The Tolerance of Gut Commensal Faecalibacterium to Oxidative Stress Is Strain Dependent and Relies on Detoxifying Enzymes

ABSTRACT

Obligate anaerobic bacteria in genus Faecalibacterium are among the most dominant taxa in the colon of healthy individuals and contribute to intestinal homeostasis. A decline in the abundance of this genus is associated with the occurrence of various gastrointestinal disorders, including inflammatory bowel diseases. In the colon, these diseases are accompanied by an imbalance between the generation and elimination of reactive oxygen species (ROS), and oxidative stress is closely linked to disruptions in anaerobiosis. In this work, we explored the impact of oxidative stress on several strains of faecalibacteria. An in silico analysis of complete genomes of faecalibacteria revealed the presence of genes encoding O₂- and/or ROS-detoxifying enzymes, including flavodiiron proteins, rubrerythrins, reverse rubrerythrins, superoxide reductases, and alkyl peroxidase. However, the presence and the number of these detoxification systems varied greatly among faecalibacteria. These results were confirmed by O₂ stress survival tests, in which we found that strains differed widely in their sensitivity. We showed the protective role of cysteine, which limited the production of extracellular O₂^•− and improved the survival of Faecalibacterium longum L2-6 under high O₂ tension. In the strain F. longum L2-6, we observed that the expression of genes encoding detoxifying enzymes was upregulated in the response to O₂ or H₂O₂ stress but with different patterns of regulation. Based on these results, we propose a first model of the gene regulatory network involved in the response to oxidative stress in F. longum L2-6.

IMPORTANCE Commensal bacteria in the genus Faecalibacterium have been proposed for use as next-generation probiotics, but efforts to cultivate and exploit the potential of these strains have been limited by their sensitivity to O₂. More broadly, little is known about how commensal and health-associated bacterial species in the human microbiome respond to the oxidative stress that occurs as a result of inflammation in the colon. In this work, we provide insights regarding the genes that encode potential mechanisms of protection against O₂ or ROS stress in faecalibacteria, which may facilitate future advances in work with these important bacteria.

INTRODUCTION

Faecalibacterium, a genus of strictly anaerobic bacteria, is among the most dominant taxa in the colon of healthy individuals (1). Recently, the taxonomy of this genus has been reassessed, with the classification of certain strains into the following six species: Faecalibacterium prausnitzii, Faecalibacterium butyricigenerans, Faecalibacterium duncaniae, Faecalibacterium gallinarum, Faecalibacterium hattori, and Faecalibacterium longum (2–4). These species are key players in intestinal homeostasis and serve as general health biomarkers (5). Specifically, these bacteria efficiently reduce intestinal inflammation and improve gut barrier function by secreting metabolites that are able to block interleukin-8 (IL-8) production and NF-κB activation (6–8). Indeed, several species of Faecalibacterium produce bioactive molecules that affect inflammation and intestinal barrier function, including the microbial anti-inflammatory molecule (MAM) (9, 10). These bacteria also produce butyrate, which is involved in maintaining the intestinal mucosa and fighting inflammation (11, 12). Moreover, inulin fermentation by strain F. duncaniae A2-165 produces excess fructose, a readily available source of carbon and energy for cells of the human colonic epithelium (13, 14). Due to their potential benefits for human health, faecalibacteria are considered promising next-generation probiotics (15).

However, the beneficial health effects of these bacteria are strongly dependent on environmental conditions in the colon; indeed, one of the characteristics associated with inflammatory bowel disease (IBD) is a notable decline in the abundance of faecalibacteria in the gut (5, 6, 16, 17). A pivotal factor in IBD pathogenesis is oxidative stress, i.e., an imbalance between the generation and elimination of reactive oxygen species (ROS), such as the superoxide anion radical (O₂^•−), hydrogen peroxide (H₂O₂), or the hydroxyl radical (HO^•) (18–20). Oxidative stress, which causes extensive cellular and molecular damage, may be involved in alterations of the microbiota (21). In addition, several studies have suggested a close association between the inflammatory disease process and oxygen (O₂) tension in gut microenvironments (22–25). Indeed, the presence of anaerobic bacteria within the intestinal lumen led to the concept of physiological hypoxia in the adjacent intestinal tissue (26), which has had important implications for our understanding of host-microbiota interactions (24). A decreasing O₂ gradient exists within the healthy human intestinal tract lengthwise from the stomach to the colon (27). Simultaneously, a second gradient is present across the radial axis inward from the intestinal submucosa, as O₂ concentrations decrease to near-anoxia at the midpoint of the lumen (27, 28). Alterations in these gradients have been implicated in intestinal dysbiosis, as evidenced by a shift in bacterial communities from obligate to facultative anaerobes in the gut of IBD patients, which suggests a disruption in anaerobiosis (21).

