Polycyclic aromatic hydrocarbons (PAHs) are of great concern as environmental and foodborne pollutants. These persistent organic pollutants are known to remain in environmental compartments (air, water, soil, and food) and bioaccumulate in organisms (Douben, 2003). Benzo[a]pyrene (B[a]P) is the most well-characterized and toxic member of the PAH compound family. Because of its mutagenic and carcinogenic effects in animal models (Baird et al., 2005; Huderson et al., 2013) and the genotoxic effects observed in humans exposed to B[a]P-containing mixtures (White et al., 2016), B[a]P has been categorized as a human group 1 carcinogen by the IARC (IARC, 2012). Human exposure to PAHs mainly occurs via oral uptake from naturally contaminated food products (Martorell et al., 2012; Veyrand et al., 2013) and the consumption of charcoal-grilled, roasted or smoked food (Lee et al., 2016). However, a recent Chinese cohort study determined that the contribution of high-PAH containing foods, such as barbecued, smoked or deep-fried meats, was approximately 13% of the overall daily dietary intake because of the low associated consumption rates (Duan et al., 2016). Exposure to PAHs occurs to a lesser extent via the inhalation of polluted air particles from cigarette smoke, engine exhaust and domestic heating. Animal model experiments along with in vitro cellular studies and clinical studies have shown that the toxicity of B[a]P and B[a]P-containing mixtures targets different organs in the body and has the potential to develop adenomas. The location of tumors appears to be related to the route of exposure. Inhalation of B[a]P preferentially induces lung cancer (Mumford et al., 1993; Denissenko et al., 1996; Ueng et al., 2010), and oral administration preferentially induces tumors and cancers in the GIT (Sinha et al., 2005; Huderson et al., 2013; Labib et al., 2013), liver (Kroese et al., 2001; Ba et al., 2015) and breast (Guo et al., 2015; White et al., 2016). However, because of the difficulty of assigning a particular toxic effect to a particular route of exposure, most clinical studies are focused on environmental B[a]P exposure, including air and oral intake (Su et al., 2014; Tian et al., 2016). Furthermore, the inhalation of polluted air particles also leads to B[a]P transport to the GIT via the mucociliary clearance mechanism (Möller et al., 2004; Mutlu et al., 2011). Once in the GIT, B[a]P enters the entero-hepatic circulatory system and is metabolized by cytochrome P450-dependent monooxygenases present in intestinal enterocytes and liver hepatocytes, thus leading to the formation of diol-epoxide compounds. These toxic molecules subsequently form B[a]P-DNA adducts (DNA binding products) that have been previously characterized as necessary for B[a]P-initiated carcinogenesis (van Herwaarden et al., 2009).

Although the human enzyme metabolism of B[a]P and its impacts on human health have been well described, studies have not been performed to determine the possible impact of B[a]P on microorganisms living in the human GIT. This microbial community, known as the human gut microbiota, is considered a near-organ and plays major roles in human health. Several pathologies are associated with characteristic shifts in the structure of the gut microbiota (also called dysbiosis), such as IBD (Kostic et al., 2014) and extra-digestive disorders, which include metabolic syndromes (e.g., obesity, diabetes, etc.), autoimmune disorders and neurological disturbances (Vijay-Kumar et al., 2010; Chen et al., 2014; Mu et al., 2016; Teng et al., 2016; Vatanen et al., 2016). Gut microbiota dysbiosis can also lead to the emergence of intestinal pathogens naturally present in the GIT such as Clostridium difficile (Antharam et al., 2013). Exposure of the murine gut microbiota to xenobiotics leads to perturbations in microbial structures and functions (Zhang L. et al., 2015). Zhang L. et al. (2015) showed that dietary TCDF altered mice gut microbiota by reducing the ratio of Firmicutes to Bacteroidetes. Furthermore, AhR signaling activation induces dietary TCDF to alter many host metabolic pathways involved in hepatic lipogenesis, gluconeogenesis and glycogenolysis, bacterial fermentation, and amino acid and nucleic acid metabolism. Allais et al. (2015) exposed mice to cigarette smoke for 24 weeks and observed alterations in the bacterial community structure. Furthermore, the level of 16S rRNA expression (and therefore the activity) of Lachnospiraceae sp. strongly increased in the colon. Besides metagenomics and metatranscriptomics, volatolomics has proven to be a promising omic approach to diagnose metabolism changes in response to physiological stresses induced by pathology (Hakim et al., 2012) or xenobiotic exposure (Berge et al., 2011). Regarding gastrointestinal or inflammatory disorders, Ahmed et al. (2013, 2016) reported that changes in fecal VOC pattern may result from changes in the microbiota and/or pathologies in the GIT.

