Adaptive Evolution of Extreme Acidophile Sulfobacillus thermosulfidooxidans Potentially Driven by Horizontal Gene Transfer and Gene Loss

ABSTRACT

Recent phylogenomic analysis has suggested that three strains isolated from different copper mine tailings around the world were taxonomically affiliated with Sulfobacillus thermosulfidooxidans. Here, we present a detailed investigation of their genomic features, particularly with respect to metabolic potentials and stress tolerance mechanisms. Comprehensive analysis of the Sulfobacillus genomes identified a core set of essential genes with specialized biological functions in the survival of acidophiles in their habitats, despite differences in their metabolic pathways. The Sulfobacillus strains also showed evidence for stress management, thereby enabling them to efficiently respond to harsh environments. Further analysis of metabolic profiles provided novel insights into the presence of genomic streamlining, highlighting the importance of gene loss as a main mechanism that potentially contributes to cellular economization. Another important evolutionary force, especially in larger genomes, is gene acquisition via horizontal gene transfer (HGT), which might play a crucial role in the recruitment of novel functionalities. Also, a successful integration of genes acquired from archaeal donors appears to be an effective way of enhancing the adaptive capacity to cope with environmental changes. Taken together, the findings of this study significantly expand the spectrum of HGT and genome reduction in shaping the evolutionary history of Sulfobacillus strains.

IMPORTANCE Horizontal gene transfer (HGT) and gene loss are recognized as major driving forces that contribute to the adaptive evolution of microbial genomes, although their relative importance remains elusive. The findings of this study suggest that highly frequent gene turnovers within microorganisms via HGT were necessary to incur additional novel functionalities to increase the capacity of acidophiles to adapt to changing environments. Evidence also reveals a fascinating phenomenon of potential cross-kingdom HGT. Furthermore, genome streamlining may be a critical force in driving the evolution of microbial genomes. Taken together, this study provides insights into the importance of both HGT and gene loss in the evolution and diversification of bacterial genomes.

INTRODUCTION

For several decades, it has been acknowledged that genomic rearrangements, including gene duplication and horizontal gene transfer (HGT), play critical roles in driving the adaptive evolution of microbial genomes in response to intense environmental perturbations (1–5). Great attention has thus been paid to gene and/or genome duplication, which represent dominant forces for functional innovation, including neofunctionalization and subfunctionalization (6, 7). Furthermore, the HGT process is considered a prevalent evolutionary mechanism that contributes to genomic diversification, species-level identification, and a trophic lifestyle (8, 9). Recently, large-scale studies based on increasing genomic data have significantly expanded the spectrum of genome reduction into a pervasive source of genetic modifications that potentially cause adaptive phenotypic diversity (10). Extensive gene loss events were observed across a wide range of organisms, including prokaryotes (11–15), protists (16), fungi (17, 18), plants (19), and even animals (20–23), thereby serving as robust evidence to support the pervasiveness of gene loss in all life kingdoms (10). However, the relative contributions of different evolutionary forces that shape the organization, structure, and diversification of microbial genomes remain elusive.

Extreme environments are generally known as almost insurmountable physical and chemical barriers to most life forms (24). However, microorganisms have been widely found in habitats that are characterized by harsh attributes, such as high temperature (25), high salinity (26), low pH (27, 28), or under strictly anaerobic conditions (29). To adapt to short-term or longstanding environmental stresses, microbes may have undergone frequent gene turnover. Sulfobacillus spp., which are Gram-positive spore-forming bacteria, are taxonomically affiliated with the order Clostridiales (30). Despite their poorly characterized low abundance compared to the dominant iron-oxidizing Leptospirillum spp. or certain members of the heterotrophic Thermoplasmatales, members of the genus Sulfobacillus are metal-tolerant, mildly thermophilic, or thermotolerant acidophiles that promote sulfide mineral dissolution and ubiquitously occur in various acidic habitats (30, 31), such as acid mine drainage (32), acid thermal springs (33), hydrothermal vents (34), and industrial bioleaching operations (35). In the past several decades, a number of papers have discussed issues related to key metabolic features within several isolated Sulfobacillus species, namely, S. thermosulfidooxidans (36), S. acidophilus (33), S. thermotolerans (37), S. sibiricus (38), and S. benefaciens (39). As such, the Sulfobacillus genus represents an intriguing consortium of microbes with disparate ecological distribution and physiological preferences.

