Between 26th and 30th of August 2019, 215 non-human primates (NHPs) fecal samples (110 baboons, 95 green monkeys, and 10 chimpanzees) and 113 non-fecal environmental samples (47 termites, 42 soil samples from termite mounds, and 24 samples including fruits, other plant parts, a centipede and a maggot) were collected in four sites in Senegal. The NHPs fecal samples were collected in three sites located in southern Senegal in the Niokolo-Koba National Park: Simenti, Dar Salam, and Niokolo Poste sites. The sample collectors observed the monkeys all day to collect their fresh stool in real time so that they could be sure that it belonged to that NHPs and not to others. The non-fecal environmental samples, with the exception of the centipede and the maggot, were NHP foods, were collected in one site in south-eastern Senegal, the Dindefelo Community Nature Reserve (12°22′01.4′′N, 12°18′00.0′′W), which is located on the border with Guinea Conakry about 35 km from the town of Kedougou. The permission to collect these samples was approved by the Direction des Parcs Nationaux and Direction des Eaux, Forêts, Chasse et de la Conservation des Sols of the Senegal Ministry for the Environment and Sustainable Development (002737/DEFCCS/DGF de la Direction des Eaux, Forêts, Chasses et de la Conservation des Sols du 27/06/2019). The monkey cohort consisted of 95 green monkey samples (Chlrocebus sabaeus) (Linnaeus, 1766) and 110 Guinea baboon samples (Papio papio) (Desmarest, 1820), while the ape cohort consists of 10 chimpanzee samples (Pan troglodytes verus)(Schwarz, 1934) (Primates: Hominidae). It should be noted that the morphological and molecular identifications of the non-fecal environmental samples have previously been reported (Baron et al., 2021).

In addition, fecal samples collected in 2015 from Niokolo Koba National Park, Senegal, were added to this study. This cohort included seven Guinea baboon samples and four green monkey samples. Permission to collect samples was granted by the National Parks Direction and Direction des Eaux, Forêts, Chasse et de la Conservation des Sols of the Senegal Ministry for the Environment and Sustainable Development (001914/DEF/DGF de la Direction des Eaux, Forêts, Chasses et de la Conservation des Sols du 05/06/2016). No other permissions were required, as this research was non-invasive and the collection of the samples did not disrupt the wild fauna. Samples were directly stored at +4°C and cultured 2 days after collection.

Non-human primates samples were aliquoted and enriched in Tryptone Soy Broth (TSB, BioMérieux, Marcy l’Etoile, France) at 37°C for 72 h. Then, 20 μl was then inoculated on antibiotic-containing media: an LBJMR (Lucie Bardet Jean-Marc Rolain) plate (containing 4 μg/ml colistin and 100 μg/ml vancomycin) (Bardet et al., 2017), MacConkey (BioMérieux) + ertapenem (0.5 μg/ml) and MacConkey + cefotaxime (1 μg/ml). The critical concentrations selected were based on the European Committee on Antimicrobial Susceptibility Testing (EUCAST) guidelines (Version 11.0). Inoculation was followed by incubation at 37°C for 24 h. Colonies that grew on the different media were replicated on Columbia agar + 5% sheep blood (BioMérieux) for further analyses. Bacterial identification was performed by Matrix Assisted Laser Desorption Ionization-Time of Flight (MALDI-TOF) mass spectrometry (Bruker Daltonik, Bremen, Germany), as described previously (Seng et al., 2009). Concerning non-fecal environmental samples, cultures were initially inoculated on MacConkey + cefotaxime and MacConkey + ertapenem and described in a previous study (Baron et al., 2021). Briefly, samples were crushed and put in TSB at 37°C for 72 h, then 20 μl were inoculated on selective media. In this study, we also cultured the non-fecal environmental samples on LBJMR media. Unique and common Enterobacterales strains of NHPs and the non-fecal environmental samples were visualized by Cytoscape program to show interaction between the two ecosystems (Shannon et al., 2003).

Antibiotic Susceptibility Test (AST) were performed using the disk diffusion method following EUCAST recommendations. In case of resistance observed in disk diffusion method, minimum inhibitory concentrations (MIC) of imipenem and ertapenem were determined using the E-test method (BioMérieux), while colistin MIC was determined using the UMIC microdilution method (Biocentric, Bandol, France). The ESBL profile was detected by the observation of a champagne-cork or a keyhole between a third or fourth generation cephalosporin and clavulanic acid (Drieux et al., 2008). The ß-CARBA test (Biorad, Hercules, CA, United States) was performed to identify the strain with carbapenemase activity.

