Enterococcus is a genus of Gram-positive bacteria that is composed of more than seventy different species found in diverse environments, both free-living and in relationships with various animals (Lebreton et al., 2017). Enterococcus faecium strains have been isolated from humans and reported to have both beneficial and pathogenic properties (Van Tyne and Gilmore, 2014). Notably, antibiotic-resistant strains of E. faecium, particularly vancomycin-resistant E. faecium (VREfm), have emerged as major causes of healthcare-associated infections (Van Tyne and Gilmore, 2014; Fiore et al., 2019; García-Solache and Rice, 2019), and have been correlated with graft-versus-host disease (GVHD) and increased mortality in allogeneic hematopoietic cell transplantation patients (Stein-Thoeringer et al., 2019). E. faecium has also been recovered from ulcerative colitis and Crohn’s disease patients and has been shown to exacerbate intestinal inflammation and colitis in mouse models of inflammatory bowel disease (IBD) Barnett et al., 2010; Seishima et al., 2019. However, commensal strains of E. faecium have also been reported to enhance intestinal immunity in animal models and have been developed as probiotics (Hanchi et al., 2018). Furthermore, microbiota analysis has shown that E. faecium was enriched in immune checkpoint inhibitor (ICI) (Gopalakrishnan et al., 2018; Routy et al., 2018; Matson et al., 2018) and chimeric antigen receptor (CAR) T-cell therapy-responsive patients (Smith et al., 2022). These studies highlight the potential pathogenic and beneficial features of E. faecium.

Our laboratory previously investigated the beneficial effects of E. faecium on host physiology (Griffin and Hang, 2022). We demonstrated that secreted antigen A (SagA), a highly conserved NlpC/P60 peptidoglycan hydrolase in E. faecium, was sufficient to confer protection against enteric infections in both Caenorhabditis elegans and mice (Rangan et al., 2016). In mice, E. faecium and an E. faecalis strain engineered to express SagA both up-regulated expression of mucins and antimicrobial peptides, resulting in improved intestinal barrier function as well as tolerance to Salmonella enterica serovar Typhimurium and Clostridioides difficile pathogenesis (Pedicord et al., 2016; Kim et al., 2019). In contrast, wild-type E. faecalis, which does not express SagA, did not exhibit these effects. We then determined the X-ray crystal structure of the SagA NlpC/P60 hydrolase domain, and demonstrated that this hydrolase preferentially cleaves crosslinked peptidoglycan fragments into smaller muropeptides (such as GlcNAc-MDP), which more effectively activate the peptidoglycan pattern recognition receptor NOD2 (nucleotide-binding oligomerization domain-containing protein 2) in mammalian cells (Kim et al., 2019). Importantly, NOD2 was shown to be required for E. faecium stimulation of intestinal immunity and tolerance to infection in vivo (Pedicord et al., 2016; Kim et al., 2019). Moreover, E. faecium and SagA were sufficient to protect mice against dextran sodium sulfate-induced colitis that required the expression of NOD2 in myeloid cells (Jang et al., 2023).

The discovery of E. faecium among the microbiota of cancer immunotherapy-responsive patients then motivated our analysis of Enterococcus species and SagA in mouse tumor models. Indeed, diverse strains of E. faecium that express SagA, but none of the non-SagA-expressing E. faecalis strains evaluated, were sufficient to promote ICI (anti-PD-L1, anti-PD-1, anti-CTLA-4) anti-tumor activity against different cancer types in mouse models (Griffin et al., 2021). Furthermore, other Enterococcus species including E. durans, E. hirae, and E. mundtii, which each have SagA orthologs with greater than 80% protein similarity to that of E. faecium, were sufficient to enhance anti-PD-L1 antitumor activity (Griffin et al., 2021). The SagA orthologs from these other Enterococcus species were expressed and secreted at similar levels to E. faecium SagA and showed similar peptidoglycan hydrolase activity in vitro (Griffin et al., 2021). Importantly, heterologous expression of SagA in inactive bacterial species (E. faecalis and Lactococcus lactis) was sufficient to promote anti-PD-L1 antitumor activity and required NOD2 in vivo (Griffin et al., 2021). Collectively, these studies demonstrated that Enterococcus peptidoglycan remodeling by SagA is sufficient to enhance intestinal immunity against infection and promote cancer immunotherapy. However, the endogenous functions of SagA in E. faecium microbiology and modulation of host immunity were not investigated previously, as sagA was believed to be an essential gene in E. faecium (strain TX1330), as reported by Murray and coworkers (Teng et al., 2003). Nevertheless, our collaborative studies with the Duerkop laboratory have indicated that phage-resistant strains of E. faecium (strain Com12) containing catalytically inactivating point mutations in sagA were still viable (Canfield et al., 2023), suggesting that SagA may not be essential and could be functionally evaluated in E. faecium.

