Multidrug-Resistant Klebsiella pneumoniae Clones from Wild Chimpanzees and Termites in Senegal

ABSTRACT

Antibiotic resistance genes exist naturally in various environments far from human usage. Here, we investigated multidrug-resistant Klebsiella pneumoniae, a common pathogen of chimpanzees and humans. We screened antibiotic-resistant K. pneumoniae from 48 chimpanzee stools and 38 termite mounds (n = 415 samples) collected in protected areas in Senegal. The microsatellite method was used to identify chimpanzee individuals (n = 13). Whole-genome sequencing was performed on K. pneumoniae complex isolates to identify antibiotic-resistant genes and characterize clones. We found a high prevalence of carbapenem-resistant K. pneumoniae among chimpanzee isolates (18/48 samples from 7/13 individuals) and ceftriaxone resistance among both chimpanzee individuals (19/48) and termite mounds (7/415 termites and 3/38 termite mounds). The bla_OXA-48 and the bla_KPC-2 genes were carried by international pOXA-48 and pKPC-2 plasmids, respectively. The ESBL plasmid carried bla_CTX-M-15, bla_TEM-1B, and bla_OXA-1 genes. Genome sequencing of 56 isolates identified two major clones associated with hospital-acquired infections of K. pneumoniae (ST307 and ST147) in chimpanzees and termites, suggesting circulation of strains between the two species, as chimpanzees feed on termites. The source and selection pressure of these clones in this environment need to be explored.

INTRODUCTION

The emergence of antibiotic resistance is often attributed to excessive use of manufactured antibiotics prescribed to humans or livestock (1). Interestingly, it has recently been shown that beta-lactam resistance was not directly connected to industrial beta-lactam production (2) but that, in contrast, countries where resistance was most frequent were those that consumed the least antibiotics (2). This paradox has not yet been explained. In addition, recent work in Nairobi, Kenya, has shown the existence of multidrug-resistant bacteria in wild animals (3). The authors suggested that they had access to human waste from patients treated with beta-lactams. However, these latter hypotheses fall short while Kenya is not a country where the use of beta-lactams is higher than that of European countries (2). We recently commented on these aspects (4, 5). More recently, antibiotic resistance determinants were found to be widely distributed in people, animals, and the environment in three ethnic groups living in Tanzania (6). There is an inconsistency between the major hypotheses of resistance emergence associated with the massive prescription of industrial beta-lactam antibiotics and the data. Beta-lactams are natural products secreted by bacteria and fungi (7). They existed in the natural environment long before the industrialization of beta-lactam synthesis, and thus, bacterial resistance mediated by various mechanisms, including beta-lactamases, existed long before human industrialization. We collected stools from chimpanzees living in a protected reserve that had no direct contact with humans to look for bacteria known to be present in humans and in termites that are included in their diet. We focused our study on Klebsiella pneumoniae, which is known to be both multidrug resistant in humans and a chimpanzee pathogen (8), to assess its degree of resistance compared to that of K. pneumoniae isolates from termites that are part of the environment.

RESULTS

Overview of the Enterobacteriaceae isolated from chimpanzee and termite samples.

We identified 13 different chimpanzee individuals from the 48 stools (Fig. S2) collected in Senegal in 2016. Among the 48 samples analyzed, the growth of at least one bacterial colony was observed for 25 samples inoculated on MacConkey plus ertapenem, for 31 samples on MacConkey plus cefotaxime, for 29 samples on MacConkey plus ciprofloxacin, and for 26 samples on Lucie Bardet-Jean-Marc Rolain medium (LBJMR containing colistin and vancomycin) (Fig. 1; Table S2). Most colonies were intrinsically resistant species (Table S2). Overall, 14 samples were positive for at least one culture of K. pneumoniae, representing 7 individuals (Fig. 1; Table S3). We identified 25 isolates of K. pneumoniae using matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF), among which 5 isolates were on MacConkey plus ertapenem medium, 9 on MacConkey plus cefotaxime agar, and 11 on MacConkey plus ciprofloxacin medium. No K. pneumoniae isolates grew on LBJMR. Antimicrobial susceptibility testing was performed on the 25 isolates (Table 1). Fifteen Klebsiella spp. isolated with a specific antimicrobial profile were then sequenced to identify their resistome (Table S3).

FIG 1.

Summary of samples, isolates, and genomes of Klebsiella spp. analyzed in this study.

