Enterococcus faecalis redox metabolism activates the unfolded protein response to impair wound healing

Abstract

Enterococcus faecalis is an opportunistic pathogen that thrives in biofilm-associated infections and delays wound healing, yet how it impairs host tissue responses is unclear. Here, we identified extracellular electron transport (EET) as a previously unrecognized source of reactive oxygen species (ROS) in E. faecalis and showed that this activity directly triggers the unfolded protein response (UPR) in epithelial cells and delays epithelial cell migration. ROS detoxification with catalase suppressed E. faecalis–induced UPR and rescued epithelial cell migration, while exogenous hydrogen peroxide was sufficient to restore UPR activation in EET-deficient strains. UPR disruption by pharmacological inhibition also impaired cell migration, highlighting a critical role for UPR homeostasis in wound repair. Our findings establish EET as a virulence mechanism that links bacterial redox metabolism to host cell stress and impaired repair, offering previously unidentified avenues for therapeutic intervention in chronic infections.

E. faecalis EET generates ROS, which induces the UPR in keratinocytes, inhibiting in vitro migration.

INTRODUCTION

Enterococcus faecalis is a gut commensal and opportunistic pathogen that causes difficult-to-treat biofilm-associated infections, including catheter-associated urinary tract infection, infective endocarditis, and chronic wound infections (1, 2). In wound settings, E. faecalis infection is associated with delayed epithelial migration and immune dysregulation (3). The success of E. faecalis in these environments is often attributed to its metabolic adaptability, including survival under nutrient limitation and oxidative stress (4). However, the extent to which E. faecalis metabolism actively interferes with host repair mechanisms is poorly understood.

One way that host cells respond to environmental and infection-induced insults is through the unfolded protein response (UPR), an evolutionarily conserved signaling pathway triggered by endoplasmic reticulum (ER) stress. The UPR integrates signals related to protein misfolding, membrane perturbations, and redox imbalance to restore homeostasis or induce apoptosis if stress persists (5–8). Pathogens have evolved diverse strategies to manipulate the host UPR, often targeting its three main pathways [inositol-requiring enzyme 1 (IRE1), protein kinase R-like ER kinase (PERK), and activating transcription factor 6 (ATF6)] to subvert host cell function, modulate immune responses, or even exploit UPR-regulated products as a nutrient source (9–14). For example, group A Streptococcus (GAS) secretes streptolysins that induce the host UPR via PERK, altering host gene expression to up-regulate asparagine synthetase. This increases asparagine secretion, which is sensed by GAS, altering its gene expression to favor pathogenicity (9, 14). While some bacterial toxins and effectors can induce the UPR, in most cases, the microbial mechanisms by which the UPR is activated or dysregulated are undefined (11, 15–17).

E. faecalis generates substantial extracellular reactive oxygen species (ROS), including superoxide and hydrogen peroxide, in the absence of aerobic respiration or fumarate reduction (18, 19). This ROS can inflict DNA damage and tissue injury in infection models, as well as modulate host signaling pathways including redox-sensitive pathways such as the UPR (20–22). However, the bacterial source of ROS in E. faecalis and its mechanistic consequences for host cell function have not been fully elucidated.

In this study, we show that E. faecalis activates the host UPR during wound infection. Through a forward genetic screen and functional validation, we identify extracellular electron transport (EET) as a previously unrecognized mechanism by which E. faecalis generates ROS, which, in turn, activates the UPR in epithelial cells and impedes their migration following wounding. We show that EET and associated demethylmenaquinone (DMK) biosynthesis pathways are required for superoxide and hydrogen peroxide generation, mutants of which produce less ROS, fail to activate the UPR, and do not impair epithelial cell migration. These findings not only establish a role for EET in ROS generation but also, through its interaction with the host UPR, establish it as a metabolic virulence mechanism by which E. faecalis disrupts epithelial repair, thereby presenting previously unidentified opportunities for targeting chronic E. faecalis–driven pathologies.

RESULTS

E. faecalis infection activates the UPR in a mouse model

We previously showed that E. faecalis infection impairs wound healing (3). To gain insight into the mechanisms that influence delayed wound repair, we reanalyzed our published single-cell RNA sequencing (scRNA-seq) dataset of E. faecalis–infected mouse wounds at 4 days postinfection (dpi) (Fig. 1A, GSE229257) (22). For each class, we calculated enrichment scores for a panel of stress-response signatures (23) (table S1). Gene set enrichment for canonical UPR targets revealed that the ER stress response is not global but concentrated in immune cells (macrophages and neutrophils) and, most notably, in an infection-specific cluster of keratinocytes (Fig. 1, B and C). By contrast, oxidative stress response (OSR) genes were up-regulated not only in UPR-elevated keratinocytes and immune cells but also in fibroblasts (Fig. 1D and figs. S1, A and B), consistent with high fibroblast redox activity in infected tissue. We previously found that E. faecalis infection interferes with wound closure signatures, drives a partial epithelial-to-mesenchymal transition (EMT) in keratinocytes, and skews macrophages toward an anti-inflammatory phenotype (22). The robust UPR response in keratinocytes offers a mechanistic clue, where excessive ER stress alters EMT dynamics in these cells, undermining their migratory role and thereby hindering wound repair during E. faecalis infection.

Fig. 1. E. faecalis infection activates the UPR in a mouse model.

(A) Uniform Manifold Approximation and Projection (UMAP) of ~23,000 single-cell transcriptomes from uninfected and E. faecalis–infected wounds (22), recolored here into the six broad cell classes used for downstream stress-response analyses. (B) Per-cell enrichment score for a curated UPR gene set projected onto the UMAP in (A). (C) Same UPR enrichment as in (B) but displayed only for keratinocyte clusters. (D) Enrichment of an OSR gene set plotted for fibroblast clusters. (E) Schematic representation of the UPR in mice, where spliced Xbp1 (Xbp1s), Chop, and Herpud1 serve as downstream markers of the IRE1, PERK, and ATF6 pathways, respectively. (F) Gene expression of UPR markers (Xbp1s, Chop, and Herpud1) at 6 dpi for uninfected and E. faecalis–infected 6- to 7-week-old C57BL/6J mouse skin wounds, normalized to intact skin. FC, fold change. Significance was determined using one-way analysis of variance (ANOVA), Dunnett’s test (unwounded skin, n = 16; wounded, uninfected skin, n = 20; wounded, infected skin, n = 33; ***P < 0 0.001 and ****P < 0.0001).

