Acetate coordinates neutrophil and ILC3 responses against C. difficile through FFAR2

The intestinal microbiota plays a crucial role in the maintenance of host homeostasis by shaping and supporting multiple physiological functions, such as nutrient absorption, metabolism, and the development of the immune system (Ding et al., 2019; Richard and Sokol, 2019; Sonnenburg and Sonnenburg, 2019; Sovran et al., 2019; Zmora et al., 2019). The microbiota also outcompetes intestinal colonization by microbial pathogens and thus provides an exogenous defense mechanism against infections (Rodriguez et al., 2015; Stecher and Hardt, 2008). Indeed, quantitative and/or qualitative changes in the composition of the microbiota, known as dysbiosis, are associated with expansion of pathobionts, damage to the intestinal epithelium, bacterial translocation into deeper organs, and ultimately development of numerous pathologies (Ferreira et al., 2014; Johanesen et al., 2015; Theriot et al., 2014).

Clostridium difficile is an anaerobic bacterium that forms spores, which can be found in the intestines of up to 17% of healthy adult individuals. C. difficile is resistant to several antimicrobial agents, including clindamycin, ampicillin, and third-generation cephalosporins, and hence acquires a selective advantage over other microorganisms during antibiotic therapy (Johanesen et al., 2015; Rea et al., 2011; Theriot et al., 2014). Thus, antibiotic-induced dysbiosis can lead to C. difficile infections (CDIs) that cause intestinal disease ranging from mild diarrhea to pseudomembranous colitis (Chen et al., 2008; Rea et al., 2011; Theriot et al., 2014). C. difficile exerts pathogenicity through spore germination and production of toxins, particularly toxins A (enterotoxin) and B (cytotoxin), which cause irreversible damage to the colonic epithelium (Bibbò et al., 2014; Voth and Ballard, 2005). Loss of epithelial integrity results in increased intestinal permeability and translocation of bacteria from the gut lumen into deeper tissues (Hasegawa et al., 2012; Naaber et al., 1998). Emphasizing the importance of healthy microbiota in preventing CDI, several studies have demonstrated that patients with recurrent CDI benefit more from transplantation of fecal microbiota derived from healthy donors than from conventional antimicrobial therapies (Lawley et al., 2012; van Nood et al., 2013).

Innate immune responses are essential for host resistance against CDI. Neutrophils provide an early line of defense: once recruited into the intestinal mucosa, neutrophils carry out several protective functions, such as producing reactive oxygen species (El-Zaatari et al., 2014) and secreting IL-1β (Hasegawa et al., 2012). Depletion of neutrophils in mice results in acute mortality following infection (Jarchum et al., 2012). Two subsets of innate lymphoid cells (ILCs) have also been shown to contribute to early host defense against C. difficile (Abt et al., 2015). ILC1s secrete IFNγ, which activates macrophage defense functions; ILC3s secrete IL-22, which induces the production of antimicrobial peptides in the gut (Sadighi Akha et al., 2015) and activates the complement pathway in the lung and liver to clear translocated bacteria (Hasegawa et al., 2014). The importance of ILCs in CDI is underscored by the observation that Rag1^−/− mice, which lack T and B cells, can recover from acute infection as well as WT mice, whereas Rag1^−/− × Il2rg^−/− mice, which lack ILCs as well as adaptive lymphocytes, succumb to infection (Abt et al., 2015; Hasegawa et al., 2014).

Production of short-chain fatty acids (SCFAs) through the fermentation of dietary fibers is a major mechanism through which the microbiota promotes intestinal immunity (Corrêa-Oliveira et al., 2016). SCFAs include acetate, propionate, and butyrate; these molecules can act through disparate mechanisms, such as activating specific G protein–coupled receptors, inhibiting histone deacetylases, stimulating histone acetyltransferase, and stabilizing hypoxia inducible factor 1 (Donohoe and Bultman, 2012; Fellows et al., 2018; Kelly et al., 2015; Kim et al., 2013; Vinolo et al., 2011). Through these pathways, SCFAs influence multiple immune cell functions, including chemotaxis, energy production, gene expression, cell differentiation, and cell proliferation (Donohoe and Bultman, 2012; Fellows et al., 2018; Kelly et al., 2015; Kim et al., 2013; Tan et al., 2014; Vinolo et al., 2011). While the role of SCFAs has been studied in various proinflammatory and antiinflammatory intestinal immune responses (Donohoe and Bultman, 2012; Fujiwara et al., 2018; Furusawa et al., 2013; Kaiko et al., 2016; Kim et al., 2013; Maslowski et al., 2009), only the impact of butyrate was recently demonstrated during CDI (Fachi et al., 2019). The influence of other SCFAs on immune responses against C. difficile is presently unknown.

Here, we examined the impact of acetate on an acute mouse model of CDI. We found that administration of acetate is remarkably beneficial in ameliorating disease. Mechanistically, we show that acetate enhances innate immune responses by acting on both neutrophils and ILC3s through its cognate receptor free fatty acid receptor 2 (FFAR2). In neutrophils, acetate-FFAR2 signaling facilitates inflammasome activation and promotes the release of IL-1β; in ILC3s, acetate-FFAR2 augments expression of the IL-1 receptor (IL-1R), which boosts IL-22 secretion in response to IL-1β. We conclude that microbiota-derived acetate promotes host innate responses to C. difficile through coordinate action on neutrophils and ILC3s.