TABLE 1.

Resistome analysis of the 23 strains of K. pneumoniae isolated from chimpanzee and termite mound samples

Specimen	Sample	No.	ST	Penicillinase	ESBL gene	Carbapenemase	Aminoglycoside(s)	FQa	Phenicol	Sulphonamide(s)	Tetracycline
C1	CHZ 1	Q1449	37	bla SHV-11			aph(6)-Id/aph(3′′)-Ib	qnrS1		sul2	tet(A)
	CHZ 10	Q1445	307	blaOXA-1/blaSHV-28	bla CTX-M-15	bla OXA-48	aac(3)-IIa (GN)/aac(6′)Ib-cr (AK)	aac(6′)Ib-cr/qnrB1	catB4	dfrA14	tet(A)
	CHZ 17	Q0897	307	blaOXA-1/blaSHV-28	bla CTX-M-15	bla OXA-48	aac(3)-IIa (GN)/aac(6′)Ib-cr (AK)	aac(6′)Ib-cr/qnrB1	catB4	dfrA14	tet(A)
	CHZ 34	Q1448	147	blaTEM-1B/blaSHV-11		bla KPC-2		qnrS1		sul2
	CHZ 35	Q0898	147	blaTEM-1B/blaSHV-11		bla KPC-2		qnrS1		sul2
	CHZ 36	Q1443	147	blaSHV-1-2a/blaTEM-1		bla KPC-2		qnrS1		sul2
C4	CHZ 7	Q1447	307	blaOXA-1/blaSHV-28	bla CTX-M-15	bla OXA-48	aac(3)-IIa (GN)/aac(6′)Ib-cr (AK)	aac(6′)Ib-cr/qnrB1	catB4	dfrA14	tet(A)
C5	CHZ 6	Q0893	147	blaTEM-1B/blaSHV-11		bla KPC-2		qnrS1		sul2
C6	CHZ 3	Q1461	307	blaOXA-1/blaSHV-28	bla CTX-M-15	bla OXA-48	aac(3)-IIa (GN)/aac(6′)Ib-cr (AK)	aac(6′)Ib-cr/qnrB1	catB4	dfrA14	tet(A)
C7	CHZ 2	Q1459	307	blaOXA-1/blaSHV-28	bla CTX-M-15	bla OXA-48	aac(3)-IIa (GN)/aac(6′)Ib-cr (AK)	aac(6′)Ib-cr/qnrB1	catB4	dfrA14	tet(A)
	CHZ 30	Q1460	307	blaOXA-1/blaSHV-28	bla CTX-M-15	bla OXA-48	aac(3)-IIa (GN)/aac(6′)Ib-cr (AK)	aac(6′)Ib-cr/qnrB1	catB4	dfrA14	tet(A)
C8	CHZ 16	Q0896	147	blaTEM-1B/blaSHV-11		bla KPC-2		qnrS1		sul2
	CHZ 16	Q1444	147	blaTEM-1B/blaSHV-11		bla KPC-2		qnrS1		sul2
C9	CHZ 5	Q0892	307	blaOXA-1/blaSHV-28		bla OXA-48	aac(3)-IIa (GN)/aac(6′)Ib-cr (AK)	aac(6′)Ib-cr/qnrB1	catB4	dfrA14	tet(A)
T20	NCS 35A	Q2547	NA	bla SHV-168
T28	NCS 43A-C	Q1947	307	blaOXA-1/blaSHV-106/blaTEM-1B	bla CTX-M-15		aac(3)-IIa (GN)/aac(6′)Ib-cr (AK)	aac(6′)Ib-cr/qnrB1		sul2/dfrA14	tet(A)
	NCS 43B	Q1948	307	blaOXA-1/blaSHV-106/blaTEM-1B	bla CTX-M-15		aac(3)-IIa (GN)/aac(6′)Ib-cr (AK)	aac(6′)Ib-cr/qnrB1		sul2/dfrA14	tet(A)
	NCS 43A-M	Q2416	307	blaTEM-1B/blaSHV-106/blaOXA-1	bla CTX-M-15		aac(3)-IIa (GN)/aac(6′)Ib-cr (AK)	aac(6′)Ib-cr/qnrB1		dfrA14/sul2	tet(A)
T30	NCS 45A	Q1945	307	blaOXA-1/blaSHV-106/blaTEM-1B	bla CTX-M-15		aac(3)-IIa (GN)/aac(6′)Ib-cr (AK)	aac(6′)Ib-cr/qnrB1		sul2/dfrA14	tet(A)
	NCS 45A-M	Q2411	307	blaTEM-1B/blaSHV-106/blaOXA-1	bla CTX-M-15		aac(3)-IIa (GN)/aac(6′)Ib-cr (AK)	aac(6′)Ib-cr/qnrB1		dfrA14/sul2	tet(A)
	NCS 45B	Q2412	307	blaTEM-1B/blaSHV-106/blaOXA-1	bla CTX-M-15		aac(3)-IIa (GN)/aac(6′)Ib-cr (AK)	aac(6′)Ib-cr/qnrB1		dfrA14/sul2	tet(A)
	NCS 45 (1)	Q3079	1418	bla SHV-145
	NCS 50D	Q2551	NA	bla SHV-110