To corroborate the scRNA-seq findings, we infected full-thickness excisional wounds in mice with E. faecalis strain OG1RF and enumerated bacterial colony-forming units (CFU) from the wounds at 6 dpi, chosen to target the proliferation and remodeling phase of healing, where we observed a two-log reduction in E. faecalis bacteria burden in wounds compared to the starting inoculum, with no significant animal weight loss or attrition, consistent with our previous studies (3) (fig. S1, C and D). We quantified UPR-associated mRNA levels of spliced Xbp1 (Xbp1s), Chop, and Herpud1 as markers of Ire1, Perk1, and Atf6 activity, respectively, in whole wound tissue and unwounded skin samples (Fig. 1E). Xbp1s transcripts were significantly higher in uninfected wounds at 6 dpi compared to unwounded skin, indicating that the UPR is activated in wounds regardless of infection state (Fig. 1F). A similar observation was reported in wounded transgenic mice expressing a x-box binding protein 1-luciferase fusion (XBP1-Luc) fluorescence marker 8 to 10 days postwounding (24). By contrast, both Xbp1s and Chop transcript levels were significantly higher in E. faecalis–infected wounds at 6 dpi compared to unwounded skin samples, while Herpud1 levels remained unchanged, indicating a lack of ATF6 activation (Fig. 1F). Together with the single-cell results, these data suggest that E. faecalis infection results in UPR dysregulation, which could affect normal wound healing (25).

IRE1 activation by E. faecalis impedes keratinocyte migration in vitro

To corroborate our in vivo and in silico findings, we examined UPR activation in keratinocytes (HaCaT) and fibroblasts (NIH-3T3), which are the dominant cell types in healthy skin that contribute to wound healing (26). E. faecalis infection significantly increased mRNA expression of all three UPR pathway markers in NIH-3T3 cells (Fig. 2A), whereas in HaCaT cells, only XBP1s and CHOP were significantly up-regulated (Fig. 2B). As a positive control, we treated both cell lines with tunicamycin (Tm), which induces the UPR by inhibiting protein glycosylation in the ER leading to an accumulation of unfolded proteins (27). Tm treatment significantly up-regulated all three UPR pathway markers in both cell lines to a greater extent compared to E. faecalis infection. Since IRE1 is the most evolutionarily conserved branch of the UPR, we further examined its activation by E. faecalis by assessing the expression of XBP1s target genes (28, 29). These included the ER chaperone binding immunoglobulin protein (BiP) (encoded by HSPA5) and ER degradation-enhancing alpha-mannosidase-like protein 1 (EDEM1), which promote ER homeostasis and are hallmarks of IRE1 activation (30, 31). Infected cells exhibited a significant increase in EDEM1 transcripts along with elevated levels of XBP1s and BiP proteins (Fig. 2, C and D), confirming that E. faecalis activates the conserved IRE1 pathway in vivo and in vitro.

Fig. 2. IRE1 activation by E. faecalis impedes keratinocyte migration in vitro.

(A and B) Gene expression of UPR markers (Xbp1s, Chop, and Herpud1) in infected with E. faecalis at MOI of 800 (A) or 600 (B) or Tm treated NIH-3T3 mouse fibroblasts (A) or HaCaT human keratinocytes (B) (n = 3). See Materials and Methods for multiplicities of infection (MOI) optimization description. (C) Gene expression of IRE1 downstream gene (EDEM1) in cells treated as in (A) (n = 3). (D) Quantitative analysis and representative immunoblots showing levels of XBP1s and BiP in HaCaT cells treated as in (A). (E) Scratch wound assay quantification for uninfected, infected, and Tm-treated cells (positive control). Shaded areas represent 1 SD for each condition (n = 4 biological replicates). (F) Gene expression of XBP1s from HaCaT cells treated with the dimethyl sulfoxide (DMSO) control (open circles) or the IRE1 inhibitor (IRE1i) 4μ8c (closed circles) (n = 3, one-way ANOVA, Dunnett’s test). (G) Scratch wound assay quantification for uninfected and infected cells treated with 0.5% DMSO control (open circles) or IRE1i (close circles) (n = 4 biological replicates). Shaded areas represent 1 SD for each condition. Significance was determined using one-way ANOVA Dunnett’s test [(A) to (C)], two-way ANOVA Tukey’s test [(E) and (G)], or one-way ANOVA Tukey’s test (F) (*P < 0.05, **P < 0.01, ***P < 0 0.001, and ****P < 0.0001).

We next investigated the impact of infection-induced UPR on wound closure using an in vitro HaCaT scratch wound assay. By 15 hours postinfection (hpi), both E. faecalis–infected and Tm-treated cells exhibited significantly slower migration compared to uninfected cells, with the difference becoming more pronounced by 27 hpi (Fig. 2E, fig. S2A, and movie S1). Notably, neither infected nor Tm-treated cells displayed appreciable migration throughout the assay period, and infected cells even showed signs of wound edge retraction as early as 6 hpi, consistent with cell shrinkage or detachment that can be early indicators of apoptosis. To determine whether UPR induction via IRE1 was responsible for impaired migration, we treated both uninfected and infected cells with 4μ8c, an IRE1 inhibitor (IRE1i) that blocks the ribonuclease activity of IRE1 (32). Treatment with 50 μM IRE1i was sufficient to block E. faecalis–induced IRE1 in HaCaT cells (Fig. 2F). IRE1i also slowed migration in uninfected wells, with the delay evident by 18 hpi and more pronounced at 27 hpi (Fig. 2G, fig. S2B, and movie S2). In infected cells, migration was minimal regardless of IRE1i treatment, showing no significant migration at 27 hpi relative to the baseline at 3 hpi.

We did not use proliferation inhibitors, such as mitomycin C, which are typically used to differentiate between proliferation and migration in these assays, because the compound caused widespread cell detachment in infected cells during preliminary experiments. Nonetheless, we attribute the observed wound closure primarily to cell migration rather than proliferation for several reasons: (i) the short 24-hour duration of the assay, (ii) the long doubling time of confluent HaCaT cells (~32 to 36 hours), and (iii) previous studies establishing that migration is the dominant factor in similar short-term scratch assays (25). Thus, our findings suggest that neither UPR hyperactivation (during infection) nor hypoactivation (after IRE1 inhibition) alone fully explains keratinocyte migration; rather, both extremes of UPR activity are implicated.

EET drives E. faecalis UPR activation and migration arrest

To dissect how E. faecalis induces the UPR, we designed a high-throughput assay for UPR induction: a NIH-3T3 reporter line (3T3R) that fluoresces when IRE1 splices a 26-nt intron from a truncated human XBP1 fused to mApple (Fig. 3, A and B). 3T3R was generated from NIH-3T3 cells transduced with lentiviral particles containing pLVX-XBP1-mApple-nuclear localization signal (NLS). While IRE1 is quiescent, a premature stop codon between XBP1 and mApple blocks translation, causing cells to remain nonfluorescent. When IRE1 is activated, intron excision shifts the reading frame, removes the stop codon, and allows production of the full XBP1s-mApple fusion, generating a red signal quantifiable by wide-field microscopy. Using this reporter, we screened a defined transposon (Tn) library of 14,976 E. faecalis OG1RF mutants (Fig. 3A) to identify mutants that failed to trigger UPR activation.

Fig. 3. EET drives E. faecalis UPR activation and migration arrest.