Here, we report the generation of a ΔsagA strain of E. faecium (Com15). This strain, generated using RecT-mediated recombineering methods developed by our laboratory (Chen et al., 2021), exhibited significantly impaired growth (Figure 1a) and sedimentation in liquid culture (Figure 1—figure supplement 1b). We then performed whole-genome sequencing to assess the fidelity of the ΔsagA strain’s genome to that of wild-type. While ΔsagA contained 14 mutations, most were outside of open-reading frames (ORFs) or were in genes that are unrelated to peptidoglycan metabolism (Supplementary file 1). One notable mutation, however, was in the predicted glycosidase EFWG_00994. This mutation, L200F, alters a leucine residue that is conserved among GH73 family members and the other putative Com15 glucosaminidases (Supplementary file 1). To confirm that the ΔsagA growth defect was truly SagA-dependent and was not an outcome of suppressor mutations, we generated a complementation strain (ΔsagA/ psagA) that contained a plasmid expressing sagA under control of its native promoter. Western blot analysis with ɑ-SagA polyclonal serum confirmed the deletion and complementation of SagA (Figure 1b). Differential interference contrast (DIC) microscopy revealed that ΔsagA cells form irregular clusters, as opposed to the typical wild-type morphology of diplococci or short chains (Figure 1c). This cell morphology suggested that dividing ΔsagA cells may be unable to separate from one another during binary fission. Indeed, transmission electron microscopy (TEM) of chemically fixed samples confirmed that ΔsagA cells have defects in cell separation (Figure 1d). Division septa could form and mature in these cells; however, daughter cells failed to separate, resulting in clusters of unseparated cells. Cells on the periphery of these clusters were highly strained, and in some cases appeared to lyse and leave behind strands of undegraded peptidoglycan (Figure 1d). Importantly, the complementation largely restored the log-phase growth rate and stationary-phase OD of the ΔsagA strain to those of wild-type (Figure 1a). It also reversed the liquid-culture sedimentation phenotype (Figure 1—figure supplement 1b) and reduced the cell clustering (Figure 1c) and defective septal separation (Figure 1d).

To investigate the significance of peptidoglycan hydrolase activity in SagA, we generated a catalytically inactive point mutant (C431A) in our sagA complementation plasmid. Consistent with our previous biochemical results on recombinant SagA NlpC/P60 domain activity (Kim et al., 2019; Espinosa et al., 2020), the C431A mutant was unable to complement the sagA deletion (Figure 1a–d), despite Western blot analysis confirming that the mutant SagA was expressed and secreted properly (Figure 1b). Together, these results demonstrated that catalytically active SagA is required for proper E. faecium growth.

Based on the growth defect of ΔsagA and the previously reported antibiotic sensitization effect of catalytically inactivating point mutants in the SagA of E. faecium strain Com12 (Canfield et al., 2023), we chose to investigate the impact of the sagA deletion on the effectiveness of various cell wall-acting antibiotics. Using minimum inhibitory concentration (MIC) test strips of common antibiotics (vancomycin, linezolid, daptomycin, tigecycline, telavancin, fosfomycin, ampicillin, ceftriaxone, and imipenem), we observed increased sensitivity to the β-lactams (ampicillin, ceftriaxone, and imipenem), moderately increased sensitivity to daptomycin, tigecycline, fosfomycin, and linezolid and no change in sensitivity to telavancin and vancomycin (Figure 1—figure supplement 2a, Supplementary file 2). MIC analysis of ampicillin in liquid culture, further demonstrated that ΔsagA has increased susceptibility to β-lactam antibiotics, which was abrogated with psagA expression (Figure 1—figure supplement 2b).