^aFQ, fluoroquinolone.

Termite species were identified morphologically, as well as by molecular amplification of the mitochondrial cox1 gene (Table S4; Fig. S3). Of the 415 samples analyzed from the 38 termite mounds, we isolated 278 strains on MacConkey plus ertapenem and 229 on MacConkey plus cefotaxime medium (Fig. 1). Among these isolates, 10 were enterobacteria, including 4 Enterobacter cloacae isolates, 3 Citrobacter freundii isolates, and 3 K. pneumoniae isolates, that were isolated on MacConkey agar plus cefotaxime. No K. pneumoniae grew on MacConkey agar plus ertapenem. The three positive samples carrying K. pneumoniae isolates came from two termite mounds, of which two were environmental, i.e., fungus (NCS43B) and fungus comb (NCS43A) from one termitary of Macrotermes sp. (Fig. S3), and one was a worker termite belonging to the Ancistrotermes cavithorax species (NCS45A). In parallel, we inoculated these 415 samples on MacConkey alone to study the clonal diversity of K. pneumoniae in termite mounds. Of these, we isolated 553 Gram-negative bacteria (GNB), including 495 enterobacteria (Table S4). Among these, 38 Klebsiella spp. were isolated from 14 termite mounds that were sequenced to study their clonal diversity. Overall, 66 Klebsiella spp. were isolated from the various samples (25 from chimpanzees and 41 from termite mounds), and 56 were sequenced (Fig. 1; Table S3). Assembly statistics are available in Table S3. Genome sizes ranged from 4,934,591 bp to 5,623,147 bp (mean 5,329,288 bp), with a GC% ranging from 54.1 to 58.5% (mean 56.6%). Genomes were assembled in 2 to 86 contigs (mean 36), with a mean coverage ranging from 6 to 79 (mean 21.4) (Table S3).

Circulation of K. pneumoniae clones in diverse ecological niches.

Of the 56 Klebsiella pneumoniae isolates sequenced, 19 isolates were identified as Klebsiella quasipneumoniae subspecies quasipneumoniae (1 from a chimpanzee sample and 18 from termite samples), 5 as Klebsiella quasipneumoniae subspecies similipneumoniae, and 8 as Klebsiella quasivariicola. One Klebsiella aerogenes isolated in termite mound samples was sequenced as an outgroup (Table S3). Twenty-three K. pneumoniae sensu stricto strains were confirmed and belonged to six different sequence types (STs) (ST307, ST147 ST37, ST1418, and two unknown STs). Interestingly, 13 isolates belonged to the same ST, ST307, and were closely related (Table 1; Fig. 2), with single nucleotide polymorphism (SNP) of the core genome varying between 0 and 189 SNP differences. The smallest variations were observed between the chimpanzee isolates (0 to 25 SNPs) and between the termite mound isolates (10 to 26 SNPs). One ape isolate (Q1459) had only 0 to 41 SNP differences with the 12 other isolates. Seven of them were isolated from chimpanzee samples, and six were isolated from termite samples. These six ST307 clones were isolated from two samples belonging to the same termite mounds (NCS43A-C, NCS43B-C, and NCS45A-C were isolated on MacConkey plus cefotaxime, whereas NCS43A-M, NCS45A-M, and NCS45B-M were isolated on MacConkey). The ST307 clones from chimpanzees carried resistance to all tested beta-lactams, including carbapenems, fluoroquinolones, aminoglycosides (gentamicin), and doxycycline (tetA gene). Antibiotic susceptibility tests of the six K. pneumoniae isolates from termite mounds showed an extended-spectrum beta-lactamase (ESBL) profile with resistance to amoxicillin, amoxicillin plus clavulanic acid, cefotaxime, and cefepime but susceptibility to carbapenems (Table 1). These isolates were also resistant to gentamicin, ciprofloxacin, co-trimoxazole, and doxycycline. The other STs were ST147 for 6 isolates from chimpanzees, ST37 from one chimpanzee, ST1418 from Ancistrotermes cavithorax, and 2 unknown STs from termite mounds. The ST147 clones were resistant to beta-lactams, including carbapenems and fluoroquinolones, and had between 7 and 44 SNP differences among their core genomes.