(A) E. faecalis OG1RF Tn screen in which a library of 14,976 mutants was screened against an NIH-3T3 cell line expressing the Xbp1-mApple reporter system (3T3R). (B) Representative epifluorescence microscopy images of 3T3R under different conditions (uninfected, WT, and Tm-treated) at 21 hpi. Scale bar, 100 μm. (C) Diagram showing the pathways in which a subset of UPR-defective mutants (genes/proteins with red font) were identified. DHNA, 1,4-dihydroxy-2-naphthoic acid; GPP, geranyl diphosphate; HPP, heptaprenyl diphosphate; ETC, electron transport chain. (D) Validation of UPR-defective mutants with 3T3R (n = 3). (E) UPR induction by WT and ΔT7SS in a 3T3(R) background (n = 3). (F and G) Scratch wound assay quantification for HaCaT infected with (F) WT and ΔEET, which were also (G) treated with either 0.5% DMSO control (open circles) or IRE1i (close circles). WT data are identical to Fig. 2 (E and G) and replicated here for ease of comparison (n = 4 biological replicates). The data underlying Fig. 2 (E and G) and (F) and (G) were generated in the same experimental sets and can therefore be directly compared against each other. Shaded areas represent 1 SD for each condition. (H) UPR induction at 24 hpi in HaCaT after treatment with 0.5% DMSO control (open circles) or IRE1i (closed circles) under uninfected, WT-infected, and ΔEET-infected conditions. Uninfected and WT-infected findings are identical to Fig. 2F and replicated here for ease of comparison (n = 3). Significance was determined using one-way ANOVA Dunnett’s test [(D) and (E)], two-way ANOVA Tukey’s test [(F) and (G)], or one-way ANOVA Tukey’s test (H) (***P < 0 0.001 and ****P < 0.0001).

Applying an upper threshold of 30% XBP1s-mApple–positive (XBP1s⁺) cells in a given population, we identified 457 UPR-defective mutants, corresponding to 369 distinct genetic loci (fig. S3A and table S2). We filtered out mutants with multiple Tn insertions, insertions outside of coding regions, and those with significant growth defects after overnight culture in brain-heart infusion (BHI) medium. Pathway analysis did not show consistent enrichment patterns; however, we noticed that a substantial number of mutants had insertions in genes associated with carbohydrate catabolism and respiration, key components of redox metabolism. This observation prompted us to focus on pathways involved in the synthesis of components and cofactors that facilitate electron flow from carbohydrate catabolism to terminal electron acceptors. Tn insertions mapped to three respiratory pathways of E. faecalis: heme-dependent aerobic respiration (ndh2 and cydD), fumarate reduction (frd), and EET (ndh3 and eetB) (Fig. 3C). Additional insertions were identified in genes involved in the biosynthesis of the quinone electron carrier DMK (menF and menE), as well as upstream precursors such as chorismate and geranyl pyrophosphate, generated either by the shikimate pathway (aroD) or the mevalonate pathway (ispA). DMK is an integral part of all three respiratory pathways by mediating electron transfer between their membrane-associated components.

We validated mutants of these respiratory pathways identified in the primary screen and found that only mutants disrupted in quinone electron carrier synthesis (menE, menF, and aroD) and EET (ndh3 and eetB) displayed reduced UPR induction compared to wild-type (WT) E. faecalis (Fig. 3D). Since these genes are involved in central metabolic processes, we quantified the growth of each mutant to ensure that reduced UPR induction was not simply due to impaired bacterial replication during infection. However, all Tn mutants grew similarly to WT in cell culture medium (fig. S3, B and C), eliminating growth defects as a confounding factor. Given the strong association between EET and UPR induction, we also tested a deletion mutant lacking entire EET operon (ΔEET) (Fig. 3, C and D), which has minimal EET activity via ferric reductase assay (fig. S3D). As expected, the ΔEET mutant was defective in UPR induction yet retained growth and antibiotic susceptibility profiles similar to the parental OG1RF WT strain (fig. S3, B, C, and E). To rule out the possibility that lower UPR induction by ΔEET could be due to higher cytotoxicity causing cell loss resulting in weaker XBP1s-mApple fluorescent signals, we assessed cytotoxicity in HaCaT cells at 3 and 24 hpi with WT and ΔEET. There was no difference in cytotoxicity between uninfected, WT-infected, or ΔEET-infected cells at 3 hpi (fig. S3F). However, WT- but not ΔEET-infected cells had significantly higher cytotoxicity compared to uninfected cells at 24 hpi. The higher cytotoxicity in WT-infected cells at a delayed time point of 24 hpi but not immediately after infection at 3 hpi suggests that WT infection was not directly causing cell death but rather indirectly via UPR hyperactivation. The lack of difference at 24 hpi between uninfected and ΔEET-infected cells also confirms the UPR-defective nature of ΔEET, as delayed cell death is a hallmark characteristic of chronic UPR hyperactivation (33, 34). To further confirm that the loss of UPR induction was specific to the EET pathway and not to secreted toxins, as is the case for GAS (14, 35), we tested a deletion mutant of the type 7 secretion system (ΔT7SS), encoding predicted secreted toxins (36). The ΔT7SS mutant induced the UPR to the same extent as WT (Fig. 3E), supporting the conclusion that the phenotype of the ΔEET mutant is a direct consequence of its function in redox metabolism.

On the basis of these characteristics, ΔEET was selected as the model UPR-defective mutant for downstream studies. We used this mutant to assess whether E. faecalis UPR induction via the EET pathway contributes to the inhibition of keratinocyte cell migration. Unlike WT infection, ΔEET did not significantly impair migration, instead showing similar migration comparable to uninfected controls (Figs. 2G and 3F, fig. S3G, and movie S3). However, treating ΔEET-infected cells with IRE1i resulted in significantly slower migration at the 27-hpi end point (Fig. 3G, fig. S3H, and movie S4). These findings suggest that lack of UPR induction by ΔEET allows for physiological levels of UPR induction that support normal cell migration. Furthermore, IRE1i treatment induces UPR hypoactivation in ΔEET-infected cells (Fig. 3H), dysregulating UPR homeostasis and impairing cell migration, similar to that observed in uninfected cells treated with IRE1i (Figs. 2G and 3G). The recovery of cell migration with ΔEET infection and its reversal by IRE1i treatment demonstrate that E. faecalis EET is associated with UPR induction and impaired cell migration.

EET-derived ROS is sufficient to activate the UPR, disrupting epithelial migration

Disruption of DMK synthesis in E. faecalis impairs both EET function and extracellular superoxide generation (18, 37). OG1RF mutants disrupted in genes involved in quinone electron carrier synthesis (aroE, aroC, aroA, menB, menD, and menE) produce less superoxide (O₂^·−) than WT (19). Superoxide radicals undergo pH-dependent spontaneous dismutation to generate hydrogen peroxide (H₂O₂), which can then participate in Fenton chemistry to generate hydroxyl radicals and other ROS in the presence of transition metals (Fig. 4A). H₂O₂ has been shown to induce the UPR in myotubule and epithelial cells at 50 and 200 μM, respectively (20, 38). However, this may not be a universal response among host cells, as another group reported no significant increase in XBP1s expression for fibroblast when treated with 1 mM H₂O₂ (39). We hypothesized that superoxide generation via EET drives UPR induction in epithelial cells via H₂O₂. To test whether H₂O₂ alone is sufficient to induce the UPR in our models, we treated 3T3R cells with increasing concentrations of H₂O₂ for 3 hours, followed by a 21-hour recovery period. XBP1s fluorescence increased significantly at 250 and 500 μM (Fig. 4B). This finding confirms that NIH-3T3 cells can mount a UPR in response to H₂O₂, supporting the idea that ROS generated via EET contributes to UPR induction during infection.

