Acquired Genetic Mechanisms of a Multiresistant Bacterium Isolated from a Treatment Plant Receiving Wastewater from Antibiotic Production

Abstract

The external environment, particularly wastewater treatment plants (WWTPs), where environmental bacteria meet human commensals and pathogens in large numbers, has been highlighted as a potential breeding ground for antibiotic resistance. We have isolated the extensively drug-resistant Ochrobactrum intermedium CCUG 57381 from an Indian WWTP receiving industrial wastewater from pharmaceutical production contaminated with high levels of quinolones. Antibiotic susceptibility testing against 47 antibiotics showed that the strain was 4 to >500 times more resistant to sulfonamides, quinolones, tetracyclines, macrolides, and the aminoglycoside streptomycin than the type strain O. intermedium LMG 3301^T. Whole-genome sequencing identified mutations in the Indian strain causing amino acid substitutions in the target enzymes of quinolones. We also characterized three acquired regions containing resistance genes to sulfonamides (sul1), tetracyclines [tet(G) and tetR], and chloramphenicol/florfenicol (floR). Furthermore, the Indian strain harbored acquired mechanisms for horizontal gene transfer, including a type I mating pair-forming system (MPF_I), a MOB_P relaxase, and insertion sequence transposons. Our results highlight that WWTPs serving antibiotic manufacturing may provide nearly ideal conditions for the recruitment of resistance genes into human commensal and pathogenic bacteria.

INTRODUCTION

Antibiotics are crucial in preventing and treating infections worldwide. However, the extensive use of antibiotics has resulted in antibiotic-resistant bacteria that can cause refractory infections, prolonged illnesses with increased risks of transmission, more pervasive morbidity, and higher incidences of death (1). Bacterial resistance to antibiotics stems from multiple molecular mechanisms, including the cell wall permeability to the drugs, lack of an essential target or of compatibility for the drugs to the cellular targets, and expression of efflux pumps in the cell membrane as well as of enzymes capable of inactivating the drugs (2). Susceptible bacteria can gain resistance traits by genetic alterations, either through mutations in preexisting chromosomal genes or by acquiring mobile resistance genes through horizontal gene transfer (HGT). The majority of the clinically relevant antibiotics derive from naturally occurring small molecules. Consequently, environmental bacteria have evolved against an extensive array of antimicrobial challenges and maintain a large reservoir of resistance genes that can be mobilized into pathogenic bacteria (3–7).

Wastewater treatment plants (WWTPs) receiving human sewage constitute a setting wherein environmental, commensal, and potentially pathogenic bacteria meet in the presence of antibiotics. A fraction of administered antibiotics is not metabolized but rather is excreted from treated patients; therefore, antibiotic concentrations as high as microgram/liter levels can be detected in hospital sewage and municipal wastewater (4). The levels of antibiotic residues in municipal WWTPs are usually well below therapeutic concentrations but still might be sufficient to provide selection pressures for the development of antibiotic resistance in bacteria (8). While resistant, nonpathogenic bacteria residing in the environment do not pose a direct risk to human health, they may spread antibiotic resistance genes horizontally into human pathogens. Therefore, environmental antibiotic pollution, in addition to clinical usage, has been identified as a potential contributor to the promotion of antibiotic-resistant pathogens (8, 9), and WWTPs have been recognized as risk environments (10, 11).

Recently, high concentrations of antibiotics have been measured in WWTPs associated with pharmaceutical production. Industrial effluents contaminated with antibiotics at close to or above therapeutic levels have been described in India (quinolones [12, 13]), China (oxytetracycline [14] and metabolites of penicillin G [15]), Croatia (sulfonamides [16]), South Korea (lincomycin [17]), and Norway (bacitracin [18]). For several years, we have studied a WWTP in Patancheru, near Hyderabad, India, that received industrial effluent from approximately 90 pharmaceutical industries. In 2007, Larsson et al. reported unusually high concentrations of pharmaceuticals, particularly quinolones, in the final effluent leaving the treatment plant, which was released into a local stream (13). The levels of ciprofloxacin exceeded 28,000 μg/liter, and other quinolones, such as enrofloxacin (780 to 900 μg/liter), norfloxacin (390 to 420 μg/liter), lomefloxacin (150 to 300 μg/liter), enoxacin (150 to 300 μg/liter), and ofloxacin (150 to 160 μg/liter), were also detected at high concentrations. The effluent has been shown to be toxic to a wide range of organisms (13, 19–22), and antibiotic resistance genes are exceedingly prevalent in the downstream sediment (23).