FIG 2.

Maximum likelihood phylogenetic tree of the K. pneumoniae complex strains isolated in this study and isolates of ST147 and ST307 isolated in South and West Africa available on NCBI.

Evidence of lateral gene transfer.

We analyzed plasmids of one ST307 (Q1445) and one ST147 (Q1448) strain isolated from a chimpanzee stool sample and of one ST307 strain isolated from a termite mound sample (Q1948). Plasmids were reconstituted using short- and long-read technologies (Illumina and Oxford Nanopore) and hybrid assembly using SPAdes. In the ST147 chimpanzee isolate (Q1448), we identified an IncFIB plasmid of 100,727 bp that carried a bla_KPC-2 gene inserted between an ISKpn7 and an ISKpn6. This plasmid shared 100% coverage and 99.98% identity with a pKPC-2 plasmid (MN657251) found in clinical isolates of K. pneumoniae ST37 and ST307 in Germany (9) (Fig. 3). The ST147 isolate also carried a 30,586-bp plasmid that had resistance genes against penicillin (bla_TEM-1B), quinolones (qnrS1), sulfonamides (sul2), and chloramphenicol (floR). In the ST307 isolate from chimpanzee (Q1445) we identified one IncL/M plasmid of 63,848 bp carrying bla_OXA-48 (Fig. 3) that presented 99% coverage and 100% identity with the pOXA-48 plasmid of 63,587 bp previously described (CP018315) (10). This bla_OXA-48 gene was located next to the transcriptional regulator DmlR and flanked by an insertion sequence (IS) transposase. Additionally, we identified an IncFIB/FII plasmid of 172,847 bp that carried a bla_CTX-M-15 gene and two bla_OXA-1 genes. This plasmid also carried genes conferring resistance to aminoglycosides [aac(3)-IIa and two aac(6’)Ib-cr], quinolone (qnrB1), tetracycline (tetA), and heavy metals (arsenic, copper, and silver) (Fig. 3). Interestingly, in the ST307 termite mound strain (Q1948), we found one single IncFIB/FII plasmid of 238,147 bp that shared 68% coverage and 100% identity with an IncFIB/FII plasmid of 172,487 bp from the Q1445 isolate. These two plasmids Q1948 and Q1445 shared 92% coverage and 88% coverage, respectively, and 99.9% identity with an ESBL plasmid of 212 kbp (pKPC3-307-TypeC KY271406) that has been previously found in ST307 K. pneumoniae clones (11, 12). Compared to the pKPC3-307-TypeC plasmid, our plasmids both acquired the bla_CTX-M-15 genes, and the Q1948 plasmid also acquired a bla_TEM-1B gene (Fig. 3). Regarding the other genomes of Klebsiella spp., they carried only the intrinsic antibiotic resistance genes, i.e., the bla_OKP-B, oqxAB, and fosA genes, for the 21 K. quasipneumoniae species (including the subspecies similipneumoniae) and bla_LEN-26, oqxAB, and fosA for the 6 K. quasivariicola species. Two K. quasipneumoniae species carried plasmids. One IncFIB and one IncFII/R plasmid were found in the Q2415 isolate that did not carry antibiotic resistance, and one ColRNAI/Col440II plasmid was found in the Q3055 isolate that also did not carry any resistance genes.

FIG 3.

Comparison of plasmids carrying KPC (left), OXA-48 (right), and ESBL (middle) with closest published plasmids.

Introduction of internationally known plasmids in ST307 and ST147 strains in West Africa.

