Natural Hot Spots for Gain of Multiple Resistances: Arsenic and Antibiotic Resistances in Heterotrophic, Aerobic Bacteria from Marine Hydrothermal Vent Fields

Abstract

Microorganisms are responsible for multiple antibiotic resistances that have been associated with resistance/tolerance to heavy metals, with consequences to public health. Many genes conferring these resistances are located on mobile genetic elements, easily exchanged among phylogenetically distant bacteria. The objective of the present work was to isolate arsenic-, antimonite-, and antibiotic-resistant strains and to determine the existence of plasmids harboring antibiotic/arsenic/antimonite resistance traits in phenotypically resistant strains, in a nonanthropogenically impacted environment. The hydrothermal Lucky Strike field in the Azores archipelago (North Atlantic, between 11°N and 38°N), at the Mid-Atlantic Ridge, protected under the OSPAR Convention, was sampled as a metal-rich pristine environment. A total of 35 strains from 8 different species were isolated in the presence of arsenate, arsenite, and antimonite. ACR3 and arsB genes were amplified from the sediment's total DNA, and 4 isolates also carried ACR3 genes. Phenotypic multiple resistances were found in all strains, and 7 strains had recoverable plasmids. Purified plasmids were sequenced by Illumina and assembled by EDENA V3, and contig annotation was performed using the “Rapid Annotation using the Subsystems Technology” server. Determinants of resistance to copper, zinc, cadmium, cobalt, and chromium as well as to the antibiotics β-lactams and fluoroquinolones were found in the 3 sequenced plasmids. Genes coding for heavy metal resistance and antibiotic resistance in the same mobile element were found, suggesting the possibility of horizontal gene transfer and distribution of theses resistances in the bacterial population.

INTRODUCTION

The ocean floor near the Azores is constituted in part by the Mid-Atlantic Ridge and a series of extensive hydrothermal vents. The largest hydrothermal vent field known at this time is located at 37°18.5′N, 32°16.5′W, averages 1,700 m in depth, and is named “Lucky Strike” (1). Here, 21 active hydrothermal vents are connected to the thermal anomalies in the subsea floor at the Mid-Atlantic Ridge that are spread over 150 km². The field is primarily made of basalt but also spikes interest due to rich deposits of ores. Temperatures of the field vary from 170° to 324°C (2), diminishing with the distance from the vent outlet due to water circularization with the surrounding cold temperatures characteristic of these depths (around 4°C), which therefore harbor a distinct microbial community. Furthermore, hydrothermal activity at Lucky Strike seems to have been episodic over a long time (2).

The nature of the bedrock and the geological events at Lucky Strike vent sites result in smaller concentrations of Rn and heavy metals (Cd, Hg, Cu, Pb, Zn, Fe, and Ag) and less acidic fluids, making it one of the least extreme environments, in a biological perspective, of the Mid-Atlantic Ridge compared to neighboring vent sites (3).

Antibiotics in the environment may result from adaptive phenotypic and genotypic responses of the microbiota. Recently, antibiotic-producing bacteria have been found in deep-sea environments. Strains belonging to the Streptomyces genus, a genus producing two-thirds of the clinically useful natural antibiotics, were reported as part of the microbial community of hydrothermal vents (53). In very stressful environments such as this one, where conditions change drastically within millimeters, antibiotic producers, which are able to kill competitors and get an ecological advantage, are not so unexpected (5, 54). Resistance to toxic compounds such as antibiotics and metals is commonly a result of resistance determinants located on mobile genetic elements. The mobility of such resistances leads to their dissemination in a microbial population and to the occurrence of tandem arrays of genetically linked resistance genes, commonly seen as coresistance, in which genes responsible for two or more resistances are adjacent to one another on a mobile genetic element (6). These tandem arrays are manifested many times in the form of multidrug resistance (MDR) systems (7). Resistance can also be related with cross-resistance, a selection process in which heavy metals and antibiotic resistance mechanisms are coupled physiologically (8). A resistance pool originating from heavy metal contamination pressure will confer resistance to any other contaminant as long as these are connected through some form of coselection (6, 9).

Many heavy metals have been associated to deep-sea hydrothermal vents. Arsenic (As), a well-known environmental toxin (4), is also a significant element in the formation of hydrothermal ore deposits.

Arsenic and antimony (Sb) are usually copresent in the environment. They have similar geochemical properties and toxicity. The amount and speciation of arsenic in hydrothermal deposits is a consequence of source rock composition, redox potential, pH, and temperature of water, as well as the concentrations of other elements in solution such as iron and sulfur (5, 10–12). In phosphate-poor systems, such as ocean environments, arsenate resistance or detoxification pathways might be essential (13, 14). In the case of antimony, little was yet explored in marine environments.

Since arsenic is environmentally prevalent, microbes developed mechanisms to handle arsenic and avoid its toxicity by several mechanisms, including extrusion mediated by arsenite efflux pumps (ArsB and ACR3). These systems also confer resistance to Sb.

Some studies demonstrated that microorganisms from extreme environments with multiple heavy metal resistances, namely, arsenate and antimonite, are also resistant to antibiotics that, surprisingly, are not found in the environment from which the organisms originated (15, 16). Resistances are commonly coupled in genetic elements that, despite being common on plasmids, can occur also in the chromosomes (17, 18). The present work focused on determining the arsenic resistance and antibiotic resistance profile of microorganisms isolated in pristine environments such as hydrothermal vents, adding to the knowledge of the extent of the microbial diversity that may contribute to antibiotic resistance, with the outcome of understanding the arsenic-antibiotic coresistance in microorganisms.

MATERIALS AND METHODS

Sample collection and isolation procedure.