The DNA of enterobacterial isolates was extracted with the BioRobot EZ1 (Qiagen, Courtaboeuf, France) using a commercial EZ1 DNA extraction kit (Qiagen) and quantified by a Qubit assay (Life Technologies, Carlsbad, CA, United States). De novo sequencing was performed using MiSeq technology (Illumina Inc., San Diego, CA, United States) in a paired-end strategy. Libraries were prepared using the Nextera XR DNA sample prep kit (Illumina). Briefly, DNAs were fragmented and tagged with adapters and dual-index barcodes. Then, DNAs were purified using the AMPure XP beads (Beckman Coulter, Inc., Fullerton, CA, United States) and pooled in equimolar concentrations. An Illumina-generated PhiX control libraries was added to the libraries that was then 2 × 250 bp paired-end sequenced on an Illumina Miseq. Genomes have been deposited under the Bioproject number PRJNA738374 on the NCBI Database. The three K. pneumoniae strains (Q1947, Q1948, and Q1945) described previously (Baron et al., 2021) were deposited under genome accession numbers GCA_903166445.1, GCA_903166485.1, and GCA_903166415.1, respectively. They were then assembled using Spades (Galaxy Version 3.12.0 + galaxy1) (Bankevich et al., 2012), and annotated with Prokka (Galaxy Version 1.14.6 + galaxy1) (Seemann, 2014) using the Galaxy platform.¹ The detection of AMR genes, plasmids and virulence genes was performed with Abricate (Galaxy Version 1.0.1) using Resfinder, Plasmidfinder and VFDB databases, respectively. The sequence type (ST) of the isolates was determined using the pubMLST² and Pasteur Institute databases.³ Pangenome analysis was performed using the Roary software (version 3.13.0) while preserving the default settings (95%: the minimum percentage of identity and 99%: a gene needs to be in to be core). The analysis included 51 genomes of sub-Saharan African strains, isolated from human, animal and the environment, that were available in PATRIC database (version 3.6.12) (accession numbers and genome details are available in Supplementary Table 1). We then used Fast tree software (version 2.1.10) to build a maximum likelihood phylogenetic tree with predefined parameters. It was computed using iTOL.⁴ A core single nucleotide polymorphism (SNP) was performed using snippy (Galaxy Version 4.6.0 + galaxy0) for the mostly related strains. We sequenced the genomes of two strains, Q2160 (ST224) and Q2170 (ST10), using a MinION sequencer (Oxford Nanopore Technologies Inc., Oxford, United Kingdom) and assembled genomes using both Illumina and MinION reads to reconstitute the IncY plasmid and compare it to the transposon of the Q2170 isolate. Bacteriophages were detected using PHASTER (Arndt et al., 2016). The reconstitution of the genetic environment was performed using the Easyfig program (Sullivan et al., 2011).

This experiment was done using the Q2160 E. coli strain containing the IncY plasmid as a donor strain and an acid-resistant E. coli J53 (F- pro Azi r) as the receiver strain. The strains were enriched in TSB for 24 h then mixed in the same tube (1 ml of the donor strain in 9 ml of the receiver strain) for another 24 h. Transconjugants were selected on LB agar, supplemented by 120 μg/mL of sodium acid and 1 μg/mL of cefotaxime (Sennati et al., 2016).

From the 226 NHP samples, we isolated 617 bacteria including 590 Gram negative bacteria (GNB) and 27 Gram positive bacteria (GPB). The detail of the number of bacterial isolates on the different media is shown in Supplementary Table 2 and Figure 1. Among these GNB, 234 were Enterobacterales, of which 29 that are not naturally resistant to the antibiotics in consideration, were studied in this work. Briefly, two of these 29 Enterobacterales grew on MacConkey + ertapenem medium: one E. coli, and one Proteus sp. On MacConkey + cefotaxime media, we isolated 16 Enterobacterales (12 E. coli and four Morganella morganii). Finally, on LBJMR (Lucie Bardet Jean-Marc Rolain) medium, we isolated three E. coli and eight Enterobacter sp. (Three E. cloacae, three E. quasiroggenkampii, one E. asburiae, and one E. bugandensis) (Supplementary Tables 3, 4).