We compared the genomes of the strains isolated in this work with 14 genomes of strains belonging to the same ST as that previously isolated in West and South Africa available on NCBI (Table S1; Fig. 2). Phylogenetic analysis showed that strains belonging to the ST307 clustered together, as did strains belonging to ST147. The ST307 K. pneumoniae strains previously isolated in Africa already carried the IncFIB/FII plasmids; however, none of them carried the IncL/M plasmid. One strain isolated in 2018 from South Africa was secreting an OXA-181 carbapenemase (OXA-48 like) (13). A strain of ST307 isolated in Guinea from rat samples carried a plasmid close to pKPC3-307-TypeC (KY271406) (12); moreover, the isolate was producing a CTX-M-15 and a TEM-1B, suggesting that the plasmid was close to that of the strains isolated from termites. However, the reconstitution of the plasmid had been performed by mapping on the pKPC3-307-TypeC plasmid and thus not finding the CTX and TEM enzymes. In contrast, none of the previously deposited K. pneumoniae ST147 isolates carried the KPC-producing plasmid in this part of the world.

DISCUSSION

Most previous studies on antibiotic resistance in wildlife have been limited to Escherichia coli. Few studies have been performed on K. pneumoniae; however, it is known that this species is present in all ecosystems, including the environment and gut of animals (14). Its genomic plasticity gives it a high capacity for horizontal gene transfer, including the acquisition of antibiotic resistance genes, with a wide number of species of environmental origin (14). In this work, analysis of Klebsiella species genomes showed a great diversity of Klebsiella species, including K. pneumoniae present in chimpanzee and termite mounds and K. quasipneumoniae in termite mounds. This distribution was detected with the use of whole-genome sequencing (WGS) conversely to MALDI-TOF mass spectrometry that failed to identify correctly species belonging to the K. pneumoniae complex.

In our study, only K. pneumoniae isolates harbored acquired antibiotic resistance genes. Strains of K. quasipneumoniae and K. quasivariicola, although present in the same ecosystem, carried only natural resistance genes. These observations emphasize the unique ability of K. pneumoniae to accumulate resistance genes against multiple antibiotics. Thus, we identified three clones of K. pneumoniae, including two clones producing carbapenemase. These clones have frequently been reported in medical diseases. The ST307 and ST147 clones are international multidrug-resistant clones (15, 16), carrying mostly ESBL genes (17) and carbapenemases (18–20). They are associated with hospital-acquired infections and have never been reported in the gut of wild animals to our knowledge (16, 18). There are limited reports in Senegal, but previous studies in West and South Africa showed that these clones are circulating (12, 21), which is also shown by our phylogenetic analysis. Moreover, the carbapenemase genes bla_OXA-48 and bla_KPC-2 were carried by international plasmids (pOXA-48 and pKPC-2), suggesting future spread from this new source of resistance. SNP variations among isolates showed both clonal transmission and evolution of these clones, such as for ST307, which seems to be sustainably present in this Senegalese nature reserve. Moreover, the high level of similarity and synteny between plasmids from chimpanzees and termites suggests lateral gene transfers between these ecosystems. These results raise questions about the role of the wildlife reservoir of these clones responsible for human infections. Moreover, two K. quasipneumoniae isolates from soil substrate (NCS36) and termites (NCS38) were susceptible to antibiotics but carried plasmids that did not rule out the possibility of future antibiotic or heavy metal resistance gene transfer in this species.

Thus, we demonstrated here that a pathogen common to both chimpanzees and humans, K. pneumoniae, some clones of which are found in both species, may carry antibiotic resistance genes that are currently considered emerging and selected by antibiotic use in humans. These chimpanzees have no direct contact with manufactured products or humans (Fig. 1), so their resistance may therefore be selected by a different ecosystem from that in humans. Moreover, we found the same ST307 clone carrying ESBL resistance genes in termite mounds; chimpanzees are known to consume several species of termites as a source of protein, and thus these insects might also be the source of Klebsiella spp. (22). However, it is of note that the chimpanzees in our study have some indirect contact with areas of human activity which might also be a source of contamination: (i) livestock sometimes enter the reserve, share water sources, and leave feces, (ii) green monkeys (Chlorocebus sabaeus), which eat human crops and have closer contact with humans, are eaten by chimpanzees, and (iii) human waste is sometimes found in the reserve (23, 24).