Water and sediment samples were collected from the Lucky Strike hydrothermal vent area (37°17′N, 32°16′W; depth, 1,695 m) in August and September 2009 during the oceanographic mission EMEPC/Luso/2009 promoted by the Task Group for the Extension of the Continental Shelf (EMEPC). Samples were recovered with the remote operated vehicle, rated to 6,000 m and operated by EMEPC (Fig. 1). The procedure for sample collection uses a robotic arm with a suction device. The suction device is buried in the sea floor sediment, and the sample is drawn. The samples were collected from vents by using 0.75-liter titanium bottles and stored at −80°C for subsequent analysis studies. Sediment samples were broken apart, homogenized, and partitioned. A quantity of 9 ml of sediments from collected sample L09D23S3 was mixed with 10 ml of glycerol (30%).

FIG 1.

Remote operated vehicle images from EMEPC at the Lucky Strike site (37°17′N, 32°16′W; depth, 1,695 m). Hydrothermal vents (A, B, and C) expel hot fluids and sediments, which are deposited on the bottom of the sea floor; samples were collected (D) for subsequent analysis.

Bacterial strains and growth media.

In order to create a collection of aerobic heterotrophic bacterial strains from the Lucky Strike vent site, aliquots from sediment samples were inoculated in Reasoner's 2A (R2A) agar (Oxoid, Thermo Scientific, United Kingdom) containing (amounts in grams/liter are given in parentheses) yeast extract (0.5), proteose peptone (0.5), casein hydrolysate (0.5), glucose (0.5), starch (0.5), sodium pyruvate (0.3), di-potassium hydrogen phosphate (0.3), magnesium sulfate anhydrous (0.024), and agar-agar (15), and prepared with seawater filtrate (SWF) collected from coastal seawater. Each dilution was incubated at 15°C for up to 1 month. During incubation, as colonies appeared, all those with different colony morphologies were selected, purified by repeated streaking, and preserved at −80°C in Luria Broth containing 15% glycerol. Isolates were also recovered in the same R2A agar with SWF and in the presence of the following heavy metals: 5 mM arsenic(V), 2 mM arsenic(III), and 0.5 mM antimony(III) (Sigma-Aldrich, USA).

DNA extraction and 16S rRNA gene amplification and phylogenetic analyses.

Total DNA was isolated from hydrothermal vent field sediments using the E.Z.N.A. Soil DNA kit D5626-01 (Omega Bio-Tek, USA) according to the manufacturer's instructions and used for amplification of arsenic genetic determinants. DNA from each isolate was obtained using a standard freeze-thaw method (19) and used for typing, plasmid recovery, amplification of resistance genetic determinants, and phylogenetic identification. The extracted DNA was stored at −20°C. Isolates were grouped by random amplification of polymorphic DNA (RAPD) typing, using primer OPA-03 (5′-AGT CAG CCA C-3′) (Operon Technologies, Inc., Alameda, CA, USA). DNA profiles were grouped on the basis of visual similarities of the fragments analyzed by electrophoresis in a 2% agarose gel in Tris-acetate EDTA (TAE) buffer stained with ethidium bromide.

Amplification of the nearly full-length 16S rRNA gene sequence of representative strains selected from each RAPD group was performed by PCR with universal primers (20). The obtained sequences were matched with the existing sequences using the BLAST program in EzTaxon (21, 22).

Sequences were aligned using the Mega 5 package (23) for construction of phylogenetic dendrograms using the neighbor-joining algorithm with the following parameters: Jukes-Cantor correction model for nucleotides and 1,000 bootstraps (24). In order to identify the isolates that belonged to the genus Sulfitobacter, the 16S rRNA gene sequences were aligned with the sequences from all the type strains of all the species of the genus, using the ARB software package (sequences obtained from EZtaxon). A phylogenetic tree was constructed, using PhyML included in the ARB software package; the isolates grouped with the type species in phylogenetic clusters and were identified according to the species of the type strain of the cluster (see Fig. S1 in the supplemental material).

Isolates' heavy metal resistance determination.

Metal resistance of the isolates was determined by incubation at 22°C for 48 h, in R2A+SWF agar with a concentration of the following heavy metals: AsO₄³⁻ (5 mM), NaSb(OH)₄ (0.5 mM), CdSO₄ (0.2 mM) or Na₂CrO₄ (0.5 mM) (Sigma-Aldrich); AsO₃³⁻ (1 mM), C₄H₆O₆U · 2H₂O (1 mM), ZnSO₄ (1 mM), CoCl₂ · 2H₂O (0.5 mM), or CuSO₄ · 5H₂O (1 mM) (Merck).

Antibiotic resistance testing of the isolates.

Strains' resistance to antibiotics was determined by evaluating bacterial growth in media containing antibiotics in various (increasing) concentrations. A suspension of each bacterial strain in 0.9% sodium chloride, with an optical density of 0.8, was plated and incubated at 22°C for 48 h.

For this assay, 12 antibiotics representing different classes and mechanisms of action were selected. Two concentration thresholds of the antibiotics were used for determining the bacterial antibiotic resistance, which were, for each antibiotic, the critical concentration and the limit of solubility of the antibiotic in the medium (25). The antibiotics tested were gentamicin (16 to 20 μg/ml), tetracycline (8 to 12 μg/ml), kanamycin (16 to 20 μg/ml), vancomycin (20 to 30 μg/ml), rifamycin B (16 to 32 μg/ml), penicillin G (16 to 20 μg/ml), nalidixic acid (16 to 24 μg/ml), erythromycin (4 to 8 μg/ml), cephalosporin C (32 to 40 μg/ml), trimethoprim (8 to 16 μg/ml), ampicillin (16 to 20 μg/ml), and polymyxin B (2 to 5 μg/ml) (Sigma-Aldrich).

Detection of arsenic genetic resistance.