We have isolated a large number of extensively drug-resistant bacteria from inside the Indian WWTP (24). In particular, the strain Ochrobactrum intermedium CCUG 57381, isolated from the equilibrator tank, was observed to be resistant to nearly all of the antibiotics screened. O. intermedium is phylogenetically related to Brucella spp., is considered to be an opportunistic pathogen, and has been isolated from human clinical samples, industrial environments, and soil (25). The species is also a human commensal present in the gut (26), making it a possible link in the recruitment of antibiotic resistance genes from environmental bacteria into the human microbiome (27). In this study, we have characterized the resistance phenotype and genotype of the multiresistant strain O. intermedium CCUG 57381, with the aim to link acquired resistance phenotypes with genomic alterations. In comparison with the type strain of the species, O. intermedium LMG 3301^T, the Indian strain showed markedly increased resistance against several classes of antibiotics, as measured by their MICs. Massively parallel pyrosequencing (454) was used to characterize the whole genome of the Indian strain, which revealed acquired mobile resistance genes and resistance-associated mutations in chromosomal genes. Our results indicate that WWTPs may serve as an important environment for the recruitment of resistance genes into commensal and pathogenic bacteria.

MATERIALS AND METHODS

The bacterium Ochrobactrum intermedium CCUG 57381, investigated in this study, was isolated as a part of a larger study in which several strains were obtained from samples taken from within the WWTP operated by Patancheru EnviroTech Ltd. near Hyderabad, India (24). The WWTP serves approximately 90 industries, which are primarily bulk drug manufacturers. The strain originated from a water sample collected on 8 March 2007 from the equilibrator tank, which mixed the different industrial effluents prior to treatment. Technical information about the wastewater treatment plant was described by Larsson et al. (13). The collected samples were diluted in sterile phosphate-buffered saline (PBS) (pH 7.2), plated on Luria-Bertani (LB) agar, tryptic soy agar (TSA), and R2A agar, and incubated at 30°C for 24 to 48 h, after which individual clones were picked for subsequent phenotypic and genotypic characterization. The isolates were restreaked and stored as glycerol stocks at −70°C. Full details of the isolation steps have been described by Marathe et al. (24). The strain was archived into the Culture Collection University of Gothenburg (CCUG) (www.ccug.se) under accession number CCUG 57381.

The strain O. intermedium CCUG 57381 was classified to the species level by comparative sequence analysis of the 16S rRNA gene and the gene for recombinase subunit A (recA). The 16S rRNA genes were amplified by PCR and sequenced, targeting the hypervariable region V3 using universal primers R518 and F357, as described previously (28). The recA sequence was determined after PCR amplification using Ochr_recAF/R primers (29, 30). The type strain O. intermedium LMG 3301^T, whose genome had been sequenced and the draft assembly made publically available, was selected as the reference for this study. No indication of exposure of O. intermedium LMG 3301^T to antibiotics has been recorded.

The resistance phenotypes of the Indian strain and the reference strain were quantified by determining the MICs of 47 antibiotics, using Etest strips (31, 32) as recommended by the manufacturer (bioMérieux SA) (Table 1; see Table S1 in the supplemental material). The antibiotics belonged to several different classes, including aminoglycosides, β-lactams (carbapenems, cephalosporins, and penicillins), macrolides, quinolones, sulfonamides, tetracyclines, and others. The MICs were recorded after 24 to 30 h of incubation at 37°C on Mueller-Hinton agar medium. The Etest strip protocols were validated by confirmation of the results for the quality control strains, Escherichia coli CCUG 17620 and Staphylococcus aureus CCUG 15915.

Table 1.