In the advent of high-throughput sequencing and development of computationally derived approaches, an increasing number of available genomes of acidophiles have been sequenced, thereby providing the first glimpses into the genomic characteristics of acidophilic life under a range of environmental conditions (40). To date, the genomic features of two S. acidophilus strains (34, 41) and three S. thermosulfidooxidans strains (31, 42) have been investigated. Furthermore, the draft genomes of several Sulfobacillus strains have been assembled from metagenomic data (30), including a strain of S. benefaciens, a strain of S. thermosulfidooxidans, and three strains with no cultured representatives. Comparative genomics have yielded valuable insights into the physiological diversity, ecological roles, and genome evolution of environmental microorganisms (9, 43, 44). Genome-guided studies have revealed novel perspectives on the genetic features within genus Sulfobacillus; however, our understanding of their evolutionary adaptation and the underlying mechanisms of species-level identification is limited.

Here, we present a detailed genomic comparison of various Sulfobacillus species, including three genomes that were sequenced in this study and five other genomes that have been previously submitted to a public database. Our study provides evidence that these genomes underwent horizontal gene transfer (HGT) and functional recruitment. Findings also showed the prevalence of gene loss coupled with genome streamlining. Taken together, our results suggest that both HGT and gene loss processes represent important driving forces that may have contributed to the evolution of microbial genomes.

RESULTS AND DISCUSSION

Phylogeny and overview of genome features.

The constructed dendrogram based on the 16S rRNA gene sequences grouped the three newly sequenced strains into a phyletic cluster that included S. thermosulfidooxidans (≥99% 16S rRNA gene similarity; Fig. 1 and Table S1). Accordingly, five Sulfobacillus genomes were selected for phylogenomic analysis (Table 1). Comparison of average nucleotide identity (ANI) based on BLAST (ANIb) and MUMmer (ANIm), as well as regression of the tetranucleotide composition (Tetra), was conducted to further infer their phylogenetic relationship. The high ANIb, ANIm, and Tetra values between the new strains (DX, ZBY, and ZJ) and recognized S. thermosulfidooxidans strains Cutipay and ST strongly indicated their affiliation. Furthermore, the ANIb, ANIm, and Tetra values of strain CBAR-13 supported the hypothesis that this strain belongs to another Sulfobacillus species that is distinct from S. thermosulfidooxidans (Table 1). Based on these phylogenetic indicators, the three novel isolated strains were designated S. thermosulfidooxidans strains DX, ZBY, and ZJ, and S. thermosulfidooxidans CBAR-13 was thereafter referred to as Sulfobacillus sp. CBAR-13.

FIG 1.

Phylogeny showing relationships among three newly sequenced strains and published Sulfobacillus species using their 16S rRNA sequences. In the maximum likelihood phylogenetic tree, nodes with greater than 50% bootstrap support are shown. These three new strains used in this study are shown in bold.

TABLE 1.

Genome-based phylogenetic indicators of Sulfobacillus strains^a

Organism	ANIb (%)			Tetra			ANIm (%)
	DX	ZBY	ZJ	DX	ZBY	ZJ	DX	ZBY	ZJ
Sulfobacillus thermosulfidooxidans
Cutipay	97.10	97.11	97.10	0.994	0.994	0.994	97.30	97.30	97.30
ST	96.99	96.99	96.99	0.999	0.999	0.999	97.22	97.21	97.21
CBAR-13	87.41	87.41	87.41	0.989	0.989	0.989	88.98	88.98	88.97
Sulfobacillus acidophilus
DSM 10332	66.37	66.43	66.45	0.697	0.696	0.696	91.99	91.78	91.78
TPY	65.76	65.79	65.76	0.697	0.697	0.697	92.06	92.06	92.04

^aThe average nucleotide identity (ANI) based on BLAST (ANIb) and MUMmer (ANIm), as well as oligonucleotide signature frequencies (Tetra), were calculated using the program JSpecies. Values of ANIb and ANIm below 96% and TETRA values below 0.99 indicate that strains used in this study belong to different species.