To further investigate the growth defect and increased β-lactam antibiotic sensitivity of ΔsagA, we performed cryo-electron tomography (cryo-ET) on frozen-hydrated samples to quantify factors such as peptidoglycan thickness, septum thickness, and placement of divisome components at a higher resolution (Figure 2—figure supplement 1a and b). There was no statistically significant change in peptidoglycan thickness and septum thickness in ΔsagA and ΔsagA/ psagA cells compared to wild-type cells (Figure 2—figure supplement 1c and d). Similar to previous studies of Bacillus subtilis (Khanna et al., 2021), the divisome machinery was directly observable as concentric rings in cross-sectional images generated by cryo-ET in E. faecium (Figure 2—figure supplement 2a–d). We quantified the distances from these rings in order to assess potential differences in divisome placement (Figure 2—figure supplement 2e). The distance between the septal membrane and the distal ring of the divisome showed a slight increase, but we did not observe any statistical difference between the distance from the septal membrane to the proximal ring and the ratio between the proximal and distal rings in wild-type, ΔsagA, and complementation (Figure 2—figure supplement 1e, Figure 2—figure supplement 2f and g). However, we observed that the placement and projection angle of growing septa were significantly altered in ΔsagA but were mostly restored by complementation (Figure 2a–f). Both room temperature TEM and cryo-ET revealed that sagA complementation largely restored the wild-type cell morphology of ΔsagA to that of wild-type (Figure 2, Figure 2—figure supplements 1 and 2). The complemented cells separate from one another and mostly take the form of diplococci. These results suggest that the ultrastructure of the peptidoglycan cell wall is unchanged as a result of sagA deletion, but peptidoglycan cleavage, divisome angle, and septal resolution are impaired.

We then investigated ΔsagA peptidoglycan composition and activation of NOD2 immune signaling. To evaluate peptidoglycan composition, E. faecium sacculi were isolated, subjected to mutanolysin digestion, analyzed by liquid chromatography mass spectrometry (Figure 3—figure supplement 1) and quantified (Figure 3a and b), as previously described (Kim et al., 2019). The ΔsagA strain showed decreased levels of small muropeptides (peaks 2, 3, and 7 with green asterisk) compared to E. faecium Com15, which was restored in ΔsagA/ psagA (Figure 3a). Moreover, we observed increased amounts of crosslinked peptidoglycan fragments (peaks 13 and 14 with purple asterisk) (Figure 3a). Notably, GlcNAc-MDP, which we previously demonstrated more effectively activates NOD2 (Kim et al., 2019), was significantly decreased in ΔsagA (Figure 3b). We next evaluated live bacterial cultures with mammalian cells to determine their ability to activate the peptidoglycan pattern recognition receptor NOD2. Indeed, our analysis of these bacterial strains using HEK-Blue NF-κB reporter cells demonstrated that ΔsagA exhibits significantly decreased NOD2 activation compared to E. faecium WT. This activation was restored in ΔsagA/ psagA (Figure 3c). For these assays, we used comparable numbers of bacteria to account for defects in ΔsagA growth, which was confirmed by colony-forming unit analysis of bacteria per well (Figure 3d). These results were consistent with our in vitro analysis of recombinant SagA with purified peptidoglycan fragments, which showed that the generation of small muropeptides (GlcNAc-MDP) more potently activated NOD2 compared to crosslinked muropeptides (Kim et al., 2019). Our results also demonstrated that while many enzymes are required for the biosynthesis and remodeling of peptidoglycan in E. faecium, SagA is essential for generating NOD2-activating muropeptides ex vivo.

We next investigated if SagA is crucial for E. faecium colonization and immune modulation in vivo using well-established mouse models of ICI cancer immunotherapy (Figure 4a). We had previously demonstrated that heterologous expression of SagA was sufficient to promote E. faecalis and L. lactis ICI antitumor activity and required Nod2 expression in mice (Griffin et al., 2021). However, activity of wild-type E. faecium in Nod2^-/- mice was not determined. Indeed, oral administration of E. faecium to microbiota-depleted/antibiotic-treated C57BL/6 mice promoted anti-PD-1 antitumor activity against MC-38 tumor growth in Nod2^+/-, but failed to do so in Nod2^-/- mice (Figure 4b). To determine if SagA is required for E. faecium function in vivo, antibiotic-treated mice were colonized with E. faecium Com15 or ΔsagA. Notably, the antitumor activity of E. faecium was abolished in ΔsagA strain (Figure 4c) even though both bacterial strains colonized mice at comparable levels, as judged by fecal E. faecium levels (Figure 4d). We then performed immune profiling of tumor infiltrating lymphocytes (TILs) using flow cytometry (Figure 4—figure supplement 1). Compared to E. faecium Com15 colonized mice, ΔsagA colonized mice showed no difference in the total amount of the tumor-infiltrating CD45⁺ cells (Figure 4e) or CD4⁺ FoxP3+ regulatory T cells (Figure 4f), but resulted in fewer CD3⁺ CD8⁺ T cells (Figure 4g). Furthermore, granzymeB⁺ CD8⁺ T cells, Ki67⁺ CD8⁺ T cells, and PD-1⁺ CD8⁺ T cells levels were also lower in ΔsagA colonized mice (Figure 4h–j). We evaluated the activity ΔsagA/psagA strain in these experiments (data not shown), but the results were inconclusive likely due to instability of psagA plasmid in E. faecium in vivo. We note that by continuous oral administration in the drinking water, live E. faecium and soluble muropeptides that are released into the media during bacterial growth may both contribute to NOD2 activation in vivo. Nonetheless, these results demonstrate SagA is not essential for E. faecium colonization, but is required for promoting the ICI antitumor activity through NOD2 in vivo.

