Wild apes are considered to be the most serious reservoir and source of zoonoses. However, little data are available about the gut microbiota and pathogenic bacteria in gorillas. For this propose, a total of 48 fecal samples obtained from 21 Gorilla gorilla gorilla individuals (as revealed via microsatellite analysis) were screened for human bacterial pathogens using culturomics and molecular techniques. By applying culturomics to one index gorilla and using specific media supplemented by plants, we tested 12,800 colonies and identified 147 different bacterial species, including 5 new species. Many opportunistic pathogens were isolated, including 8 frequently associated with human diseases; Mycobacterium bolletii, Proteus mirabilis, Acinetobacter baumannii, Klebsiella pneumoniae, Serratia marcescens, Escherichia coli, Staphylococcus aureus and Clostridium botulinum. The genus Treponema accounted for 27.4% of the total reads identified at the genus level via 454 pyrosequencing. Using specific real-time PCR on 48 gorilla fecal samples, in addition to classical human pathogens, we also observed the fastidious bacteria Bartonella spp. Borrelia spp., Coxiella burnetii and Tropheryma whipplei in the gorilla population. We estimated that the prevalence of these pathogens vary between 4.76% and 85.7%. Therefore, gorillas share many bacterial pathogens with humans suggesting that they could be a reservoir for their emergence.
The African great apes, including gorillas, are a reservoir and source of human pathogens1,2,3,4,5,6. Calvignac-Spencer et al. identified 16 viral genera from wild great apes, including 8 genera that could be transmitted between humans and apes7. Among these different transmissible viruses, the Ebola and Marburg viruses are the most virulent pathogens and have caused multiple human outbreaks due to direct handling of gorillas and chimpanzees2,8. Furthermore, there is strong evidence that the human immunodeficiency virus originated from simian immunodeficiency viruses in chimpanzees and gorillas4. In addition to viruses, parasites, including Plasmodium falciparum and Plasmodium vivax, nematodes (such as Trichuris, Ascaris, Oesophagostomum, Strongyloides, and Trichostrongylus) and the Anoplocephalid cestode can infect both humans and great apes3,9.
However, there are limited data on pathogenic bacteria in the gorilla. These bacteria include Staphylococcus aureus, Escherichia coli, Rickettsia felis, Bacillus anthracis-like bacteria, Salmonella spp., Campylobacter spp. and Shigella spp.9,10,11,12. Additionally, Mycobacterium tuberculosis and leprosy caused by Mycobacterium leprae have been observed in nonhuman primates13,14. Furthermore, Rwego et al. demonstrated the possibility of gastrointestinal bacterial transmission between humans and gorillas sharing the same habitat15. Therefore, studying the composition of bacterial communities in the gorilla gastrointestinal tract is important.
Although pyrosequencing is a rapid and efficient method for determining the bacterial phyla, it is limited, especially for bacterial identification at the genus and species levels16.
Microbial culturomics (large-scale culture conditions followed by mass spectrometry or 16S rRNA identification of the isolated colonies) was proved to be an efficient method to explore the gut microbiota not only because it is able to isolate a high number of bacterial species including new species but also because it is more sensitive than metagenomic methods to detect minority populations including pathogens16,17,18.
In this study, we explored the prevalence of human bacterial pathogens in wild gorillas from southern Cameroon. We exhaustively analyzed one stool sample using culturomics and pyrosequencing to detect the commensal bacteria that are potential human pathogens19. Forty-eight additional fecal samples from wild gorillas were screened via real-time PCR to examine the prevalence of the human bacterial pathogens that were identified in the first stool sample using culturomics and of other human pathogens, including fastidious bacteria that are tested routinely in our lab for the diagnosis of infections in Africa.
Source of fecal samples, microsatellite analyses and genetic identification of gorilla individuals
Fecal samples from western lowland gorillas (Gorilla gorilla gorilla) were collected at two sites in south-central Cameroon: near the Minton village (this sample was selected and used for exhaustive examination using culturomics, pyrosequencing and real-time PCRs because of its enough volume required for these analyses), and 47 fecal samples were collected near the Messok village (these samples were used for the molecular screening of human bacterial pathogens via real-time PCRs). A microsatellite analysis of 48 fecal samples enabled the identification of 21 gorilla individuals: 19 males and 2 females (Supplementary Table 1).
