Enterococcus faecalis Antagonizes Pseudomonas aeruginosa Growth in Mixed-Species Interactions

ABSTRACT

Enterococcus faecalis is often coisolated with Pseudomonas aeruginosa in polymicrobial biofilm-associated infections of wounds and the urinary tract. As a defense strategy, the host innately restricts iron availability at infection sites. Despite their coprevalence, the polymicrobial interactions of these two species in biofilms and under iron-restricted conditions remain unexplored. Here, we show that E. faecalis inhibits P. aeruginosa growth within biofilms when iron is restricted. E. faecalis lactate dehydrogenase (ldh1) gives rise to l-lactate production during fermentative growth. We find that an E. faecalis ldh1 mutant fails to inhibit P. aeruginosa growth. Additionally, we demonstrate that ldh1 expression is induced under iron-restricted conditions, resulting in increased lactic acid exported and, consequently, a reduction in local environmental pH. Together, our results suggest that E. faecalis synergistically inhibits P. aeruginosa growth by decreasing environmental pH and l-lactate-mediated iron chelation. Overall, this study emphasizes the importance of the microenvironment in polymicrobial interactions and how manipulating the microenvironment can impact the growth trajectory of bacterial communities.

IMPORTANCE Many infections are polymicrobial and biofilm-associated in nature. Iron is essential for many metabolic processes and plays an important role in controlling infections, where the host restricts iron as a defense mechanism against invading pathogens. However, polymicrobial interactions between pathogens are underexplored under iron-restricted conditions. Here, we explore the polymicrobial interactions between commonly coisolated E. faecalis and P. aeruginosa within biofilms. We find that E. faecalis modulates the microenvironment by exporting lactic acid which further chelates already limited iron and also lowers the environmental pH to antagonize P. aeruginosa growth under iron-restricted conditions. Our findings provide insights into polymicrobial interactions between bacteria and how manipulating the microenvironment can be taken advantage of to better control infections.

INTRODUCTION

Many infections are often polymicrobial in nature (1–3) and include wound infections (4–7), periodontitis (8, 9), otitis media (10, 11), urinary tract infections (UTI) (12–15), and cystic fibrosis (16–20). Biofilms are also implicated in all these infections (21–37). Polymicrobial biofilms can better tolerate antibiotic treatment and escape from host immune responses, enabling the survival and persistence of the infecting bacteria (38, 39). Hence, understanding how pathogens interact in biofilms may inform improved treatment strategies.

Iron is an essential nutrient for almost all microbial species. In humans, iron regulation functions as a host innate immune mechanism against invading pathogens (40). In the human body, iron is scarcely available to pathogens due to the sequestration of most iron intracellularly such that only a small amount of free iron (approximately 10⁻²⁴ M) is accessible in the absence of infection (41). During an infection, additional iron-withholding mechanisms further restrict iron availability to pathogens (40). For example, immune cells producing hepcidin (42) or lactoferrin (43), and siderocalin/lipocalin-2 (43, 44) modulate iron availability at the infection site. As a result, when developing in vitro polymicrobial biofilm interaction models, it is critical to take into account the iron availability in the environment.

Enterococci are opportunistic pathogens implicated in several types of infections (45), and enterococcal infections in humans are mostly caused by Enterococcus faecalis and Enterococcus faecium (46, 47). E. faecalis is often coisolated with Pseudomonas aeruginosa in biofilm-associated infections such as wound infections, urinary tract infections, and periodontitis (4, 48–53). As such, understanding the polymicrobial interactions between these two species is of interest. The biofilm-forming potential of E. faecalis and P. aeruginosa is well studied individually (54–56). However, despite their cooccurrence as well as increasing efforts made to understand the polymicrobial interactions between bacterial species, there have been no reports examining E. faecalis and P. aeruginosa polymicrobial interactions both in biofilms and under iron-restricted conditions.

In this work, we show that E. faecalis inhibits P. aeruginosa growth within biofilms when iron is restricted. The growth inhibition is a consequence of increased l-lactate produced by E. faecalis, catalyzed by lactate dehydrogenase (ldh1) from pyruvate. We also show that ldh1 expression is upregulated when iron is restricted. l-Lactate produced by E. faecalis is exported from the cell as lactic acid (57), whereupon it is deprotonated to l-lactate, releasing a hydrogen ion (H⁺) and in turn lowering the pH in the surrounding environment. We demonstrate that this lowered environmental pH and l-lactate-mediated chelation of iron ultimately contribute to P. aeruginosa growth inhibition by E. faecalis under iron-restricted conditions. Together, our work highlights the possibility of manipulating the microenvironment to antagonize specific bacterial species within biofilms.

RESULTS

E. faecalis inhibits P. aeruginosa growth under iron-restricted conditions.

The compound 2,2′-dipyridyl (22D) is widely used as a neutral ligand for the chelation of metal ions and is a high-affinity chelator of iron (58). Therefore, 22D was added as a supplement into TSBG (tryptone soya broth with glucose) growth medium to restrict iron availability. We first investigated the interactions between 12 different E. faecalis clinical isolates and P. aeruginosa PAO1 BAA-47 using static biofilm assays. In iron-restricted medium (TSBG supplemented with 1 mM 22D), PAO1 BAA-47 growth was inhibited in all of the mixed-species biofilms compared to PAO1 BAA-47 single-species biofilm (Fig. 1A). In contrast, E. faecalis growth was similar in the single- and mixed-species biofilms (Fig. 1B). These data indicate that E. faecalis inhibition of P. aeruginosa is not strain specific. We next performed the biofilm assay, now with E. faecalis OG1RF and eight different P. aeruginosa clinical isolates, to examine whether P. aeruginosa susceptibility to E. faecalis-mediated inhibition was strain specific. We observed that the growth of all P. aeruginosa isolates was inhibited in the mixed-species biofilms compared to their respective single-species counterpart (Fig. 1C), while E. faecalis growth in the single- and mixed-species biofilms remained unaffected (Fig. 1D). These data demonstrate that all tested E. faecalis and P. aeruginosa clinical isolates engage in mixed-species antagonism under iron-restricted conditions.

FIG 1.