Amplification of the gene encoding the ACR3 efflux pump was performed by PCR with the following degenerate primer set: dACR5F (5′-TGA TCT GGG TCA TGA TCT TCC CVA TGM TGV-3′) and dACR4R (5′-CGG CCA CGG CCA GYT CRA ARA ART T-3′) (26). PCR included an annealing temperature gradient with 10 cycles from 57 to 52°C (−0.5°C per cycle) followed by 25 cycles at 52°C. Each annealing step lasted 45 s. For the amplification of the arsB gene, the degenerate primer set used was darsB1F (5′-GGT GTG GAA CAT CGT CTG GAA YGC NAC-3′) and darsB1R (5′-CAG GCC GTA CAC CAC CAG RTA CAT NCC-3′) (26). The PCR was performed with a temperature of annealing of 55°C. Positive PCRs were sequenced, and the sequences were used in the phylogenetic analysis of the clone library for comparison to the data obtained from the community DNA.

Clone libraries were generated from the PCR amplicons of the ACR3 and arsB genes. Clones were generated using pGEM-T Easy (Promega, USA) and transformed into Escherichia coli DH5α. Using the blue/white screening test on LB media supplemented with ampicillin, IPTG (isopropyl-β-d-thiogalactopyranoside), and X-Gal (5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside), clones were randomly selected, checked for inserts by PCR, and sequenced (Macrogen Europe, Amsterdam, Netherlands). Operational taxonomic units (OTUs) were taxonomically identified using BLASTx (National Center for Biotechnology Information, Bethesda, MD, USA). Sequences with poor sequence alignment, no BLAST hits, or BLAST hits to genes that were not the ACR3 gene or arsB were removed from downstream analyses. OTUs and known sequences were aligned using the BioEdit tool. Mega 5 was used for construction of maximum likelihood phylogeny dendrograms using the Dyhoff model, considering uniform rates and a partial deletion (95% cutoff) on gaps/missing data.

Plasmid isolation and sequencing.

All isolates were screened for native plasmid content. Native plasmids from isolates were recovered with the Midiprep plasmid isolation kit (NZYTech, Portugal) from a cell suspension in 30 ml R2A agar. Plasmids were visually confirmed in 1% agarose electrophoresis gel, 30 min at 90 V, and quantified by Nanodrop (Thermo Scientific, USA). Plasmids were linearized by the following restriction enzymes: EcoRI, BamHI, and HindIII (TaKaRa). Reaction mixtures were incubated for 2 h 30 min at 37°C, and the results were observed in 1% agarose electrophoresis gel.

Strains carrying plasmids were compared, and those strains carrying the higher number of phenotypic resistances (to antibiotics and to metals) had their plasmids selected to sequence. A second selection was performed, and the plasmids with the highest values of purity and concentration were selected for sequence by Illumina, CA, USA.

Plasmid sequence assembly and gene annotation.

Raw data from the Illumina sequencing were assembled using the assembly program Edena V3, and assembly coverage of >8 was the standard for the assessment of high-quality assembly. Gene annotation was performed using the “Rapid Annotation using Subsystems Technology” (RAST) server. By using RAST, genes were annotated according to functional domains, recognized in multiple databases, grouped in functional subsystems, and divided into hypothetical and nonhypothetical genes. Genes without functional domains, in databases accessible using RAST, were not grouped in functional subsystems. BLAST software was used in the search for antibiotic and heavy metal resistance determinants and for mobility-conferring elements (Inc systems, pylus-encoding genes, and replication initiators). The results of sequencing were also screened for the presence of chromosomally carried genes (16S rRNA, RNA polymerase-encoding gene), as a quality control for the presence of chromosomal DNA.

Nucleotide sequence accession numbers.

The 16S rRNA gene sequences of the isolates reported in this study have been deposited in the EMBL database under the accession numbers KJ862079 to KJ862112, and the sequences from the amplified arsB and ACR3 genes reported have been deposited under the accession numbers KM516031 to KM516054.

RESULTS

Heterotrophic aerobic bacteria can be isolated from deep-sea hydrothermal vent fields.

A total of 35 strains able to grow in antimonite (0.5 mM), arsenate (5 mM), or arsenite (1 mM) were obtained from deep-sea samples at the Lucky-Strike site.

RAPD typing showed that these strains consisted of 14 different clonal groups isolated in the presence of arsenate, 6 groups in the presence of arsenite, and 3 groups in the presence of antimonite (data not showed). From each RAPD clonal group, one isolate was chosen for phylogenetic characterization by 16S rRNA gene sequencing. The majority of the isolates belonged to Alphaproteobacteria (68%), and the rest belonged to Flavobacteria (32%). In detail, the genus Sulfitobacter included the majority of the isolates, with 12 strains from the species Sulfitobacter pontiacus (3 arsenate-, 3 arsenite-, and 6 antimonite-resistant strains) and 5 strains identified as Sulfitobacter dubius (3 arsenate- and 2 antimonite-resistant strains) (Table 1). The highest number of species isolated was from arsenate (8 species), followed by arsenite (2 species) and antimonite (2 species) (Table 1). After S. pontiacus, the second species with the largest isolate representation was Dokdonia donghaensis (four arsenate- and three arsenite-resistant strains). In antimonite- and arsenite-containing media, a lower diversity was observed than with arsenate-containing media. In addition to the species previously described, Maribacter dokdonensis (one isolate), Aurantimonas coralicida (one isolate), Oceanibulbus indolifex (four isolates), Mesoflavibacter zeaxanthinifaciens (three isolates), and Erythrobacter citreus (one isolate) were isolated from media containing arsenate.

TABLE 1.