The selection of these multidrug-resistant clones in a theoretically antibiotic-free ecosystem raises the question of the genesis and selection of these strains in the apes of our study. The antibiotics to which they are resistant, including thienamycin and other carbapenems, are natural antibiotics that are secreted by many bacteria, including bacteria commonly found in plants or soil, such as Erwinia carotovora, Streptomyces cattleya, and Photorhabdus luminescens (25). It is possible that there is a source in the diet of chimpanzees in this area, including thienamycin in its natural state, promoting bacteria that carry this resistance mechanism. It is also possible that this resistance is selected by another unknown factor. In any case, this result is consistent with the findings that there is not necessarily a direct link between antibiotic consumption and the prevalence of resistance in countries in the humid intertropical zone, probably due to the natural secretion of antibiotics to which resistance is developing. There is an urgent need to identify the natural source of antibiotics to better understand the emergence and spread of antibiotic resistance to specific sites.

MATERIALS AND METHODS

Chimpanzee sample collection.

In August 2016, chimpanzee stools were collected in the Dindefelo Community Nature Reserve (hereafter Dindefelo) at 12°22′01.4′′N 12°18′00.0′′W, located in southeastern Senegal, about 35 km from the town of Kedougou on the border with Republic of Guinea (Fig. S1). This area is located on the last foothills of the Fouta Djallon massif. The habitat is a Sudano-Guinean savanna woodland mosaic. The climate is hot and highly seasonal, with the dry season lasting 7 months. Annual mean temperature is 28.5°C. Dindefelo chimpanzees belong to the critically endangered West African chimpanzee subspecies Pan troglodytes verus (Schwarz, 1934) (Primates: Hominidae) (23, 26). Feces were collected underneath night nests and other places within the home range of the chimpanzees (Fig. S1). A total of 48 samples were collected from three sites. The first two sites rendered 10 old fecal samples (CHZ01 to CHZ10), but the third site provided 38 fresh fecal samples (CHZ11 to CHZ48).

The permit to collect samples was given by the Environment and Durable Development Ministry of Senegal for chimpanzee stool samples (001914/DEF/DGF de la Direction des Eaux, Forêts, Chasses et de la Conservation des Sols du 05/06/2016). No other permit was required, as this research was noninvasive and the collection of samples did not disrupt the wild fauna.

The identification of specimens was based on microsatellite studies directly on stools. A QIAamp DNA stool minikit (Qiagen, Venlo, Nederland) was used to extract DNA from the chimpanzee stool samples according to the manufacturer’s instructions. Then, PCR amplification of the patr-B gene (∼400 bp) was performed as previously described (27). Fragments of this size were successfully amplified in the microsatellite typing system previously applied to chimpanzee stool DNA. Then, PCR amplification and sequencing targeting exons 2 and 3, which encode the most polymorphic and functionally relevant sites of the Patr-B molecule, were performed as previously described. Exon 2 and 3 sequences were concatenated and aligned by CLUSTALW software. Phylogenetic inferences were performed with the maximum likelihood and neighbor-joining methods under the Kimura 2-parameter model and complete deletion with MEGA software.

Termite sample collection.

The study was carried out in August 2019 at four areas in Senegal: three in central Senegal in Niokolo-Koba National Park (Simenti, Dar Salam, and Niokolo Poste sites) and one in Dindefelo (Fig. S1) in the vicinity of the sites where the chimpanzee samples were collected. Termite mounds were identified visually, and then the samples were collected using a shovel and pincer. From each termitary, the soil substrate, the fungus comb (where available), and adult termites (both soldiers and workers) were collected in a plastic ventilated box where they were stored during transport at ambient temperature. On arrival in the laboratory in France, the live termites were separated from the substrate and fungus comb, washed, disinfected as previously described (28), and then stored at –80°C. In total, 38 termite mounds were examined that were divided into 415 samples prepared for the bacterial culture (Table S1). A total of 311 samples contained termites (soldiers and/or workers), soil substrates, and/or fungus combs. A total of 93 samples contained only soil substrate and were also screened (48 samples inoculated two times), and 8 samples contained only fungus combs (2 samples from 4 termite mounds) (Fig. S1; Table S1).

Morphological identification of termites was confirmed by molecular characterization. A portion of the mitochondrial cox1 gene was amplified using the newly designed set of primers, Termites1490 5′-TCAACIAAYCAYAARGAYATTGG-3′ and Termites2198 5′-TAIACTTCAGGGTGICCRAARAAYCA-3′. These primers target the same portion as that of widely used Folmer’s primers (29). Each termite was individually cut into pieces and resuspended in 200 μl of G2 buffer. DNA extraction and mix preparation were performed as previously described (28). Cycler conditions for PCR analysis included an initial denaturation step at 95°C for 15 min, followed by 40 cycles of 1 min at 95°C, 30 s annealing at 50°C, 1 min at 72°C, and a final extension step for 5 min at 72°C. PCR products were sequenced as previously described (28). Molecular phylogenetic and evolutionary analyses were conducted in TOPALi 2.5 (http://www.topali.org/).