The defining trait of obligate anaerobes is that O₂ blocks their growth, and yet the underlying mechanisms remain unclear (29). Even in anaerobes, O₂ is a source of endogenous ROS (30). Several studies have revealed that strict anaerobes, such as Bacteroides, Desulfovibrio, Pyrococcus, and Clostridium spp., have highly sophisticated mechanisms for dealing with O₂ (29, 31, 32). Similarly, although faecalibacteria are described as extremely O₂ sensitive, the strain F. duncaniae A2-165 can grow in the presence of low levels of O₂ by using an extracellular electron shuttle of flavins and thiols to transfer electrons to O₂ (33). This process probably involves a flavin reductase, which might regenerate NAD⁺ from NADH and reduce O₂ to H₂O₂ (34). In addition, in certain patients with gut inflammation, a species-level dysbiosis was reported with F. longum L2-6 dominating over F. duncaniae A2-165. This finding suggests that the L2-6 strain might be better than A2-165 at coping with O₂ and oxidative stress in an inflamed colonic environment (35, 36).

In order to fully understand the potential of faecalibacteria as a tool for the prevention and treatment of inflammatory intestinal pathologies, fundamental questions about its biology must be answered. To this end, this study aimed to describe candidate genes involved in the survival and/or adaptation of different strains of the Faecalibacterium genus to oxidative stress generated during gut inflammation. Unfortunately, no genetic tool currently exists for faecalibacteria either to construct mutants or even to transform them with a plasmid. First, we performed an in silico analysis to identify genes coding for scavenging enzymes that might offer protection from O₂ and/or ROS. We were particularly interested in F. longum L2-6 because its relative abundance remains high in certain patients with gut inflammation (35, 36). Moreover, a gene cluster encoding N-acetylgalactosamine use is uniquely present in F. longum L2-6, which raises questions about its involvement in mucin degradation in a damaged gut environment (36). Using strain F. longum L2-6 as a model, we then studied their expression patterns and detected the existence of at least two profiles of induction and likely two regulons that are upregulated when cells fight against O₂ or H₂O₂ stress. Based on this information, we propose the first model of a transcriptional regulatory network involved in the response to oxidative stress in F. longum.

RESULTS

In silico prediction of O₂ and ROS scavenging enzymes.

Using published genomes of 15 isolated and sequenced strains representing several species of the Faecalibacterium genus, we searched for homologs of enzymes that have been shown to scavenge O₂ or ROS in aerobic or anaerobic bacteria, including catalase (Kat), superoxide dismutase (SOD), alkyl hydroperoxide reductase (Ahp), flavodiiron protein (FDP), iron-containing superoxide reductase (SOR), rubrerythrin (Rbr), and reverse rubrerythrin (revRbr). The target sequences for Kat were taken from heme catalases, manganese catalases, and catalase-peroxidases (37). The query sequences for SOD and Ahp were taken from Bacillus subtilis, while those of FDP, SOR, Rbr, and revRbr were from Clostridium spp. and Desulfovibrio spp. In the genomes of faecalibacteria, we identified open reading frames (ORFs) homologous to FDP, SOR, Rbr, and revRbr, which all play important roles in protecting cells from O₂, oxidative, or nitrosative stress (Fig. 1A; Table 1).

FIG 1.

Prediction of enzymatic systems for the detoxification of O₂ and/or ROS in faecalibacteria. (A) Bacterial systems of detoxification of O₂ and ROS. FDPs, revRbrs, and Rbrs reduce O₂ to H₂O thanks to their O₂ reductase activity. Like peroxidases, revRbrs and Rbrs are also able to reduce H₂O₂ to H₂O through their H₂O₂ reductase activity. Only SORs and SODs can reduce the superoxide anion, O₂^•−, into H₂O₂. (B) Scheme of domain functions of O₂- and/or ROS-detoxifying enzymes present in faecalibacteria. (C) Number of genes coding for different detoxification enzymes in 15 isolated and sequenced strains of faecalibacteria. Species names are indicated on the left side (2–4), some strains belong to a distinct cluster C (55), and the lack of a species name indicates that the strain has not yet been precisely classified within the genus Faecalibacterium. Ahp is a potential alkyl peroxidase only found in strain L2-6. The flavin reductase system was reported to reduce O₂ to H₂O₂ in the presence of riboflavins and thiols (33, 34).

