoethanol, and RNA was isolated using the RNeasy kit (QIAGEN). DNA contamination was eliminated using either the Turbo DNA-free kit (Invitrogen) or the on-column RNase-free DNase set (QIAGEN). RNA was reverse transcribed using M-MLV reverse transcriptase and random primers (Invitrogen), in the presence of the RNase inhibitor, RnaseOUT (Invitrogen). Approximately 30 ng cDNA was used as a template in qPCR reactions with 0.5 μM of forward and reverse primers [52] and Power SYBR Green PCR Master mix (Applied Biosystems). Control samples were prepared that lacked reverse transcriptase to confirm genomic DNA was eliminated from samples. qPCR reactions were carried out using the StepOnePlus Real-Time PCR system, and comparisons were obtained using the ΔΔC_T and 2^-ΔCt method (Applied Biosystems) for relative expression. Kits were used according to manufacturers’ protocols.

RNA-seq

Overnight cultures of WT Y. pseudotuberculosis were diluted 1:100 into fresh media and grown at 37°C in the presence or absence of 0.1μg/ml Dox. At 2h and 4h post-treatment, approximately 10⁵ bacterial cells were collected from cultures, and RNA was isolated as described above. RNA was prepared for RNA-seq analysis using a RNAtag-seq approach [35]. Following sequencing, reads were aligned using the Y. pseudotuberculosis strain YPIII fully annotated genome (genome: NC_010465, virulence plasmid: NZ_LT596222). Relative abundance changes in transcript levels were assessed using a DESeq2 method of pairwise comparisons between treatment groups [59]. An adjusted p-value of 0.05 was considered significant, and we focused on genes differentially regulated ≥2 fold for validation experiments and downstream assays.

Murine model of systemic infection

Jackson Laboratories (Bar Harbor, ME) provided 6–8 week-old female C57BL/6 mice and the animal experiments were approved by the Johns Hopkins University Institutional Animal Care and Use Committee (Protocol #: MO19H418). Mice were inoculated intravenously via tail vein injection with 10³ CFU Y. pseudotuberculosis in 100μl for all experiments [52]. At 48 hours post-inoculation, mice were treated with 40 mg/kg doxycycline (720μg in 100μl sterile PBS) via intraperitoneal injection [18]. Spleens were harvested at the indicated timepoints post-treatment and either homogenized to quantify CFUs, stabilized in RNALater for RNA isolation, or fixed in 4% paraformaldehyde (PFA) for histology. Tissue was fixed by incubating 24h in 4% PFA at 4° C. Animals were euthanized by lethal isoflurane asphyxiation followed by cervical dislocation and vital organ removal.

qRT-PCR to detect bacterial transcripts from infected tissues

At the indicated timepoints, spleens were harvested and immediately submerged in RNALater to stabilize transcripts, incubated at 4°C three weeks, then tissue was stored at –80°C prior to RNA isolation. Tissue was then thawed, added to Buffer RLT (QIAGEN) + ß-mercaptoethanol (200μl/10mg tissue), and homogenized. Cell debris was pelleted, and lysate was used for RNA isolation as described above. DNA contamination was eliminated using the Turbo DNA-free kit (Invitrogen). The MicrobEnrich kit (Invitrogen) was used to deplete mouse mRNAs, tRNAs and rRNAs. Approximately 500ng RNA was used in reverse transcription reactions, and approximately 25ng cDNA were used in qPCR reactions, as described above.

Generation of deletion and rescued strains

osmB and ompF deletion strains were generated by amplifying the start codon + 3 downstream codons, the 3’ terminal 3 codons + the stop codon and fusing these fragments to generate a start + 6 aa + stop deletion construct for each gene. Deletion constructs were amplified with 500 base pairs flanking sequence on each side, cloned into the suicide vector, pSR47S, and conjugated into Y. pseudotuberculosis [60]. Sucrose selection was used to select for bacteria that had incorporated the desired mutation after a second recombination event [60]. PCR, sequencing, and qRT-PCR were used to confirm deletion strains. To mark mutant cells with fluorescence, we also generated deletion strains that contained the previously described, yopE::mCherry reporter [52,53]. Rescued strains were generated by amplifying the gene of interest with 500 base pairs flanking sequence on each side, inserting this into pSR47S, conjugating into Y. pseudotuberculosis deletion strains, and selecting bacteria that had integrated the WT copy of the gene into the chromosome. Rescue was confirmed by PCR and sequencing.

Generation of the TetON reporter strains and P_tetA::tusB construct

The TetON reporter (P_tetA::mCherry-ssrA) was previously described [18], and was constructed by fusing tetR through the P_tetA promoter to a ssrA-tagged destabilized variant of mCherry [61,62]. The TetON reporter was expressed from the low copy pMMB67EH plasmid and transformed into WT and mutant strains by electroporation [60]. The constitutive GFP plasmid used alongside TetON in mouse infections has been previously described [18,52,53]. This GFP plasmid was also used to mark the WT strain for in vitro and in vivo competition experiments with the osmB and ompF deletion strains. P_tetA::tusB was generated the same way as the TetON reporter; the tetR through P_tetA fragment was fused to tusB. P_tetA::tusB was inserted into the low copy pACYC184 plasmid and electroporated into WT Y. pseudotuberculosis, as previously described [60]. To mark the P_tetA::tusB strain with fluorescence, we also transformed the P_tetA::tusB plasmid into a WT yopE::mCherry reporter strain [52,53].