We tested the clinical course of CDI in WT mice fed a soluble fiber-rich diet (10% pectin) in comparison to mice fed control diet (AIN93M; Fig. 1 A). Mice fed the pectin-rich diet had improved clinical scores and lost less weight than mice fed the control diet (Fig. 1, B and C). The protective effect of the fiber-rich diet was associated with an increase in intestinal bacteria content (Fig. S1 A), as well as higher concentrations of acetate in serum (Fig. 1 D). Because acetate is a metabolite produced by intestinal bacteria through soluble fiber fermentation, this suggests that bacterial-derived acetate may be protective during CDI. Corroborating this hypothesis, addition of 150 mM acetate to the drinking water before and throughout the infection was sufficient to protect mice against CDI, resulting in improved clinical scores and less-severe weight loss (Fig. 1, E–G) without impacting intestinal bacterial content (Fig. S1 B). Histologically, we observed increased inflammatory infiltrate in acetate-treated mice 2 d postinfection (p.i.) as well as attenuated histopathology and mucosal hyperplasia 4 d p.i. (Fig. 1 H). Since acetate did not inhibit C. difficile growth (Fig. S1 C) or toxin production in vitro or in vivo (not depicted) and had rapid course of action by improving clinical scores and weight loss even when it was administered at the time of infection (Fig. S1, D–F), we conclude that acetate improves host resistance to C. difficile rather than directly interfering with bacterial replication.

We next examined the impact of acetate on the host response to CDI. Neutrophils have been shown to be crucial early during the host response to C. difficile (Jarchum et al., 2012). In accordance with the histological observations, as early as 1 d p.i., mice treated with acetate had more neutrophils infiltrating the colon, but not inflammatory monocytes, than untreated controls (Figs. 1 Hand 2, A and B). This initial effect of acetate treatment was associated with more colonic content of CXCL1, a neutrophil chemoattractant produced by intestinal epithelia and neutrophils themselves (Fig. 2 C). The colons of acetate-treated mice also contained more IL-1β (Fig. 2 C), a cytokine produced by intestinal epithelia and neutrophils that has a prominent role in impeding the translocation of gut bacteria and their systemic dissemination (Hasegawa et al., 2012). The content of proinflammatory TNFα was slightly reduced (Fig. 2 C). Consistent with the increase in neutrophil infiltration and IL-1β production, acetate-treated mice showed a detectable reduction of bacteria translocation into mesenteric lymph nodes (mLNs) on day 1 p.i., which became obvious in several organs on day 2 p.i. (Fig. 2, D and E). Further substantiating a role in protecting the intestinal mucosa from CDI-induced damage, acetate also curtailed systemic permeability of FITC-dextran after oral administration in infected mice (Fig. 2 F). In vitro, acetate did not prevent intestinal epithelial cell death by C. difficile toxins, as measured by propidium iodide and calcein-AM staining (Fig. 2, G and H). Together, these data suggest that acetate bolsters host resistance to C. difficile–mediated intestinal damage by enhancing early intestinal immune responses, such as neutrophil accumulation and production of IL-1β.

It has been shown that C. difficile toxins trigger IL-1β production through inflammasome activation. C. difficile toxins glucosylate RhoA-GTPase, activating the Pyrin inflammasome (Xu et al., 2014). In addition, they trigger the NLRP3 inflammasome through an as-yet-undefined mechanism, perhaps by inducing reactive oxygen species (Ng et al., 2010). To determine whether acetate enhances inflammasome activation, we examined the impact of acetate on CDI in mice lacking either the effector caspase-1 or one of various inflammasomes implicated in caspase-1 activation. The beneficial effect of acetate on clinical score, weight change, and intestinal permeability during CDI was abrogated in Casp1/11^−/− mice (Fig. 3, A–C), as well as Nlrp3^−/− mice, which lack one of the inflammasomes that sense C. difficile toxins (Fig. S1, G–I). In contrast, acetate maintained its protective effect in Nlrc4^−/− mice and in large part, in Nlrp6^−/− mice (Fig. S1, J–P). Acetate also failed to prevent translocation of bacteria into the liver, spleen, or mLNs of Casp1/11^−/− mice (Fig. 3 D). Additionally, acetate did not enhance the content of IL-1β, CXCL1 (Fig. 3 E), or neutrophil infiltration (Fig. 3 F) in the colons of Casp1/11^−/− mice. Since caspase-1 activation also triggers IL-18 production, we examined the impact of acetate on IL-18 during CDI. Increased amounts of IL-18 were detected in the colons of WT mice on day 1 p.i. (Fig. S2 A). However, acetate did not enhance colonic IL-18 content in Casp1/11^−/− mice, providing further evidence that acetate acts by promoting inflammasome-regulated cytokine production during CDI.

Given the major role of neutrophils in IL-1β secretion and in controlling bacterial translocation during CDI (Hasegawa et al., 2012; Fig. 2), we wanted to examine the effects of acetate on inflammasome activation by C. difficile in neutrophils. Stimulation of neutrophils purified from the bone marrow of WT mice with acetate markedly potentiated the production of IL-1β in response to C. difficile or its supernatant containing a defined amount of toxins (Fig. 3 G), without affecting neutrophil viability (Fig. S2 B). The enhancing effect of acetate on IL-1β secretion was evident whether or not neutrophils were primed with LPS (Fig. 3 G). Moreover, acetate effect was blocked by an inhibitor of caspase-1, whereas it was unaffected by an inhibitor of neutrophil elastase (Fig. 3 H). We conclude that acetate augments inflammasome-regulated IL-1β secretion by neutrophils in response to C. difficile toxins.

Among various mechanisms, acetate can modulate cell functions through activation of FFAR2 (also known as GPR43) in epithelial cells and neutrophils (Corrêa-Oliveira et al., 2016). Given that acetate generates faster outcomes through FFAR2 signaling than through other mechanisms requiring chromatin remodeling, and that acetate acts rapidly during CDI (Fig. S1, D–F), we sought to examine the impact of FFAR2 in acetate-mediated protection against CDI. We found that the beneficial effects of acetate on clinical score, weight change, systemic translocation of bacteria, and intestinal permeability during CDI were abrogated in Ffar2^−/− mice (Fig. 4, A–D). Additionally, the increased colonic content of CXCL1, IL-1β, and IL-18; the enhanced infiltration of neutrophils in the colon; and the reduced systemic bacterial translocation mediated by acetate in WT mice were not evident in Ffar2^−/− mice (Fig. 4, E–G; and Fig. S2 C). Together, these findings suggest that the effects of acetate on host resistance to CDI depend on FFAR2.