E. faecalis inhibits P. aeruginosa growth under iron-restricted conditions. (A to D) Enumeration of PAO1 BAA-47 (A), E. faecalis clinical isolates (B), P. aeruginosa clinical isolates (C), and OG1RF (D) from 24-h biofilms with single or mixed inocula grown in TSBG medium supplemented with 1 mM 22D. Dotted lines represent inoculum of bacteria spotted. n ≥ 3 with 3 technical replicates; error bars represent standard deviation from the mean. Statistical analysis was performed using the Mann-Whitney U test: *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001. For panel A, statistical significances are for all strains compared to PAO1 BAA-47. (E) Enumeration of PAO1 BAA-47 from 24-h cultures grown with fresh TSBG medium, water, or cell-free supernatant obtained from 24-h biofilms of PAO1 BAA-47, OG1RF, and PAO1 BAA-47 mixed with OG1RF. The water and respective supernatants were mixed at a 1:1 ratio with fresh TSBG medium supplemented with 1 mM 22D for PAO1 BAA-47 growth. n = 4 with 3 technical replicates; error bars represent standard deviation from the mean. Statistical analysis was performed using the Mann-Whitney U test: *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001. (F and G) Enumeration of PADP6 (F) and OG1RF (G) from 24-h macrocolonies with single or mixed inocula grown in TSBG medium supplemented without and with 2 mM 22D. Bacterial species were mixed at a 1:1 ratio for mixed-species macrocolonies. Dotted lines represent inoculum of bacteria spotted. n = 3 with 3 technical replicates; error bars represent standard deviation from the mean. Statistical analysis was performed using the Mann-Whitney U test: ns, not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001.

We next performed supernatant transfer assays in which either E. faecalis or P. aeruginosa single- or mixed-species biofilms were grown in iron-restricted media, and their cell-free biofilm supernatants were collected and supplied to P. aeruginosa at a 1:1 ratio with fresh medium for subsequent growth assays. There were minimal differences in PAO1 BAA-47 growth when supplemented with biofilm supernatant obtained from single-species PAO1 BAA-47 biofilm compared to supplementation with water (Fig. 1E). However, when PAO1 BAA-47 was grown with supernatant obtained from single-species OG1RF or mixed PAO1 BAA-47 and OG1RF biofilms, we observed a significant inhibition of PAO1 BAA-47 growth compared to control supplementation (Fig. 1E), suggesting that P. aeruginosa inhibition is mediated by the presence of E. faecalis.

Next, to validate the above findings and to determine the mechanistic basis of polymicrobial interactions between E. faecalis and P. aeruginosa under iron-restricted conditions, we performed mixed-species macrocolony biofilm assays (59, 60), initially using E. faecalis OG1RF and P. aeruginosa PAO1 for our experiments. However, PAO1-WT (wild type) was sensitive to iron restriction at 22D concentrations greater than 1 mM (see Fig. S1A and B in the supplemental material) and had a MIC of 0.8 mM to 22D (data not shown), while there was minimal effect on OG1RF up to 3 mM 22D (Fig. S1C). As such, a P. aeruginosa PAO1 spontaneous mutant that was resistant to 22D chelation was generated and named PADP6. Whole-genome sequencing of this mutant revealed a single nucleotide polymorphism in the nalC repressor gene (4166561G>T, R15L). This mutation in PADP6 restored growth in 2 mM 22D to similar levels as those of PAO1-WT macrocolonies grown in unchelated media (Fig. S1A), and PADP6 was less sensitive to 22D iron restriction than PAO1-WT (Fig. S1B and D). Mutation in nalC causes an overexpression of the iron-regulated mexAB-oprM operon encoding the MexAB-OprM efflux pump upon severe iron restriction (61, 62). Additionally, the MexAB-OprM efflux pump is implicated in pyoverdine siderophore secretion (62, 63). Hence, to understand how the single nucleotide polymorphism in nalC enhances 22D resistance in PADP6, we quantified the expression of mexA, mexB, and oprM as well as pyoverdine production in PAO1-WT and PADP6 when grown in iron-restricted media. As expected, we observed higher expression of mexA, mexB, and oprM in PADP6 than in PAO1-WT in both unchelated and 0.5 mM 22D-chelated media (Fig. S2A, B, and C). Pyoverdine secretion was also higher in PADP6 than in PAO1-WT when grown in 0.5 mM 22D-chelated media (Fig. S2D), suggesting that PADP6 confers tolerance to iron starvation by increasing pyoverdine secretion. Moving forward, E. faecalis OG1RF and P. aeruginosa PADP6 were used for all subsequent experiments.

To validate that PADP6 was also susceptible to E. faecalis-mediated growth inhibition when iron was restricted (supplemented with 2 mM 22D), we performed the macrocolony biofilm assay and observed that PADP6 growth was inhibited in mixed-species macrocolonies compared to PADP6 single-species macrocolonies (Fig. 1F), whereas OG1RF growth in the single- and mixed-species macrocolonies was unaffected (Fig. 1G). Further, the addition of ferric chloride (FeCl₃) to 22D-chelated media restored PADP6 growth in the mixed-species macrocolonies to levels similar to PADP6 single-species growth in chelated media without FeCl₃ supplementation (Fig. S3A and B), indicating that PADP6 growth inhibition in mixed-species macrocolonies is specific to the presence of E. faecalis and iron restriction. The supplementation of other trace metals to 22D-chelated media was unable to rescue growth inhibition of P. aeruginosa (Fig. S3C and D). We also explored the polymicrobial interactions between E. faecalis and P. aeruginosa in 2 mM 22D-chelated TSB growth medium (lacking the supplemental glucose of TSBG). However, we did not observe an antagonistic relationship between these two bacteria (Fig. S4A and B), suggesting that nutritional differences contribute to the antagonistic relationship between E. faecalis and P. aeruginosa. Subsequently, TSBG medium was used for all experiments as we wanted to understand and exploit the mechanistic basis of the antagonism between E. faecalis and P. aeruginosa for future possibilities in treatment applications. Hence, based on the mixed-species antagonism observed in TSBG medium, we hypothesized that E. faecalis OG1RF produces a factor, or modulates the local environment, such that it is unfavorable for the growth of PADP6 under iron-restricted conditions.

Planktonic growth was also examined to determine whether PADP6 growth inhibition in the presence of OG1RF was specific to biofilms. In unchelated media, we observed PADP6 growth inhibition by approximately 1 log in coculture with E. faecalis OG1RF compared to PADP6 alone, while OG1RF growth in coculture remained unaffected (Fig. S5). However, PADP6 was further inhibited by more than 2 logs in the presence of 22D. Therefore, E. faecalis OG1RF inhibition of P. aeruginosa PADP6 growth is not a biofilm-specific phenotype.