Characterization of strains: plasmid content and antibiotic and heavy metal resistance

Strain ID	Genus and species	Presence of plasmids	Antibiotic(s)a to which strain shows resistance	Metal(s)b to which strain shows resistance
As(V) 1	Dokdonia donghaensis	−	KAN, NAL, CPC, TMP, PMB	As(V), As(III), Zn, Cu, Sb, Cd
As(V) 3	Sulfitobacter pontiacus	−	TET, KAN, NAL, CPC, TMP, PMB	As(V), As(III), Cu, Sb, Cd
As(V) 4	Sulfitobacter dubius	+	PMB	As(V), Cr, Cu, Sb, Cd
As(V) 5	Dokdonia donghaensis	−	TET, KAN, CPC, TMP	As(V), As(III), Zn, Cu, Sb, Cd
As(V) 6	Dokdonia donghaensis	+		As(V), As(III), Zn, Cu, Sb, Cd
As(V) 7	Oceanibulbus indolifex	+	COL, TET, KAN, VAN, PEN, NAL, ERY, CPC, AMP	As(V), Zn, Cu, Sb, Cd
As(V) 8	Sulfitobacter pontiacus	−	KAN, NAL, TMP	As(V), As(III), Cu, Sb, Cd
As(V) 9	Sulfitobacter dubius	−	TET, KAN, NAL, CPC, PMB	As(V), Cr, Zn, Cu, Sb, Cd
As(V) 10	Oceanibulbus indolifex	−	KAN, NAL, CPC, PMB	As(V), Zn, Cu, Cd
As(V) 11	Mesoflavibacter zeaxanthinifaciens	−	KAN, PMB	As(V), Zn, Cu, Sb, Cd
As(V) 12	Oceanibulbus indolifex	−	KAN, PMB	As(V), Zn, Cu, Cd
As(V) 13	Sulfitobacter pontiacus	+	NAL, TMP	As(V), As(III), Cu, Sb, Cd
As(V) 14	Dokdonia donghaensis	−	TET, KAN, TMP, PMB	As(V), As(III), Zn, Cu, Sb, Cd
As(V) 15	Mesoflavibacter zeaxanthinifaciens	+	KAN, TMP, PMB	As(V), Zn, Cu, Sb, Cd
As(V) 16	Erythrobacter citreus	+	TET, KAN, VAN, RIF, NAL, ERY, TMP, PMB	As(V), As(III), Zn, Cd
As(V) 17	Sulfitobacter dubius	−	COL, TET, VAN, RIF, NAL, ERY, TMP, PMB	As(V), Cr, Cu, Sb, Cd
As(V) 18	Oceanibulbus indolifex	−	KAN, NAL, PMB	As(V), Zn, Cu, Cd
As(V) 19	Mesoflavibacter zeaxanthinifaciens	−	KAN, TMP, PMB	As(V), Zn, Cu, Sb, Cd
As(V) 20	Oceanibulbus indolifex	−	NAL, TMP, AMP	As(V), Zn, Cu, Cd
As(V) 21	Maribacter dokdonensis	+	COL, TET, KAN, VAN, PEN, NAL, CPC, TMP, AMP, PMB	As(V), As(III), Zn, Cu
As(V) 22	Aurantimonas coralicida	+	COL, TET, KAN, VAN, PEN, NAL, CPC, TMP, AMP, PMB	As(V), Cr, U, Zn, Sb
As(III) 1	Sulfitobacter pontiacus	−		As(V), As(III), Cu, Sb, Cd
As(III) 2	Dokdonia donghaensis	−		As(V), As(III), Zn, Cu, Sb, Cd
As(III) 3	Sulfitobacter pontiacus	−	KAN, PEN, NAL, TMP, AMP	As(V), As(III), Cu, Sb, Cd
As(III) 4	Dokdonia donghaensis	−	KAN, NAL, TMP, AMP, PMB	As(V), As(III), Zn, Cu, Sb, Cd
As(III) 5	Sulfitobacter pontiacus	−	KAN, PEN, NAL, CPC, TMP, AMP, PMB	As(V), As(III), Cu, Sb, Cd
As(III) 6	Dokdonia donghaensis	−	COL, TET, KAN, VAN, RIF, PEN, NAL, ERY, CPC, TMP, AMP, PMB	As(V), As(III), Zn, Cu, Sb, Cd
Sb 1	Sulfitobacter pontiacus	−	TET, KAN, RIF, PEN, NAL, CPC, TMP, AMP	As(V), As(III), Cu, Sb, Cd
Sb 2	Sulfitobacter pontiacus	−	TET, KAN, RIF, PEN, NAL, CPC, AMP	As(V), As(III), Cu, Sb, Cd
Sb 3	Sulfitobacter pontiacus	−	KAN, TMP	As(V), As(III), Cu, Sb, Cd
Sb 4	Sulfitobacter pontiacus	−	TET, KAN, RIF, PEN, NAL, CPC, TMP, AMP	As(V), As(III), Cu, Sb, Cd
Sb 5	Sulfitobacter dubius	−	COL, TET, KAN, VAN, RIF, PEN, NAL, ERY, CPC, TMP, AMP, PMB	As(V), Cr, Cu, Sb, Cd
Sb 6	Sulfitobacter dubius	−		As(V), Cr, Cu, Sb, Cd
Sb 7	Sulfitobacter pontiacus	−	TET, KAN, RIF, PEN, NAL, CPC, AMP	As(V), As(III), Cu, Sb, Cd
Sb 8	Sulfitobacter pontiacus	−		As(V), As(III), Cu, Sb, Cd

^aCOL, chloramphenicol (20 μg/ml); TET, tetracycline (12 μg/ml); KAN, kanamycin (20 μg/ml); VAN, vancomycin (30 μg/ml); RIF, rifampin (32 μg/ml); PEN, penicillin G (20 μg/ml); NAL, nalidixic acid (24 μg/ml); ERY, erythromycin (15 μg/ml); CPC, cephalosporin C (40 μg/ml); SXT, trimethoprim (16 μg/ml); AMP, ampicillin (20 μg/ml); PMB, polymyxin B (5 μg/ml).