We sought to determine whether acetate enhances neutrophil production of IL-1β through FFAR2. We prepared neutrophils by density gradient centrifugation from the bone marrow of WT and Ffar2^−/− mice, primed them in vitro with LPS in the presence or absence of acetate, stimulated them with C. difficile toxins, and then measured IL-1β production. The ability of acetate to augment IL-1β production by neutrophils was entirely dependent on FFAR2 (Fig. 4 H). Similar results were obtained using neutrophils highly purified from bone marrow by cell sorting (Fig. S2 D). We also tested whether acetate could enhance IL-1β production by neutrophils that are stimulated with a known inflammasome activator, the potassium (K⁺) ionophore nigericin. Indeed, acetate enhanced the K⁺ efflux–induced IL-1β production, and this effect was dependent on FFAR2 (Fig. 4 I). Thus, acetate enhances inflammasome activation and IL-1β secretion in neutrophils through FFAR2.

We further corroborated that acetate-FFAR2 signaling enhances IL-1β production in neutrophils through inflammasome activation. Treatment of neutrophils with BAPTA-AM (1,2-bis(o-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid acetyloxymethyl ester), which inhibits NLRP3 by blocking Ca²⁺ influx, abrogated the ability of acetate to potentiate IL-1β production through FFAR2 signaling in response to LPS and C. difficile toxins or nigericin (Figs. 4 J and S2 E). A similar result was obtained by treating neutrophils with extracellular KCl, which inhibits NLRP3 by blocking K⁺ efflux (Figs. 4 K and S2 E). In contrast, incubation of neutrophils with NaCl, which boosts inflammasome activation by promoting K⁺ efflux, facilitated acetate enhancement of IL-1β production through FFAR2 (Fig. 4 K). Taken together, these results suggest that the ability of acetate–FFAR2 axis to potentiate the production of IL-1β in response to C. difficile toxins depends on inflammasome activation.

It has been shown that ILC3s protect from CDI through IL-22 production (Abt et al., 2015; Hasegawa et al., 2014). Moreover, ILC3s express high amounts of FFAR2 (Fig. S3 A and http://www.immgen.org). Therefore, we asked whether acetate also impacts this protective mechanism. We found that acetate treatment increased the expression of Rorc and Il22 in the colon of WT mice 2 d p.i. (Fig. 5 A), as well as IL-22 target genes (Bcl2, Ccnd1, Reg3g, Muc1, and Muc4) 5 d p.i. (Fig. S3 B). Increased amounts of IL-22 protein were detected in the colons of WT mice but not Ffar2^−/− mice on day 5 p.i. (Fig. 5 B), suggesting that acetate may promote IL-22 production through FFAR2. Although IL-22 was previously shown to be primarily produced by ILC3s during CDI (Abt et al., 2015), whether acetate enhances IL-22 secretion by ILC3s and/or T cells was not clear. To address this question, we examined the impact of acetate in Rag2^−/− and Rag2/Il2rg^−/− mice with CDI. The protective effect of acetate was evident in Rag2^−/− mice, which lack T and B lymphocytes but retain ILCs, whereas it was undetectable in Rag2/Il2rg^−/− mice, which lack both innate and adaptive lymphocytes (Fig. 5, C–F). Reduction of intestinal permeability by acetate was also maintained in Rag2^−/− but not Rag2/Il2rg^−/− mice (Fig. 5, G and H). Thus, acetate-mediated protection of infected mice depends on ILCs.

We next examined the impact of acetate on the abundance of ILC subsets on day 5 p.i. Acetate-treated mice showed a moderate increase in the number of ILC3s, but not of other ILCs, in the colon (Fig. 5 I) and small intestine (Fig. S3 C). To validate that acetate-mediated protection is dependent on ILC3s, we infected AhR^f/f Rorc-Cre mice, which have a strong reduction of ILC3s (Song et al., 2015). We found that the beneficial effects of acetate on clinical score, weight change, colon length, intestinal permeability, and systemic translocation of bacteria during CDI were strongly reduced in the absence of ILC3s (Fig. 6, A–F). These mice also did not show clinical recovery and mucosal hyperplasia on day 5, in contrast to what normally occurs in WT mice (Fig. 6, A and B; and Fig. S3 D), indicating that ILC3s are important not only to mitigate the disease but also for a complete resolution.

Since ILC3s produce IL-22, which promotes epithelial cell defense mechanisms and mucus production, we asked whether acetate-mediated protection against CDI depends on IL-22 production. To test this, acetate-treated and untreated mice received an i.p. dose of anti–IL-22 neutralizing antibody or isotype control on day 1 and 3 after CDI (Fig. 6 G). Neutralization of IL-22 entirely reduced the protective effect of acetate, raising the clinical score and body weight loss of acetate-treated mice to levels comparable to those seen in untreated mice that also received anti-IL-22 (Fig. 6, H and I). Similarly, beneficial effects of acetate on clinical score, body weight, bacterial translocation, and epithelial permeability during CDI were strongly reduced in Il22^−/− mice (Fig. 6, J–M). Conversely, newly generated mice with genetic disruption of IL-22 binding protein (Il22ra2), which acts as an IL-22 decoy (Huber et al., 2012), were as resistant to CDI as acetate-treated WT mice. Acetate treatment did not further improve CDI resistance of Il22ra2^−/− mice (Fig. S4 A). Together, these data highlight the importance of IL-22 production by ILC3s in the protective effect of acetate.