E. faecalis ldh1 is responsible for P. aeruginosa growth inhibition under iron-restricted conditions.

An E. faecalis mariner transposon library screen was performed to identify E. faecalis mutants that did not inhibit PADP6 planktonic growth under iron-restricted conditions. We identified six E. faecalis mutants that did not inhibit PADP6 growth when cocultured; however, upon validation of these six mutants in the macrocolony assay, three of them were validated for failure to inhibit PADP6 growth: ldh1, gloA3, and guaB (Table S1). Here, we focused on the transposon insertion in ldh1 (ldh1::Tn). In E. faecalis, there are two copies of ldh (ldh1 and ldh2) that encode l-lactate dehydrogenase (LDH) to catalyze the reduction of pyruvate to l-lactate (64). Of the two copies, ldh1 accounts for the majority of l-lactate produced in E. faecalis in laboratory settings (64, 65).

We then created an ldh1 deletion mutant (OG1RF Δldh1) and validated the transposon screening results in a macrocolony biofilm assay with PADP6 and OG1RF Δldh1. Specifically, PADP6 growth was not inhibited in the mixed PADP6 and OG1RF Δldh1 macrocolonies compared to mixed PADP6 and OG1RF macrocolonies (Fig. 2A). Importantly, OG1RF Δldh1 growth in single- and mixed-species macrocolonies was unaffected (Fig. 2B). Upon ldh1 chromosomal complementation in OG1RF Δldh1, PADP6 growth was inhibited to similar levels as when grown with OG1RF (Fig. 2A), suggesting that ldh1 plays a role in inhibiting PADP6 growth under iron-restricted conditions.

FIG 2.

l-Lactate produced by E. faecalis inhibits P. aeruginosa growth in iron-restricted media. (A and B) Enumeration of PADP6 (A) and OG1RF, Δldh1, and Δldh1::ldh1 (B) from 48-h macrocolonies with single or mixed inocula grown in 2 mM 22D-chelated TSBG medium. Bacterial species were mixed at a 1:1 ratio for mixed-species macrocolonies. Dotted lines represent inoculum of bacteria spotted. n = 4 with 3 technical replicates; error bars represent standard deviation from the mean. Statistical analysis was performed using the Mann-Whitney U test: ns, not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001. (C) Quantification of l-lactate exported from 48-h single- and mixed-species macrocolonies grown in TSBG medium supplemented without and with 2 mM 22D. n = 3 with 2 technical replicates; error bars represent standard deviation from the mean. Statistical analysis was performed using two-way analysis of variance with Tukey’s test for multiple comparisons: ns, not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001. (D) Quantification of iron when 200 μM iron(II) sulfate heptahydrate (Fe²⁺) and iron(III) chloride hexahydrate (Fe³⁺) were supplemented without and with various concentrations of 22D (0.5, 1, and 2 mM) or l-lactate (10 and 20 mM). n ≥ 3 with 2 technical replicates; error bars represent standard deviation from the mean. Statistical analysis was performed using the Mann-Whitney U test: ns, not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001. (E and F) Enumeration of PADP6 (E) and OG1RF and Δldh1 (F) from 48-h macrocolonies with single or mixed inocula grown in 2 mM 22D-chelated TSBG medium without and with increasing lactic acid concentrations (2.5, 5, and 10 mM). Bacterial species were mixed at a 1:1 ratio for mixed-species macrocolonies. Dotted lines represent inoculum of bacteria spotted. n = 3 with 3 technical replicates; error bars represent standard deviation from the mean. Statistical analysis was performed using the Mann-Whitney U test: ns, not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001.

E. faecalis-derived l-lactate is responsible for P. aeruginosa growth inhibition in mixed-species macrocolonies when iron is restricted.

In a previous study, the amount of lactate exported in the supernatant of E. faecalis V583 Δldh1 was lower than that for E. faecalis V583 (65). Thus, to investigate the possible role of l-lactate in inhibiting PADP6 growth in mixed-species macrocolonies grown under iron restriction conditions, we quantified the extracellular l-lactate. We detected significantly more l-lactate from OG1RF single-species macrocolonies grown under iron-restricted than under unchelated conditions (Fig. 2C), indicating that OG1RF increased l-lactate production in iron-restricted media is independent of PADP6 presence. Increased l-lactate production was further supported by transcriptomic data, in which we observed upregulation of ldh1 (log2FC [fold change] = 0.57) in OG1RF single-species macrocolonies grown in iron-restricted media compared to those grown in unchelated media (Table 1). Next, we detected a significant reduction of l-lactate in mixed PADP6 and OG1RF macrocolonies compared to OG1RF single-species macrocolonies in unchelated media, whereas l-lactate levels were comparable between single- and mixed-species macrocolonies under iron-restricted conditions (Fig. 2C). Based on transcriptomic data, we observed an upregulation of ldh1 (log2FC = 3.27) in mixed PADP6 and OG1RF macrocolonies grown in iron-restricted media compared to those grown in unchelated media (Table 1). Together, these data indicate that ldh1 is upregulated when iron is restricted, leading to increased production and export of l-lactate. In previous studies, lactate was found to chelate iron (66–68), which we confirmed under our assay conditions for ferric iron (Fig. 2D). Consequently, even mild iron-chelating effects of E. faecalis-derived l-lactate could further restrict iron availability in the environment.

TABLE 1.

E. faecalis l-lactate dehydrogenase differentially regulated under iron-restricted conditions^a

Comparison conditions	Locus tag	Name	Description	log2FC	P value	FDR
OG1RF single-species macrocolonies grown in iron restriction relative to OG1RF single-species macrocolonies grown in unchelated conditions	OG1RF_10199	ldh1	l-Lactate dehydrogenase	0.57	8.75E−05	1.68E−03
Mixed OG1RF and PADP6 macrocolonies grown in iron restriction relative to mixed OG1RF and PADP6 macrocolonies grown in unchelated conditions	OG1RF_10199	ldh1	l-Lactate dehydrogenase	3.27	9.02E−25	8.91E−24

^aThe complete table can be found in supplemental file 1.