MICs for the screened O. intermedium strains CCUG 57381 and LMG 3301^T

Antibiotica	MIC (μg/ml)		Clinical breakpointsd
	Indian strainb	Reference strainc	S, ≤	R, >
Aminoglycosides
Amikacin	24	24	8	16
Gentamicin	1.5	1.5	2	4
Streptomycin (+)	32	8
Tobramycin (−)	6	16	2	4
Carbapenems
Ertapenem (+)	0.75	0.38	0.5	1
Imipenem	6	6	2	8
Meropenem	1	1	2	8
Cephalosporins
Cephalothin (first generation)	>256	>256
Cefaclor (second generation)	>256	>256	IE	IE
Cefotaxime (third generation)	>32	>32	1	2
Ceftazidime (third generation)	>256	>256	4	8
Ceftriaxone (third generation)	>32	>32	1	2
Macrolides
Azithromycin (+)	32	2	IE	IE
Clarithromycin (+)	>256	8	IE	IE
Erythromycin (+)	>256	24	IE	IE
Penicillins
Augmentin	>256	>256	2	8
Piperacillin/tazobactam	>256	>256	4	16
Amoxicillin	>256	>256	2	8
Ampicillin	>256	>256	2	8
Amdinocillin	>256	>256	IE	IE
Temocillin (−)	8	16
Penicillin G	>32	>32	0.25	2
Phenicols
Chloramphenicol (+)	32	16	IE	IE
Quinolones
Nalidixic acid (+)	>256	16	IE	IE
Ciprofloxacin (+)	>32	0.25	0.5	1
Ofloxacin (+)	>32	0.5	0.5	1
Levofloxacin (+)	>32	0.25	1	2
Gatifloxacin (+)	>32	0.5
Moxifloxacin (+)	>32	2	0.5	1
Sulfonamides and dihydrofolate reductase inhibitors
Sulfamethoxazole (+)	>1,024	2
Trimethoprim-sulfamethoxazole (+)	>32	0.064	IE	IE
Trimethoprim	>32	>32	IE	IE
Tetracyclines and glycylcyclines
Doxycycline (+)	24	1.5	IE	IE
Tetracycline (+)	24	1.5	IE	IE
Tigecycline	3	3

^aAntibiotics with a difference in MICs are highlighted with a plus sign (+) if the Indian strain was more resistant than the reference strain and with a minus sign (−) if it was more susceptible.

^bO. intermedium CCUG 57381.

^cO. intermedium LMG 3301^T.

^dEUCAST non-species-related breakpoints. IE, insufficient evidence that the species in question is a good target for therapy with the drug.

Genomic DNA from O. intermedium CCUG 57381 was prepared, following the method of Marmur (33). Extracted DNA was resuspended in water. The residual RNA and protein contaminations were verified by measuring absorbances at 260, 230, and 280 nm, using a NanoDrop ND-1000 UV-visible spectrophotometer (Thermo Fisher Scientific Inc.). The DNA was sequenced using massively parallel pyrosequencing on the Genome Sequencer FLX system with titanium chemistry at GATC Biotech's facilities (34). All reads shorter than 50 bp and with more than 10% ambiguous bases were discarded before the analysis (1,250 reads, 0.5%).

The genome sequence of the type strain of O. intermedium LMG 3301^T was downloaded from the Pathosystems Resource Integration Centre (35) (PATRIC) to be used as a reference in the mapping of the sequence data of the Indian strain. The genome of the type strain consists of two chromosomes (GenBank accession numbers ACQA01000001.1 and ACQA01000002.1) and two plasmids (ACQA01000003.1 and ACQA01000004.1) (36). High-quality reads were aligned to the four reference sequences, using the BLAST-like alignment tool (BLAT) with the parameter −fine (37). From the results, it was clear that the reference strain's smaller plasmid (ACQA01000004.1) was not present in the Indian strain (see Fig. S1 in the supplemental material). Therefore, only the two chromosomes and the larger plasmid of the reference strain were used as a reference genome in this study. The draft assembly of the genome of the Indian strain was performed as follows. A reference-based consensus sequence was determined by evaluating the BLAT result of each position of the reference genome. If the coverage in a specific position was at least 5-fold and the frequency of one particular base was more than 85%, that base was recorded for that position; if not, the position was set to ambiguous (N, total of 2,833 bp). Reads that did not align to any part of the reference genome were assembled de novo, using 454's GS De Novo Assembler software with the default parameters.