From the 113 non-fecal environmental samples, we isolated 206 bacteria including 201 GNB and 5 GPB. The detail of the number of bacterial isolates on the different media is shown in Supplementary Table 5 and Figure 1. Among the GNB, 84 were Enterobacterales, of which 43 non-naturally resistant were studied in this work. Briefly, ten Enterobacterales grew on MacConkey + ertapenem, including four M. morganii, three Enterobacter sp. (one E. quasiroggenkampii, one E. cloacae, and one E. hormaechei), one Citrobacter cronae, one Klebsiella aerogenes, and one Serratia marcescens. On MacConkey + cefotaxime, we isolated three K. pneumoniae, two C. cronae, and two Enterobacter sp. (one E. quasiroggenkampii and one E. hormaechei). Finally, on LBJMR, 25 Enterobacter sp. (nine E. cloacae, eight E. quasiroggenkampii, four E. bugandensis, two E. roggenkampii, and two E. sichuanensis), one K. aerogenes and one K. pneumoniae were isolated. Identical strains from the same origin are reported once. It should be noted that culture results from non-fecal environmental samples on MacConkey + ertapenem, MacConkey + cefotaxime, and MacConkey media alone have previously been described (Baron et al., 2021), but the strains of interest were sequenced for this study. Specific and common species between samples were represented in Figure 2.

A total of 72 Enterobacterales were isolated from NHPs and non-fecal environmental samples. We confirmed the phenotypic resistance of these isolates by antibiotic susceptibility tests (AST) (Supplementary Table 4). Sixteen isolates were confirmed to be resistant to cefotaxime (N = 16/339 samples, 5%), two resistant to ertapenem (N = 2/340 samples, 0.6%), and 37 resistant to colistin (N = 37/339 samples, 11%). The ß-carba-test was negative for the two ertapenem-resistant isolates. These antibiotic-resistant isolates were isolated from different NHP species and the environment. In chimpanzees, we isolated one ESBL-producing E. coli and six colistin-resistant Enterobacter sp. In green monkeys (Chlorocebus sabaeus), we identified seven E. coli and one Morganella morganii which produced ESBL, and one colistin-resistant Enterobacter cloacae. Four ESBL-producing and two colistin-resistant E. coli as well as one Enterobacter asburiae were found in Guinea baboons (Papio papio). From the non-fecal environmental samples, we isolated two ertapenem-resistant M. morganii, 27 colistin-resistant Enterobacterales including 25 Enterobacter sp., one K. pneumoniae and one K. aerogenes and three ESBL-producing K. pneumoniae.

Of these 72 Enterobacterales, 23 had a multidrug resistant profile (MDR) (Magiorakos et al., 2012) and were resistant to four or more antibiotic families, 46 to fewer than four antibiotic families, and three had no resistance. Resistance to third generation cephalosporins (3GC) was 29% (N = 21/72) whereas colistin resistance represented 65% (47/72) of the total strains. 3GC-resistant isolates, had co-resistance with tetracycline family (52%, N = 11/21) (doxycycline), fluoroquinolone (43%, N = 9) (ciprofloxacin) and trimethoprim-sulfamethoxazole (62%, N = 13). Resistance to amoxicillin was found in 75% of the E. coli strains (N = 12/16).

We sequenced the genome of the 72 Enterobacterales and studied clonal population and AMR genes. Details of the assembly results are provided in Supplementary Table 4. The population distribution of Enterobacterales was mainly represented by new sequence types (STs), indicating the existence of a population specific to the ecosystem analyzed but we found clonal complexes that are associated with human infections for E. coli and K. pneumoniae. Of the 16 E. coli, six belonged to ST10, two to ST224, and nine to other unique STs (ST212, ST202, ST196, ST469, ST2803, ST642, ST10648, ST3580, and ST6611). For K. pneumoniae, three belonged to ST307 and had previously been described (Baron et al., 2021), but one had a new ST5460. Of the 37 Enterobacter sp. sequenced, only three strains belonged to a known ST (ST1084- Q3805, ST113- Q2141, and ST565-Q2153), while the remaining 34 isolates belonged to 29 new different STs. Interestingly, one group was containing strains belonged to the chimpanzee and non-fecal environmental samples. It consisted of 11 STs (ST1545, ST1547, ST1548, ST1549, ST1552, ST1554, ST1563, ST1565, ST1566, ST1567, and ST1570) combining strains of the environment origin (N = 7) and strains from chimpanzees (N = 4). Finally, two K. aerogenes and two C. cronae from the non-fecal environmental samples belonged to new STs ST225 and ST226 for K. aerogenes and ST574 and ST575 for C. cronae (Supplementary Table 4).