To confirm this, we compared PADP6 and OG1RF transcriptomes of mixed PADP6 and OG1RF macrocolonies to mixed PADP6 and OG1RF Δldh1 macrocolonies grown in iron-restricted media. Unfortunately, due to the low PADP6 cell numbers in the mixed PADP6 and OG1RF macrocolonies and therefore lesser numbers of PADP6 raw counts, we were unable to draw conclusions regarding iron availability based on the gene expression profile of PADP6 (Table S2). However, OG1RF iron acquisition genes, such as ABC transporters (69), were upregulated in the mixed PADP6 and OG1RF macrocolonies, suggesting that iron availability was restricted when l-lactate levels were high (Table S3). Consistent with this, increased l-lactate levels in the environment were inversely correlated with PADP6 growth in the mixed PADP6 and OG1RF macrocolonies under iron-restricted conditions (Fig. 2A and C).

Due to the anionic nature of l-lactate at all metabolic pH, it cannot pass through the E. faecalis cell membrane freely (64). As a result, l-lactate is exported out of E. faecalis as lactic acid (57) with a pK_a of 3.86 (70–72), which is lower than the surrounding environmental pH. Therefore, upon export, lactic acid is deprotonated to l-lactate and releases H⁺ into the environment, resulting in a lowered environmental pH. To further investigate the role of l-lactate, we added increasing amounts of lactic acid as a supplement to mixed PADP6 and OG1RF Δldh1 macrocolonies in 22D-chelated media. We observed a dose-dependent inhibition of PADP6 growth in the mixed-species macrocolonies with increasing lactic acid concentrations from 2.5 mM to 10 mM (Fig. 2E), while OG1RF Δldh1 growth in the mixed-species macrocolonies remained relatively unchanged (Fig. 2F).

l-Lactate is necessary, but not sufficient, for inhibiting P. aeruginosa growth under iron-restricted conditions.

To investigate whether l-lactate alone was sufficient for inhibiting PADP6 growth in iron-restricted media, we grew PADP6 single-species macrocolonies supplemented with increasing 22D and lactic acid concentrations. When supplemented with 10 mM or 20 mM lactic acid in unchelated media, PADP6 growth remained relatively unchanged compared to that in unchelated media without lactic acid supplementation (Fig. 3A). Based on the lactic acid supplementation results obtained in Fig. 2E and F, we expected that PADP6 growth would be inhibited when supplemented with 10 mM lactic acid under iron-restricted conditions. However, upon supplementation of 10 mM lactic acid, we observed a significant inhibition of PADP6 growth only when it was grown in 4 mM 22D-chelated media, but not in 2 mM and 3 mM 22D-chelated media compared to unchelated media (Fig. 3A), whereas, when media were supplemented with 20 mM lactic acid, we observed significant PADP6 growth inhibition at all tested 22D concentrations, compared to unchelated media (Fig. 3A). A possible explanation for why more lactic acid was needed to inhibit single-species PADP6 growth in 2 mM 22D could be that E. faecalis ldh2 is also contributing to l-lactate production in OG1RF Δldh1, and hence, a smaller amount of lactic acid was sufficient to inhibit PADP6 growth in the mixed-species macrocolonies. Taken together, these data demonstrate that l-lactate is necessary, but not sufficient, for PADP6 growth inhibition in iron-restricted media.

FIG 3.

P. aeruginosa growth inhibition is due to lowered environmental pH under iron-restricted conditions. (A) Enumeration of PADP6 from 48-h single-species macrocolonies grown in TSBG medium supplemented without and with increasing 22D concentrations (2, 3, and 4 mM), which is then further supplemented without and with lactic acid (10 and 20 mM). Dotted lines represent inoculum of bacteria spotted. n ≥ 3 with 3 technical replicates; error bars represent standard deviation from the mean. Statistical analysis was performed using the Mann-Whitney U test: ns, not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001. (B) Enumeration of PADP6 from 48-h single-species macrocolonies grown in 2 mM 22D-chelated TSBG medium supplemented without and with 20 mM lactic acid (pH unadjusted and pH adjusted to pH 6.60) or 20 mM citric acid (pH unadjusted and pH adjusted to pH 6.61). Dotted lines represent inoculum of bacteria spotted, and dashed lines represent limit of detection. n = 3 with 3 technical replicates; error bars represent standard deviationl from the mean. Statistical analysis was performed using the Mann-Whitney U test: ns, not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001. (C) Planktonic growth of PADP6 in pH-unadjusted (pH 7.29) and pH-adjusted (pH 6.52, 5.50, 4.52, and 3.54) TSBG medium supplemented with 2 mM 22D. n = 3 with 3 technical replicates; error bars represent standard deviation from the mean.

Decreased environmental pH under iron restriction inhibits P. aeruginosa growth.

Next, we investigated why PADP6 viability was lost in the presence of l-lactate under iron-restricted conditions. Upon export of lactic acid by E. faecalis, it is deprotonated into l-lactate and H⁺. We therefore examined whether a lowered environmental pH contributes to PADP6 growth inhibition when iron is restricted. Supplementation of iron-restricted media with 20 mM pH-unadjusted lactic acid (pH 2.69) resulted in significant PADP6 growth inhibition compared to the absence of lactic acid supplementation (Fig. 3B). In contrast, PADP6 growth was unaffected when we supplemented the media with 20 mM pH-adjusted lactic acid (pH 6.60) using sodium hydroxide (NaOH) when iron was otherwise restricted (Fig. 3B). To examine if other organic acids had the ability to inhibit P. aeruginosa growth, we also supplemented the iron-restricted media with citric acid to lower the environmental pH. Supplementation with pH-unadjusted citric acid (pH 2.41) significantly inhibited PADP6 growth to below the limit of detection, while no significant growth difference was observed upon supplementation with 20 mM pH-adjusted citric acid using NaOH (pH 6.61) or in the absence of citric acid supplementation (Fig. 3B). Moreover, alleviation of the low pH in iron-restricted media with PIPES (1,4-piperazinediethanesulfonic acid disodium salt), MOPS [3-(N-morpholino)propanesulfonic acid], or HEPES buffer partially rescued P. aeruginosa growth in the mixed-species macrocolonies (Fig. S6). To further assess the impact of low pH on P. aeruginosa growth, we grew PADP6 planktonically in pH-unadjusted and pH-adjusted 22D-chelated media. When grown in 22D-chelated media at pH 5.50 and lower, we observed minimal to no PADP6 growth (Fig. 3C), suggesting that P. aeruginosa growth is generally inhibited when the bacterium is grown in low-pH environments as previously shown (73, 74). These data also help in explaining an interesting observation seen in Fig. 1F and Fig. 2A, whereby when PADP6 was grown in mixed-species macrocolonies with OG1RF, the CFU per macrocolony were ~2 × 10⁷ for 24-h macrocolonies (Fig. 1F) and ~9 × 10⁵ for 48-h macrocolonies (Fig. 2A), suggesting that P. aeruginosa grows from its initial inoculum of ~5 × 10⁵ CFU but eventually dies off likely due to the drop in pH and competitive depletion of iron. Altogether, these data show that the lowered environmental pH as a consequence of E. faecalis lactic acid export, coupled with l-lactate-mediated chelation of iron, plays a critical role in P. aeruginosa growth under iron-restricted conditions.