TABLE 1.

Strains studied in this work and their genes encoding putative O₂- and/or H₂O₂- detoxifying enzymes

Straina	Gene encoding:
	SOR	FDP	Rbr	revRbr	AhpCF	Flavin reductase
CNCM4543		peg.3223		peg.636		peg.2304
A2-165		peg.2868	peg.2386	peg.331		peg.1599
M21/2	peg.488; peg.1000; peg.669	peg.2188; peg.662; peg.666	peg.670	peg.2154		peg.386
CNCM4541	peg.1592	peg.1798	peg.2042	peg.1060		peg.2324
L2-6	peg.1595; peg.3138	peg.38	peg.743	peg.1991	peg.541; peg.542	peg.698

^aGenomes are annotated according to the PATRIC database.

FDPs are known to reduce O₂ to water or NO to nitrous oxide (38). They are modular proteins with a two-domain core and several additional modules with putative extra redox centers. FDP proteins encoded by faecalibacteria correspond indeed to class A FDPs, which contain only the two canonical domains (Fig. 1B) (38). To act as O₂/NO reductases, class A FDPs receive electrons in general from a small protein, rubredoxin, that itself is reduced by an NAD(P)H:rubredoxin oxidoreductase (39). However, we did not identify proteins homologous to rubredoxins and NAD(P)H oxidoreductases in faecalibacteria, suggesting the existence of another electron transfer mechanism in these bacteria.

SOR enzymes function as oxidoreductases to convert superoxide into H₂O₂ (Fig. 1A) (40). Rbrs act as peroxidases through NAD(P)H oxidation and revRbrs have been reported to have O₂-reductase activity (41–44). Rbrs and revRbrs require NAD(P)H:rubredoxin oxidoreductase as a redox partner. As stated above for FDPs, the redox partner of Rbrs and revRbrs in faecalibacteria remains to be identified. Interestingly, genes encoding the two subunits of Ahp (AhpF and AhpC) were detected only in F. longum L2-6. Ahp enzymes are thiol-specific peroxidases and are the primary scavengers of endogenous H₂O₂ in aerobic bacteria (Fig. 1A) (45).

Overall, examination of these faecalibacterial genomes revealed several candidates for the scavenging of O₂ or ROS (H₂O₂ and O₂^•−). In addition, all the genomes contained a gene encoding a flavin reductase, which may be involved in the extracellular electron shuttle that reduces O₂ to H₂O₂ (34). However, the analyzed genomes differed in the type and number of putative detoxifying enzymes present (Fig. 1C). In particular, strains F. duncaniae CNCM4543 and CNCM4574 contained only two genes encoding one FDP and one revRbr, whereas strain F. prausnitzii M21/2 and CNCM4546 contained many detoxifying enzymes, with several copies of genes encoding FDPs and SORs (Fig. 1C). This finding raised the question of whether the disparity in the content of detoxifying enzymes among strains might correlate with a difference in O₂ and ROS tolerance.

Difference of air tolerance among faecalibacterial strains.

To investigate this hypothesis, we compared different strains of faecalibacteria with respect to their ability to survive in ambient air (about 20% O₂). Five strains with different sets of detoxifying enzymes were selected, as follows: F. longum L2-6, F. duncaniae CNCM4543 and A2-165, F. prausnitzii M21/2, and the unclassified strain CNCM4541. These strains were grown under anaerobic conditions and exposed to atmospheric O₂ for different durations (from 2 to 20 min). As illustrated in Fig. 2A, almost none of the tested strains could withstand prolonged exposure to air, but different mortality trends were observed. The most sensitive strain, namely, CNCM4543, reached 100% mortality after 5 min of air exposure. On average, about 5% of cells of strains A2-165, CNCM4145, or L2-6 survived after 5 min of exposure to atmospheric O₂; after 10 min, this percentage decreased to 0.05% (A2-165) or around 0.5% (L2-6 and CNCM4541), and no cells survived after 20 min. In contrast, 0.15% of M21/2 cells were able to survive 20 min of air exposure (Fig. 2B). Notably, the high sensitivity to O₂ demonstrated by strain NCM4543 correlated with a low number of scavenging enzymes (Fig. 1C), while the increased resistance of strain M21/2 correlated with the presence of many genes encoding putative O₂- or ROS-detoxifying enzymes (Fig. 1C).

FIG 2.