Fig. 4. EET-derived ROS is sufficient to activate the UPR, disrupting epithelial migration.

(A) Diagram showing the dismutation of superoxide radical (O₂^·−) into hydrogen peroxide (H₂O₂) via the catalytic activity of superoxide dismutase (SOD) or spontaneously via a pH-dependent process. H₂O₂ can be converted into a highly reactive hydroxyl radical in the presence of transition metals such as ferrous ions (Fe²⁺) via the Fenton reaction or neutralized into water and oxygen if catalase is present. (B) Dose-dependent H₂O₂-mediated UPR induction in 3T3R (n = 3). (C) In-frame deletion mutants corresponding to the Tn mutants of the EET and DMK synthesis pathways that were identified as UPR-defective hits during the Tn screen. The ΔMEN mutant had its entire operon deleted, which contained the menF/D/C/E/B genes (n = 3). (D to F) UPR induction in 3T3(R) for (D) uninfected (E) WT infection, and (F) ΔEET infection, in the presence (closed circles) or absence (open circles) of H₂O₂, SOD, and catalase (n = 3). (G to I) UPR induction in uninfected (G), WT-infected (H), and ΔEET-infected (I) HaCaT cells in the presence (closed circles) or absence (open circles) of H₂O₂ and catalase (n = 3). (J) Lipid oxidation in uninfected, WT-infected, and ΔEET-infected HaCaT cells using BODIPY 581/591 C11 (n = 3). (K to M) Scratch wound assay quantification in uninfected (K), WT-infected (L), or ΔEET-infected (M) HaCaT cells treated with catalase and/or H₂O₂. Shaded areas represent 1 SD for each condition (n = 3 biological replicates). Significance was determined using one-way ANOVA, Dunnett’s test [(B) and (J)], one-way ANOVA, Tukey’s test [(C) to (I)], or two-way ANOVA, Tukey’s test (K to M) (*P < 0.05, **P < 0.01, ***P < 0 0.001, and ****P < 0.0001).

Next, to test whether EET generates ROS, we quantified superoxide generation by validated UPR-defective mutants (ΔEET and Tn insertion mutants in menE, menF, aroD, ndh3, and eetB). At the same time, we generated in-frame deletion mutants for each of these genes. All exhibited significantly lower superoxide generation compared to WT and non-UPR–defective mutants (ispA, frd, ndh2, and cydD), and the in-frame deletion mutants were similar to their respective Tn mutants (Fig. 4C and fig. S4A), supporting a link between EET-dependent superoxide production and UPR activation.

It has been previously shown that hemin supplementation attenuates superoxide generation in heme-auxotrophic E. faecalis by activating the heme-dependent cytochrome bd pathway (19). It was hypothesized that the oxidation of divalently reduced DMK via cytochrome bd reduces spontaneous oxidation of univalently reduced DMK radicals, thereby attenuating superoxide generation (19). Therefore, to confirm that superoxide generation is mediated by the EET and not by spontaneous univalent oxidation of reduced DMK radicals, we also quantified superoxide generation with 10 μM hemin. If superoxide generation is DMK mediated, then heme supplementation should not decrease superoxide generation in the DMK mutants (ΔMEN, ΔmenE, and ΔmenF) but will decrease superoxide generation in the EET mutants (ΔEET, Δndh3, and ΔeetB) since they have an intact DMK synthesis pathway. However, if superoxide generation is EET mediated, then heme supplementation should not decrease superoxide generation in the EET mutants but will decrease superoxide generation in the DMK mutants since they have an intact EET pathway. Comparing hemin-free and hemin-supplemented medium, superoxide generation was significantly reduced in WT and DMK mutants but not in any of the EET mutants (Fig. 4C). This validated the EET as the primary pathway for superoxide generation in E. faecalis.

Superoxide dismutase (SOD) and catalase are antioxidants that catalyze the dismutation of superoxide radicals into H₂O₂ or H₂O₂ into water and oxygen, respectively (Fig. 4A). To determine whether E. faecalis–derived ROS drives UPR induction, we treated infected 3T3R cells with exogenous SOD and catalase. In control experiments, neither enzyme altered XBP1s expression in uninfected cells, whereas treatment with 250 μM H₂O₂ to simulate ROS-induced stress robustly increased XBP1s expression (Fig. 4D). In cells stimulated with H₂O₂, the addition of catalase significantly reduced XBP1s expression, whereas SOD had no effect. This result confirms that H₂O₂ is the specific ROS driving UPR induction in this assay.

In WT-infected cells, catalase again significantly reduced XBP1s expression, while SOD had no effect (Fig. 4E). Adding H₂O₂ to WT-infected cells did not further increase UPR activation, suggesting that E. faecalis–generated ROS levels are already saturating, whereas cotreatment with catalase reversed this effect. By contrast, catalase had no impact on ΔEET-infected cells, consistent with their low ROS generation (Fig. 4F). However, functional complementation of the ΔEET with H₂O₂ partially restored UPR induction, which was again reversed by catalase, confirming that H₂O₂ is sufficient to rescue the UPR-defective phenotype when EET is disrupted. These findings were replicated in HaCaT cells (Fig. 4, G to I). Since lipid peroxidation is a known cause of UPR activation (40), we next measured it in infected cells using a BODIPY 581/591 C11 fluorescent probe. WT-infected cells exhibited significantly more lipid peroxidation than uninfected controls, while cells infected with the ΔEET mutant showed no significant change (Fig. 4J). These data support a model where EET-derived ROS from E. faecalis induces the UPR via lipid peroxidation.

To determine whether E. faecalis–generated ROS impairs wound healing, we assessed in vitro wound closure after treatment with H₂O₂ or/and catalase. H₂O₂ treatment of uninfected cells significantly slowed cell migration at 27 hpi, although no wound edge retraction was observed, unlike in WT-infected cells (Fig. 4K, fig. S4B, and movie S5). Cotreatment with catalase restored cell migration to physiological levels. Similarly, catalase treatment of WT-infected cells significantly improved cell migration, which approached that of uninfected controls and without retraction at 27 hpi (Fig. 4L, fig. S4C, and movie S6). Adding H₂O₂ to untreated WT-infected cells had no further effect, consistent with UPR saturation seen earlier (Fig. 4, E and H). Cotreatment with H₂O₂ and catalase mirrored catalase-treatment alone, reinforcing the role of E. faecalis–derived H₂O₂ in cell migration arrest (Fig. 4L, fig. S4C, and movie S6). Despite H₂O₂ functionally complementing ΔEET in terms of UPR induction and inhibiting cell migration in uninfected cells, adding H₂O₂ to untreated ΔEET-infected cells did not significantly slow cell migration at 27 hpi (Fig. 4M, fig. S4D, and movie S7), suggesting that cell migration is influenced by other host pathways activated by the presence of E. faecalis besides H₂O₂-mediated UPR induction (41–43). Nevertheless, these findings demonstrate that the EET pathway of E. faecalis generates ROS, which oxidizes lipids in epithelial cells, hyperactivates the UPR, and inhibits cell migration.