Acetate did not increase IFNγ expression in the colon or IFNγ production by primary ILC3 ex vivo (Fig. S4, B–D), suggesting no obvious impact on ILC1s. We conclude that ILC3s production of IL-22 has a major role in acetate-induced protection against CDI.

Since ILC3s express FFAR2, and acetate protection was dependent on these cells and FFAR2, we decided to test the direct effect of acetate on the production of IL-22. We incubated primary ILC3s in vitro with acetate, C. difficile supernatant, and IL-23 and measured the percentage of IL-22–producing cells by intracellular staining. Acetate had no impact on IL-22 production in these cultures (Fig. 7 A). Similar results were obtained using the ILC3 line MNK3 (Allan et al., 2015; Fig. 7 B). However, the percentage of IL-22–producing ILC3s was increased in acetate-treated mice 5 d p.i. (Fig. 7 C). Similarly, acetate treatment increased the expression of IL-22 mRNA in ILC3s FACS sorted from antibiotic-treated mice (Fig. 7 D). It is known that ILC3s secrete IL-22 in response to IL-1β and IL-23 (Takatori et al., 2009; Vivier et al., 2018). We sought to determine whether acetate stimulated ILC3s indirectly through enhanced expression of IL-23 receptor or IL-1R. We found that acetate increased the expression of IL-1R in (a) MNK3 cells (Fig. 7, E and F); (b) primary ILC3s isolated from C. difficile–infected mice 5 d p.i. (Fig. 7 G); (c) primary ILC3s isolated from antibiotic-treated mice (Fig. 7 H); and (d) primary ILC3s treated with acetate in vitro (Fig. 7 I). These results suggested that acetate may increase ILC3 responsiveness to IL-1β. To confirm that IL-1R expression is induced by acetate via FFAR2, we assessed the impact of synthetic FFAR2 agonist and antagonist on the stimulatory activity in primary ILC3s. FFAR2 agonist increased the percentages of IL-1R–expressing primary ILC3s, as did acetate, while FFAR2 antagonist inhibited such effect (Fig. 7 I). To corroborate that acetate increases the responsiveness of ILC3s to IL-1β, we examined IL-22 production of FACS sorted-ILC3s (Fig. 7 J) and MNK3 cells (Fig. S4 E) stimulated in vitro with different combinations of acetate and IL-1β. Acetate-FFAR2 signaling clearly enhanced the percentage of IL-22–producing cells induced by IL-1β (Fig. 7 J). Further, we also found that acetate does not change the expression of Ffar2 and Ffar3 in the colon of mice 5 d p.i. (Fig. S4 F), as well as both MNK3 (Fig. S4 G) and ILC3s FACS-sorted from antibiotic-treated mice (Fig. S4 H).

We sought to directly demonstrate that acetate activates primary ILC3s through IL-1β produced by neutrophils. We isolated ILC3s from the intestinal lamina propria of Rag1^−/− mice, which lack T and B cells, and incubated them ex vivo with IL-23 together with culture supernatants from neutrophils that had been stimulated with LPS and C. difficile toxins, with or without acetate, or with synthetic FFAR2 agonist (Fig. 8 A). ILC3s stimulated with supernatant of neutrophils treated with acetate or FFAR2 agonist produced more IL-22 than did ILC3s cultured with supernatant of neutrophils treated with LPS and C. difficile toxins only. This effect could be blocked with a neutralizing anti-IL-1β antibody, but not by a FFAR2 antagonist (Fig. 8 A), demonstrating that ILC3 production of IL-22 is induced by neutrophil-derived IL-1β and not FFAR2 activation of ILC3s. In contrast to the stimulating effect on cytokine secretion, the supernatant of neutrophils had no impact on the abundance of total ILC3s or any of the CCR6⁺NKp46⁻, CCR6⁻NKp46⁻, or CCR6⁻NKp46⁺ subsets included in the ILC3 population purified from the intestine using a standard cell sorting protocol (Fig. S4, I and J).

Finally, examination of mice with a deletion of the Ffar2 gene in neutrophils (Ffar2^ΔS100) provided in vivo demonstration that acetate-mediated protection against CDI depends on neutrophils. The beneficial effect of acetate treatment on clinical score and weight change in CDI was reduced in Ffar2^ΔS100 mice compared with Ffar2^f/f mice (Fig. 8 B). Notably, Ffar2^ΔS100 mice exhibited no clinical recovery on day 5 in contrast to Ffar2^f/f mice. Moreover, acetate-treated Ffar2^ΔS100 mice showed decreased expression of Il22, Rorc, and Il1b in the colon compared with acetate-treated Ffar2^f/f mice 5 d p.i. (Fig. 8 C), as well as reduced expression of IL-22 target genes (Bcl2, Ccnd1, Reg3g, and Muc1) in the colon (Fig. 8 D), demonstrating that ILC3-mediated protection during CDI is at least in part due to acetate-FFAR2 mediated activation of neutrophils. Altogether, these data corroborate that acetate protects from CDI by promoting neutrophil-ILC3 cross-talk through IL-1β.

The presence of commensal gut microbiota is essential to suppress C. difficile expansion and pathogenicity (Carroll and Bartlett, 2011; Koenigsknecht and Young, 2013; Lewis and Pamer, 2017). Our study demonstrates that the protective impact of the microbiota depends at least in part on the production of acetate through fiber fermentation, which activates the FFAR2 signaling pathway in both neutrophils and ILC3s coordinating their protective functions. Acetate-FFAR2 signaling in neutrophils enhances inflammasome-mediated secretion of active IL-1β, which provides a first line of defense against C. difficile toxins. Acetate-FFAR2 signaling in ILC3s augments expression of IL-1R, and the elevated levels of IL-1R boost IL-1β–induced production of IL-22, which elicits antimicrobial and repair mechanisms in intestinal epithelial cells. Thus, microbiota-derived acetate promotes coordinate activation of neutrophils and ILC3s through FFAR2, which bolsters host innate inflammatory and repair responses to C. difficile.