Bacterial diversity of gorilla gut using culturomics
In this study, a total of 86 culture conditions were tested: the optimal conditions applied for human gut exploration (Supplementary Table 2)16 and innovative media, which were developed from plants (Supplementary Table 2). From one fecal sample, a total of 12,800 colonies were tested and 147 bacterial species were observed (Table 1 and Supplementary Table 2). The bacteria belonged to 47 genera divided into 4 phyla; Firmicutes (62.6%), Actinobacteria (24.5%), Proteobacteria (12.2%) and Fusobacteria (0.7%) (Table 1 and Supplementary Table 2). A comparative analysis showed that 113/147 (76.9%) of the bacterial species observed in the gorilla gut using culturomics were also observed in the human gut using culturomics (Table 1). Among the 86 different culture conditions tested, the most effective was a culture vial supplemented with rumen fluid in an anaerobic atmosphere followed by incubation for 11 days; this isolated 19 bacterial species (Supplementary Table 2). The incubation of the stool sample in aerobic or anaerobic enriched culture vials yielded 60 bacterial species (40.8%) including 50 (34%) anaerobic species (Supplementary Table 2).
Bacterial analyses of gorilla gut using specific plant media
Because gorillas are principally herbivores and because the gut microbial diversity may be influenced by diet, we designed novel media using tobacco leaves and tropical plants including the mango, papaya and banana fruits (see Methods and Supplementary Table 2). Of the 147 bacterial species observed, 60 (40.8%) grew on media supplemented with plant extracts, including 28 strains (19.04%) that were isolated exclusively using these media (Supplementary Table 2). The best plant condition was a culture vial supplemented with banana in an anaerobic atmosphere, used here for the first time, which enabled the isolation of 17 bacterial species (Supplementary Table 2).
New bacterial species and human pathogens in the gorilla gut
In addition to previously known bacteria (Table 1 and Supplementary Table 2), five new species were isolated in the fecal sample analyzed via culturomics, and their genomes were sequenced, including one new genus, Gorillibacterium massiliense20, and 4 new species: Bacillus massiliogorillae21, Microbacterium gorillae, Paenibacillus gorillae22 and Paenibacillus camerounensis (Table 1 and Supplementary Table 2). The latter two species were recovered only from media supplemented with mango fruit (Supplementary Table 2). Notably, several potential human pathogens, including Proteus mirabilis, Klebsiella pneumoniae, Mycobacterium bolletii, S. aureus, Clostridium botulinum, Acinetobacter baumannii, E. coli, Enterococcus faecalis, Enterococcus faecium, Clostridium perfringens and Serratia marcescens, were isolated from this stool sample (Table 1 and Supplementary Table 2). C. botulinum was isolated twice using culture vials supplemented with rumen fluid and thioglycolate, and this species identification was performed by sequencing. M. bolletii was isolated using the MOD4 medium23 and identified by sequencing the rpoB gene24.
Bacterial diversity of gorilla gut via pyrosequencing
A total of 75,595 reads were obtained from pyrosequencing the fecal sample analyzed via culturomics. Of these, 36,463 reads with high-quality sequencing were analyzed and distributed into 11 phyla using the RDP classifier. Firmicutes composed the largest fraction (46.8% of the total reads), followed by Actinobacteria and Bacteroidetes (20% and 18.6%, respectively). Spirochaetes and Verrucomicrobia comprised 7.4% of the total reads (approximately 3.7% each). The remaining phyla, including Chloroflexi, Cyanobacteria, Fibrobacteres, Fusobacteria, Proteobacteria and Tenericutes, were represented by rare sequences. Finally, 4.6% of the reads were ascribed to unclassified bacterial phyla in the RDP database. Approximately 13% of the reads were identified at the genus level, yielding 38 genera. Prevotella (Bacteroidetes), Treponema (Spirochaetes) and Bifidobacterium (Actinobacteria) were the most abundant genera obtained from the 454 pyrosequencing (accounting for 40.8%, 27.4% and 11.3%, respectively, of the total reads identified at the genus level). Using the BLASTn tool in the NCBI website and setting the sequence similarity threshold to 98.7%, only 316 reads were identified at the species level, leading to the identification of 16 bacterial species.
Comparing the bacterial phyla found in the gorilla gut using culturomics and pyrosequencing with those of previous studies
Firmicutes followed by Actinobacteria comprised the majority of the detected bacteria in the fecal sample through both methods (i.e., culturomics and pyrosequencing). This is similar to the bacteria cultured from fecal samples from humans in Africa and Europe16. The high abundance of Firmicutes was reported previously in wild mountain gorillas25, western lowland gorillas26 and also in other primates including old world monkeys27, chimpanzees28 and wild pygmy loris29. Actinobacteria was the second most predominant phylum in our sample; this result is in agreement with the study by Vlckova et al. in captive western lowland gorillas26. Although Proteobacteria was the third most isolated phylum via culturomics, it was recovered from rare reads using pyrosequencing. Additional phyla such as Bacteroidetes, Cyanobacteria, Verrucomicrobia, and Tenericutes were identified in this fecal sample exclusively through pyrosequencing. All of these microbial divisions have been previously detected in human and nonhuman primates16,26,27. The remaining phyla found via metagenomics, including Spirochaetes, Chloroflexi and Fibrobacteres, are usually absent from the human gut30 but are present in herbivore guts such as gorillas, colobuses (Colobus guereza and Piliocolobus tephrosceles) and guenons (Cercopithecus ascanius)25,27.