The genomes of the Indian strain, O. intermedium CCUG 57381, and the reference strain, O. intermedium LMG 3301^T, were compared to find mutations, acquired antibiotic resistance genes, and mechanisms for horizontal gene transfer. The sequences of both genomes were matched against the Antibiotic Resistance Genes Database (ARDB) (38) using BLASTx with default filters masking any low-complexity sequences (e.g., repetitive regions) turned off (parameter −F F) (39). The nonoverlapping hits with the most matching amino acids (aa) were assessed manually, using BLASTx against GenBank, and the best hits were validated with manual literature searches. To verify the presence of the β-lactam antibiotic resistance gene and its transcriptional regulator annotated in the reference strain (ampC and ampR, respectively), the genome sequence was compared to all verified ampC genes listed by Jacoby (40) and the annotated and characterized ampR in Ochrobactrum anthropi, using BLASTx (parameter −F F). The best full-length hit for each of the two genes was extracted and used as references when mapping all reads with 454's GS Mapper (default parameters). The resulting contiguous sequences (contigs) were translated using transeq (bacterial codon table). Plasmid-mediated quinolone resistance genes were identified using a method developed by Boulund et al. (41). Mutations conferring resistance to quinolones were identified by comparing the quinolone target proteins (GyrA, GyrB, ParC, and ParE) from the reference strain with the draft genome using tBLASTn (parameter −F F). The procedure was also used to find mutations in the ribosomal protein S12 and the 16S rRNA gene that confer resistance to streptomycin, as well as mutations in the 23S rRNA gene conferring resistance to macrolides; the latter two analyses were performed using BLASTn (parameter −F F). The reference sequences for the proteins GyrA, GyrB, ParC, ParE, and S12, as well as the nucleotide (nt) sequences for 16S and 23S rRNAs, were retrieved from the database PATRIC (GenBank accession numbers: GyrA, EEQ95723.1; GyrB, EEQ94812.1; ParC, EEQ93589.1; ParE, EEQ93589.1; S12, EEQ95962.1; locus tag 16S rRNA, OINT_2001241; and locus tag 23S rRNA, OINT_1001994). By aligning the sequences with the corresponding sequences in Escherichia coli (GyrA, NP_416734.1; GyrB, YP_026241.1; ParC, NP_417491.1; ParE, NP_417502.1; S12, NP_417801.1; 16S rRNA, JWR0092; and 23S rRNA, JWR0095), the differences between the Indian strain and the reference strain were compared to known resistance mutations in E. coli. The identification of genes involved in conjugation was performed using CONJscan (with heuristics off) (42). Insertion sequences (ISs) were identified using the IS-finder search tool with BLASTx and the filter turned off (http://www-is.biotoul.fr) (43), and insertion sequence common regions (ISCRs) were detected using BLASTn (parameter −F F) against all ISCRs listed by Toleman et al. (44).

Nucleotide sequence accession number.

The raw sequence data have been submitted to the SRA database under accession number SRR866623.

RESULTS

Ochrobactrum intermedium CCUG 57381 was identified genotypically by comparative 16S rRNA and recA gene sequence analyses, since the 16S rRNA gene alone is insufficient for classification at the species level within the Ochrobactrum genus (29). The 16S rRNA gene sequence was most similar to that of the type strain of O. intermedium (accession number AM490623 [45], 99.5% nucleotide [nt] identity), and the best match for the recA sequence was also O. intermedium (AM422944, 97.7% nt identity). From these data, the Indian strain could be identified with confidence as a strain of the species O. intermedium. The strain exhibited substantially higher MICs, indicating greater resistance, to several classes of antibiotics, than the reference strain (O. intermedium LMG 3301^T) (Table 1), as well as a number of environmental and clinical strains of O. intermedium (25). Compared to the reference strain, the Indian strain was more resistant to quinolones (>128-fold increase in MICs), as well as the sulfonamide sulfamethoxazole, both as a single substance and in combination with trimethoprim (>512- and >500-fold increases, respectively). Increased resistances toward macrolides (up to a >32-fold increase), tetracyclines (16-fold), and the aminoglycoside streptomycin (4-fold) were also detected. There were also slight (2-fold) increases in the resistances to chloramphenicol and the carbapenem ertapenem. The reference strain was, conversely, more resistant to the aminoglycoside tobramycin (a 3-fold difference) and the penicillin temocillin (a 2-fold difference). The Indian strain would be classified as resistant to all tested antibiotics, using the EUCAST non-species-related clinical breakpoint MIC values, except for the carbapenems and the aminoglycoside gentamicin (46).

To link the resistance phenotype to genotypic resistance mechanisms, the genome of the Indian strain was sequenced. Massively parallel pyrosequencing (454) produced 90.6 Mb of DNA sequence data distributed over 249,120 reads with an average length of 364 bp, which covered 99% of the three sequences of the reference genome (ACQA01000001.1 to ACQA01000003.1), at an average coverage of 16-fold (see Fig. S1 in the supplemental material). The remaining 1% (133 kb in ACQA01000001.1, 322 kb in ACQA01000002.1, and 0.2 kb in ACQA01000003.1) consisted mainly of larger uncovered regions, implying that the genome of the Indian strain contained a number of putative deletions. The de novo assembly of the 44,529 unmapped reads (18% of all reads) resulted in 199 contigs that were 0.5 to 87 kb long (5 kb on average). The contigs had a cumulative length of 1 Mb, suggesting that the Indian strain had acquired larger horizontally transferred elements. No circular contigs were identified, implying that any acquired plasmids were likely to be of substantial size.