By using newly generated Ffar2^ΔS100 mice, we conclusively demonstrated that acetate provides protection against CDI by activating neutrophil FFAR2. Since FFAR2 acts as a chemotactic receptor (Schulenberg and Maslowsky, 2009; Vinolo et al., 2011), its engagement by acetate may be important to increase the initial recruitment of neutrophils after CDI and/or their survival at C. difficile–damaged sites. Moreover, we found that acetate-FFAR2 signaling augmented IL-1β production by neutrophils in response to C. difficile toxin–mediated activation of the inflammasome. Indeed, the impact of acetate on IL-1β was negligible in Ffar2^−/− as well as Casp1/11^−/− and Nlrp3^−/− mice. How acetate-FFAR2 signaling enhances C. difficile toxin–induced inflammasome activation remains to be investigated. It is possible that acetate-FFAR2 signaling facilitates K⁺ efflux and/or Ca²⁺ influx that trigger NLRP3 activation. Consistent with this, engagement of FFAR2 on colonic epithelial cells by acetate was shown to stimulate NLRP3 through Ca²⁺ influx and cell hyperpolarization (Macia et al., 2015). Acetate-FFAR2 signaling may also activate other downstream mediators that interact with inflammasomes, such as microfilaments, tubulin, and protein kinase C μ–dependent integrin-linked kinase, as shown for other G protein–coupled receptors that impact inflammasome activation (Boro and Balaji, 2017). Although not tested in our study, acetate may also enhance pyrin inflammasome activation in response to C. difficile toxins. While inflammasome activation was mainly detected in neutrophils, we do not exclude that intestinal macrophages may contribute to caspase-1–mediated IL-1β production. We also noticed that acetate can enhance caspase-1–mediated production of IL-18. This effect may be relevant in other models of colitis, such as dextran sulfate sodium–induced colitis, in which IL-18 has a prominent role in promoting epithelial repair (Macia et al., 2015).

In addition to boosting the IL-1β response, acetate enhanced IL-22 response during CDI. IL-22 promotes antimicrobial and repair mechanisms in epithelial cells (Ouyang and Valdez, 2008; Pickert et al., 2009), as well as systemic immune responses via activation of the C3 complement pathway (Hasegawa et al., 2014). IL-22 induction contributed considerably to the overall protection mediated by acetate, which indeed was markedly reduced in Il22^−/− mice, as well as WT mice treated with a neutralizing anti-IL-22 antibody. We confirmed a previous report that ILC3s are a major source of IL-22 during CDI (Abt et al., 2015) and demonstrated that mice specifically lacking ILC3s have a defect in control of the acute phase and in the resolution of CDI. We further examined how acetate induces IL-22 in ILC3s. Although ILC3s express FFAR2, we found no direct effect of acetate-FFAR2 signaling on IL-22 production. However, the acetate-FFAR axis enhanced ILC3 expression of IL-1R, thereby heightening ILC3 sensitivity to IL-1β, a known inducer of IL-22 that is produced by neutrophils during CDI and perhaps by other myeloid cells. Thus, acetate stimulates IL-22 through a coordinate increase in IL-1β bioavailability and ILC3 responsiveness to IL-1β. These results complement the recent observation that FFAR2 signaling promotes in situ proliferation of colonic ILC3 (Chun et al., 2019). It was shown that butyrate, but not acetate or propionate, suppresses ILC2 function and ameliorates ILC2-driven airway inflammation through histone deacetylase inhibition and H3 acetylation, independent of FFAR2 and FFAR3 (Thio et al., 2018). Whether acetate impacts epigenetic regulation of ILC3s, neutrophils, or other immune cells responding to C. difficile remains to be explored.

Currently, the first line treatment for C. difficile–infected patients includes vancomycin, fidaxomicin, and/or metronidazole (McDonald et al., 2018). However, patients suffer relapsing disease refractory to standard antibiotic treatment benefit from fecal transplant (McDonald et al., 2018). Our discovery of the impact of acetate treatment on CDI in mice may further contribute to innovate the treatment of C. difficile. Future studies will be necessary to determine whether acetate is similarly effective in humans. In summary, our data demonstrate a marked protective effect of microbial-derived acetate against CDI and show that acetate activates FFAR on both neutrophils and ILC3s, resulting in a coordinate increase of inflammasome-induced production of IL-1β in neutrophils and responsiveness to IL-1β in ILC3s. Overall, this study highlights the considerable impact and the complex mechanisms of action of intestinal microbiota metabolites in the control of dysbiosis and intestinal infections.

8-wk-old C57BL/6, Rag2/Il2rg-deficient, Rag1^−/−, and S100-Cre male mice were purchased from Jackson Laboratory or the Multidisciplinary Centre for Biological Investigation. Caspase1/11, Nlpr3, and Rag2-deficient mice were purchased from the Centre for Development of Experimental Models for Medicine and Biology of the Federal University of Sao Paulo. Nlrp6^−/−, Nlrc4^−/−, Il22^−/−, and Ffar2^−/− mice were provided by our collaborators from the Institute of Biological Science of the Federal University of Minas Gerais. AhR^fl/fl RorcCre mice were previously reported (Song et al., 2015). Ffar2^fl/fl mice were provided by our collaborator from the University of Illinois at Chicago, IL. All strains were maintained in a C57BL/6 background and were kept in regular filter-top cages with free access to sterile water and food. Animal procedures were approved by the Ethics Committee on Animal Use of the Institute of Biology (protocol numbers 3230-1, 3742-1, and 4886-1) and by Washington University Animal Studies Committee.