Survival test in ambient air of several strains of faecalibacteria. (A) Pictures of colonies of faecalibacteria exposed to air. A total of 10 μL of serial dilutions of cultures (nondiluted [ND] to −4) were deposited on BHIS medium agar plates and exposed to air for 2, 5, 10, or 20 min. Notably, −1, −2, −3, and −4 indicate 10-fold, 10²-fold, 10³-fold, and 10⁴-fold dilution, respectively. (B) Log of the percentage of Faecalibacterium colonies calculated from the number of CFU/mL that survived exposure to air, depending on the time of exposure. For each strain, four or five independent tests were carried out.

Air exposure and survival in liquid medium for F. longum L2-6.

For this experiment, we focused on F. longum L2-6 because of its interesting behavior in certain patients with gut inflammation (35, 36). First, we developed a protocol to quantify cell survival in liquid medium after exposure to air. In our procedure, we took aliquots of air-exposed cultures and quickly returned them to anaerobic conditions using a preconditioned anaerobic medium in a polypropylene plate (see Fig. S1 in the supplemental material). In this way, we were able to monitor the biomass (optical density at 600 nm [OD₆₀₀]) and to quantify the survival of exponentially growing cells (CFU/mL) exposed to air for 5, 10, 15, or 30 min, without agitation (Fig. 3A and B). F. longum L2-6 showed 100% survival after 30 min of air exposure, suggesting that the cells have efficient O₂ detoxification systems (Fig. 3B). However, cell growth did not resume after the restoration of anaerobic conditions (Fig. 3A).

FIG 3.

Effect of exposure to ambient O₂ or H₂O₂ on the growth and survival of F. longum L2-6. (A) Growth curves of strain L2-6 in BHIS before and after 30 min of exposure to air without agitation. The arrows indicate the time of removal from (O₂) and reintroduction (anaerobiosis) to the anaerobic chamber. The blue portion of the curve corresponds to the duration of air exposure. (B) Counts of CFU/mL for plates of strain L2-6 exposed to air for up to 30 min without agitation. (C) Growth curves of strain L2-6 before and after 30 min of exposure to air with agitation and in the presence or absence of added cysteine. The arrow (O₂) indicates removal from the anaerobic chamber. The dark portion of the curve corresponds to growth in the anaerobic chamber in BHIS medium, the green portion corresponds to exposure to air in the BHIS medium containing cysteine, and the orange portion corresponds to exposure to air in the culture medium without cysteine added. (D) Counts of CFU/mL for plates of strain L2-6 inoculated from liquid cultures exposed to air for up to 30 min with agitation and in the presence or absence of added cysteine. Green bars, survival in the cysteine-containing BHIS medium; orange bars, survival in the BHIS medium without the addition of cysteine. For each condition, four independent experiments were done. (E) Quantification of O₂^•− produced by bacterial cells. O₂^•− production was monitored by following the formation of XTT formazan at 470 nm. At time zero (T0), cells were exposed to air. (F) Effect of H₂O₂ on the growth of strain L2-6 cultivated in BHIS under anaerobic conditions. The arrow indicates the addition of 50 or 100 μM H₂O₂ to the bacterial culture. Curves are representative of three independent experiments.

We then tested the growth and survival of F. longum L2-6 under more challenging conditions by exposing cells to air with agitation and assessed the impact of the absence of cysteine in the culture medium. Indeed, in the presence of O₂, cysteine was proposed to be employed by F. duncaniae A2-165 in a flavin-cysteine redox shuttle to transfer electrons to O₂ (33). This process may involve a flavin reductase, which is present in F. longum L2-6 (Fig. 1C). When cysteine was added to the medium (green curves in Fig. 3C), strain L2-6 survived up to 10 min of aeration (Fig. 3D). Instead, when cysteine was not added, the OD₆₀₀ decreased very sharply (orange curves in Fig. 3C) and was accompanied by a very short survival time with a 3-log decrease of CFU after 5 min (Fig. 3D). Simultaneously, we evaluated the bacterial production of O₂^•− by measuring the absorption of 2,3-bis (2-methoxy-4-nitro-5-sulfophenyl)-2H-tetrazolium-5-carboxanilide (XTT). As depicted in Fig. 3E, after air exposure, O₂^•− gradually accumulated over time when cysteine was added to the medium. When cysteine was absent, the O₂^•− accumulation upon air exposure was 4-fold higher and accelerated sharply in the first few minutes of oxygenation. It thus appeared that the cysteine added to the medium limited the production of O₂^•− and played a protective role that enabled longer survival of F. longum L2-6 under high O₂ tension.