E. faecium is a prominent microbiota species and pathogen in animals and humans. While E. faecium identification and dominance in fecal microbiota have been correlated with health and disease outcomes, the underlying mechanisms have only begun to emerge. Our previous gain-of-function studies with recombinant protein and engineered bacterial strains demonstrated that SagA, a unique secreted peptidoglycan hydrolase that is highly conserved in E. faecium strains and other Enterococcus species, but not E. faecalis, is sufficient to activate NOD2 and promote host immunity in vivo (Rangan et al., 2016; Pedicord et al., 2016; Kim et al., 2019; Griffin et al., 2021). However, the endogenous functions of SagA in E. faecium were unknown. Based on the results described above, we now show that SagA peptidoglycan hydrolase activity is not required for E. faecium viability, but is essential for proper growth and specifically for septal separation following cell division (Figures 1 and 2). These results are consistent with key peptidoglycan hydrolases in other bacterial species, such as AmiA/B in Escherichia coli (Yang et al., 2011), CwlO/LytE in Bacillus subtilis (Meisner et al., 2013; Wilson et al., 2023), PcsB in Streptococcus pneumoniae (Sham et al., 2011) and RipC in Mycobacterium tuberculosis (Mavrici et al., 2014). Although the mechanisms of regulation may differ amongst peptidoglycan hydrolases and remain to be determined for SagA, the sagA promoter contains sequence motifs that may be regulated by WalRK two-component systems (Brogan and Rudner, 2023) and the SagA protein contains a N-terminal coil-coil domain that may interact with E. faecium divisome components, which warrants further investigation in the future. Notably, the deletion of sagA in E. faecium Com15 also renders this strain more susceptible to cell wall-acting antibiotics (Figure 1—figure supplement 2), akin to inactive SagA variants in E. faecium Com12 (Canfield et al., 2023), suggesting SagA and its related peptidoglycan hydrolases may be potential antibacterial targets to use in combination with existing antibiotics. Interestingly, clade B strains of E. faecium including Com12 and Com15 have been suggested to be reclassified as Enterococcus lactis based on genomic and phenotypic similarity (Belloso Daza et al., 2022; Belloso Daza et al., 2021). Nonetheless, SagA orthologs have greater than 90 percent protein sequence identity within the C-terminal NlpC/P60 hydrolase domain of ICI therapy-promoting Enterococcus species (E. faecium-clade A and B, E. durans, E. hirae and E. mundtii) (Griffin et al., 2021). Notably, all the Enterococcus strains and species that express active SagA orthologs that we have analyzed promote ICI cancer immunotherapy in mouse models (Griffin et al., 2021).

Beyond intrinsic functions in E. faecium microbiology, we have also demonstrated for the first time that SagA is essential for the activation of host immunity. Notably, we showed that ΔsagA does not generate significant amounts of non-crosslinked muropeptides (i.e. GlcNAc-MDP) that are sensed by the peptidoglycan pattern recognition receptor NOD2 (Figure 3). Indeed, ΔsagA failed to activate NOD2 ex vivo (Figure 3) and promote ICI antitumor activity in mouse models compared to wild-type E. faecium (Figure 4). Collectively, our results provide important mechanistic insight on SagA in peptidoglycan remodeling for bacterial cell separation and reveal an essential feature of E. faecium (Figure 5) and other related Enterococcus species that are associated with human health, disease, and response to therapy.