Comparative genomics identified several chromosomal genes containing mutations that have been associated previously with antibiotic resistance (see Table S2 in the supplemental material). Mutations causing amino acid substitutions in the quinolone targets were detected: four in GyrA (A83V, S84P, S206A, and A865G, using Escherichia coli sequence numbering) three in ParC (A80V, E205D, and D439E), and one in ParE (R135H); none were detected in GyrB (Fig. 1). Among these, the amino acid substitutions 83V and 84P in GyrA, as well as substitutions in position 80 in ParC, have been linked previously to quinolone resistance in E. coli. The 23S rRNA gene, which is known to harbor mutations resulting in macrolide resistance, had an insertion of a guanine after position 1656 (E. coli numbering), but in only 10% of the reads covering the position, and seven nucleotide mutations in a region that is present in only one of the three 23S rRNA genes in the reference genome. There were no mutations causing amino acid substitutions in the ribosomal protein S12 and no nucleotide differences in the 16S rRNA gene, both of which have been linked to streptomycin resistance. The chromosomally encoded AmpC β-lactamase had one mutation causing an amino acid substitution (T127I [E. coli numbering]), while its transcriptional regulator, AmpR, had five (P114R, K118N, S119R, G274E, and L285V [O. anthropi numbering]).

Fig 1.

Amino acid substitutions in the DNA gyrase (A) and topoisomerase IV (B) of the quinolone-resistant Indian strain (O. intermedium CCUG 57381) compared to the sensitive reference strain (O. intermedium LMG 3301^T) and E. coli K-12.

Three of the de novo assembled contigs contained horizontally transferred genes that have been associated previously with antibiotic resistance (Table 2; Fig. 2). Resistance region A coded for the class G tetracycline efflux pump Tet(G), its regulator protein TetR (47), and the chloramphenicol/florfenicol resistance pump FloR (50). The region also encoded TraG, which is involved in the horizontal transfer of plasmids, two open reading frames (ORFs) (orfA2 and orfA5) similar to the LysR transcriptional regulator family, and three ORFs of unknown function (orfA1, orfA3, and orfA4). TraG is one of the 24 Tra proteins used for the conjugational transfer of DNA between bacteria and is responsible for pilus formation and aggregate stability (48). A high sequence similarity was observed between orfA1 and a previously identified hypothetical ORF fused with a truncated sulfonamide resistance gene (sul1), although the sul1 fraction was not present in resistance region A (100% amino acid [aa] identity over the non-sul1 part of ACX47979.1). A portion of the region (4,562 bp) showed a high sequence similarity (97% nt identity) to a region of Salmonella genomic island 1 (SGI1-J, called SGI2 by Levings et al. [49]) (Fig. 2). The similarity began at orfA1, ended in the middle of the LysR-like orfA2, and included all of the resistance genes in the contig. Few reads matched to the remaining parts of SGI1-J, implying that the entire genomic island was not present in the Indian strain.

Table 2.

Acquired resistance genes in the Indian strain

Resistance region	Gene name	Protein name	Description (reference)	Protein length (aa)	GenBank accession no.	Identity (%)	Location within contig (strand)	MIC(s) (μg/ml) when expressed in E. coli (reference)
A	floR	Chloramphenicol efflux protein	Confers resistance to chloramphenicol and florfenicol (64)	404	ABZ01840.1	99.5	1187–2398 (+)	Chloramphenicol, 64; florfenicol, 16 (62)
	tet(G)	Tetracycline resistance protein, class G	Resistance to tetracycline by an active tetracycline efflux; this is an energy-dependent process that decreases the accumulation of the antibiotic in whole cells; this protein functions as a metal-tetracycline/H+ antiporter (64)	393	ABZ01843.1	95.5	3387–4511 (+)	Oxytetracycline, 200; doxycycline, 50; minocycline, 25; tetracycline, 50 (50)
	tetR	Tetracycline repressor protein, class G	Repressor of the tetracycline resistance element; its N-terminal region forms a helix-turn-helix structure and binds DNA; binding of tetracycline to TetR reduces the repressor affinity for the tetracycline resistance gene [tet(G)] promoter operator sites (64)	208	ABZ01841.1	95.7	2611–3234
B	floR	Florfenicol export protein	Appears to provoke a reduction of the content of the major porins OmpA and OmpC (50)	404	AAL33886.1	94.8	3685–4896 (+)	Chloramphenicol, 64; florfenicol, 16 (62)
C	sul1	Dihydropteroate synthase type 1	Catalyzes the formation of the immediate precursor of folic acid; it has been implicated in resistance to sulfonamide; the type II enzyme is stable, whereas type I loses its activity rapidly (63)	279	AAC44317.1	99.3	2644–3480 (+)	Sulfathiazole, >128 (63)

Fig 2.