DISCUSSION

In this study, we sought to characterize the polymicrobial interactions between commonly coisolated E. faecalis and P. aeruginosa under an iron-restricted condition. We show that E. faecalis inhibits P. aeruginosa growth in biofilms when iron availability is restricted in an LDH-dependent manner. Since E. faecalis Ldh1 catalyzes the reduction of pyruvate to l-lactate, the increased ldh1 expression translates to an increased l-lactate production in E. faecalis. The l-lactate produced is then exported as lactic acid which then gets deprotonated into l-lactate, releasing H⁺ in the surrounding environment. Together, the chelation of iron by l-lactate and the lowered-pH environment contribute to P. aeruginosa growth inhibition under iron-restricted conditions.

An outstanding question from this work is how iron restriction leads to an upregulation of E. faecalis ldh1. E. faecalis possesses two copies of the ldh gene, ldh1 and ldh2, both encoding l-lactate dehydrogenase (64). Both isoenzymes contribute to l-lactate production through catalyzing the reduction of pyruvate to l-lactate (64). The activity of both isoenzymes is regulated by fructose-1,6-bisphosphate, intracellular phosphate, and pH levels (75). The principal l-lactate dehydrogenase, encoded by ldh1, was suggested to be posttranscriptionally regulated upon different growth rates (76). The transcription of ldh1 is also activated by CcpA, a global transcription regulator of carbon catabolite repression, by binding to a catabolite-responsive element (cre) box identified upstream of the ldh1 gene (77). However, little is known about whether iron levels play any role in regulating ldh1 expression. We show that E. faecalis ldh1 expression is increased in iron-restricted media, and this is supported by an increased amount of l-lactate measured in the surrounding environment of the macrocolonies grown. Ferric uptake regulator (Fur) is a transcription factor involved in regulating iron uptake and homeostasis (78). The DNA-binding sequence of Fur is well studied (69, 79, 80), and we tried using different DNA Fur binding motifs to identify possible Fur binding sites upstream of ldh1, but we did not find any that resemble known binding motifs. Despite this, a previous study reported that ldh1 is differentially expressed between E. faecalis OG1RF and OG1RF Δfur mutant, suggesting that ldh1 expression is directly or indirectly influenced by Fur (69). An upregulation of ldh1 and several genes involved in iron transport is also observed in another transcriptome study done when E. faecalis was exposed to urine (81). It is therefore consistent that in iron-restricted environments, such as urine or E. faecalis macrocolonies growing in iron-restricted media, ldh1 expression is induced. This observation is not limited to E. faecalis as ldh1 expression is similarly induced under iron-restricted conditions for the anaerobe Clostridium acetobutylicum (82). Although it remains unclear how iron levels influence ldh1 expression in E. faecalis, it is tempting to speculate that there is an interplay between iron levels and energy metabolism during growth under iron-restricted conditions.

Lactic acid exported by E. faecalis is deprotonated to l-lactate and H⁺. We show that the consequent lowered environmental pH contributes to P. aeruginosa growth inhibition when iron is restricted. This is not surprising as P. aeruginosa growth is generally affected at low pH and the bacterium prefers to grow in a more neutral pH range (73, 74). In fact, E. faecalis V583 adopts a similar strategy of lowering environmental pH as a result of lactic acid export to inhibit Klebsiella pneumoniae growth in polymicrobial biofilms (83). P. aeruginosa siderophores pyoverdine and pyochelin have an iron formation stability constant of approximately 10²⁴ and 10⁵ M⁻¹, respectively (84, 85), while l-lactate has a stability constant of approximately 10² M⁻¹ (66). The binding affinity for iron and zinc of the Escherichia coli Nissle siderophore yersiniabactin changes according to pH, in that yersiniabactin preferentially binds to zinc as pH increases (86). In addition, a high concentration of H⁺ is required for dechelation of iron-siderophore complexes (87). As such, the low-pH environment around E. faecalis likely affects the binding affinity of iron and promotes dissociation of iron from its siderophore complexes before it can be taken up by specific outer membrane receptors into P. aeruginosa. Moreover, varying the pH of the medium alters the production of pyoverdine and pyochelin (88). Even though a P. aeruginosa PADP6 strain that is able to grow under iron-restricted conditions is used to study the polymicrobial interactions with E. faecalis, the lowered iron availability resulted from 22D-mediated iron chelation and l-lactate-mediated iron chelation is likely an added stress apart from the lowered-pH environment. After all, iron is an essential element for many cellular and metabolite processes (89, 90). Together, the low-pH environment created by E. faecalis might possibly affect P. aeruginosa siderophore-mediated iron uptake and consequently negatively impact its growth due to insufficient intracellular iron.