^bAs(V), arsenate (5 mM); As(III), arsenite (1 mM); Cr, chromium (0.5 mM); U, uranium (1 mM); Zn, zinc (1 mM); Co, cobalt (0.5 mM); Cu, copper (1 mM); Sb, antimonite (0.5 mM); Cd, cadmium (1 mM).

Isolates showed ubiquitous resistance to arsenate.

All isolates were tested for heavy metal resistance by media containing one of the nine heavy metals and metalloids: AsO₄³⁻, NaSb(OH)₄, CdSO₄, Na₂CrO₄, AsO³⁻, C₄H₆O₆U · 2H₂O, ZnSO₄, CoCl₂ · 2H₂O, and CuSO₄ · 5H₂O. All isolates were able to survive in the tested concentrations of arsenate, but none grew in the presence of cobaltous ions (Co²⁺).

The highest number of heavy metal resistances (six) was found in S. pontiacus strains As(V) 5, 6, and 14 and in D. donghaensis strains As(III) 2, 4, and 6 (Table 1). These were followed by the following, with five heavy metal resistances: A. coralicida strain As(V) 22; S. pontiacus strains As(V) 3 and 8, As(III) 1, 3, and 5, and Sb 1, 2, 3, 4, 7, and 8; S. dubius strains As(V) 4, 9, and 17 and Sb 5 and 6; and M. zeaxanthinifaciens strains As(V) 11, 15, and 19. All other strains were resistant to four heavy metals. Additionally, only the Sulfitobacter isolates were not resistant to Zn.

It is noteworthy that A. coralicida strains As(V) 22, S. dubius strains As(V) 4, 9, and 17 and Sb 5 and 6, M. zeaxanthinifaciens strains As(V) 11 and 15, and O. indolifex strain As(V) 7 were able to grow in antimonite but not in arsenite. Arsenite transport systems can also extrude antimonite, which is the only known resistance mechanism for antimonite. Only one strain, A. coralicida strain As(V) 2, was able to survive the uranyl concentrations used.

Antibiotic resistance can be found in isolates from deep-sea environments.

Sensitivity and/or resistance to a particular antibiotic was determined by growth tests on solid medium plates containing antibiotics (Table 1).

The antibiotic tests showing the greatest numbers of resistant isolates were the kanamycin test with 26 isolates, followed by the nalidixic acid test with 22 isolates; 21 isolates were found to be resistant for trimethoprim, 20 for polymyxin B, 15 for tetracycline and cephalosporin C, 13 for ampicillin, 10 for penicillin G, 9 for rifampin, 7 for chloramphenicol and vancomycin, and finally 5 isolates for erythromycin. The microorganisms were preferably able to overcome the inhibition of the protein synthesis (30S ribosomal subunit), followed by the DNA replication inhibition (topoisomerase) and the folic acid synthesis inhibition (dihydrofolate reductase).

On one hand, two isolates showed resistance to all antibiotics tested: D. donghaensis strain As(III) 2 and S. dubius strain Sb 5 (Table 1). On the other hand, the isolates from D. donghaensis strains As(V) 6 and As(III) 2, S. dubius strain Sb 6, and S. pontiacus strains As(III) 1 and Sb 8 showed no resistance to any antibiotic tested (Table 1).

Multiple antibiotic and heavy metal resistance phenotypes were found in isolates from hydrothermal vent fields.

For each strain, the number of phenotypic resistances to heavy metals and to antibiotics was assessed. The objective was to preview the possible existence of genetic coresistances. None of the isolates was resistant to all the antibiotics and heavy metals tested (Table 1). Furthermore, different strains of the same species showed different antibiotic resistance profiles, in contrast to equal metal resistance profiles, as was observed in D. donghaensis strains As(V) 5 and As(V) 6, which differ in their resistance profile to tetracycline, kanamycin, cephalosporin C, and trimethoprim. Antibiotic resistance seemed to be more strain specific than heavy metal resistance.

The comparison of the numbers of resistances accumulated by the different isolates was one of the methods used to select organisms for plasmid sequencing. Strain As(V) 6 was not included since it does not have plasmids, as was found for several strains with high numbers of resistances. M. zeaxanthinifaciens As(V) 22 and O. indolifex As(V) 7 had plasmids and were selected for plasmid sequencing, but we were not able to obtain plasmidic DNA with sufficient quality to be sequenced.

Plasmids may contain clues to some heavy metal and antibiotic resistance mechanisms.

Plasmid DNA was isolated from 7 of the 35 isolates (Table 1). The purity of the 7 plasmid DNAs was confirmed by enzymatic digestion after isolation and electrophoresis (data not shown). After considering the cross-resistance profiles, 3 plasmids from three strains, M. zeaxanthinifaciens As(V) 15, S. dubius As(V) 4, and M. dokdonensis As(V) 21, were selected for Illumina sequencing (Table 2). Newly assembled contigs were annotated using the RAST tool. All plasmids gave coverage of >22.

TABLE 2.