To generate Il22ra2^tm1b (null LacZ) and Il22ra2^tm1c (flox) mice, two Il22ra2^{tm2a(EUCOMM)Wtsi} ES cell clones (ES line JM8A3.N1) were obtained from European Conditional Mouse Mutagenesis Consortium; one correctly targeted clone, confirmed by Southern blot analysis, was introduced into B6-albino (C57BL/6J-Tyr^c-2J/J) eight-cell embryos by laser-assisted injection. Male chimeras were initially bred to B6-albino mice to assess germline transmission; those transmitting were bred to CMV-Cre transgenic mice (B6.C-Tg(CMV-Cre)1Cgn/J; >99% C57BL/6) to delete exon 3 of Il22ra2 and the neomycin-resistance cassette. To delete lacZ as well as the neomycin-resistance cassette and generate mice with loxP sites flanking exon 3 of Il22ra2, chimeras were bred to CAG-FLPe C57BL/6 mice. The CMV-Cre and FLPe transgenes were subsequently bred out of the lines.

Toxigenic C. difficile VPI 10463 strain was cultivated in brain-heart infusion (BHI) blood agar supplemented with hemin (5 µg/ml) and menadione (1 µg/ml) at 37°C in anaerobic atmosphere (AnaeroGen, Oxoid; Thermo Fisher Scientific) in jars.

Human colon carcinoma cells (HCT116) were cultivated in DMEM supplemented with 10% FBS, 2 mM l-glutamine, 100 U ml⁻¹ penicillin, and 100 µg/ml streptomycin (Vitrocell Embriolife) until the 10th passage. The MNK3 cell line was previously described and provides an attractive in vitro system to study the function of ILC3/LTi cells (Allan et al., 2015). MNK3 cells were cultured in DMEM with 10% FBS, 2 mM GlutaMAX, 1 mM sodium pyruvate, 55 µM 2-mercaptoethanol, 50 µg/ml gentamicin, and 10 mM Hepes, from HyClone with 4% IL-7 and 2% IL-2.

Mice infections were performed as previously described (Chen et al., 2008). Mice were pretreated with antibiotic mixture (0.4 mg/ml kanamycin, 0.035 mg/ml gentamicin, 0.035 mg/ml colistin, 0.215 mg/ml metronidazole, and 0.045 mg/ml vancomycin; Sigma-Aldrich) added to drinking water for 4 d. Next, mice received clindamycin (10 mg/kg, i.p.; Sigma-Aldrich). After 1 d, mice were infected with 10⁸ CFU of C. difficile by gavage. Mice were weighed and monitored daily during the entire protocol with a clinical severity score that varied from 0 (normal) to 15, as described (Li et al., 2012; Table S1).

Animals received oral pretreatment with acetate at 150 mM or placebo in the drinking water, as reported in other studies (Vieira et al., 2017). Acetate treatment started 1 d before addition of antibiotics and continued throughout the protocol. Acetate solution was prepared using acetic acid (Sigma-Aldrich) at a concentration of 1.5 M in water (stock solution 10×, pH adjusted to 7.2–7.4, filtered at 0.22 µm). Mice had ad libitum access to water and food during the entire protocol. In dietary experiments, mice were maintained in diets with different amounts of fiber: a control diet, based on American Institute of Nutrition (AIN93) recommendations containing 5% cellulose; and a diet that in addition to 5% cellulose was supplemented with 10% of soluble fiber pectin from citrus (Vieira et al., 2017). Mice were prefed with the different diets for 21 d (Fig. 1 A).

Blood samples were harvested from mice, and the serum was used for measurement of SCFA concentration. Samples were prepared as previously described (Fellows et al., 2018). Chromatographic analyses were performed using a GCMS-QP2010 Ultra mass spectrometer (Shimadzu; Thermo Fisher Scientific) and a 30 m × 0.25 mm fused-silica capillary Stabilwax column (Restek Corp.) coated with 0.25 µm polyethylene glycol. Samples (100 µl) were injected at 250°C using a 25:1 split ratio. High-grade pure helium was used as carrier gas at 1.0 ml/min constant flow. Mass conditions were as follows: ionization voltage, 70 eV; ion source temperature, 200°C; full scan mode, 35–500 m/z with 0.2-s scan velocity. The runtime was 11.95 min.

Fecal samples (50 mg) were used for extraction of microbial genomic DNA using the PureLink Microbiome DNA Purification kit (Thermo Fisher Scientific). For bacterial load measurement, DNA was quantified by quantitative PCR (qPCR) using primers complementary to Eubacteria 16S rDNA conserved region (E338F sense 5′-ACT​CCT​ACG​GGA​GGC​AGC​AGT-3′ and U1407R anti-sense 5′-ATT​ACC​GCG​GCG​GCT​GCT​GGC-3′; Durand et al., 2010). A standard curve was constructed with serial dilutions of Escherichia coli genomic DNA. Results were normalized by percentage of the control.

For the in vitro assay with C. difficile and acetate, bacteria were cultured 3 d earlier on BHI blood agar supplemented with hemin (5 µg/ml) and menadione (1 µg/ml) in anaerobic chamber at 37°C. Initially, bacteria were resuspended and washed twice with sterile PBS. After that, the bacterial concentration was adjusted and maintained at 0.5 × 10⁷ CFU/ml in a final volume of 5 ml BHI broth with or without acetate at different concentrations. Cultures were incubated for 72 h in anaerobiosis at 37°C and had the optical density of the medium read at 600 nm wavelength. To verify possible contaminations of the liquid culture, the medium was plated on BHI blood agar, and bacterial growth and gram staining of the colonies were observed.