We noticed that 53 of the 72 Enterobacterales had four or more resistance genes, some had the same association of genes, such as E. coli and K. pneumoniae (Supplementary Table 4 and Table 1). We found a CTX-M-15 type β-lactamases in 12 E. coli from NHPs, three ST307 K. pneumoniae isolated from the environment (Figure 3). Four of the 12 E. coli (2 ST224, 1 ST202, and ST469) harbored an IncY plasmid of 91,959 bp length (GC% = 52.06%) carrying resistance genes to beta-lactam (bla_CTX–M–15, bla_TEM1B), sulfonamide (sul2, dfrA14), fluroquinolones (qnrs), tetracycline (tetA, tetR) and aminosides [aph(3′′), aph(6)]. The plasmid IncY was non conjugative, did not have a complete conjugative apparatus, and conjugation to an E. coli J53 failed on three attempts. It was found in three green monkey and one baboon samples (Figure 3). A Tn2 transposon of 30,340 bp (GC% = 52.05%) containing the same AMR genes found on the IncY plasmid [bla_CTX–M–15, bla_TEM1B, sul2, dfrA14, qnrs1, aph(3′′), aph(6), tetA, and tetR] was observed in the chromosome of four of the six ST10 isolates (Q2156, Q2170, Q2165, and Q2157). These four strains differed from 1 to 14 SNPs in their core genome suggesting they had the same origin (Figure 3). Interestingly, this transposon was similar to that present in the plasmid IncY (Figure 4) but was inserted in a SfII-like prophage (KC736978) of 35.5 bp (GC% = 50.23%) in the chromosome of the ST10 isolate (Q2170). On the 5′–3′ locus, an insertion sequence (IS)26 inserted in an integrase (INT) of the phage, causing a deletion of 8 bp in the integrase sequence, shifting the reading frame and leading to the synthesis of 187 amino acid residues of the integrase protein against 216 amino acid residues in the reference phage. On the 3′–5′ locus, the IS26 inserted before the tail fiber assembly protein (TfaE), leading to the truncation of a hypothetical protein located between the integrase (INT) and the TfaE protein. Interestingly, the intact SFII-like prophage was found in a β-lactamases susceptible E. coli isolate (Q3820) that was only resistant to colistin (Figure 4). Finally, the four other E. coli isolates carrying CTX-M-15 enzymes belonged to different STs [two ST10 (Q2157, Q2164), one ST3580 (Q2155), and one ST212 (Q3821)], had no IncY plasmid and did not have the same transposon. Three of them carried one to five plasmids (IncF, Col, and IncB plasmids, details in Supplementary Table 4), but the location of bla_CTX–M–15 on the plasmids could not be identified in these genomes. The conjugation experiment showed negative results as no growth has been detected for a J53 E. coli receiving the cefotaxime resistant gene.

Regarding other Enterobacterales, three C. cronae genomes harbored a fluroquinolone-resistance gene (qnrB12) as well as the chromosomal beta-lactamase bla_{CMY –98_1.} In the 37 Enterobacter spp., we identified four different chromosomal ampC genes, namely bla_CMH–3–1 (N = 10), bla_ACT–6–1 (N = 10), bla_CMG–1 (N = 14), and bla_MIR–1 (N = 2). None of these strains had an inducible expression of their AmpC β-lactamase with a phenotypic resistance to 3 GC. In the eight M. morganii genomes isolated from green monkeys and non-fecal environmental samples, we found the ampC β-lactamase (bla_MOR–2, bla_DHA12), conferring resistance to 3 GC. The summary of antibiotic resistance genes in these strains is presented in Table 1. The bla_CMG gene shows 98.34% similarity with bla_ACT–62 suggesting that they could be allelic variants of the same AmpC gene.

We compared the genomes of E. coli, K. pneumonia, and Enterobacter spp. isolated in our work with 51 genomes from different origin isolated in sub-Saharan Africa (Figure 3). Enterobacter isolates clustered together, far from the human isolates which were consistent with the ST results. E. coli and K. pneumoniae isolates were more related with human isolates but their core genome differed by more than 20,000 SNPs suggesting there were not directly related. We have also included in this analysis two strains of E. coli (131i and 104) that were isolated from animals use for human consumption in Nigeria (Sharma et al., 2020) that were interesting because they carried an IncY plasmid, the antibiotic resistance genes included in the transposon [bla_CTX–M–15, bla_TEM1B, sul2, dfrA14, qnrs1, aph(3′′), aph(6), tetA and tetR] and a SfII-like prophage. However, these isolates did not cluster with isolates that contained the transposon (Q2154, Q2156, Q2160, Q2161, Q2163, Q2165, Q2157, and Q2170). Unfortunately, we were not able to find the antibiotic resistance genes location in these two isolates.