Although studies have not focused on the impact of B[a]P on the human gut microbiota, B[a]P metabolization has been demonstrated by this microbial consortium. van de Wiele et al. (2005) showed, in vitro, that the human gut microbiota can biotransform B[a]P into estrogenic metabolites, and they also identified 7-hydroxybenzo[a]pyrene as a B[a]P derivative.

To highlight the underlying mechanisms and associated consequences of the interaction between B[a]P and gut microbiota, we performed 16S barcoding, metatranscriptomic and volatolomic analysis to determine the impact of B[a]P on two distinct human fecal microbiota.

B[a]P is a ubiquitous environmental pollutant that bioaccumulates in organisms and is formed de novo by certain food processing methods, such as smoking and high cooking temperatures. Although digestive and extra digestive pathologies have been linked to compositional or functional alterations of the gut microbiota, B[a]P-induced perturbations of the human gut microbiota have not been previously studied.

The B[a]P dosages chosen for this work were informed by two previous studies carried out on the interaction between the B[a]P and the gut microbiota allowing us to make comparisons and discussions (van de Wiele et al., 2005; Ribière et al., 2016). These doses are higher than the typical daily consumption of B[a]P; however, B[a]P exposure is chronic throughout a person’s life. The experiment included two clearly distinct FM structures which might reflects the large individual gut microbiota variations found in the human population. Thus similar responses of the two FM toward the B[a]P exposure could not be attributed to a restricted microbial composition.

Surprisingly, B[a]P did not induce a significant change in the microbial structure, although slight modifications were observed at the family level, such as increases in the Sutterellaceae (formerly Alcaligenaceae), Tissierellaceae and Ruminococcaceae families and reductions in the Rikenellaceae and Lachnospiraceae families. Our laboratory previously investigated the impact of 28 days of oral B[a]P exposure on the fecal and intestinal mucosa-associated bacteria in mice (Ribière et al., 2016). Similar to the results of the present work, we showed that B[a]P did not have a significant impact on the bacterial richness (Shannon index) and diversity (chao1 index, observed_otus) during the course of exposure. However, we showed that B[a]P induced significant shifts in the composition and relative abundance of the stool and mucosa associated bacterial communities. These results differed from those presented here because significant variations in the bacterial structure were not observed, which might have been caused by various aspects of the experimental design. Murine models are widely used in gut microbiota studies; however, discrepancies between human and mice gut bacterial composition have been reported. Although Bacteroidetes and Firmicutes are the two major phyla dominating the human and mouse gut microbiota (Eckburg et al., 2005; Ley et al., 2005), 85% of the bacterial genera found in mouse gut microbiota are not present in humans (Ley et al., 2005). Finally, several environmental factors may shape the mice gut microbiota, specifically cage effects (including coprophagy), dietary factors (standardized chow diet composed mainly of plant materials), and genetic background (including inbreeding to preserve genetic homogeneity).