Phenotypic and genotypic characterization of antibiotic and heavy metal resistances of strains with sequenced plasmids

Plasmid, organism	Phenotypic resistances	Genotypic reference	Gene name	Resistance determinant	Reference(s)
pAs(V)21, Maribacter dokdonensis	COL, TET, KAN, VAN, NAL, PEN, CPC, TMP, AMP, PMB, As(V), As(III), Zn, Cu	CDC.320912.7.peg.851	blaA	β-Lactams	42
		CDC.320912.7.peg.468	blaC		43
		CDC.320912.7.peg.829	parE	Fluoroquinolones	27
		CDC.320912.7.peg.1428	gyrB		27, 44
		CDC.320912.7.peg.248	tolC	MDR	45
		CDC.20912.7.peg.843	MATE MDR pump family gene		45, 46
		CDC.200912.7.peg.246	macA		47
		CDC.320912.7.peg.247	macB		46, 47
		CDC.320912.7.peg.1346	omlA, nodT family		48
		CDC.320912.7.peg.734	RND, MFP subunit gene		49
		CDC.320912.7.peg.883	czcD	Cobalt-zinc-cadmium resistance	50
		CDC.320912.7.peg.922	cusA czsA		51
		CDC.320912.7.peg.202	cusB czsB		51
pAs(V)4, Sulfitobacter dubius	PMB, As(V), Cr, Cu, Sb, Cd	CDC.218673.8.peg.440	blaC	β-Lactams	43
pAs(V)15, Mesoflavibacter zeaxanthinifaciens	KAN, TMP, PMB, As(V), Zn, Cu, Sb, Cd	CDC.393060.6.peg.41	chrA	Chromium	52
		CDC.393060.6.peg.52	czcD	Cobalt-zinc-cadmium resistance	47

Genes that indicate the presence of chromosomes (16S rRNA and RNA polymerase genes) were not found in any of the plasmids. RAST analysis successfully identified plasmid replication (repA, repB, repC) and transfer (IncI/IncF transposon) genes. The plasmid from strain M. dokdonensis As(V) 21 [pAs(V) 21] showed 1,664 coding sequences and 23 noncoding RNAs sequences and included mobile element genes as the IncI and IncF plasmid conjugative transfer systems, as well as replication initiators repA and repFIB and several type IV secretion genes (virB1, virB4, virB5, virB6, virB8, virB9, virB10, virB11). The plasmids from strains S. dubius As(V) 4 [pAs(V)4] and M. zeaxanthinifaciens As(V) 15 [pAs(V)15] displayed comparatively fewer numbers of coding sequences (440 and 450) and <10 noncoding RNAs sequences. For pAs(V)4, replication initiators repA and repB were found, as they were found for pAs(V)15 along with a parA operon and several type IV secretion genes (virB2, virB4, virB8). The preliminary analysis of the three novel plasmids' sequences illustrates the limitation in plasmid genome information since for all plasmids RAST annotation was capable of clustering only 11 to 26% of genes in functional subsystems as defined by Aziz and collaborators (55). In all plasmids, there was a large clustering of membrane transport-encoding genes, such as tripartite ATP-independent periplasmic (TRAP) transporters and MDR efflux pumps, in the “membrane transport” functional subsystem, followed by genes encoding enzymes involved in surface polysaccharide biosynthesis in the subsystem “cell wall and capsule” in pAs(V)4 and pAs(V)15. There are also a large number of genes involved in carbohydrates metabolism and transport, such as the glycerol and glycerol-3-phosphate uptake and utilization cassette in the subsystem “carbohydrates” in pAs(V)4 and pAs(V)21.

We have found plasmidic antibiotic resistance determinants for β-lactams in plasmids pAs(V)21 and pAs(V)4 and for fluoroquinolones in pAs(V)21 (Table 2). The gyrB and parE genes are housekeeping genes, coding for DNA gyrase subunit B and topoisomerase IV subunit B, respectively. These genes were signaled as subject to mutations, coding for proteins with certain amino acid substitutions that confer resistance against fluoroquinolones to the bacteria carrying these plasmids (27). When analyzing the sequence obtained from the translated amino acid sequence of gyrB gene, a mutation was detected, i.e., the amino acid substitution at position 459 from asparagine (N) to histidine (H). The same analyses of the translated amino acid sequence of parE gene detected the amino acid exchange of valine (V) in position 327 for an isoleucine (I). In the parE translated protein sequence, other mutations where observed, but in this case, the amino acid exchanges did not correspond to the conferring of resistance mutations described in literature (27). In the case of determinants for specific resistance to heavy metals, those were found for only two metals: copper and chromium (Table 2). For copper, the plasmid pAs(V)21 possessed the CIA (copper translocating P-type ATPase) and MO (multicopper oxidase) genes. For chromium, plasmid pAs(V)15 was found to have a chrA (chromium transporter) gene. Notably, highly specific uptake systems for iron (e.g., siderophores) and for tungsten (tupABC, encoding an ABC transporter specific for tungstate) were also found (data not shown). Additionally, plasmids pAs(V)21 and pAs(V)15 encoded unspecific antibiotic and heavy metal resistance systems of multidrug resistance (MDR) transport. The MDR transporters detected were the inner membrane system MacAB, a resistance-nodulation-division multidrug efflux pump, an outer membrane TolC efflux protein, and the CzcD cation exporter (Table 2).

In plasmid pAs(V)21, the gene encoding the β-lactamase BL is located between two mobile elements characterized as a TnpA transposase and a resolvase, enzymes involved in the transposition and replication of the Tn3 transposon group (28). Also on pAs(V)21, the gene for the unspecific macrolide transporter macA is located immediately downstream from a sequence encoding the subunit B of the transposase orfAB.

Correlating plasmid genotype with strain phenotype reveled that only few resistances are plasmid encoded. In strain As(V) 21, phenotypically resistant to ampicillin, penicillin G, and cephalosporin C (β-lactams) as well as to nalidixic acid (fluoroquinolones), plasmidic genetic determinants for β-lactams and fluoroquinolone were found, as well as plasmid-encoded copper resistance, which correlates with its copper resistance phenotype (Table 2). On the plasmids from the other strains, genetic determinants for the phenotypic resistances were not observed, suggesting that the phenotype could be a result of multidrug resistance, another unspecific resistance, a new resistance system, and/or a chromosome-encoded resistance.

