Opportunistic fungal pathogen Candida glabrata circulates between humans and yellow-legged gulls

The opportunistic pathogenic yeast Candida glabrata is a component of the mycobiota of both humans and yellow-legged gulls that is prone to develop fluconazole resistance. Whether gulls are a reservoir of the yeast and facilitate the dissemination of human C. glabrata strains remains an open question. In this study, MLVA genotyping highlighted the lack of genetic structure of 190 C. glabrata strains isolated from either patients in three hospitals or fecal samples collected from gull breeding colonies located in five distinct areas along the French Mediterranean littoral. Fluconazole-resistant isolates were evenly distributed between both gull and human populations. These findings demonstrate that gulls are a reservoir of this species and facilitate the diffusion of C. glabrata and indirect transmission to human or animal hosts via environmental contamination. This eco-epidemiological view, which can be applied to other vertebrate host species, broadens our perspective regarding the reservoirs and dissemination patterns of antifungal-resistant human pathogenic yeast.

. Frequency distribution of the 129 MLVA genotypes identified for the 190 C. glabrata samples isolated from yellow-legged gulls or humans. *A singleton is a genotype that has been found only once in the study population.
yellow-legged gull isolates was comparatively higher than among the human isolates. In particular, differentiation due to genetic structure was particularly high between the population from the Riou Archipelago and those from all other study sites. The overall highest pairwise F ST value was 0.615 (P < 0.001) between Riou and La Grande-Motte (Fig. 3, Table 3). The Mantel test showed that geographical distances explained 14.4% of C. glabrata genetic differentiation (P = 0.023).
In vitro fluconazole susceptibility was assessed in 54 C. glabrata isolates, all of which were collected within the same time period in the Marseille area, including 25 samples collected from yellow-legged gull breeding colonies on the Frioul and Riou Archipelagos and 29 isolates collected from patients at the university hospital of Marseille. Overall, 23 isolates were classified as fluconazole resistant (minimal inhibitory concentration ≥ 64 mg/L); 9 and 14 (36.5%, 95% confidence interval (CI) [18.0-57.5%] vs. 50%, 95%CI [29.5-67.5%], P = 0.53) were isolated from gull or human hosts, respectively. The absence of genetic clustering according to the host or fluconazole susceptibility is depicted in the MLVA-based MST tree (Fig. 4).

Discussion
Overall, this study highlights the absence of significant genetic differentiation between C. glabrata populations in humans or yellow-legged gulls. We also demonstrated that antifungal-resistant isolates are present within the gut mycobiota of yellow-legged gulls. The low differentiation between human and gull C. glabrata populations is in agreement with a previous study that has shown that C. dubliniensis populations isolated from herring gulls (Larus argentatus) or humans were genetically similar 30 . In contrast, geographic location of the collection site was the major factor in genetic variance. Furthermore, Mantel test analysis showed a trend of increasing genetic differentiation with increasing geographical distance. Similarly, de Meeûs et al. 31 have shown using both multilocus enzyme electrophoresis and randomly amplified polymorphic DNA that C. glabrata populations isolated from patients in Paris and Montpellier, which are 800 km apart, displayed genetic differentiation (F ST = 0.11, P = 0.054). Using the same MLVA scheme as in the present study, Dhieb et al. 32 have shown highly significant genetic differentiation (F ST = 0.359, P < 10 −5 ) between C. glabrata populations isolated from patients in France or Tunisia. Although the geographical scale of our study was much more limited than in the previous studies, we detected either relatively high or low genetic differentiation according to the study sites. As we found evidence for both dispersion and differentiation, our study was indeed adequately scaled to dissect transmission profiles and detect reservoirs.
C. glabrata is a component of the human gut mycobiota. We hypothesize that yellow-legged gulls inadvertently ingest yeast such as C. glabrata with their food, which might be contaminated with human excreta in highly  Table 2. Allelic richness, diversity and evenness of the 111 MLVA genotypes identified for the 190 C. glabrata samples isolated from yellow-legged gulls or humans were estimated for each study site. Allelic richness was specified using the number of multilocus genotypes observed per population (MLG) and the number of expected MLG at the smallest sample size based on rarefaction (eMLG). Genetic evenness was estimated using the E5 index. Diversity was calculated using the Simpson's (lambda) index corrected by the number of isolates in a population.