DISCUSSION

Our findings identify E. faecalis EET as a virulence mechanism that links bacterial redox metabolism to host stress responses and impaired tissue repair. We show that EET-dependent production of ROS, particularly H₂O₂, activates the UPR in epithelial cells. UPR dysregulation by E. faecalis disrupts normal epithelial function and cell migration, revealing a direct mechanistic connection between bacterial energy metabolism and host healing processes.

E. faecalis is known to produce ROS, which has been previously linked to host cellular injury and even carcinogenesis (21). In this study, we refine the genetic basis of this activity in terms of a dedicated electron transport system. Previous studies showed that E. faecalis generates extracellular superoxide through a process that requires DMK, which is diminished when terminal quinol/cytochrome oxidases are functional, suggesting that respiratory disruption changes the dynamics of electron flow to one that favors the reduction of oxygen to superoxide, resulting in ROS production (19). In this model, reduced form of nicotinamide adenine dinucleotide (oxidized form) incompletely reduces DMK, producing semiquinone intermediates that, in the absence of classical aerobic or anaerobic respiration, donate electrons univalently to molecular oxygen, generating superoxide, which spontaneously dismutates into hydrogen peroxide that can contribute to host oxidative stress (19). This movement of electrons may be mediated by soluble shuttles, such as flavins (44, 45). In this study, we refine the mechanistic basis of E. faecalis ROS generation by identifying a critical role for EET in this process.

While disrupting aerobic respiration is required for ROS generation and EET function (19, 37), our data show that the disruption of aerobic respiration alone is not sufficient and that ROS generation also depends on an intact EET system. However, the mechanistic contribution of specific EET components involved in ROS generation remains to be determined. Whether these components overlap with those required for extracellular metal reduction or electrode respiration or represent a distinct branch of EET machinery activated under redox stress remains an important question for future work.

The requirement for EET in E. faecalis ROS production aligns with growing evidence that extracellular electron transfer in Gram-positive bacteria extends beyond classical anaerobic respiration. While canonical diderm systems such as Shewanella and Geobacter use cytochromes and conductive pili to transfer electrons to external acceptors (46, 47), recent studies show that monoderm bacteria, including Listeria monocytogenes and Lactobacillus plantarum, use flavin-based EET to support energy conservation, redox homeostasis, and virulence under host-relevant conditions (44, 45, 48). In particular, E. faecalis and L. monocytogenes rely on EET for fitness in the mouse gastrointestinal tract (44, 49), and L. plantarum uses EET to enhance adenosine 5′-triphosphate yield via substrate-level phosphorylation in the absence of a classical respiratory chain (50). We have previously shown significant up-regulation in EET genes (ndh3, eetB, and dmkB) at 8 hpi and significant fitness defects for the same mutants at 3 dpi in a mouse wound infection model (51). Our findings position E. faecalis within this emerging framework and reveal a distinct facet: In the presence of oxygen and fermentable sugars, EET contributes to univalent electron transfer from reduced DMK to oxygen, generating superoxide and hydrogen peroxide. This is noteworthy, as the use of oxygen as an electron acceptor is normally associated with canonical aerobic respiration in E. faecalis, in which heme-dependent cytochrome bd functions as a terminal oxidase that transfers electrons to oxygen to generate water as well as a proton-motive force. However, E. faecalis EET uses oxygen as an electron acceptor to generate copious amounts of ROS. This noncanonical aerobic respiration allows E. faecalis, a heme auxotroph, to use oxygen as an electron acceptor to regenerate nicotinamide adenine dinucleotide when exogenous heme is absent, bestowing it with a greater degree of metabolic versatility without having to invest in metabolically costly heme synthesis pathways. Furthermore, EET-mediated superoxide generation is only moderately attenuated when heme is present, suggesting that this phenotype could be physiologically and clinically relevant in a much broader range of E. faecalis–associated pathologies beyond wound infections. Unlike environmental microbial EET systems that avoid oxygen to prevent radical formation, E. faecalis appears to exploit this interaction, thereby linking EET to oxidative stress at the host-pathogen interface.

The UPR is increasingly recognized as a central node in host stress responses during infection, particularly in epithelial and immune cells. Multiple pathogens including Salmonella enterica, Helicobacter pylori, Pseudomonas aeruginosa, Streptococcus pyogenes, and Brucella melitensis have been shown to manipulate the UPR through secreted toxins or effector proteins, often to promote intracellular survival or dampen immune responses (9–11, 14, 52). Here, we demonstrate that E. faecalis selectively activates the IRE1 and PERK arms of the UPR both in vivo and in vitro, manifesting in impaired epithelial cell migration. Unlike previously described examples, E. faecalis induces the UPR independent of dedicated virulence factors, instead leveraging metabolic ROS production via EET. Similar redox-based virulence strategies have been described in other pathogens. Streptococcus pneumoniae produces hydrogen peroxide via SpxB, contributing to epithelial damage (53, 54), and P. aeruginosa phenazines generate intracellular ROS in airway cells (55, 56). However, E. faecalis is the first example, to our knowledge, where a defined EET system is shown to drive ROS production that directly alters host stress signaling and function.

A compelling hypothesis arising from our data is that the observed UPR hyperactivation and impaired cell migration are consequences of ferroptosis. Ferroptosis is a regulated, iron-dependent form of cell death distinct from apoptosis, driven by the catastrophic accumulation of lipid peroxides. This process is typically restrained by the antioxidant enzyme glutathione peroxidase 4, and its failure leads to membrane damage (57, 58). Ferroptosis is increasingly recognized as a critical factor in diverse pathologies and host-pathogen interactions (59–61). The accumulation of lipid peroxides, the biochemical hallmark of ferroptosis, is a known trigger of severe ER stress, providing a direct mechanistic link to UPR activation (40, 62, 63). Our findings align remarkably well with this framework. The lipid peroxidation we observed in infected keratinocytes is the defining feature of ferroptosis, and the subsequent cell retraction is a classic morphological correlate of this death pathway (57, 58, 64). This model mechanistically connects E. faecalis EET-driven ROS production to the downstream cellular pathologies of lipid stress, UPR activation, and inhibition of cell migration. While our experiments with catalase confirm ROS as the primary trigger, they do not exclude ferroptosis as the ultimate executioner pathway. We therefore propose that the E. faecalis–host interaction may represent a model of infection-induced ferroptosis. Testing this hypothesis, for which our study provides the foundational rationale, will be a critical next step and could be directly addressed by employing specific inhibitors like ferrostatin-1.