Based on their complete genomes, we then excluded Sulfobacillus acidophilus strains TPY and DSM 10332 from our evaluation of genome completeness. Other genome assemblies of Sulfobacillus strains were assessed by CheckM, suggesting that these were near complete. The details on the Sulfobacillus genomes are shown in Table 2. Genome sizes and predicted protein-coding sequence (CDS) counts significantly varied among Sulfobacillus strains. Further inspection indicated slight variations in the total genome size of S. thermosulfidooxidans strains DX, ZBY, and ZJ (approximate size, 3.18 Mb), much smaller than those seen in other Sulfobacillus strains (Table 2). S. thermosulfidooxidans Cutipay had the largest genome size (3.86 Mb) and number of CDSs (4,250), whereas S. thermosulfidooxidans DX had the smallest genome size (3.18 Mb) and number of CDSs (3,211). These findings indicated that numerous accessory genes in S. thermosulfidooxidans Cutipay were likely acquired by HGT, similar to earlier comparative genomic analyses of Sinorhizobium strains (45, 46).

TABLE 2.

General features of bacterial genomes of Sulfobacillus strains

Organism	Geographic origin	Genome sequence status	Genome size (bp)	Coverage (×)	Completeness (%)a	No. of contigs	GC content (%)	Maximum contig length (bp)	Minimum contig length (bp)	N50 length (bp)	N90 length (bp)	No. of RNA genes				No. of predicted CDSsb	Reference or source
												5S rRNA	16S rRNA	23S rRNA	tRNA
Sulfobacillus thermosulfidooxidans
ZBY	Copper mine tailings, Chambishi, Zambia	Draft	3,180,484	140	99.00	24	48.47	716,967	218	404,938	88,978	3	1	1	51	3,211	This study
ZJ	Copper mine tailings, Fujian, China	Draft	3,180,801	136	99.00	20	48.47	688,926	231	452,891	108,775	2	1	1	51	3,213	This study
DX	Copper mine tailings, Jiangxi, China	Draft	3,180,071	132	99.00	33	48.47	688,914	1,285	171,016	52,886	1	1	1	51	3,227	This study
ST	Acid hot spring, Yunnan, China	Draft	3,325,386		98.00	97	48.35	408,384	201	139,115	34,608	0	1	1	48	3,408	31
Cutipay	Acidic mining environments, northern Chile	Draft	3,858,948	116	99.00	47	49.32	671,675	1,007	383,918	38,694	5	5	8	51	4,250	42
Sulfobacillus sp. CBAR-13	Escondida mine, Antofagasta, Chile	Draft	3,828,023		99.00	3	48.91	3,195,503	25,061	3,195,503	607,459	3	3	3	52	4,123	Unpublished datac
Sulfobacillus acidophilus
TPY	Hydrothermal vent, Pacific Ocean	Complete	3,551,206	26		1	56.76	3,551,206	3,551,206	3,551,206	3,551,206	0	5	5	52	3,894	34
DSM 10332	Coal spoil heap, United Kingdom	Complete	3,557,831	168		2	56.75	3,472,898	84,933	3,472,898	3,472,898	0	5	5	53	3,891	41

^aThe completeness of draft genome sequences were evaluated by CheckM.

^bGene prediction was performed using the platform RAST.

^cUnpublished data of S. Marin-Eliantonio, A. Moya-Beltran, M. Acosta-Grinok, F. Issotta, P. Galleguillos, J. P. Cardenas, V. Zepeda, L. G. Acuna, D. Cautivo, D. S. Holmes, R. Quatrini, and C. Demergasso.

Comparison of Sulfobacillus genome contents.