Mouse colons were harvested, opened longitudinally, and fixed in 4% formalin/0.1% glutaraldehyde. Tissues were processed into historesin, and 5-µm sections were prepared for staining with hematoxylin and eosin or Giemsa/Rosenfeld solution. Slides were analyzed using an Olympus microscope (model U-LH100HG). Samples were analyzed blindly using histological scores for each parameter (Table S2). Overall scores were the sums of each component and varied from 0 to 30.

The lamina propria of the colon was collected and processed for neutrophil analysis. First, the colon was extracted and washed twice with 1× HBSS supplemented with 5% FBS and 2 mM EDTA in a 50-ml Falcon tube under horizontal shaking in an orbital mixer at 250 rpm for 20 min at 37°C. Then, enzymatic digestion was performed in solution containing 1× HBSS supplemented with 2 mM EDTA and collagenase IV (40 U/ml) at 200 rpm for 15 min at 37°C. The obtained cells were then separated from the tissue using a 70-µm filter (Cell Strainer; BD Bioscience). Samples were counted in Neubauer’s chamber, and 10⁶ cells were labeled with specific antibodies coupled to different fluorochromes and analyzed by flow cytometry. Anti-Ly6G FITC (Gr-1) clone 1A8-Ly6g, anti-Ly6C PE clone HK1.4, and anti-CD11b APC clone M1/​​70 anti-mouse antibodies (eBioscience, Thermo Fisher Scientific) were used for analysis of the neutrophil population on the lamina propria. Data were analyzed using FlowJo software.

Colon samples (100 mg) were homogenized in PBS containing protease inhibitors (Thermo Fisher Scientific). Samples were centrifuged for 10 min at 2,000 g, and supernatants used for measurement of IL-1β, TNF-α, CXCL1, CXCL2, IL-10, IL-22, and IL-18 using the Duo Set ELISA kit (R&D Systems).

Spleen, liver, and mLNs were harvested on day 1 or 2 of infection. Bacterial 16S rDNA were extracted using the PureLink Microbiome DNA Purification kit (Thermo Fisher Scientific), and gene levels were quantified by qPCR using primers complementary to Eubacteria 16S rDNA conserved region (Table S3). The bacterial load was determined by a standard curve with serial dilutions of E. coli genomic DNA, and the CFU per gram of tissue was determined by dividing gene levels by sample weight.

Mice received 200 µl FITC-dextran (70,000 D; Sigma-Aldrich) suspension (250 mg/kg) by gavage on day 2 of infection. After 4 h, mice were anesthetized, blood was collected by caudal puncture, and fluorescence readings were performed in a Multi-Mode Microplate Reader (Synergy HT) at 485/528 nm (excitation/emission). A standard curve was prepared with serial dilutions of FITC-dextran in PBS.

Human colon carcinoma cells (HCT116) were cultivated in 1:500 bacterial supernatant and 10, 25, or 50 µM of acetate in 96-well plates (1.0 × 10⁵ cells/well). After 48 h, cells were washed gently using Dulbecco’s PBS (DPBS), and medium was replaced by 100 liters Calcein-AM and propidium iodide (2 µM) in DPBS and incubated for 30 min at 37°C under 5% CO₂. Images were obtained using the Cytation 5 Cell Imaging Multi-Mode Reader, and green fluorescence (485/530 nm, excitation/emission) of viable cells and red fluorescence (530/645 nm) of dead cells were quantified using Gen5 software (Biotek). As a positive control, cells were preincubated with 0.1% Triton X-100 for 15 min. To obtain C. difficile supernatants, the toxigenic strain VPI 10463 was cultured 24 h at 37°C in anaerobic conditions in BHI medium supplemented with hemin and menadione. The culture was centrifuged at 10,000 g for 5 min, and supernatant was used for treatment of HCT116 culture (1:500 ratio).

Total RNA was extracted from tissue using the PureLink RNA kit (Ambion). RNA was converted to cDNA using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems), and qPCR was performed using Power SYBR Green PCR Master Mix (Applied Biosystems) and primers indicated in Table S3. Quantification of gene expression was performed using the 2^{ΔΔ Ct} method, with β2-microglobulin as a reference gene.

Bone marrow neutrophils were isolated using a Percoll (GE Healthcare) gradient (55/65%). The purity of this preparation was >70%. FACS-sorted Ly6G⁺CD11b⁺ cells were also used in some analyses. Cells (4 × 10⁶ cells/ml) were plated at 37°C in RPMI (Vitrocell) containing 10% FBS and without antibiotics. Initially, cells were incubated for 2 h with or without addition of LPS (0.1 µg/ml, E. coli 0111:B4) and 25 mM acetate. C. difficile (1:1) or its supernatant (1:500) was then added to the culture for 4 h. Subsequently, IL-1β cytokine released in the culture was quantified by ELISA. To elucidate the pathways of acetate on inflammasome activation, after incubation in vitro with LPS ± acetate, neutrophils were treated with a known inflammasome activator, the potassium (K⁺) ionophore nigericin (1 µM), or with BAPTA-AM (25 µM), which inhibits NLRP3 by blocking Ca²⁺ influx, ± C. difficile culture supernatant (1:500). Similarly, neutrophils were followed by stimulation with extracellular KCl (45 mM), which inhibits NLRP3 by blocking K⁺ efflux, or NaCl (45 mM), which boosts inflammasome activation by promoting K⁺ efflux. We also performed an experiment with either caspase-1 (10 µM Q-VD-Oph) or neutrophil elastase (10 µM Sivelesat sodium salt) inhibitors. To determine the cellular viability of neutrophils, BioLegend’s FITC Annexin V Apoptosis Detection Kit with 7-aminoactinomycin D (7-AAD) was used, and the supplier’s instructions were followed.