Transcriptional response to oxidative stress.

To further investigate the impact of oxidative stress in F. longum L2-6, we examined the changes in gene expression that occurred in response to stress caused by air as well as H₂O₂. We compared the expression level of six genes that encode putative detoxifying enzymes (Fig. 4 and 5) as well as the gene encoding a putative flavin reductase (34). Expression of the reference peg.52 gene, which encodes a translation elongation factor (TefG), was not modified during oxidative stress (see Fig. S2 in the supplemental material).

FIG 4.

Expression patterns of genes likely involved in O₂ or ROS detoxification after air exposure in F. longum L2-6. Gene expression was analyzed by qRT-PCR and normalized to the reference gene peg.52-TefG. The results were expressed as a ratio after air exposure compared with the nonstressed condition (time at 0 min). (A to F) Expression of the six genes of interest after 0, 2, 5, 10, or 15 min of exposure to air without agitation. The average and standard deviation of three independent experiments are shown. Statistically significant increases in expression compared with time zero are indicated with asterisks (Mann-Whitney; *, P < 0.03). Note that the ordinate scale is different for each panel.

FIG 5.

Expression patterns of genes in F. longum L2-6 in response to H₂O₂ stress. Gene expression was analyzed by qRT-PCR and normalized to the reference gene peg.52-TefG. The results were expressed as a ratio after H₂O₂ exposure compared with the nonstressed condition (time at 0 min). (A to F) Expression of the six genes of interest at 0, 5, 15, or 30 min after the addition of 50 μM H₂O₂ to the BHIS medium. Black bars, expression levels without stress; gray bars, expression following the addition of H₂O₂. The average and standard deviation of three independent experiments are shown. Statistically significant increases in expression compared with the control are indicated with asterisks (Mann-Whitney; *, P < 0.03). Note that the ordinate scale is different for each panel.

Bacteria were exposed to air for 2, 5, 10, or 15 min without agitation. After 10 min of exposure to air, expression of the genes encoding the two SORs was upregulated more than 150-fold (Fig. 4A and B). After 15 min, the Rbr-encoding gene was induced 30-fold (Fig. 4C), while the genes encoding the revRbr, FDP, and AhpC were only slightly but significantly upregulated (Fig. 4D to F). Instead, there was no change in the expression in the gene encoding the flavin reductase under these conditions (Fig. S2). Taken together, these results suggested that the two SORs are crucial enzymes for the protection of faecalibacteria to air likely through the detoxification of endogenous O₂^•−, which accumulates upon air exposure (Fig. 3E).

We also studied the transcriptional response of these genes following the exposure of F. longum L2-6 to a sublethal concentration of H₂O₂ (50 μM) (Fig. 3F) for 5 to 30 min. Reverse transcription-quantitative PCR (qRT-PCR) analysis revealed two expression profiles (Fig. 5). The genes encoding the two SORs, the Rbr, and the revRbr were transiently upregulated with a peak in expression 5 min after stress exposure that was followed by a gradual decrease in expression (Fig. 5A to D). In particular, the Rbr-encoding gene was upregulated 25-fold at 5 min after stress (Fig. 5C), suggesting that the Rbr might be an important enzyme for H₂O₂ detoxification. In addition, the genes encoding the FDP and AhpC enzymes showed a gradual increase in expression until 30 min after H₂O₂ stress exposure (Fig. 5E and F). However, the level of induction remained low (3-fold). Finally, we did not observe any change in the regulation of the gene encoding the flavin reductase (Fig. S2).

In silico analyses of promoter sequences and proposal of a regulatory model.

The similarity of the transcriptional response to H₂O₂ stress exhibited by the genes encoding the Rbr, revRbr, and the two SORs suggested that they belong to the same regulon. Based on this hypothesis, we sought to identify the presence of a common nucleotide motif in the promoter regions of these four genes using the MEME suite 5.4.1 (46). To do this study, we examined the 300 bp upstream of the translation start sites of the four genes and identified a common 22-bp consensus motif in each promoter region (Fig. 6A). This motif is located upstream of putative −10 and −35 elements recognized by the σ^A factor (Fig. 6B). This motif was absent upstream of the genes encoding FDP, Ahp, and the flavin reductase that have a different profile of expression upon H₂O₂ exposure or an absence of induction.

FIG 6.