The three contigs containing antibiotic resistance genes in the draft assembly of the genome of the Indian strain (O. intermedium CCUG 57381). (Top) Genes encoded in resistance region A and a partial depiction of SGI1-J. (Middle) Genes encoded in resistance region B and a partial depiction of plasmid R55. (Bottom) Genes encoded in resistance region C.

Resistance region B showed similarities to region A and also contained the genes for a chloramphenicol/florfenicol efflux pump, FloR, and a plasmid conjugation protein, TraG (90% and 88% aa identities to the genes in region A, respectively). Similarities (>80% nt identity) were also detected between orfA1 and orfB1, the LysR-like orfA2, orfA5 and orfB4, orfB5, although the genes were arranged in a different order (Fig. 2). Additionally, the contig included two hypothetical ORFs (orfB2 and orfB3) of unknown function. Resistance region B showed homology (91% nt identity) to the resistance plasmid R55, initially isolated from Klebsiella pneumoniae (50, 51). However, few reads matched the remaining portions of R55, indicating that the Indian strain did not contain the entire plasmid.

Resistance region C contained two genes: the sulfonamide resistance gene sul1 (63) and the transferase of an insertion sequence (IS) (52). The transferase exhibited 96% amino acid identity to ISApr9, which belongs to the family IS1380 (see Fig. S2 in the supplemental material) (43). The inverted repeats marking the edges of the IS were both found upstream of the sul1 gene. Furthermore, several other assembled regions from the genome of the Indian strain contained genes associated with the horizontal transfer of genetic material (Table 3). In addition to the IS in resistance region C, the Indian strain had acquired ISRle4 belonging to subgroup IS407 in family IS3. There were also six genes involved in conjugation (see Table S3 in the supplemental material), encoding a MOB_P relaxase, mating pair formation proteins of type I (traI, traM, traN, and traY), and an ATPase (traU) (42). No insertion sequence common region (ISCR) transposases were detected (44).

Table 3.

Elements involved in horizontal gene transfer acquired by O. intermedium CCUG 57381

Elementa	Lengthb	Alignment coverage (%)	E value	Identity (%)
				nt	aa
Conjugation system
MPFI
traI	273	86	1.20 × 10−58
traM	229	86	1.40 × 10−44
traN	336	98	1.60 × 10−53
traY	745	86	1.50 × 10−58
Relaxase
MOBP	509	98	9.40 × 10−30
ATPase
traU	1,008	99	1.30 × 10−265
IME
SGI1-J	42,415 nt	11		97
Plasmid
R55	170,810 nt	3		91
IS
IS1380
ISApr9	450 aa	100			96
IS3
ISRle4	396 aa	97			87

^aIME, integrative mobilizable element; IS, insertion sequence.

^bValues for the conjugation system are profile lengths.

DISCUSSION

We have sequenced the genome of the extensively antibiotic-multiresistant Ochrobactrum intermedium CCUG 57381, isolated from the equilibrator tank of a WWTP receiving industrial wastewater from bulk drug industries in Patancheru, India. A draft assembly of the genome revealed not only chromosomal mutations and horizontally transferred genes associated with resistance but also acquired genetic elements involved in the mechanisms for horizontal gene transfer. The resistance phenotypes of the Indian strain and the reference strain (type strain O. intermedium LMG 3301^T) were compared, and considerable increases in the resistance against quinolones, sulfonamides, tetracyclines, macrolides, and the aminoglycoside streptomycin were observed. The majority of the observed differences could be linked to acquired resistance mutations in autochthonous chromosomal genes or resistance genes acquired by horizontal gene transfer.