Similarly to our present work, Maurice et al. (2013) showed that a short period of exposure (4 h) with six host-targeting drugs did not alter the microbial active community structure of the human gut microbiota. A prior work from Sawulski et al. (2014) demonstrated that in the soil environment, alterations in the bacterial community structure may be correlated to the pattern of degradation of a given PAH. B[a]P was degraded much more slowly than less complex PAHs and to a lower extent, which was reflected in slow changes of the bacterial community structure (no significant variations were observed after 2 days of exposure). Changes of the gut microbiota structure after xenobiotic exposure has previously been reported, and most studies in which xenobiotic exposure (antibiotics excluded) altered the mice microbial structure lasted at least several days and often weeks (Breton et al., 2013; Chassaing et al., 2015; Zhang Y. et al., 2015; Ribière et al., 2016). Thus, to test such a chronic exposure, a 28-days experiment in continuous fermentor will be carried out in the laboratory.

Although the two FM structures appeared to remain stable after a short exposure time, the bacterial activity may have adapted to the presence of the toxic compound. The volatolome profiling has been rapidly emerging in disease diagnosis (Arasaradnam et al., 2014), and changes in VOC patterns have been reported in many physiological and pathological states, including gut dysbiosis in gastrointestinal diseases. Seven VOCs were found to have significantly altered concentrations in both FMs tested under exposure to B[a]P (Table 1). There are no reported links between these compounds and known microbial metabolic processes. As end-products, these compounds may arise from diverse microbial metabolic pathways. Control samples with the pollutant but without gut microbiota were assessed in the SPME-GC-MS analysis. VOCs that are identified as shifting in the manuscript were either not found in control samples or were present in control samples but showed no abundance variation between vehicle and B[a]P samples. We did not look at the fungi community. However, using the SortmeRNA software (Kopylova et al., 2012) on the RNA-Seq data, we did not see any ribosomal eukaryote sequences. Thereby, excluding an abundance below the detection limit, VOCs seem to arise from prokaryote activity.

The depletion of volatolome esters like butyl butanoate has recently been reported in response to gastrointestinal metabolic disorders (De Preter et al., 2015). Accordingly, Ahmed et al. (2015) suggested that differences in levels of fecal esters may reveal changes in the fermentation process of the gut microbiota. Regarding changes in benzaldehyde, 2-methyl-phenol, 3-octanone, 2-pentylfuran, 2-hexylfuran, De Preter et al. (2013) reported that the depletion or augmentation of metabolites belonging to benzenoids, ketones and furan derivatives would probably be a consequence of the disruption of the normal bacterial ecology in pathologies like IBD. It could be hypothesized that these same compounds would also trigger a gut metabolic response to chemical agents such as B[a]P.

In the present study, B[a]P exposure alters the gut microbiota volatolome in a dose-dependent manner. Furthermore, although the bacterial composition of the two studied microbiota were clearly different, their metabolic responses were similar with seven markers significantly altered by B[a]P exposure in both microbiota. This shift in the volatolome shows that B[a]P induces a deviation in the microbial metabolism. At present, no systematic studies have been conducted on the identification of gut microbiota volatolome after xenobiotic exposure. Nonetheless, Berge et al. (2011) noted a shift in the volatolome of liver cells of chickens exposed to various dietary chemical contaminants, including a mixture of three PAHs among which was B[a]P. Therefore, the present work may enlarge again the field of potential applications of volatolomics, which already includes cancer diagnosis (Hakim et al., 2012; Broza et al., 2015), food authentication (Engel et al., 2007; Sivadier et al., 2008), and ecotoxicology (Ratel and Engel, 2009; Berge et al., 2011).

Finally, to further characterize the influence of B[a]P on the active gut microbiota, we analyzed gene expression using RNA sequencing. Transcripts related to four KEGG metabolic pathways were found to be upregulated in both fecal metatranscriptomes. These pathways include vitamin and cofactor metabolism, cell wall compound metabolism, DNA repair and replication systems, and aromatic compound metabolism.