Analysis of gene promoter regions in F. longum L2-6. (A) Identification of a consensus sequence of 22 bp in the promoter regions of the genes peg.743 (P value of 8.30e-11), peg.1991 (P value of 7.85e-12), peg.1595 (P value of 7.19e-11), and peg.3138 (P value of 3.22e-10). The size of the nucleotide at each position correlates with its relative prevalence in sequences used in the MEME algorithm (46). (B) Promoter regions of the genes of interest. Orange sequences, 22-bp motifs; red underlined sequences, putative -35 and -10 binding sites likely recognized by the σ^A factor associated with RNA polymerase, as predicted with the BPROM tool (56); black underlined sequences, ribosomal binding sites; blue letters, translation start codons. (C) Alignment of 22-bp motifs identified in promoter regions of genes encoding putative O₂- or ROS-detoxifying enzymes in several faecalibacterial species using the MEME algorithm.

In order to evaluate the robustness of this motif, we searched its presence in the promoter regions of genes encoding Rbr, revRbr, and SOR in the genomes of other faecalibacterial species, including F. duncaniae A2-165, F. hattori CNCM4575, F. prausnitzii M21/2, and the strain KLE1255, which belongs to the cluster C (Fig. 1C). Interestingly, the 22-bp motif was also identified in several promoter regions of strains A2-165, CNCM4575, and KLE1255 (Fig. 6C; see Fig. S3 in the supplemental material), which reinforces its potential role in the transcriptional regulation of downstream genes.

Based on our results, we developed a regulatory model (Fig. 7) that is the first to be proposed for any member of faecalibacteria. In this model, we assume that the identified nucleotide motif is the recognition binding site of a still uncharacterized transcriptional regulator that responds to H₂O₂ stress and coordinately regulates the expression of at least four genes.

FIG 7.

Proposed model of transcriptional regulation of genes involved in O₂ and/or ROS detoxification in F. longum strain L2-6. An unknown transcriptional regulator binds to the identified 22-bp motif to regulate the expression of genes in response to H₂O₂. A MerR-like regulator may be involved in the strong activation of the SOR-encoding genes in response to air exposure.

DISCUSSION

In the gut of IBD patients, the relative paucity of strictly anaerobic bacteria, such as faecalibacteria, strongly suggests a disruption in anaerobiosis and points to a role for O₂ in intestinal dysbiosis. In this study, we show that diverse strains of faecalibacteria exhibit a broad range of air sensitivity and possess different O₂ and ROS detoxification systems. It is interesting to note that the CNCM4543 strain, which contains only one FDP and one revRbr, was the most sensitive to O₂ (2 min of resistance to air), while the strain with the highest O₂ tolerance, M21/2, had many genes that encode putative O₂- or ROS-detoxifying enzymes (Fig. 1C). The pattern observed for the other strains is less straightforward. However, this study used drastic conditions (20% O₂ tension) that are far from the physiological environment encountered in a healthy colon (0.1 to 0.4% in the lumen) (27). It would therefore be useful to repeat this experiment using conditions that more closely resemble physiological O₂ tensions.

The differences we observed among faecalibacteria strains raise many questions. Sensitivity to oxidative stress is a very complex phenomenon (47) that does not depend solely on the presence and effectiveness of detoxification enzymes. O₂ and ROS react nonspecifically with many redox enzymes/proteins causing damage to their catalytic centers, Fe/S groups, or flavinic cofactors. It is possible that the wide range of O₂ tolerance demonstrated by faecalibacteria could also be due to differences in the efficiency of detoxification enzymes and/or differences in the sensitivity to oxidation of certain metabolic enzymes. Another hypothesis could be that different faecalibacterial strains do not have the same ability to repair oxidized thiol groups, and indeed, the systems involved in this process deserve further attention. Finally, it should be noted that in an inflamed intestine, the redox potential is also modified (48), which is an observation that has been reported for many diseases associated with alterations of the gut microbiota. At the cellular level in the host, a link has been demonstrated between the regulation of the redox potential and activation of the NF-κB pathway (49). The ability of F. longum L2-6 to persist in inflamed colon could be related to its ability to grow at a higher oxidation-reduction potential than other faecalibacteria.