Quinolones inhibit the type 2 topoisomerases DNA gyrase and topoisomerase IV, which are essential for DNA replication and transcription in the bacterial cell (53). Both enzymes are tetramers (X₂Y₂), where DNA gyrase is encoded by the genes gyrA and gyrB, and topoisomerase IV is encoded by parC and parE. Mutations in the quinolone resistance-determining regions (QRDRs) of the genes are the most common causes of high degrees of resistance to quinolones in bacteria (53). The genome of the Indian strain possessed missense mutations, causing four amino acid substitutions in GyrA, none in GyrB, three in ParC, and one in ParE (Fig. 1). Two of the substitutions in GyrA (A83V and S84P [E. coli numbering]) were in positions that have been reported previously to result in quinolone resistance in E. coli (53). Of the three altered positions in ParC, one has been linked to resistance in E. coli (position 80 [E. coli numbering]), while the substitution in ParE was outside the QRDR. Additionally, the Indian strain did not contain any known or putative mobile quinolone resistance genes (qnr genes) (41, 54). The mutations present in the quinolone target genes are thus a likely explanation for the considerable increase in MICs of all tested quinolones.

The decreased susceptibility of sulfonamides and tetracyclines observed in the Indian strain could be linked to horizontally acquired resistance genes. The sul1 resistance gene, located on resistance region C, encodes a dihydropteroate synthetase that is unaffected by sulfonamide inhibition (63). The gene has been linked previously to high-level resistance against sulfonamides and thus is likely to be responsible for the high MIC for sulfamethoxazole observed in the Indian strain. Both O. intermedium strains tested were resistant to the dihydrofolate reductase inhibitor trimethoprim. Therefore, sul1 should also result in an increased resistance to the combination of sulfamethoxazole and trimethoprim, which is consistent with the observed phenotypes (Table 1). Resistance region A contained the tetracycline resistance gene tet(G) together with its repressor regulatory gene tetR. The tet(G) operon is known to provide resistance to tetracycline and doxycycline (55) but not to glycylcyclines, such as tigecycline (47). This finding is in line with the resistance phenotype, which demonstrated a 16-fold increase in the resistance to tetracycline and doxycycline but no change for tigecycline (Table 1).

The Indian strain also exhibited an increased resistance to macrolides (>8-fold) and the aminoglycoside streptomycin (4-fold). However, the genome did not contain any known genes for resistance to either macrolides or aminoglycosides. Mutations were found in the ribosomal 23S rRNA gene (see Table S2 in the supplemental material), but they were located far from the previously identified macrolide resistance mutations in E. coli (positions 754, 2057, 2058, 2059, 2505, and 2611) (56, 57). Furthermore, no mutations were found in the 16S rRNA gene or the ribosomal protein S12, both of which are linked to streptomycin resistance. The difference between the resistance phenotypes of the Indian strain and the reference strain thus could not be linked to any known resistance genotype. It should be noted, however, that O. intermedium isolates have been shown to exhibit high degrees of variation in resistance to both aminoglycosides and macrolides. A study of 17 environmental and clinical isolates of O. intermedium reported MICs for the macrolide azithromycin of between 2 and >256 μg/liter and MICs for the aminoglycosides amikacin and gentamicin of 4 to 128 μg/liter and 1 to 6 μg/liter, respectively (25). The MIC values measured for both the Indian and the reference strains (32 and 2 μg/liter for azithromycin, respectively, 24 μg/liter for amikacin, and 1.5 μg/liter for gentamicin) are all within these ranges and, conceivably, could be explained by the natural variation caused by different gene expressions of general purpose efflux pumps or differences in cell wall permeability.

The Indian strain had acquired two resistance genes against chloramphenicol/florfenicol (floR). The FloR protein encoded in resistance region A was 99% identical to that of SGI1-J and 90% identical to the FloR encoded by resistance region B, which was 95% identical to that of plasmid R55. The FloR fragment of R55 has been shown to increase the MICs of chloramphenicol and florfenicol in E. coli to 32 μg/liter and 128 μg/liter, respectively (50). A difference in the chloramphenicol resistance was indeed observed, where the Indian strain demonstrated a slightly increased resistance (32 μg/liter, compared to 16 μg/liter for the reference strain). However, both MIC values are within the range of the reported effect of FloR (50), suggesting that there were other, possibly intrinsic, chloramphenicol resistance mechanisms in place before the Indian strain acquired the floR genes. Furthermore, the MICs for chloramphenicol are also in the range of those in previously screened O. intermedium strains (25), implying that the acquired resistance genes had only a limited impact or were not expressed.