Molecular detection of human pathogenic bacteria and estimation of their prevalence in the gorilla gut population tested
A total of 48 fecal samples from 21 gorillas were examined via specific real-time PCR to analyze the prevalence of pathogens identified using culturomics and other fastidious pathogens commonly tested in our center (Supplementary Table 3). Our molecular tests showed that the most prevalent bacteria in the gorilla gut were (in decreasing order of estimated prevalence) K. pneumoniae (80.95%), Acinetobacter spp. (76.19%), Borrelia spp. (47.6%), S. aureus (42.86%), Bartonella spp. (38%), Pseudomonas aeruginosa (38%), S. marcescens (33.33%), Mycobacterium abscessus (28.57%), E. faecalis (28.57%), Coxiella burnetii (23.8%), E. coli (19%), A. baumannii (19%) and Tropheryma whipplei (4.76%) (Table 2). However, M. tuberculosis, Treponema pallidum, B. anthracis, Streptococcus pneumoniae, Bordetella pertussis, Neisseria spp. and Francisella tularensis were not detected via rtPCR in any fecal sample (Table 2).
In this study, we used the culturomics approach because it has greater sensitivity than molecular methods and can be used to detect even minority bacterial populations including pathogens16,17,18. Many opportunistic human pathogens were observed using culturomics, including 8 pathogens frequently associated with human disease (Table 1)19,31. Among these, C. botulinum is also present in herbivore and carnivore hosts, including pigs, horses, cattle, giant pandas and tigers32,33,34,35,36. In contrast, M. bolletii is an emerging pathogen for which no animal source is known31.
Our molecular examination of gorilla feces via real-time PCR confirmed these observations and demonstrated that wild gorillas are important reservoirs of emerging human pathogens, including fastidious bacteria such as C. burnetii, Bartonella spp., T. whipplei and Borrelia spp. (Table 2). This is the first observation of T. whipplei in a non-human primate host. Humans are the predominant reservoir and source of this bacterium37. Moreover, a high prevalence of pathogens, including K. pneumoniae, Borrelia spp., S. aureus, P. aeruginosa and S. marcescens was observed (Table 2).
The presence of these pathogens in samples from different sources collected at different periods as well as their high prevalence indicate that they are part of the commensal microbiota of the gorilla gut (Table 2)12,15. It is important to consider whether these opportunistic human pathogens are virulent under certain conditions in gorillas. The pathogenicity of certain bacteria in gorillas and chimpanzees has been observed; these bacteria include B. anthracis-like bacteria, S. aureus, M. leprae and Pasteurella multocida11,12,13,38.
Furthermore, new culture media were developed based on the herbivore behavior of gorillas; these media were generated using 4 different plants. We observed that applying adapted culture media enhances the growth of bacteria that are difficult to cultivate. For example, rumen extracts mimicking the gut habitat were used to isolate a high proportion of human gut microbiota16. Here, approximately one-fifth of the recovered bacterial species were isolated specifically in the novel media supplemented with plant extracts, and more than one-third of all observed species, including 2 new species, grew in these media (Supplementary Table 2). It will be important to test other gorilla fecal samples and to develop further novel media using other plants eaten by gorillas, for instance, to mimic the ecological conditions that can promote bacterial growth39.
In conclusion, we observed that many human bacterial species, including pathogens, are also present in the gorilla gut. This type of non-invasive analysis of gorilla feces could facilitate the surveillance and discovery of potential human pathogens and unknown bacteria, especially in areas where human and gorilla habitats overlap and because of the increasing presence of humans in the African equatorial forests. Furthermore, similar studies on non-human primates would contribute to the detection of emerging diseases circulating in wild fauna.
Gorillas' fecal samples from two sites in south-central Cameroon (near the Minton village and near the Messok village) were collected as previously described40 (about 1 g of dung was collected in a 50-ml tube containing RNAlater™ (Ambion, Austin, TX)) and preserved at ambient temperature for a maximum of 3 weeks in the field and subsequently stored at −80°C, except the sample used in culturomics, for which about 25 g of dung was collected and stored directly at −80°C without the RNAlater reagent. The time between defecation and collection was estimated at <24 h, according to the physical texture of the samples. No experimentation was conducted on these animals. The collection of fecal samples from the soil was approved by the Ministry of Scientific Research and Innovation of Cameroon. No other permit was required, as this research was non-invasive work, and the collection of the samples did not disrupt wild fauna.