Using F. longum L2-6 as a model, we investigated the transcriptional expression of seven genes, which encode proteins potentially involved in the response to oxidative stress, including enzymes homologous to Rbr, revRbr, FDP, SOR, and Ahp, as well as a putative flavin reductase. Our results show that the induction of these genes differs between conditions of O₂ and H₂O₂ stress, suggesting that F. longum L2-6 might use different scavenging strategies for O₂ and ROS. In addition, the two different expression profiles triggered by H₂O₂ exposure suggest the possible existence of two regulons. One regulon controls the four genes that encode the two SORs, Rbr, and revRbr. Their transcription is transiently and rapidly upregulated in response to H₂O₂ stress, and a common 22-bp motif is present upstream of the −10 and −35 boxes of these genes. This finding suggests the involvement of a common transcriptional regulator, which is likely an activator that binds to the identified motif and responds directly or indirectly to H₂O₂ (Fig. 7). In addition, the presence of the 22-bp motif is conserved in the promoter regions of genes encoding Rbr, revRbr, or SOR in the genome of F. duncaniae A2-165, F. hattori CNCM4575, and strain KLE1255, suggesting its involvement in a conserved regulatory network across faecalibacterial species.

Interestingly, a gene encoding a MerR-type transcriptional regulator (peg.1594) is located upstream of the peg.1595_SOR gene. The majority of regulators in the MerR family react to environmental stimuli, including oxidative stress (50, 51). To complete our regulatory model, we hypothesize that the strong activation of the two SOR-encoding genes could be under the control of a MerR-like regulator that responds directly to O₂ or indirectly to O₂^•−, which accumulates upon air exposure (Fig. 7).

Our proposed regulatory model highlights how much remains to be discovered about gene regulatory networks in Faecalibacterium, one of the main bacterial genera in the human gut microbiota. We reported previously the presence of complex regulatory networks in the strain A2-165 (52). As a comparison, in strictly anaerobic Clostridioides difficile, O₂ stress is known to affect the metabolism of carbon, amino acids, sulfur, iron, or cofactors, as well as the repair of molecules damaged by oxidation, such as thiol groups and DNA (32, 53). The future use of -omics approaches in faecalibacteria may be able to generate a comprehensive overview of the biological processes affected by O₂ and ROS stress. Such work could highlight the genes or proteins that are directly involved in detoxification processes—in particular, the partner proteins of detoxification enzymes for the transfer of electrons to NADP— and the repair of damaged molecules, as well as the transcriptional regulators that are involved in the stress response and the reorientation of metabolic pathways. These efforts may help to explain the disappearance, maintenance, or domination of certain strains within an inflamed colonic environment.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

The strains of faecalibacteria used for the experiments presented in this study are described in Table 1. They were cultured in brain heart infusion-supplemented (BHIS) medium containing 37 g/L brain heart infusion, 20 g/L yeast extract, 0.027 mol/L sodium acetate tri-hydrate, 0.5 mg/mL l-cysteine, and 2 mg/mL d-fructose. The medium was solidified via the addition of 15 g/L agar. Cultures were grown in a Coy chamber under anaerobic conditions (5% H₂, 5% CO₂, and 90% N₂). During the experiments, strains were cultured in 15-mL or 50-mL centrifuge tubes (polypropylene; Sarstedt AG & Co).

Test of survival on plates in ambient O₂.

An isolated colony of a single faecalibacterial strain was resuspended in 1 mL of BHIS medium and incubated at 37°C overnight. This suspension was used to inoculate a new culture at a ratio of 1:50, which was then incubated at 37°C. When the culture reached the exponential phase (OD₆₀₀ of 0.5), serial dilutions from 10⁻¹ to 10⁻⁴ were done. A total of 10 μL of each dilution was deposited on BHIS plates. Each plate was exposed to ambient air for different durations (0, 2, 5, 10, or 20 min); the exposure time zero min corresponds to the control grown in anaerobiosis. The air-exposed plates were then returned to the anaerobic chamber and incubated at 37°C for 24 to 48 h. For each strain, four or five independent tests were carried out.

Test of survival in liquid medium in ambient O₂.

In this experiment, we used 96-well deep-well polypropylene plates (780210; Greiner Bio-One). Overnight cultures were used to inoculate a culture in BHIS medium at a 1:50 ratio. When the culture reached the exponential phase (OD of 0.5), the sample was transferred to an Erlenmeyer flask with a screw cap. The flask was taken out of the anaerobic chamber at the same time as a preconditioned deep-well plate that contained in each well 450 μL of BHIS coated with 200 μL of mineral oil (M8410; Sigma). The uncapped culture flask was placed in a Microbiological Safety Station hood without agitation. At different durations of exposure to the ambient air, 50 μL of culture was removed from the flask and resuspended under anoxic conditions in a well (under the drop of mineral oil) of the preconditioned deep-well plate. Once the test was complete, the culture flask and the deep-well plate were returned to the anaerobic chamber. To quantify bacterial survival, 10 μL of serial dilutions from 10⁻¹ to 10⁻⁴ were performed of each culture, which were used to inoculate BHIS plates. The plates were incubated for 24 to 48 h at 37°C. Four independent experiments were performed.