Interestingly, P. aeruginosa did not inhibit E. faecalis growth in the presence or absence of iron restriction. P. aeruginosa is known to produce antimicrobial metabolites that are toxic toward other Gram-positive bacteria such as Staphylococcus aureus. For example, the antimicrobial metabolite 2-heptyl-4-hydroxyquinoline N-oxide (HQNO) inhibits both the growth of S. aureus (91) and the activity of succinate:quinone oxidoreductase from Bacillus subtilis (92). Moreover, P. aeruginosa produces rhamnolipids which alter the cell surface of S. aureus, resulting in increasing membrane permeability (93). Despite these antimicrobial metabolites, P. aeruginosa and S. aureus are often coisolated in infections (48, 94) as P. aeruginosa attenuates its virulence when cocolonized with S. aureus and hence is more permissive to S. aureus growth (95, 96). We hypothesize that the inability of P. aeruginosa to inhibit E. faecalis growth may also be due to a similar attenuation of P. aeruginosa virulence when grown with E. faecalis. In addition, rather than being in an antagonistic relationship, P. aeruginosa and E. faecalis could be symbiotic when iron is not restricted or under different environmental conditions. There is evidence between E. faecalis and E. coli, as well as between P. aeruginosa and S. aureus, whereby interspecies metabolic feeding benefits one species and promotes polymicrobial biofilms and infections (59, 97). Thus, although P. aeruginosa possesses the ability to produce antimicrobial metabolites, E. faecalis might be cross-feeding a metabolite(s) that is beneficial to P. aeruginosa, and as a result, P. aeruginosa may tolerate E. faecalis in the right environment.

Even though we showed that E. faecalis antagonizes P. aeruginosa under iron-restricted conditions in vitro, they are often coisolated in vivo (48–50, 52). The contrasting observations that we made in vitro and in vivo could be due to in-host bacterial or host mechanisms that negate or override the antagonism. Our findings show that, in addition to iron-restrictive effects, P. aeruginosa growth inhibition is largely dependent on the effects of pH arising from export of E. faecalis lactic acid into the environment. As such, coexistence of E. faecalis and P. aeruginosa in vivo may be due to host influence on E. faecalis l-lactate production or to the host buffering the environmental pH, negating E. faecalis l-lactate effects (98). Another possibility could be differences in in vitro and in vivo spatial organization of E. faecalis and P. aeruginosa. When P. aeruginosa and E. faecalis were grown in vitro, they exhibited distinct spatial separation in which P. aeruginosa formed a structured biofilm above the E. faecalis biofilm (99). However, during polymicrobial wound infection (P. aeruginosa, E. faecalis, S. aureus, and Finegoldia magna), P. aeruginosa is seen throughout the wound bed as well as at the leading edge of the wound (51). Host factors might also contribute to and affect spatial organization as this is evident in gut microbiome spatial organization (100, 101). Spatial structuring that keeps the two bacteria physically separated could therefore blunt any local pH and iron competition effects.

Based on the current findings, we propose a working model of E. faecalis and P. aeruginosa polymicrobial interactions in vitro (Fig. 4). During coculture of E. faecalis and P. aeruginosa in unchelated media, l-lactate produced by E. faecalis is exported out via a symporter with H⁺ as lactic acid. Since the pK_a of lactic acid is lower than the pH of the environment, lactic acid gets deprotonated into l-lactate and H⁺, acidifying the environment. The l-lactate in the environment then chelates iron in the media (Fig. 4A). In contrast, during coculture under iron-restricted conditions, there is an upregulation of E. faecalis ldh1 expression which translates to increased l-lactate production and lactic acid exported, further acidifying the environment. As E. faecalis continues to grow and lactic acid accumulates over time, the l-lactate in the environment further restricts iron availability in the iron-restricted media and the acidity of the environment subsequently exceeds a pH threshold at which P. aeruginosa can no longer grow (Fig. 4B).

FIG 4.

Proposed working model of E. faecalis and P. aeruginosa in vitro polymicrobial interactions. Interactions between E. faecalis and P. aeruginosa under unchelated (A) and iron-restricted (B) conditions. (A) In unchelated conditions, l-lactate produced in E. faecalis is exported with hydrogen ions via a symporter (purple) as lactic acid, which is then deprotonated in the environment into l-lactate and hydrogen ions (H⁺). This l-lactate then chelates iron in the environment. (B) In iron-restricted conditions, E. faecalis ldh1 expression is upregulated. Consequently, as E. faecalis grows, l-lactate production and lactic acid secretion increase. This further chelates iron under iron-restricted conditions and lowers the environmental pH to a point at which P. aeruginosa cannot grow. The figure was created with BioRender.com.

Many infections are often polymicrobial, and our work emphasizes the importance of how changes in the microenvironment such as iron or pH levels can significantly influence the interactions between two bacterial species. Despite the contrasting observations for E. faecalis and P. aeruginosa antagonism in vitro and in vivo, exploration of the mechanistic basis of antagonistic relationships between bacteria is informative because knowledge of such in vitro antagonism between bacteria has the potential to be used as a basis for additional control strategies against specific bacterial pathogens in the management of infections.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

Bacterial strains used in this study are listed in Table S4 in the supplemental material. Unless stated, all P. aeruginosa and E. faecalis bacterial strains were grown at 37°C under shaking or static conditions for 16 to 18-h, respectively. Cells were harvested by centrifugation at 12,000 × g for 5 min, and cell pellets were washed twice with 1 mL of 1× sterile phosphate-buffered saline (PBS). The final pellet was resuspended in 3 mL of 1× sterile PBS prior to measurement of optical density at 600 nm (OD₆₀₀). Cell suspensions were then normalized to the required cell number for different experimental assays. For Pseudomonas selection, bacteria were spotted onto Pseudomonas isolation agar (PIA) (Difco BD, USA) supplemented with 100 µg/mL ampicillin (Sigma-Aldrich, USA). For E. faecalis OG1RF selection, bacteria were spotted onto tryptone soya broth (TSB) (Oxoid, Canada) solidified with 1.5% agar (Oxoid technical no. 3) and supplemented with 10 mM glucose (TSBG), 10 µg/mL colistin (Sigma-Aldrich, USA) and 10 µg/mL nalidixic acid (Sigma-Aldrich, USA), respectively.

Planktonic, static biofilm, and macrocolony biofilm assays.