Our data further show that UPR activation must be tightly regulated for effective wound healing. These findings are consistent with previous studies in aged keratinocytes and fibroblasts displaying higher baseline levels of UPR markers and slower in vitro wound cell migration, which could be reversed upon treatment with 4-phenylbutyrate, a broad-acting UPR inhibitor (25). Similarly, we show that pharmacologic inhibition of IRE1, the key UPR sensor, led to hypoactivation and impaired cell migration even in uninfected cells, underscoring the importance of physiological UPR signaling during repair. Regulated UPR induction is also necessary for other wound healing processes such as the differentiation of dermal fibroblasts to myofibroblasts, which promote wound contracture and collagen deposition (65). Furthermore, UPR inhibitors especially PERK inhibitors have demonstrated cytotoxicity to pancreatic islet cells that depend on mild UPR induction to perform their secretory function (66). Even if UPR induction is not lowered to below baseline levels, UPR inhibition will still be counterproductive in restoring homeostasis in pathologies where host cells are already dependent on some level of UPR induction to perform physiological processes. Therefore, targeting the source of UPR dysregulation, i.e., bacterial ROS production, rather than the host UPR, may constitute a more effective therapeutic strategy.

In summary, this work reveals that E. faecalis leverages its respiratory machinery not only for metabolic flexibility but also to perturb host cell physiology. While EET has been linked to efficient infection of the gastrointestinal tract, this work presents molecular details that may contribute to its role in pathogenesis. Future studies should examine the role of EET in vivo, its regulation and contribution within polymicrobial settings, and the potential for targeting redox metabolism to mitigate E. faecalis infections that are increasingly recalcitrant to antibiotic therapy.

MATERIALS AND METHODS

Bacterial strains and growth conditions

All bacterial strains used in this study are listed in table S3. E. faecalis strains were routinely cultured on BHI (Neogen, #NCM0016A) agar plates and grown in BHI broth. Escherichia coli strains, used for DNA isolation and plasmid manipulation, were cultured in Luria-Bertani (LB) broth or on LB agar plates at 37°C. When required, antibiotics were added at the following final concentrations: erythromycin, 500 μg/ml for E. coli and 25 μg/ml for E. faecalis; rifampin, 25 μg/ml; and chloramphenicol, 10 μg/ml. To prepare overnight cultures, a single E. faecalis colony was inoculated into a 14-ml tube containing 4 ml of BHI broth, with the lid tightly sealed. Cultures were grown statically for 18 to 24 hours at 37°C. The following day, the overnight culture was centrifuged (4000g for 10 min), and the bacterial pellet was washed once with phosphate-buffered saline (PBS) before being resuspended in 1 ml of complete Dulbecco’s modified Eagle’s medium (DMEM) (see the “Cell culture” section). The bacterial suspension was then normalized by optical density at 600 nm (OD₆₀₀) to a concentration equivalent to 8 × 10⁸ CFU/ml (OD₆₀₀ = 1) and further adjusted depending on the specific experimental application.

Mouse wound excisional model

All in vivo procedures were approved by the Institutional Animal Care and Use Committee at Nanyang Technological University, Singapore (protocol no. ARF SBS/NIEA-0314), in accordance with national guidelines. Male C57BL/6J mice (6 to 7 weeks old, 22 to 25 g; InVivos, Singapore) were housed under specific pathogen–free conditions. The wound infection model was adapted from a previous study (3). Briefly, mice were anesthetized with 3% isoflurane, and dorsal hair was removed using clippers and a depilatory cream (Nair). The skin was disinfected with 70% ethanol, and a 6-mm full-thickness excisional wound was created using a sterile biopsy punch (Miltex, Integra, #33-36). The wound was immediately inoculated with 10 μl of an E. faecalis OG1RF suspension containing 2 × 10⁶ CFU. The wound was then sealed with a transparent dressing (Tegaderm, 3M, #7100252702) to prevent contamination. Postprocedure, mice were housed individually to prevent wound disruption. At the experimental end point, mice were euthanized, and a 1-cm² piece of skin tissue centered on the wound was excised. For bacterial enumeration, tissues were collected in sterile PBS, homogenized, and plated on BHI agar supplemented with rifampin to confirm infection by the inoculated strain. For molecular analysis, tissues were homogenized in either TRIzol reagent (Thermo Fisher Scientific, #15596026) for RNA extraction or ice-cold radioimmunoprecipitation assay (RIPA) buffer (Thermo Fisher Scientific, #89900) supplemented with a protease inhibitor cocktail (Roche, #11697498001) for protein extraction.

scRNA-seq integration and downstream analysis

Single-cell datasets (GSE229257) were reprocessed in R 4.3.2 (67) following the workflow established in our original study (22). Raw matrices were imported with Seurat 5.1.0 (68–71). Cells expressing <200 genes, >6000 genes, or >12% mitochondrial reads were removed; genes detected in <5 cells were discarded. Library size variation was normalized with SCTransform (method = “glmGamPoi,” vst.flavour = “v2”). Batch effects between biological replicates (uninfected versus E. faecalis–infected wounds) were corrected with Seurat’s reciprocal principal components analysis (PCA) integration (30 PCs). Principal components (n = 30) were used for FindNeighbors/FindClusters (resolution = 0.4) and RunUMAP (dims = 1 to 30). Broad cell classes were assigned on canonical markers, as previously reported (22), and subclustering of keratinocytes and fibroblasts used a second round of PCA/Uniform Manifold Approximation and Projection (UMAP) at resolution = 0.6. All UMAPs use color-blind-safe palettes generated with RColorBrewer 1.1-3 (72) (brewer.pal, palette = “Set2”).

Differential expression between infected and uninfected cells within each cluster was ranked by the Wilcoxon area under curve statistic using wilcoxauc (73) (presto 1.0.0). Ranked lists served as input for gene set enrichment analysis with fgseaMultilevel (fgsea 1.28.0, minSize = 15, maxSize = 5,000, eps = 0) (74, 75). Gene sets for UPR, OSR, and heat shock response were curated from MSigDB (v2023.1). Enrichment was considered significant at Benjamini-Hochberg false discovery rate of <0.05 (76). For each cluster, the positive, significant normalized enrichment scores for infected cells were projected onto the UMAPs.

RNA extraction, reverse transcription, and quantitative real-time polymerase chain reaction

Total RNA was extracted from cell lines and homogenized mice wound samples using the EZ-10 DNAaway RNA Miniprep Kit (Biobasic, #BS88136) following the manufacturer’s protocol. RNA concentrations were quantified using the Qubit Broad Range RNA Quantification Assay (Thermo Fisher Scientific, #Q10210), together with the Qubit 3 Fluorometer following the manufacturer’s protocol. cDNA was synthesized from extracted RNA normalized to 1 μg per sample using the RevertAid Reverse Transcriptase (Thermo Fisher Scientific, #EP0441) following the manufacturer’s protocol. Quantitative polymerase chain reaction (qPCR) was performed using the Luna Universal qPCR Master Mix (New England Biolabs, #M3003E) together with a CFX-96 (Bio-Rad) or a QuantStudio3 (Thermo Fisher Scientific) Real-Time PCR system following the manufacturer’s protocol. Each 20 μl of reaction has a final concentration of cDNA (2.5 ng/μl) and primer pairs (0.25 μM; table S4) for target genes. Relative mRNA was normalized to the housekeeping gene Gapdh/GAPDH using the 2^−∆∆Ct method (77).