To investigate the effect of culture shaking and the presence of cysteine, another experiment was performed. Liquid cultures were prepared as described above. Cells cultivated under anaerobic conditions in BHIS medium containing cysteine were divided into two parts. After centrifugation (2 min at 5,000 rpm), one bacterial pellet was resuspended in BHIS medium containing cysteine, while the second bacterial pellet was resuspended in BHIS without cysteine. Cultures were then exposed to ambient air with agitation. At the time of exposure to ambient air, cultures were agitated (110 rpm) to further aerate the medium. Survival and growth were quantified as described above.

Measurement of O₂^•− free radicals.

A 2,3-bis (2-methoxy-4-nitro-5-sulfophenyl)-2H-tetrazolium-5-carboxanilide (XTT; Sigma) assay was performed to measure the bacterial production of superoxide radical ions (O₂^•−). No light adsorption was observed for BHIS medium alone. Subsequent O₂^•− production was monitored by following the formation of XTT formazan at 470 nm. XTT was dissolved in phosphate-buffered saline (PBS; pH 7) and added to the bacterial culture to a final concentration of 0.4 mM. Absorbance at 470 nm was monitored over time using an Infinite M200 luminescence reader (Tecan, Germany).

Effect of H₂O₂ on growth.

Overnight cultures were diluted in BHIS medium at a 1:50 ratio. Once the culture reached OD₆₀₀ of 0.5, 50 μM, or 100 μM H₂O₂ was added. The OD₆₀₀ of the culture was monitored every hour. To study transcriptional regulation in response to H₂O₂, 2-mL samples of the culture were taken before or at 5, 15, and 30 min after the addition of 50 μM H₂O₂. The samples were centrifuged for 2 min at 20,000 × g at 4°C. Bacterial pellets frozen in liquid nitrogen were stored at −80°C until RNA extraction.

RNA extraction and reverse transcription-quantitative PCR (qRT-PCR).

RNA was extracted using the RNeasy minikit from Qiagen. The concentration of the extracted RNA was measured at 260 nm with a NanoDrop spectrometer, and RNA quality was checked using a bioanalyzer system (Agilent). RNA was reverse transcribed using the high-capacity cDNA reverse transcription kit (Invitrogen) and amplified in Brilliant III ultra-fast SYBR green qPCR master mix (600828; Agilent) using the StepOne real-time PCR system (Applied Biosystems) with the primer pairs in Table 2. The following cycling parameters were used: 10 min at 95°C and 40 cycles of 15 s at 95°C and 1 min at 60°C. Each sample was quantified in technical triplicates. The peg.52 gene from strain L2-6 was used as endogenous control for each sample, and all threshold cycle (C_T) values were normalized to that of peg.52. The differential expression between experimental conditions was determined by Livak’s comparative method, also called ΔΔC_T (54).

TABLE 2.

List of qPCR primers used in this study

Primer name	Sequence	Target
TB021	GGTGAGCTATGAGTCCACCG	Forward peg.38
TB022	TGCCCACGATGTTCAGGTAG	Reward peg.38
TB025	TTGCTGCCAACCTCAAGGAG	Forward peg.743
TB026	AAGTAGCTCTGGGGATGTGC	Reward peg.743
TB029	ACAAGCGCTATGCCTACGAG	Forward peg.1991
TB030	TCTTGGCCAGGTCCATCTTG	Reward peg.1991
TB031	CCCATGGCGGATGTTCACTA	Forward peg.1595
TB032	GCAGTATGCGTAAACAGCCA	Reward peg.1595
TB033	TCCAGTGGCTCTTTGTGGAG	Forward peg.3138
TB034	TCGTCATCCACAGACCATGC	Reward peg.3138
TB071	CCGACACACATTTCGTGCAT	Forward peg.541
TB072	AATGAAGCTTCCGCGCTCT	Reward peg.541
TB073	AGCTGGCATACTTCCGTGTC	Forward peg.52
TB074	AGCAGCAATATCACCTGCGT	Reward peg.52
SA01	CAACATGACCAGCTTCATGG	Forward peg.698
SA02	GAACTCGGGGTACTTCACCA	Reward peg.698

Statistical analysis.

Mean and standard deviation values of the bacterial count (CFU/mL) are represented. For qRT-PCR, the statistical analysis was performed using the nonparametric Mann-Whitney test using GraphPad Prism software (V9.4; San Diego, CA).