Bacterial cultures were normalized to 1 × 10⁸ to 2 × 10⁸ CFU/mL in 1× PBS. Macrocolonies were produced by inoculating 5 μL of the respective bacterial cultures onto the surface of TSB or TSBG solidified with 1.5% agar and incubated at 37°C for either 24-h or 48-h. For mixed-species macrocolonies, bacterial species were mixed at a 1:1 ratio. When appropriate, the TSB or TSBG agar was further supplemented with or without 2,2′-bipyridyl (22D) (Sigma-Aldrich, USA), iron(III) chloride hexahydrate (Merck, USA), citric acid (Merck, USA), lactic acid, iron(II) sulfate heptahydrate, copper(II) chloride anhydrous, manganese(II) sulfate, magnesium(II) chloride, zinc chloride anhydrous, 1,4-piperazinediethanesulfonic acid disodium salt (PIPES), 4-morpholinepropanesulfonic acid, 3-(N-morpholino)propanesulfonic acid (MOPS), or HEPES buffer (all purchased from Sigma-Aldrich, USA). Macrocolonies were excised and resuspended in 2 mL of sterile 1× PBS, followed by bacterial enumeration on the respective selection agar. For supernatant transfer and static biofilm assays, single- and mixed-species inocula were prepared as described above and inoculated in TSBG medium for 24-h at 37°C unless stated otherwise. For static biofilm assays, CFU enumeration was performed first by scraping the wells of 6-well microtiter plates and then pipetting to mix homogenously and serially diluting mixtures for plating on selective agar plates. For preparation of supernatant media, 24-h biofilms were first scraped, suspended in conical tubes (cells and spent medium together), and centrifuged at 4,000 rpm for 20-min to pellet the cells. Spent medium was transferred to a new tube and filter sterilized to obtain cell-free supernatant. The cell-free supernatant was then mixed with fresh TSBG medium or water and then supplemented with 1 mM 22D, prior to inoculating for subsequent growth at 37°C under static conditions for 24-h. For planktonic assay, 5 μL of the respective bacterial cultures was inoculated into TSBG medium and incubated at 37°C under shaking conditions for 24-h.

Construction of PADP6 and PADP6-mCherry strains.

Overnight cultures of PAO1-WT were diluted to 10⁹, 10⁸, 10⁷, and 10⁶ CFU/mL in 1× PBS, and 300 µL of each cell suspension was plated onto LB Lennox agar (Difco BD, USA) supplemented with 1.5, 2, 2.5, 3, and 4 mM 22D. Plates were incubated at 37°C for 24 to 36-h. PAO1-WT and a few colonies that grew under the 3 mM 22D-chelated condition were isolated, and genomic DNA (gDNA) was extracted using the Wizard genomic DNA purification kit (Promega, USA) for use in whole-genome sequencing. PADP6 was chromosomally tagged with mCherry through triparental conjugation using PADP6 as a recipient with delivery plasmid pUC18-miniTn7-P_tac-mCherry (E. coli) and helper plasmid pTNS1 (E. coli), resulting in PADP6-mCherry (102–104).

Genome sequencing and analysis.

Raw reads were imported into CLC Genomics Workbench 8.0 (Qiagen, Germany), followed by quality trimming to remove bad-quality reads. The trimmed reads were then mapped to the reference genome before the Basic Variant Detection module was used to detect mutations using the default parameters. The mutations detected in the isolate were then filtered against the PAO1-WT control to determine the mutations acquired for survival under iron-restricted conditions.

Planktonic growth assay.

Bacterial cultures were normalized to an OD₆₀₀ of 0.01 in the respective media and inoculated into 24-well microtiter plates. All microtiter plates were incubated at 37°C under shaking conditions. Planktonic growth was measured by recording the OD₅₉₅ between 30-min and 1-h intervals using a Tecan Infinite M200 Pro spectrophotometer (Tecan Group Ltd., Switzerland) until early stationary growth phase was reached.

RNA extraction from planktonic cultures and absolute quantification by RT-qPCR.

P. aeruginosa cultures were normalized to an OD₆₀₀ of 0.01 in the respective media and inoculated into 24-well microtiter plates. All microtiter plates were incubated at 37°C under shaking conditions. Planktonic growth was measured at OD₆₀₀ at regular intervals using a Tecan Infinite M200 Pro spectrophotometer (Tecan Group Ltd., Switzerland) until an OD₆₀₀ of approximately 0.5 was reached. Bacteria were then harvested in RNAprotect bacterial reagent (Qiagen, Germany) and incubated at room temperature for 5-min before centrifugation at 10,000 × g for 10-min. The supernatant was decanted, and bacterial pellets collected were subjected to total RNA extraction using an RNeasy minikit (Qiagen, Germany) according to the manufacturer’s protocol. Extracted RNA samples were subsequently treated with DNase (Turbo DNA-free kit; Invitrogen, USA) for removal of contaminating genomic DNA before the RNA was purified with the Monarch RNA cleanup kit (New England Biolabs, USA). The concentration of RNA and potential DNA contamination were quantified using Qubit RNA BR and Qubit double-stranded DNA (dsDNA) HS assay kits, respectively (Invitrogen, USA). The extracted RNA was also quality checked using a TapeStation instrument (RNA ScreenTape; Agilent Technologies, USA). RNA samples with a maximum of 10% DNA contamination and a RNA integrity number equivalent (RINe) value of ≥7.5 were used for reverse transcription-quantitative PCR (RT-qPCR). Equivalent amounts of RNA across all samples were converted to cDNA using SuperScript III First-Strand Synthesis SuperMix (Invitrogen, USA). Following cDNA synthesis, absolute quantification of mexA, mexB, and oprM was performed using the Kapa SYBR Fast qPCR master mix (2×) kit (Kapa Biosystems, USA). The primers used for amplification of mexA, mexB, and oprM are listed in Table S5 in the supplemental material.

Pyoverdine quantification.

P. aeruginosa cultures were normalized to an OD₆₀₀ of 0.01 in the respective media and inoculated into 24-well microtiter plates. All microtiter plates were incubated at 37°C under shaking conditions. Planktonic growth was measured at OD₆₀₀ at regular intervals using a Tecan Infinite M200 Pro spectrophotometer (Tecan Group Ltd., Switzerland) until an OD₆₀₀ of approximately 0.5 was reached. Cell-free supernatant was then obtained from these samples. Pyoverdine secretion was quantified by measuring the fluorescence intensity of the supernatants (excitation at 400 nm and emission at 450 nm) and normalizing it to the respective OD₆₀₀ of each sample.