Arsenic resistance determinants were found only in arsenic-resistant bacteria.

Total DNA from the sediment sample L09D23S3 was extracted and subjected to genetic determination of ACR3 and arsB genes. Additionally, isolates were also screened for the presence of these genes. Positive amplifications were sequenced and phylogenetically analyzed.

Amplicons from total DNA resulted in 28 ACR3 gene sequences (13 sequences were identical) and 19 arsB sequences (14 sequences were identical) (Table 3). These sequences were submitted to the NCBI library using blastx.

TABLE 3.

Clone library of ACR3 and arsB genes from total DNA and ACR3 and arsB genes amplified from isolate DNA

Genetic determinant	Gene, protein (organism)
	From total DNA	From isolate DNA
ACR3 gene	ACR3-1 gene, arsenite efflux pump (Ochrobactrum tritici)	ACR3-As(V)8 gene, arsenite efflux pump (Sulfitobacter sp. EE-36)
	ACR3-2 gene, arsenical resistance protein (Afipia sp. 1NLS2)	ACR3-As(V)13 gene, arsenite efflux pump (Sulfitobacter sp. EE-36)
	ACR3-3 gene, arsenite efflux pump and related permease (Magnetospirillum magneticum AMB-1)	ACR3-As(V)16 gene, arsenic transporter (Novosphingobium pentaromativorans)
	ACR3-4 gene, arsenical resistance protein ACR3 (Vibrio furnissii CIP 102972)	ACR3-As(V)20 gene, arsenite efflux pump (Sulfitobacter sp. EE-36)
	ACR3-5 gene, arsenite efflux pump and related permease (Mariprofundus ferrooxydans PV-1)
	ACR3-8 gene, arsenical resistance membrane ACR3 family protein (Congregibacter litoralis KT71)
	ACR3-9 gene, arsenical resistance protein (Oligotropha carboxidovorans OM5)
	ACR3-11 gene, arsenic transporter (Methylococcus capsulatus)
	ACR3-12 gene, arsenical resistance protein (Oligotropha carboxidovorans OM5)
	ACR3-13 gene, arsenical resistance protein (gammaproteobacterium NOR5-3)
	ACR3-14 gene, arsenite efflux pump (Nitrobacter winogradskyi Nb-255)
	ACR3-16 gene, arsenical resistance protein (Rhodopseudomonas palustris DX-1)
	ACR3-21 gene, arsenical resistance protein (Thiocapsa marina 5811)
	ACR3-22 gene, arsenical resistance protein (Thiocapsa marina)
	ACR3-25 gene, arsenite efflux pump (Ochrobactrum tritici)
arsB	arsB1, arsenite efflux protein ArsB (Sulfurimonas autotrophica DSM 16294)
	arsB2, arsenical pump membrane protein (Arcobacter nitrofigilis DSM 7299)
	arsB14, arsenic efflux pump protein (Sulfurovum sp. NBC37-1)
	arsB15, arsenic efflux pump protein (Escherichia coli EC4100B)
	arsB16, arsenical efflux pump membrane protein (Arcobacter sp. L)

In the isolates, only the amplification of the ACR3-encoding gene was positive. This occurred for S. pontiacus strains As(V) 8 and 13, E. citreus strain As(V) 16, and O. indolifex strain As(V) 20 (Table 3).

The phylogenetic tree of translated ACR3 amino acid sequences showed that ACR3 amino acid sequences from the isolates clustered separately compared with ACR3 amino acid sequences obtained from total DNA (Fig. 2).

FIG 2.

(A) Phylogenetic tree of 16S rRNA genes obtained from deep-sea isolates. Sequences were aligned using Mega 5 for construction of phylogenetic dendrograms using the neighbor-joining algorithm with the following parameters: Jukes-Cantor correction model for nucleotides and 1,000 bootstraps. (B) Phylogenetic tree of the deduced amino acid sequences encoded by bacterial ACR3-like sequences from the hydrothermal vent environment and from the isolates. ACR3-like sequences were obtained from microbial total DNA of sediments of L09D23S3. Maximum likelihood phylogeny reconstruction used the Dyhoff model, considering uniform rates and a partial deletion (95% cutoff) on gaps/missing data. Putative horizontal gene transfer events (labeled with frames) have been compared based on the inconsistency of the ACR3 amino acid tree (B) and the 16S rRNA tree (A).

Phylogenetic analysis suggested that most ACR3 arsenite efflux proteins were consistent with 16S rRNA gene sequences. However, an exception included the ACR3 sequence from E. citreus As(V) 16, which clustered with ACR3 sequences from organisms with different phylogenies. The ACR3s separated into two major clades in previous studies (26, 29). In this study, 2 major clusters were also observed; one included most sequences isolated from the total DNA from the hydrothermal vent fields sediments, and the other included the ACR3 from the isolates and 3 sequences from sediments. In the cluster including the sequences from the isolates, 2 subgroups formed; one included only sequences from Alphaproteobacteria and all the isolates except ACR3-As(V) 16, and the other included sequences from Gammaproteobacteria and Cytophagales along with the 3 sequences from sediments and the remaining isolate.

DISCUSSION

The hydrothermal Lucky Strike field sediments in the Azores archipelago was shown to harbor a diversity of bacteria able to grow under aerobic conditions. The species recovered on R2A+SWF medium with As(V), As(III), or Sb(III) belonged mostly to the Alphaproteobacteria class and to the Flavobacteria, according to 16S rRNA gene sequence analyses. Aerobic chemolithotrophic bacteria are known to exist in deep-sea environments (30). The results obtained from the Lucky Strike sediments, using cultivation methods, are similar to those found in a variety of pelagic marine environments using culture-independent methods (31), where members of the Alphaproteobacteria are the major population. In Lucky Strike, as in the other marine environments, the genus Sulfitobacter included the majority of the isolates (32).