C. difficile–infected mice, treated or not with acetate in the drinking water, received an i.p. dose of 50 µl PBS with 150 g anti-mouse IL-22 neutralizing antibody (clone 8E11; Genentech) or an equivalent amount of isotype control IgG2a (BioXcell) on days 1 and 3 p.i. Mice were clinically evaluated until the fifth day of infection.

Adult small intestine or colon samples were harvested from C57BL/6 male mice, and mesenteric adipose tissue, Peyer’s patches, and intraepithelial lymphocytes were first removed by dissection and two EDTA extraction washes. Intestinal lamina propria immune cells were isolated using Collagenase 4 (Sigma-Aldrich) digestion (40 min) and were enriched at the interface between a gradient of 40% and 70% of Percoll (GE Healthcare) in HBSS. For functional experiments, small intestine lamina propria cells were cultured in a 96-well plate in complete medium and stimulated ex vivo in the presence of Golgi Plug (BD) for 3 h at 37°C. Following incubation, Live⁺CD45⁺LIN^– (CD3^–CD5^–CD19^–B220^–) CD45^lowThy1.2^high ILC3s were stained for surface molecules, fixed, and stained for intracellular IL-22 (clone Poly5164; BioLegend). For IL-1R expression, cells were stained with anti-mouse CD121a (IL-1R, clone JAMA-147; BioLegend). Data were analyzed using FlowJo software. The MNK3 cell line was previously described (Allan et al., 2015). MNK3 was stimulated and stained for intracellular IL-22 as described for primary ILC3s.

Analyses were performed using GraphPad software 5.0. All data are presented as means ± SEM, and n represents the number of samples per group. The exact value of n in each experiment is indicated in the corresponding figure legends. Differences were considered significant for P < 0.05. Results were first analyzed using D’Agostino/Shapiro–Wilk normality tests and compared by Student’s t test or Mann–Whitney U test, as appropriate. For more than two groups, differences were compared by one-way analysis of variance followed by Tukey’s post hoc test.

Fig. S1 shows the relative bacterial load in the feces on days 0 and 2 d p.i., in vitro growth of C. difficile after treatment with acetate, and clinical scores and body weight changes of mice treated with acetate after infection and of infected Nlrp3^−/−, Nlrc4^−/−, and Nlrp6^−/− mice. Fig. S2 shows IL-18 content in the colon of WT and Casp1/11^−/− and Ffar2^−/− mice, neutrophils viability before and after in vitro incubation, and IL-1β production by neutrophils isolated from the bone marrow and stimulated in vitro. Fig. S3 shows expression of SCFA receptors in ILC populations, IL-22 target genes expression in the colon 5 d p.i., ILC subsets in the small intestine 5 d p.i., and representative histological sections of colons of ILC3-deficient mice 5 d p.i. Fig. S4 shows clinical scores, body weight changes, and intestinal permeability of IL-22 binding protein–deficient mice (Il22ra2^−/−); effect of acetate on IFNγ production; in vitro treatment of MNK3 cells with acetate and IL-1β; effect of acetate on Ffar2 and Ffar3 expression; and ILC3 subsets. Table S1 describes the clinical score criteria used to evaluate infected mice. Table S2 lists all parameters used for histopathological analyses. Table S3 includes a list of primers used to analyze the gene expression by qPCR. Table S4 lists the reagents used in our study.

We thank Drs. Jennifer K. Bando, Marina Cella, and Susan Gilfillan for critical comments. We also thank Marcelo Pires Amaral for helping with mouse experiments. We thank Tangsheng Yi (Department of Immunology Discovery, Genentech, South San Francisco, CA) and Genentech for the kind gift of anti–IL-22.

This study was supported by a research grant from Fundação de Amparo à Pesquisa do Estado de São Paulo (12/10653-9, 17/16280-3). The study was also financed by the Conselho Nacional de Desenvolvimento Científico e Tecnológico and Coordenação de Aperfeiçoamento de Pessoal de Nível Superior, Finance Code 001. J.L. Fachi, P.B. Rodrigues, and L.P. Pral are recipient of fellowships from Fundação de Amparo à Pesquisa do Estado de São Paulo (2017/06577-9, 2019/14342-7, 2018/02208-1). M. Colonna was supported by Mucosal Immunology Studies Team, National Institute of Allergy and Infectious Diseases, National Institutes of Health (U01 AI095542). B.T. Layden is supported by the National Institutes of Health under award number R01DK104927-01A1; University of Chicago Diabetes Research and Training Center (P30DK020595); and Department of Veterans’ Affairs, Veterans Health Administration, Office of Research and Development, VA merit (grant no. 1I01BX003382-01-A1).

Author contributions: Conceptualization: J.L. Fachi, M. Colonna, M.A.R. Vinolo; Methodology: J.L. Fachi, C. Sécca, P.B. Rodrigues, F.C.P. Mato, J.S. Felipe, B. Di Luccia, L.P. Pral, M. Rungue, V.M. Rocha, F.T. Sato, U. Sampaio, M. Colonna, S.R. Consonni; Investigation: J.L. Fachi, K.R. Bortoluci, M. Colonna, M.A.R. Vinolo; Original draft: J.L. Fachi, M. Colonna, M.A.R. Vinolo, Review & editing: J.L. Fachi, M.T.P.S. Clerici, H.G. Rodrigues, N.O.S. Câmara, S.R. Consonni, A.T. Vieira, S.C. Oliveira, C.R. Mackay, K.R. Bortoluci, B.T. Layden, M. Colonna, M.A.R. Vinolo; Visualization: all authors; Funding acquisition: M. Colonna, M.A.R. Vinolo; Resources: A.T. Vieira, S.C. Oliveira, H.G. Rodrigues, N.O.S. Câmara, C.R. Mackay, B.T. Layden, K.R. Bortoluci, M. Colonna, M.A.R. Vinolo; Supervision: M. Colonna, M.A.R. Vinolo.