E. faecalis transposon library screen.

An E. faecalis OG1RF mariner transposon library consisting of 14,978 mutants was cryogenically stocked in 96-well microtiter plates (105). These OG1RF transposon mutants were cultured in 180 µL brain heart infusion (BHI) broth at 37°C under static conditions for 16 to 18-h in 96-well microtiter plates using a cryoreplicator (Adolf Kühner AG, Switzerland) and spotted onto BHI agar plates for incubation at 37°C for 24-h. Following that, OG1RF transposon mutants from the BHI agar plates were cultured for primary screening in 180 µL BHI broth as described above. PADP6-mCherry cultures were grown and washed as described above. Both the OG1RF transposon mutant cultures and PADP6-mCherry were normalized to an OD₆₀₀ of 0.01 in TSBG medium supplemented with 1.2 mM 22D. A primary screen of the E. faecalis transposon library was done by mixing the normalized OG1RF transposon mutant cultures and PADP6-mCherry at a 1:1 ratio in 96-well microtiter plates (total volume of 200 µL). The microtiter plates were then incubated at 37°C under static conditions for 22-h. The growth of PADP6-mCherry was quantified by measuring mCherry fluorescence intensity (excitation = 480 nm, emission = 615 nm) using a Tecan Infinite M200 Pro spectrophotometer. Secondary validation of the OG1RF transposon mutants was performed by mixed-species macrocolony biofilm assays as described above to quantify the growth of PADP6 and each transposon mutant (CFU per milliliter) under 1 mM 22D iron-restricted conditions at 37°C for 24-h.

Molecular cloning.

The primers used in this study are listed in Table S5. Transformants were screened using respective selection agar as follows: (A) E. coli strains, LB with 500 µg/mL erythromycin (pGCP213), and (B) E. faecalis strains, BHI with 25 µg/mL erythromycin (pGCP213). Generation of E. faecalis knockout mutants was done by allelic replacement using a temperature-sensitive shuttle vector described previously (106). Vector pGCP213 was linearized using restriction enzymes (New England Biolabs, USA) for the construction of OG1RF Δldh1 and OG1RF Δldh1::ldh1. Linearized vector and inserts were ligated using the In-Fusion HD cloning kit (Clontech, TaKaRa, Japan) and transformed into Stellar competent cells. Successful plasmid constructs were verified by Sanger sequencing and subsequently extracted and transformed into OG1RF. Transformants were selected with erythromycin at 30°C and then passaged at a nonpermissive temperature at 42°C with erythromycin to select for bacteria with successful plasmid integration into the chromosome. For plasmid excision, bacteria were serially passaged at 37°C without erythromycin for erythromycin-sensitive colonies. These colonies were then subjected to PCR screening for detection of deletion mutant (OG1RF Δldh1) or chromosomal complementation of ldh1 (OG1RF Δldh1::ldh1).

Lactate-Glo assay.

l-Lactate quantification was done using the Lactate-Glo assay kit (Promega, USA). A 3-cm by 3-cm section of agar surrounding the macrocolonies was excised, resuspended in 5 mL of sterile 1× PBS, and homogenized to measure secreted l-lactate in the agar surrounding the macrocolonies. Supernatants were then collected by centrifuging the homogenate at 5,000 × g for 10-min and used for l-lactate quantification. Briefly, an equal volume of lactate detection reagent was added to the supernatant and incubated for 60-min at room temperature before luminescence was read using a Tecan Infinite M200 Pro spectrophotometer.

Total iron quantification.

Iron quantification was done using the iron assay kit (Colorimetric) (Abcam, UK) per the manufacturer’s instruction. Prior to quantification, samples were prepared by adding different concentrations of 22D or sodium l-lactate (Sigma-Aldrich, USA) as a supplement to 200 μM iron(II) sulfate heptahydrate (FeSO₄·7H₂O) and iron(III) chloride hexahydrate (FeCl₃·6H₂O). The output was measured immediately at OD₅₉₃ using a Tecan Infinite M200 Pro spectrophotometer. The iron concentration in each sample was computed based on the standard curve generated using the iron standards.

RNA extraction from macrocolonies.

Matured single- and mixed-species macrocolonies grown for 48-h on TSBG agar supplemented with and without 2,2′-bipyridyl, in biological triplicates, were first scraped into RNAprotect bacterial reagent (Qiagen, Germany) and incubated at room temperature for 5-min before centrifugation at 10,000 × g for 10-min. The supernatant was decanted, and bacterial pellets collected were then subjected to total RNA extraction using an RNeasy minikit (Qiagen, Germany) with slight modifications. Briefly, cell pellets were resuspended in Tris-EDTA (TE) buffer containing 20 mg/mL lysozyme (Sigma-Aldrich, USA), and each sample was further supplemented with 20 µL proteinase K (Qiagen, Germany). This was followed by incubation at 37°C for 1-h, and subsequent extraction steps were performed according to the manufacturer’s protocol. Extracted RNA samples were treated with DNase (Turbo DNA-free kit; Invitrogen, USA) for removal of contaminating genomic DNA before the RNA was purified with the Monarch RNA cleanup kit (New England Biolabs, USA). The concentration of RNA and potential DNA contamination were quantified using Qubit RNA BR and Qubit dsDNA HS assay kits, respectively (Invitrogen, USA). The extracted RNA was quality checked using a TapeStation instrument (RNA ScreenTape; Agilent Technologies, USA) before it was sent for sequencing. Every sample had to have a minimum RNA concentration of 40 to 80 ng/µL, a maximum of 10% DNA contamination, and a RINe value of ≥8.0, before being used for library preparation and subsequent sequencing as 100-bp paired-end reads on an Illumina HiSeq2500 at the Singapore Centre for Environmental Life Sciences Engineering (SCELSE) sequencing facility.

Transcriptomic analysis.

The raw reads obtained were checked using FastQC (version 0.11.9) and adaptor trimmed using BBDuk from BBMap tools (version 39.79) (107). Trimmed reads were then mapped using bwa-mem of BWA (version 0.7.17-r1188) with options “-T 20 -k 13” against the E. faecalis OG1RF (NCBI accession no. CP002621) or P. aeruginosa PAO1 (NCBI accession no. NC_002516) reference genome. Reads mapped to open reading frames were quantified using HTSeq-count of HTSeq (version 0.12.4) with option “-m intersection-strict” (108). Ribosomal sequences were filtered out from all data sets. Differential gene expression analysis was performed in R using edgeR (version 3.28.1) (109). The log₂ fold change values extracted were based on the false-discovery rate (FDR) of ≤0.05.

Statistical analysis.

Statistical analyses were performed with GraphPad Prism software (version 9.0.0; CA, USA) and are described in the respective figure legends.

Data availability.

All transcriptome sequencing (RNA-seq) sequences were deposited in the National Center for Biotechnology Information Gene Expression Omnibus database under accession number GSE190090.