The diversity of species isolated in the presence of arsenate include also Flavobacteriaceae, most of them belonging to the species Dokdonia donghaensis, an organism first detected in seawater (26, 33). Individuals from the same phylum are found in seawater typically as 10 to 15% of the total bacteria (34). Strains of this ecologically versatile species were shown to possess the arsC gene, coding for arsenate reductase, and proteorhodopsin, a light-dependent proton pump (35).

Our results show that strains isolated from this extreme environment can be very resistant to different metals. There were strains resistant to up to six different metals [S. pontiacus strains As(V) 5, 6, and 14 and D. donghaensis strains As(III) 2 and 4]. Even though none was resistant to Co²⁺, as will be discussed below, the czcRS operon was found in a plasmid from one of the isolates. All were resistant to high concentrations of arsenate (5 mM) and some to arsenite (2 mM). This high level of resistance may represent adaptations to the varied dissolved metal and metalloid ions resulting from hydrothermal vent fluids. Arsenic is known to deposit around hydrothermal vents as sulfide minerals or iron oxyhydroxides (36).

The high resistance to arsenate might also be related to the ubiquitous arsenate resistance genes among bacteria (26). Genetic determinants related to arsenic resistance were already found in strains from all the genera isolated except for strains of Mesoflavibacter (Uniprot database). The ars operons encode either ArsB or ACR3 as the As(III) efflux pumps conferring resistance to arsenite. In this work, 28 ACR3 sequences and 19 arsB sequences were detected in the sediments, but in the isolates only 4 ACR3 sequences were detected. This can be related to the fact that ACR3 is the most frequently found arsenite pump in Alphaproteobacteria (to which the majority of the isolates belong) (26). Phylogenetically, ACR3s separated into two clusters, one including sequences obtained directly from the sediments and the other including the sequences obtained from the isolates. This can most probably be related with the selection of a restricted group of isolates due to the methodology used.

The phylogenetic relatedness of ACR3 seems to be in accordance with the corresponding tree predicted by 16S rRNA gene comparison. This suggests that ACR3 was mostly vertically inherited from the gene pool of the respective taxonomic clade and, therefore, absent from mobile genetic elements and also indicates an ancient origin of the enzyme (29, 37). However, one strain, E. citreus As(V) 16, showed putative horizontal gene transfer (HGT) events with ACR3, suggesting that HGT also played a role during the inheritance process (38, 39).

A growing body of evidence suggests that nonpathogenic environmental organisms are a reservoir of resistance genes, including genes conferring resistance to antibiotics. These organisms may eventually transfer these to pathogens (40, 41). We report that, like surface microbes, these bacteria from a deep-sea environment, considered preserved from anthropogenic impact, were highly resistant to antibiotics. Some strains were resistant to 12 different commercially available antibiotics. Resistance to a range of structurally different antibiotics was detected, and it was not species specific. Phenotypically, and compared to nonclinical strains, the highest level of resistance was observed for aminoglycosides (only four strains were not resistant to kanamycin), but mostly genetic determinants for β-lactams and multidrug resistance systems were present in the annotated plasmids. The comparison of the plasmid genotypes with strain phenotypes suggests that the phenotype could be the result of the presence of a specific genetic determinant, as in the case of the strain M. dokdonensis As(V) 21, which is phenotypically resistant to the fluoroquinolone nalidixic acid. This strain showed mutated parE and gyrB genes, encoding the topoisomerase IV subunit B and DNA gyrase subunit B, respectively, with specific amino acid exchanges that inhibit the effect of fluoroquinolones. Also, in strain M. dokdonensis As(V) 21, the resistance to the β-lactams penicillin G, cephalosporin C, and ampicillin could be justified by the presence of plasmidic genetic determinants encoding two β-lactamases, a class B β-lactamase (BL) and a class C β-lactamase (Blc).

It cannot be overruled that resistance can also be due to multidrug resistance or some other form of unspecific resistance. Plasmids encoding unspecific antibiotic and heavy metal resistance systems of MDR transport were found in this work. This is exemplified by copper- and zinc-resistant strains M. dokdonensis As(V) 21 and M. zeaxanthinifaciens As(V) 15, both with the gene encoding the copper-zinc-cadmium resistance protein CzcD, a cation efflux system that may act as a membrane-bound sensor or a metal ion transporter. In M. dokdonensis strain As(V) 21, the presence of the outer membrane efflux pump TolC, encoded by the gene tolC in pAs(V)21, may justify the resistant phenotype to chloramphenicol, tetracycline, β-lactams, and nalidixic acid, since all these antibiotics can be extruded by the TolC pump.

Alternatively, resistance can be conferred by a chromosome-encoded resistance and/or by a system not yet described.

Noteworthy, the plasmid pAs(V)21, which carries the highest number of genetic determinants related to antibiotic and metal resistance, also encoded conjugative systems IncI and IncF, suggesting that resistance genes can be mobilized to other bacteria permitting HGT.

As found before, this work demonstrates that antibiotic resistance is widespread in the environment even in the absence of anthropogenic antibiotic use (40). It is likely that competition for resources plays a dominant role in species persistence. This competition may occur through numerous adaptations, including through the production of antimicrobials to outcompete nutritional rivals. In deep-sea environments, organisms must also cope with different metals at high concentrations. Therefore, the acquisition and development of defense mechanisms as metal-extruding pumps that, at the same time, could extrude other toxins such as antibiotics, to enable flexibility and growth in the presence of noxious compounds, could be equally important. Mobile elements could be the basis for the bacterial answer to these challenges in stressful environments such as hydrothermal vents.