Figure 1. STRUCTURE clustering (admixture) in which each isolate is represented by a single vertical line that is partitioned into K = 2 colored segments.
The segment length represents the individual's estimated membership fractions in cluster 1 (red) and cluster 2 (green). Isolates with multiple colors have admixed genotypes from each cluster.  anthropic marine and terrestrial environments. Compared with other sea birds, yellow-legged gulls are highly synanthropic. This is a major reason that gull populations have grown concomitantly with human-made environments, including human refuse sites, along the Mediterranean littoral 33 . Moreover, beach sand may also play a role as a reservoir for C. glabrata 34 . The birds become a reservoir as the yeast develops into a component of the gut microbiota. The yellow-legged gull can fly relatively extended distances along the Mediterranean littoral to feed on landfills. Therefore, garbage dumps in urbanized areas may be a potential source of clinically important yeast transmitted by gulls. Due to the high mobility of gulls, the birds facilitate the dissemination of the yeast by releasing their droppings over an expansive area of the marine and terrestrial environment. Humans may be infected with C. glabrata originating directly (via bird droppings in their direct environment) or, more frequently, indirectly by ingesting food that has been contaminated with bird droppings. The genetic homogeny of human and bird isolates clearly suggests that yellow-legged gulls play a role in the diffusion of C. glabrata acquired from an anthropic environment. In line with this hypothesis, yellow-legged gulls transmit and spread potential human pathogens in various environments. Indeed, Bonnedahl et al. 16 have demonstrated that yellow-legged gulls disseminate antibiotic-resistant Escherichia coli isolates not far from Pierre Blanche. Similarly, we demonstrated the presence of genetically homogeneous fluconazole-resistant C. glabrata populations in both gulls and humans in Marseille. Therefore, our findings show that yellow-legged gulls act as a carrier, reservoir and disseminator of C. glabrata. These birds may thus contribute to the transmission of yeast to humans and likely other animal hosts. Moreover, the likelihood of infectious disease transmission via yellow-legged gulls has therefore increased due to the marked demographic growth of gull populations along the Mediterranean littoral. Our findings also demonstrate that yellow-legged gulls represent a reservoir of antifungal-resistant Candida strains and disseminate antifungal-resistant isolates.

Conclusions
The close proximity and interaction between very dense human and yellow-legged gull populations in cities of the Mediterranean littoral facilitates the circulation of microorganisms between the two hosts. Gulls likely ingest C. glabrata by eating or drinking in environments contaminated with human excreta. The yeast eventually    evolves to be incorporated into their gut mycobiota and is spread via gull feces to human environment. Moreover, yellow-legged gulls represent a reservoir of fluconazole-resistant C. glabrata isolates, which may have a potential impact on public health, including food spoilage control and potable water source protection.