Some of these results may be linked to previous data from the literature as for the lipophilic and mutagenic properties of the B[a]P. As lipophilic compounds, PAHs have been shown to penetrate into the cytoplasmic membrane, which results in an increase in membrane fluidity that leads to the loss of membrane functionality and bacterial cell damage (Sikkema et al., 1994). One of the major adaptive mechanisms of bacteria cells to counteract this effect is to increase membrane rigidity (enhance membrane lipid saturation) to prevent compound accumulation (Murínová and Dercová, 2014). This bacterial physiological adaptation may explain the increase observed in the level of transcripts implicated in the metabolism of cell wall compounds. The mutagenic effect of B[a]P is also known to occur via DNA adducts (Baird et al., 2005). DeMarini et al. (2011) characterized the mutagenic potencies and mutation spectra of B[a]P in Salmonella strains TA98, TA100, and TA104. The results from this study showed that (i) 24% of the mutations induced by B[a]P in Salmonella led to complex frameshifts, (ii) B[a]P produced primarily GC to TA transversions and (iii) the majority (96%) of stable adducts induced were at a guanine site. Similarly, the mutagenic potencies of the B[a]P might explain the observed induction of the DNA repair and replication systems. Furthermore, within the vitamin and cofactor metabolism pathway, the transcript levels related to folate biosynthesis, which is a major cofactor in DNA synthesis (Harvey and Dev, 1975; Sangurdekar et al., 2011), increased in both metatranscriptomes.

In the FM-2 metatranscriptome, transcripts related to the steroid hormone biosynthesis pathway were also upregulated, which is consistent with prior work from van de Wiele et al. (2005). They implemented a batch fermentation system and showed the ability of the human gut microbiota to metabolize several PAHs, including B[a]P, to metabolites with estrogenic activity. Indeed, several hydroxylated-PAH metabolites are structurally similar to the steroidal hormones that bind the human estrogen receptor (ER) (Fertuck et al., 2001; Hirose et al., 2001; Sievers et al., 2013). These authors confirmed the PAH biotransformation by the identification of the metabolites 1-hydroxypyrene and 7-hydroxybenzo[a]pyrene in colon digests of pyrene and B[a]P, respectively. Finally, few pathways were downregulated, such as the glycolysis-gluconeogenesis and the bacterial chemotaxis toward simple carbohydrates. This reduction in energy metabolism may be part of the cell adaptation process. Stressful environmental conditions, such as B[a]P exposure, may lead the cells to rapidly perform coping mechanisms. Microorganisms may thus engage energy for adaptation mechanisms to ensure necessary physiological functions prior to devoting resources to energy consumption for growth and multiplication (Murínová and Dercová, 2014). The ability of gut microbial strains to bind B[a]P via physical adsorption to peptidoglycans has also been reported (Hongfei et al., 2013). Thus, another hypothesis is that steric hindrance caused by B[a]P adsorption limits the bacterial ability to sense carbohydrates in its environment.

For the first time, a model of acute B[a]P exposure in the human gut microbiota was initiated. Although significant impacts were not observed on the gut microbiota structure, the microbial activity was altered both at the metatranscriptome and at the volatolome levels that could contribute to disturb human gut homeostasis. Humans are exposed to B[a]P as well as to mixtures of PAHs and other environmental and foodborne chemicals. Thus, our results suggested that further works should consider pollutants that could contribute to the development of microbial functional dysbiosis, which will potentially impair the human host homeostasis.

All sequence data produced via 16S rRNA/rDNA amplicon and RNA sequencing are available in the NCBI Sequence Read Archive, BioProject PRJNA339203, under accession no. SRP082261.

CD, PP, EP, and EE conceived the study and design the experiments. CD and JR performed experimental procedures. CD, JR, EE, and PP analyzed data. SD provided support for batch fermentations and RB for metatranscriptomic experiments. BB helped with treatments of sequencing data analysis. CD, PP, EP, and EE interpreted data and wrote the manuscript. All authors reviewed and approved the manuscript. All authors agree to be accountable for all aspects of the work.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.