Cell culture

Murine embryonic fibroblasts (NIH-3T3) and human keratinocytes (HaCaT) were cultured in DMEM (Gibco, #11995065) supplemented with 10% heat-inactivated fetal bovine serum (Cytiva, #SV30160.03) and 4 mM GlutaMAX (Gibco, #35050061). This medium is referred to as “complete DMEM.” The lentivirus packaging line, 293FT, was cultured in complete DMEM further supplemented with 0.1 mM minimum essential medium nonessential amino acids (Gibco, #11140050), 6 mM l-glutamine (Gibco, #25030081), 1 mM sodium pyruvate (Cytiva, #SH30239.01), and geneticin (500 μg/ml; Gibco, #10131027). All cells were maintained at 37°C in a humidified 5% CO₂ incubator. Cells were washed once with PBS (Gibco, #14190144) and detached using 0.25% Trypsin-EDTA. Trypsinization times were 15 min for HaCaT, 5 min for 293FT, 4 min for 3T3R, and 3 min for NIH-3T3 cells. The reaction was neutralized with an equal volume of complete DMEM. Cells were pelleted by centrifugation, resuspended in fresh medium, and counted using a Countess 3 Automated Cell Counter. For experiments, cells were seeded in 12-well plates at densities of 2.85 × 10⁴ cells/cm² (NIH-3T3), 1.15 × 10⁴ cells/cm² (3T3R), or 1.15 × 10⁵ cells/cm² (HaCaT) and incubated for 24 hours before use. For the Tn scree, 3T3R cells were seeded in 96-well plates at 2.85 × 10⁴ cells/cm². Unless otherwise specified, the following final concentrations of reagents were used: Tm at 0.2 μg/ml (for qPCR), 2.5 μg/ml (for scratch wound assays), or 5 μg/ml (for immunoblot/microscopy); the IRE1i 4μ8c at 50 μM (MedChemExpress, #HY-19707), added 1 hour before infection; H₂O₂ at 250 μM; catalase at 100 U/ml; and SOD (Sigma-Aldrich, #S5395) at 100 U/ml.

Optimization of in vitro infection

Assay conditions were optimized across (i) multiple time points and (ii) multiplicities of infection (MOIs) to identify those that maximize in vitro XBP1s expression.

In vitro infection

Confluent NIH-3T3 and HaCaT cells were infected with E. faecalis at an MOI of 800 (800 CFU per host cell) and 600 (600 CFU per host cell), respectively. Following a 3-hour infection period, the medium was removed, and cells were washed three times with PBS. To eliminate extracellular bacteria, fresh complete medium supplemented with a gentamicin-penicillin antibiotic cocktail (50 μg/ml) was added, and the cells were incubated for an additional 21 hours (78).

Immunoblotting

Cells were lysed with RIPA buffer (Thermo Fisher Scientific, #89901) supplemented with reconstituted cOmplete protease inhibitor cocktail (Roche, #11697498001) by gentle agitation on ice for 5 min before centrifugation for 15 min at 12,000g at 4°C. A mixture of 15 μg of total proteins was separated on 10% SDS–polyacrylamide gel electrophoresis and transferred on nitrocellulose membranes. Immunoblotting was performed with appropriate primary antibodies and IRDye-conjugated secondary antibodies (table S5). Proteins were visualized using the near-infrared fluorescence system (Odyssey CLx Imaging System).

In vitro scratch wound assay model

Scratch assays were performed in 12-well plates, adapting a previously published protocol to facilitate automated microscopy (79). Confluent HaCaT cell monolayers were scratched with a sterile P200 pipette tip and subsequently infected as described in the “In vitro infection” section. Following the postinfection wash, fresh complete medium supplemented with antibiotics and 25 mM Hepes (Gibco, #15630080) was added to each well. Wound closure was monitored on a Zeiss Axio Observer 7 microscope (×10 magnification), acquiring bright-field images every 30 min for 45 hours in a controlled environment (37°C at 5% CO₂). The resulting time-lapse images were analyzed using a customized CellProfiler pipeline to quantify the scratch area (80). To ensure accuracy, images where the wound area was misidentified by the automated pipeline were manually measured in ImageJ using the Wound Healing Size Tool plugin (81).

XBP1s reporter cell line

pLVX-XBP1-mNeonGreen-NLS plasmid was a gift from D. Andrews (table S6). Codon-optimized mApple cDNA was synthesized as a gBLOCK fragment (IDT) and inserted in pLVX-XBP1-mNeonGreen-NLS to replace mNeonGreen using Gibson Assembly (New England Biolabs, #E2611L) according to the manufacturer’s protocol. The resulting plasmid, pLVX-XBP1-mApple-NLS, was sequence verified and then cotransfected into 293FT cells with the pLP1, pLP2, and pLP/vesicular stomatitis virus glycoprotein packaging plasmids (table S6). Supernatants containing lentiviral particles were harvested at 36 and 60 hours posttransfection, pooled, and filtered (0.45 μm). NIH-3T3 cells were then transduced with the filtered virus for 24 hours in the presence of polybrene (8 μg/ml; Sigma-Aldrich, #H9268). After 24 hours of recovery, infected cells were selected with 2 μM puromycin. Clonal cell lines were established by seeding single cells into 96-well plates and expanding them for 2 weeks. Last, positive clones were validated by assessing homogenous fluorescent signal upon Tm treatment. One validated clone, designated the 3T3R line, was selected for this study.

Tn screen

A high-throughput screen was performed using an established E. faecalis OG1RF mariner Tn library containing 14,976 mutants arrayed in 96-well plates (82). Following overnight growth, the OD₆₀₀ of each mutant culture was measured with a Tecan M200 microplate reader. For infection, 5 μl of each culture was added to 3T3R cells seeded in 96-well plates, and plates were centrifuged at 300g for 5 min to synchronize contact. Following infection, cells were stained with Hoechest 33342 (2.5 μg/ml; Thermo Fisher Scientific, #H21492) for 15 min, and the medium was then replaced with phenol red–free DMEM (Gibco, #31053028) supplemented with 25 mM Hepes, 1 mM sodium pyruvate, and 4 mM GlutaMAX. Plates were imaged on a Zeiss CellDiscoverer 7 microscope (×10 magnification) using two fluorescence channels to detect the XBP1s-mApple reporter [excitation/emission (Ex/Em), 570/594 nm] and Hoechst-stained nuclei (Ex/Em, 348/455 nm). The images were subsequently analyzed with CellProfiler 4.2.1 to quantify the percentage of UPR-positive (mApple-expressing) cells in each well.