The DNA extracted from 48 fecal samples was used for microsatellite analysis as previously described40. Briefly, seven loci including D18S536, D4S243, D10S676, D9S922, D2S1326, D2S1333 and D4S1627 were amplified (Supplementary Table 1). A region of amelogenin was also amplified for determining the gender40. The loci were amplified four times to exclude allelic dropout. One microliter of the amplification product was mixed with 10 μl of formamide and 0.25 μl of size marker (ROX GeneScan 400HD, Applied Biosystems, Foster city, CA). This mixture was analyzed via sequencing using the 3130xl Genetic Analyzer (Applied Biosystems). The sequencing products were visualized using the GeneMapper 3.7 software (Applied Biosystems). The number of wild gorillas was then identified by comparing the loci in the 48 samples analyzed (Supplementary Table 1).
Culturomics and bacterial identification
Two strategies were used in this study. First, 70 conditions were selected, which were effective for detecting more than 370 bacterial species, including minority pathogens16 (Supplementary Table 2). Second, new media were developed including solid media supplemented with plants. Tobacco leaves and the fruits of mango, banana and papaya were crushed and lyophilized, and solutions containing 12 mg of plant extract per ml of sterile water were prepared and filtered using 0.2-μm filters. Additionally, a solution of 14 mg of agar per ml of sterile water was prepared. Using these solutions, the following media were prepared: filtered mango solution/agar (20 ml/80 ml); filtered papaya solution/agar (20 ml/80 ml); filtered banana solution/agar (10 ml/900 ml); filtered tobacco solution/agar (10 ml/90 ml); 12 g of lyophilized tobacco + 5% sheep blood (BioMerieux, Craponne, France)/agar (14 g/l); and 12 g of lyophilized banana + 5% sheep blood (BioMerieux, Craponne, France)/agar (14 g/l).
Additionally, culture vials (BD BACTEC™ Plus Aerobic/F Medium and BD BACTEC™ Lytic/10 Anaerobic/F Medium) supplemented with 5 ml of filtered banana solution were used. Finally, we also used MOD4 agar medium (specific for the culture of Mycobacterium spp.)23 and 12 g of soil from the Timone hospital garden (Latitude: N 43° 17′ 20.151″; Longitude: E 5° 24′ 15.3822″)/agar (14 g/l).
For the fecal sample collected near Minton, the stool was divided into 1-g aliquots. Each gram was suspended in 9 ml of Dulbecco's Phosphate-Buffered Saline 1× (DPBS) and diluted from 1/10 to 1/1010; appropriate dilutions of the stool samples were plated onto different culture media at various conditions (Supplementary Table 2). The colonies isolated in different media were identified using Matrix-Assisted Laser Desorption/Ionization Time-Of-Flight Mass Spectrometry (MALDI-TOF MS) and, when necessary, using 16S rRNA gene sequencing as previously reported16.
New bacterial species
The new species Gorillibacterium massiliense, Paenibacillus gorillae, Bacillus massiliogorillae, Microbacterium gorillae and Paenibacillus camerounensis were deposited in the German collection of microorganisms (DSMZ) under the accession numbers DSM 27179, DSM 26181, DSM 26159, DSM 26203 and DSM 26182, respectively, and were submitted in the GenBank database under the accession numbers KC193239, JX650054, JX650055, JX650056 and JX650057, respectively. The description and genome sequencing of these new species were performed for three of them20,21,22 and are ongoing for those remaining.
454 FLX Titanium pyrosequencing analyses
DNA was extracted from the feces as previously described (Macherey-Nagel, Hoerdet, France). 16S rRNA amplicon pyrosequencing and pyrosequencing analyses were performed as previously reported16. A sequence file was deposited to the NCBI Sequence Read Archive under the run accession number SRR1177866 and BioProject accession number PRJNA239545.
Real-time PCR assay for the detection of human pathogens
The primers and probes used in this study were previously described or designed in our laboratory for routine diagnosis (Supplementary Table 3). The DNA was extracted from 48 gorilla fecal samples using a kit (Macherey-Nagel, Hoerdet, France) as previously described16. In each PCR, 5 μl of DNA of each fecal sample was used in a final volume of 25 μl using the QuantiTect Probe PCR Kit (Qiagen, Courtaboeuf, France). The real-time PCR was performed using a CFX96™ Real-Time PCR Detection System (Bio-Rad, Life Science, Marnes-la-Coquette, France) with the following program: 95°C for 15 min followed by 44 cycles of 95°C for 0.5 min and 60°C for 1 min. The positive samples were sequenced using the same primers.
Fadi BITTAR was supported by a Chair of Excellence IRD provided by the Institut de Recherche pour le Développement/IHU Méditerranée Infection. We thank Amandine Esteban for microsatellite analyses. This work was supported in part by grants from the National Institute of Health (RO1 AI 50529) and the Agence Nationale de Recherches sur le SIDA (ANRS 12255).