O. intermedium strains normally harbor the chromosomally encoded β-lactamase AmpC and its regulator protein AmpR, which are associated with resistance to β-lactam antibiotics. The expression of AmpC in the closely related O. anthropi has been shown to cause high-level resistance to penicillins and cephalosporins but not to carbapenems (58). Both O. intermedium strains in this study showed phenotypic resistance to β-lactam antibiotics, similar to that of O. anthropi, suggesting that AmpC is both functional and expressed. Hence, the mutations in the β-lactam genes seem to have limited or no effect. One of the penicillins tested, temocillin, is a narrow-spectrum derivative of ticarcillin, whose structure makes it stable toward AmpC and extended-spectrum β-lactamases (ESBL) (59). It is used specifically to treat infections caused by ampC-expressing Enterobacteriaceae (40), which explains the lower MIC values for temocillin than for the other penicillins.

The Indian strain had several genes associated with horizontal gene transfer that were not present in the reference strain. Four mating pair-forming (MPF) genes, traIMNY, together with the traU ATPase gene and the MOB_P relaxase gene detected in the Indian strain, suggest that the strain has obtained at least one conjugative element carrying a class I type IV secretion system (42, 60). Type I conjugative systems are usually located on plasmids of 60 to 500 kb and are relatively rare in Alphaproteobacteria, but they have been identified previously in Methylobacterium spp. and Gluconobacter oxydans (42). This finding is in line with the observation that the Indian strain has obtained a large amount of novel genetic material and that the Ochrobactrum genus has been shown previously to have a variable genome size. In addition to two chromosomes, O. intermedium isolates have been shown to carry as many as 3 additional plasmids, some of which have substantial size (36). The genome of the Indian strain also contains the insertion sequences (ISs) ISApr9, which is a part of the IS1380 family, and ISRle4 belonging to subgroup IS407 in the IS3 family (43, 52). As is commonly observed for ISs, both the IS1380 and the IS3 families carry only 1 or 2 genes responsible for the mobility of the IS, e.g., a transferase gene (52). There have been examples of ISs that are directly associated with the horizontal transfer of antibiotic resistance genes in Alphaproteobacteria (e.g., ISSm2 in Xanthobacter autotrophicus) (61). However, the inverted repeats in resistance region C were flanking the transferase gene, which suggests that it is not directly involved in mobilizing sul1, located downstream from the IS.

The Indian strain had acquired resistance to different classes of antibiotics targeting a variety of processes in the bacterial cell, including DNA replication as well as cell wall and protein synthesis. The multitude of molecular mechanisms that could cause the acquired multiresistance made traditional methods for the detection of antibiotic resistance genes, such as PCR, impractical. Instead, whole-genome sequencing provided a comprehensive exploratory approach that is able to detect simultaneously all known, previously reported resistance mutations and genes. The type strain O. intermedium LMG 3301^T is well characterized, and its genome had been sequenced. Sequencing the Indian strain at a 16-fold coverage, using the long reads produced by massively parallel pyrosequencing, was sufficient to identify genomic alterations, both chromosomal mutations and acquired mobile resistance genes. However, it should be noted that a higher sequencing coverage combined with successive closure of the assembly gaps would be necessary to fully assemble all the genetic material acquired by the Indian strain. Similarly, for bacterial isolates without a suitable reference genome, more extensive sequencing strategies would be needed. Nevertheless, this study shows that next-generation whole-genome sequencing combined with bioinformatic analysis provides an explorative and cost-effective approach to identify resistance factors and predict the resistance phenotypes of bacterial isolates.

WWTPs receiving human sewage create a setting wherein environmental bacteria meet human commensals and pathogens in the presence of antibiotics. The strain O. intermedium CCUG 57381, isolated in the industrial wastewater treatment plant in Patancheru, was able to survive in an extreme environment due to acquired resistance mutations and horizontally transferred genes. Because the strain was isolated from the equilibrator tank, before the step in which human sewage is added to the process, this consideration suggests an environmental origin for this strain other than the human gut. Although we cannot say exactly when and where the Indian strain acquired its resistance mechanisms, the extreme setting encountered at the WWTP (or before, at the industry delivering antibiotic-contaminated wastewater to the plant) provided the required selection conditions for its propagation. Our results, therefore, not only exemplify the genetic basis for how environmental bacteria can acquire multiresistance but also support the hypothesis of industrial WWTPs as potential breeding grounds for the selection of “superbugs,” such as O. intermedium CCUG 57381. The present study underlines the importance of effective management practices for treating antibiotic-contaminated waste that consider the risks for the promotion of resistance.