Construction of in-frame deletion mutants in E. faecalis

General molecular biology reagents were sourced as follows: genomic DNA from E. faecalis was isolated using the Wizard Genomic DNA Purification Kit (Promega, #A1120), while plasmid DNA was isolated from E. coli using the PureLink Plasmid Miniprep Kit (Invitrogen, #K210011). All primers used in this study (table S7) were designed on the basis of the E. faecalis OG1RF genome (NC_017316). Gene fragments were amplified with Phusion High-Fidelity DNA Polymerase (Thermo Fisher Scientific, #F530), and routine screening was performed with Taq DNA Polymerase (New England Biolabs, #M0273). T4 DNA ligase and all restriction enzymes were used according to the manufacturer’s protocols (New England Biolabs). In-frame deletion mutants were generated using the temperature-sensitive shuttle vector pGCP213 (table S6), following a previously described protocol (83). Deletion constructs were created using two main strategies. For most single genes and smaller operons, regions of ~450-bp flanking the target were amplified from OG1RF gDNA; the upstream region was amplified with primer pair P1/P2 and the downstream region with P3/P4. These fragments were then fused by overlap extension PCR using the outer primers P1 and P4 and subsequently cloned into the Pst I site of pGCP213. For the larger ΔEET and ΔT7SS operons, a gBlock Gene Fragment (IDT) containing the fused upstream and downstream flanking regions was synthesized and cloned into the vector. The resulting deletion constructs were transformed into the appropriate E. faecalis parent strain by electroporation. Transformants were first selected on BHI-erythromycin agar at the permissive temperature of 30°C. To promote chromosomal integration, colonies were then passaged at the nonpermissive temperature of 42°C with erythromycin selection. Curing of the integrated plasmid was achieved by passaging the bacteria at 30°C in antibiotic-free BHI. Last, the successful deletion of the target gene or operon was verified by colony PCR using external (Screen F/R) and internal (Intern F/R or Intern R) primer pairs (table S7).

XBP1s fluorescent reporter assay

Following infection or treatment, 3T3R cells were stained with Hoechst 33342 (2.5 μg/ml) and transferred to a phenol red–free imaging medium. For each well, a 3 by 3 grid of images was acquired on a Zeiss CellDiscoverer 7 microscope. The intensity of the XBP1s-mApple signal within each nucleus (identified by Hoechst staining) was quantified using a CellProfiler pipeline, and the average intensity per well was used to gauge the level of UPR induction.

Growth curve assay

To assess bacterial growth, overnight bacterial cultures were processed as described in bacterial strains and growth conditions. After pelleting, they were normalized to a starting OD₆₀₀ of 0.05 in phenol red–free complete medium on a 96-well plate, which was sealed with a Breathe-Easy sealing membrane (Sigma-Aldrich, #Z380059) following the manufacturer’s protocols. OD₆₀₀ readings were taken at 10-min intervals over 20 hours. The rate of change of the OD₆₀₀ readings was calculated at 50-min intervals, and the highest rate of change was used to calculate the doubling time.

Antibiotic time-kill assay

To assess antibiotic killing, the supernatant of infected HaCaT cells were collected at 4, 5, 6, and 24 hpi. These were serially diluted on 96-well plates and 5 μl of the dilutions were spotted onto BHI-Agar plates, which were incubated at 37°C for 24 hours. Plates were imaged using a ProtoCOL3 Plus system (Don Whitely Scientific), and bacterial colonies were manually enumerated on ImageJ using the multipoint tool.

Cytotoxicity assay

To quantify total cytotoxicity, both detached (floating in the medium) and attached cells were collected and analyzed from infected HaCaT cell cultures. First, to collect the detached cell fraction, the culture medium was harvested, and each well was washed once with 1 ml of PBS. This wash was pooled with the collected medium, and the mixture was centrifuged (300g for 5 min). The resulting cell pellet was carefully resuspended in 20 μl of complete medium. Next, to collect the attached cell fraction, the remaining cells in the well were trypsinized for 15 min, neutralized with complete medium, pelleted by centrifugation, and resuspended in 500 μl of fresh complete medium. The viability of both the detached and attached cell suspensions was determined separately using a Countess 3 Automated Cell Counter with trypan blue staining. Total cytotoxicity was then calculated by summing the number of nonviable cells from both fractions and dividing by the total number of cells (viable and nonviable) from both fractions (Eq. 1).

$$\text{Cytotoxicity}=\frac{\text{No}.\text{of dead cells}(\text{detached}+\text{adhered})}{\text{No}.\text{of cells}(\text{detached}+\text{adhered})}\times 100\%$$ (1)

Superoxide assay

Extracellular superoxide generation was measured by adapting a previously published cytochrome C reduction assay for a 96-well plate format (18). Briefly, E. faecalis cultures were normalized to an OD₆₀₀ of 0.005 in 200 μl of phenol red free complete medium containing 20 μM of cytochrome C (Sigma-Aldrich, #C3131). The reduction of cytochrome C was measured as the change in absorbance at 550 nm (with 650 nm as the reference wavelength) every 2 min for 90 min at 37°C using a Tecan M200 microplate reader. The rate of superoxide generation was calculated from the maximal rate of change in absorbance, after correcting for pathlength. This rate was determined using Beer’s Law with an extinction coefficient of 21.5 mM⁻¹ cm⁻¹ for reduced cytochrome C (Eq. 2). To determine the amount of superoxide specifically, the rate measured in a parallel reaction containing 100 U/ml of SOD (Sigma-Aldrich, #S5395) was subtracted from the rate measured in its absence.

$$\begin{array}{c}\text{Rate of superoxide generation}(\text{nmol}\text{per}\text{minute}\text{per}{10}^9\text{CFU})=\\ \frac{{∆}_{\text{MAX}}\text{OD}550}{21.5\times 1}\times \frac{1\times {10}^9\text{CFU}}{2\times {10}^7\text{CFU}}\times \frac{60}{600}\times \frac{{10}^9}{{10}^6}\end{array}$$ (2)

Lipid peroxidation assay

Lipid peroxidation was assessed in HaCaT cells using the Image-iT Lipid Peroxidation Kit (Thermo Fisher Scientific, #C10445) at 24 hpi following the manufacturer’s protocols. Imaging was performed on a Zeiss CellDiscoverer 7 microscope using two fluorescence channels to detect fluorescence from reduced (Ex/Em, 592/614 nm) and oxidized (Ex/Em, 495/519 nm) BODIPY 581/591 C11. The mean intensity of the reduced and oxidized fluorescent reporters was quantified using a CellProfiler pipeline, with lipid peroxidation calculated on the basis of the ratio of reduced:oxidized signals for each condition.

Statistical analysis

Statistical analyses were performed using GraphPad Prism 9 and 10. In bar-dot plots, dots represent individual replicates, and the bar height indicates the median. Statistical significance was determined using either a one-way analysis of variance (ANOVA) with Dunnett’s or Tukey’s multiple comparisons test or for scratch wound assays, a two-way ANOVA with Tukey’s multiple comparisons test. An adjusted P < 0.05 was considered significant. Unless otherwise stated in the figure legends, all experiments were performed with a minimum of three independent biological replicates.

Supplementary Materials

The PDF file includes:

Supplementary Methods

Figs. S1 to S4

Tables S3 to S7

Legends for tables S1 and S2

Legends for movies S1 to S7

References

Other Supplementary Material for this manuscript includes the following:

Tables S1 and S2

Movies S1 to S7