Microbiological procedures.

All samples were inoculated in Trypticase soy broth (TSB, bioMérieux, Marcy l’Etoile, France) and incubated for 24 h at 37°C. Then, 20 μl of TSB was plated on five specific media, MacConkey (bioMérieux), MacConkey plus 0.5 mg/liter ertapenem, MacConkey plus 1 mg/liter cefotaxime, MacConkey plus 0.5 mg/liter ciprofloxacin, and LBJMR (30) containing 4 mg/liter colistin and 50 mg/liter vancomycin, and incubated for 24 h at 37°C with slight differences between the types of samples. Chimpanzee stool samples were inoculated on every medium except MacConkey alone, whereas termite samples were inoculated on all media except ciprofloxacin and LBJMR. Except for LBJMR, antibiotic concentrations were chosen according to the critical concentration found in the European Committee on Antimicrobial Susceptibility Testing (EUCAST). Each colony growing on the media was then streaked out on Columbia agar plus 5% sheep blood (bioMérieux) for further analyses. Identification of strains was performed by matrix-assisted laser desorption ionization-time of flight mass spectrometry (MALDI-TOF) as previously described (31). Antimicrobial susceptibility testing was performed on K. pneumoniae isolates according to EUCAST recommendation (version 7.0). ESBL profile was detected by the observation of a champagne cork between a 3rd or 4th generation cephalosporin and clavulanic acid. When necessary, the MIC was determined using the Etest (bioMérieux), except for the colistin MIC, which was obtained using the UMIC microdilution method (Biocentric, Bandol, France). Carbapenemase activity was detected using the β CARBA test (Bio-Rad, Hercules, California, United States).

Genomic and bioinformatic analysis of Klebsiella species isolates.

DNA of Klebsiella isolates was purified via an automated EZ1 strategy (Qiagen, Hilden, Germany). Genomes were sequenced by MiSeq technology (Illumina Inc., San Diego, CA, USA) using a paired-end strategy and assembled with SPAdes (version 3.12.0) (32). Three isolates (Q1445, Q1448, and Q1948) were also sequenced using long-read techniques (Oxford Nanopore Technology, Oxford, United Kingdom) and hybrid assembly using SPAdes. Genomes were annotated with Prokka (33) and Abricate (https://github.com/tseemann/abricate) using Resfinder (34) and ARG-ANNOT (35) databases for antibiotic resistance detection Plasmidfinder for plasmid detection (36). Species and subspecies identification was confirmed using 16S rRNA phylogenetic tree and DNA-DNA hybridization using GGDC software (http://ggdc.dsmz.de/ggdc.php#) with BLAST+ local alignment tool (37). MLST was performed using the MLST version 2.19.0 (https://github.com/tseemann/mlst) available on Galaxy platform (https://usegalaxy.eu/). Pangenome analysis was performed with Roary (version 3.13.0) with default parameters (minimum percentage identity for blastp, 95%, percentage of isolates a gene must be in to be core, 99%) (13). For this analysis, we included 14 genomes of K. pneumoniae ST307, ST147, and ST37 isolated in sub-Saharan countries that were available on NCBI (accession number are listed in Table S1). We first used the core gene alignment output file to construct the maximum likelihood phylogenetic tree using RaXML software (version 7.7.6) with default parameters (38). Phylogenetic tree was visualized using iTOL (https://itol.embl.de/tree/). SNPs were detected using Snippy version 4.5.0 (https://github.com/tseemann/snippy), with Q1445 and Q1448 as references for ST307 and ST147 strain comparisons, respectively. A core SNP alignment was then built using the same references, and an SNP distance matrix was constructed from the multiple sequence alignment. Default parameters of the software were used for these analyses. For plasmid comparison, we identified closest plasmid by BLASTN on NCBI (https://blast.ncbi.nlm.nih.gov/Blast.cgi). We thus compared and visualized our plasmid with the reference found using CGview software (http://cgview.ca/).

Data availability.

Genomes of this project have been deposited on EBI under the project number PRJEB38163. The individual line list analyzed in this study is available online.