Antibiotic susceptibility assays

Antibiotic susceptibility of mutant strains was assessed using three in vitro approaches: growth inhibition, viability, and competitive survival. Growth inhibition was assessed by diluting overnight cultures (1:100) into fresh LB containing increasing doses of Dox (0.01–1μg/ml), and quantifying absorbance (600nm) every 2h using a microplate reader. Cultures were incubated for 8h at 37° C with rotation. Viability was assessed with exponential phase cells, which were subcultured from overnight cultures (1:100) into fresh LB and rotated at 26° C for 2h. Exponential phase cells were then transferred to fresh LB containing increasing doses of Dox (0.1–4μg/ml), and incubated 4h at 37° C with rotation. Cells were plated for viability at the time of Dox addition (0h), 2h, and 4h post-treatment. Competitive survival was assessed with exponential phase cells prepared as described above, and cells were co-cultured in wells of 96 well plates with increasing doses of Dox. Plates were incubated at 37° C on an orbital shaker (VWR). Single strain controls were grown in parallel to ensure growth and Dox sensitivity was similar in 96 well plates compared to test tubes. Cells were plated to quantify CFUs of each strain at the time of Dox addition (0h), 2h, and 4h post-treatment. Chloramphenicol (Cm) resistance was used to quantify the number of WT cells relative to total bacterial numbers for the experiments with deletion strains. For the WT vs. P_tetA::tusB experiments, the tusB strain was marked with Cm resistance.

High osmolarity experiments

Overnight cultures were considered stationary phase, and addition of 1M sorbitol, 1M NaCl, or vehicle control (equivalent volume dH₂O) were added directly to cultures. For experiments with exponential phase cells, aliquots of overnight cultures were washed in LB without NaCl (yeast extract and tryptone only) and subcultured (1:100 dilution) into LB without NaCl. Cultures were incubated 2h at 26° C with rotation before addition of the indicated amounts of NaCl (0.5M, 1M) or vehicle control (equivalent volume dH₂O). Absorbance (600nm) was used to detect changes in the density of cultures, values were normalized to the starting absorbance of each culture.

NO donor experiments

Experiments with the NO donor, DETA-NONOate (Cayman Chemical), were performed with exponential phase cells, prepared as described above. Exponential phase cells were transferred to fresh LB in the presence or absence of 2.5mM DETA-NONOate in 96 well plate. Cultures were immediately plated to determine the starting CFU, then incubated 4h at 37° C with orbital shaking, and plated to determine the final CFU. Fold change in CFU was calculated for each culture.

Fluorescence microscopy—Host tissues

After fixation with PFA, spleens were frozen-embedded in optimal cutting temperature (O.C.T.) compound (Tissue-Tek, VWR). 10 μm thick sections were obtained with a cryostat microtome (Microm HM 505e) and transferred to charged microscope slides (VWR). Slides were stored at –80° C, then thawed in PBS at room temperature, washed 3x in PBS, and stained with Hoechst (1:10,000 dilution in PBS) for 5 minutes. Slides were then washed 3x with PBS and coverslips were mounted with ProLong Gold (Life Technologies). One section was imaged for each tissue. Tissues were imaged with a Zeiss Axio Observer 7 (Zeiss) inverted fluorescent microscope with an Apotome.2 (Zeiss) with the 63x oil immersion objective. Images of the sections were captured with an Axiocam 702 mono camera (Zeiss).

Antibiotic permeability assays

Exponential phase cultures of TetON-containing strains (WT, ΔosmB, ΔompF, ΔosmB ΔompF) were incubated in the absence or presence of increasing doses of Dox (0.01–1μg/ml) for 4h at 37° C with rotation. Approximately 10⁶ cells were pelleted and fixed in 4% PFA at 4° C for 18h (overnight). Fixed cells were pelleted, PFA was removed, and cells were resuspended in 20μl sterile PBS. 5μl of each sample was imaged on 1% agarose (in PBS) gel pads. 5 fields of view were captured for each sample, and single cell fluorescence was quantified using Volocity software, as described below. This resulted in approximately 500–800 total cells imaged across three biological replicates, or approximately 150–300 cells per condition, per replicate.

Image analysis

Volocity image analysis software was used to quantify microcolony areas as previously described [18,52]. Briefly, either GFP or mCherry signal were used to detect individual bacteria and quantify total area of each microcolony. One cross-section/tissue was imaged to represent each spleen. Dox permeability was detected using TetON reporter expression, which was quantified using Volocity to measure single cell mCherry fluorescence. For in vitro experiments, relative mCherry signal was calculated by dividing the mCherry fluorescence value of each cell by the average mCherry value of untreated WT cells, which were used to set a baseline fluorescence value. For in vivo experiments, relative Dox accumulation was quantified by dividing the mean mCherry signal across a microcolony, by the mean signal of constitutive GFP, as previously described [18].