Methods
C. glabrata isolates. In this study, we analyzed 190 C. glabrata isolates. One hundred eleven samples were isolated from feces collected on the soil at five yellow-legged gull breeding colonies as previously described by Al-Yasiri et al. 29 . Briefly, the sampled breeding colonies were located in the departments of Hérault and Bouchesdu-Rhône, in the South of France. In Hérault, the colonies were located in a natural reserve at the lagoon of Pierre Blanche and in two cities, Palavas-les-Flots and La Grande-Motte. In Bouches-du-Rhône, two breeding colonies were located on the Frioul and Riou Archipelagos off the coast of Marseille. Yellow-legged gulls may yet be exposed to varying levels of anthropogenic pressure. In this study for instance, birds breeding at the lagoon of Pierre Blanche were exposed to a relatively low anthropogenic influence compared with those breeding on the building rooftops in the cities of Palavas-les-Flots, La Grande-Motte and Marseille. Riou and Frioul Archipelagos are suburban ecocline exposed to an intermediate anthropogenic influence. Seventy-nine C. glabrata samples were isolated from patients at three university hospitals in Marseille, Montpellier and Nimes, which were located in the same geographical area and sampled during the same time period as those isolated from gulls. All isolates were subcultured on Malt extract agar (Sigma Aldrich, USA). The samples were identified via the MALDI-TOF MS (matrix-assisted laser desorption/ionization-time-of-flight mass spectrometry) technique, as previously described 35 , and subcultured onto chromogenic medium plates (CHROMagar TM , Becton Dickinson, France) to verify isolate purity 36 .
Multiple-locus variable number tandem repeat analysis (MLVA). All C. glabrata isolates were typed with eight microsatellite markers (GLA2, GLA3, GLA4, GLA5, GLA6, GLA7, GLA8 and GLA9) as previously described by Brisse et al. 37 . Genomic DNA was extracted using the NucliSENS TM EasyMAG TM (bioMérieux) system 38 , eluted in 50 μ l and stored at − 20 °C. Amplification reactions were performed using a Lightcycler TM 480 (Roche Diagnostics, GmbH, Germany) instrument with Lightcycler TM 480 Probes Master (Roche Diagnostics, GmbH, Germany). The loci, primer sequences, fluorophores and hybridization temperatures are described in Table 4. The PCR products were visualized using 2% agarose gel electrophoresis in 1X of Tris borate EDTA buffer (Euromedex, France) with SYBR TM safe DNA gel stain (Invitrogen, USA). Next, 1 μ l of 1:100 diluted PCR products was mixed with a solution containing 25 μ l HiDi formamide (Life Technologies, France) and 0.5 μ l Gene Scan TM 500 LIZ TM size standard (Applied Biosystems, UK). The fragment length was determined via capillary electrophoresis using an ABI 3130 Genetic Analyzer (Applied Biosystems, France) and analyzed using GeneMapper software v4.0 (Applied Biosystems, France).
Population genetic analysis. Several indices of clonal diversity were estimated using the poppr R package 39 , including the genetic richness, i.e. the number of multilocus genotypes (MLG) observed per population and the number of expected MLG at the smallest sample size based on rarefaction (eMLG); the genetic evenness E 5 , (ranging from 0, a population composed of a single genotype, to 1, all genotypes having equal frequency); Simpson's (lambda) diversity index 40 corrected by the number of isolates in a population; and the clonal fraction 41 . STRUCTURE v. 2.3.1 was used to identify genetically distinct clusters via estimation of the proportion of membership (Q) of each isolate in each cluster. This software applies a Bayesian model-based clustering approach 42 to model K populations that are characterized by a set of allele frequencies at each locus and probabilistically assigns isolates to clusters based on the multilocus genotypes. For each K value, ranging from 1 to 10, the posterior probability of the data was measured using an admixture model with correlated allele frequencies over 10 independent runs with a burn-in period of 5000 followed by 50000 Markov Chain Monte Carlo steps. The number of clusters was estimated by the value of K that maximizes the posterior probability of the data. The highest Δ K value corresponds to the most pronounced partition of the data and indicates the likely number of clusters.
The among-and between-population differentiation due to genetic structure was estimated via Slatkin's linearized pairwise fixation index (F ST ) and Analysis of MOlecular VAriance (AMOVA) using ARLEQUIN v3.5 software. The effect of geographical distances on genetic differentiation was tested via Mantel test. MLVA-based Minimum Spanning Tree (MST) was constructed using BIONUMERICS software v7.1.  Fluconazole susceptibility testing. C. glabrata anti-fluconazole susceptibility was assessed as described by Clinical Laboratory Standards Institute (CLSI) M27-S4 document 43 . Fluconazole resistance was defined as a minimal inhibitory concentration of ≥ 64 mg/L. Each assay was validated using ATCC 22019 (Candida parapsilosis) and ATCC 6258 (Candida krusei) quality control strains.