Mira et al. (47) reported that the variations in the genome sizes of bacteria were attributable to differences in gene number, in which bacterial genomes are tightly packed, with most regions consisting of functional protein-coding sequences. Functional assignment based on COG classification revealed that the newly sequenced strains consisted of a high number of genes that could be assigned to COG categories C (energy production and conversion), E (amino acid transport and metabolism), and G (carbohydrate transport and metabolism; Table S2). Similar to most microorganisms, these essential genes allow acidophiles to efficiently take up nutrients from the environment, as well as maintain a basic lifestyle. Konstantinidis and Tiedje (48) previously reported that genes in large genomes were disproportionally enriched in regulatory and secondary metabolisms. Similarly, relatively larger genomes were predicted to harbor more CDSs that could be assigned to the COG category Q (secondary metabolites biosynthesis, transport, and catabolism) than in medium- and small-sized genomes (Table S2).

To identify the pangenome of Sulfobacillus strains, a total of 9,049 CDSs obtained from the three novel genomes plus five reference genomes were clustered by PanOCT using a 65% sequence identity cutoff. We acquired a core genome of 671 CDSs, which was much smaller than that of a pangenome, indicating a significant difference among various Sulfobacillus species. Apart from the core genome, these flexible genes are dispensable and are probably not essential to basic bacterial lifestyle, but they may confer niche adaptation. The results showed that the genomes of S. thermosulfidooxidans strains harbored relatively fewer group-specific genes than S. acidophilus strains and Sulfobacillus sp. CBAR-13 (Fig. 2A). Similar to the findings of previous studies that conducted pangenome analyses (43, 45, 49), the highest number of core genes was observed in the functional category J (translation, ribosomal structure, and biogenesis). Furthermore, functional assignment based on COG classification revealed that most genes belonged to COG categories C and E (Fig. 2B). Differences in functions encoded by group-specific genes were also identified. A large number of CDSs (1,978) were shared by S. thermosulfidooxidans strains and Sulfobacillus sp. CBAR-13, thereby highlighting the most common traits between these two groups.

FIG 2.

Pangenome analysis of Sulfobacillus strains. The shared and strain-specific genes among Sulfobacillus genomes were calculated using PanOCT (A) and were then assigned to COG categories (B). (C) Numbers of core genome and unique genes within S. thermosulfidooxidans strains and Sulfobacillus sp. CBAR-13 are depicted in the six-way Venn diagram. Also, shared and strain-specific genes were used to be aligned against the COG database.

We further observed orthologous genes within recognized S. thermosulfidooxidans strains and Sulfobacillus sp. CBAR-13. Approximately 2,649 genes were shared by all strains (Fig. 2C). Around 1,174 genes were exclusively identified in Sulfobacillus sp. CBAR-13 and thus described as strain specific, further indicating their genomic differences from other S. thermosulfidooxidans strains. In addition, the number of unique genes in S. thermosulfidooxidans genomes varied from 9 to 938. Further analysis based on COG classification showed that the four most abundant genes shared by S. thermosulfidooxidans strains and Sulfobacillus sp. CBAR-13 were assigned to COG categories E, C, G (carbohydrate transport and metabolism), and J (Fig. 2C).

Identification of inferred metabolic traits and niche adaptation.

Sulfobacillus spp., known as a cohort of mildly thermophilic or thermotolerant acidophiles, are frequently found in various acidic settings worldwide. They are facultative anaerobes that utilize energy and electrons derived from aerobic oxidation of iron and a wide range of inorganic sulfur compounds for the assimilation of organic and/or inorganic carbon, as well as other anabolic metabolisms (30, 37, 38). According to the annotation results of the KEGG Automatic Annotation Server (50), the metabolic potentials of Sulfobacillus strains were reconstructed and compared with each other to investigate shared metabolic and species- and/or strain-specific pathways (Table S3). Also, in this context, a comparison of predicted stress tolerance mechanisms was performed.

(i) Comparison of central metabolisms.

Sulfobacillus spp. are known as mixotrophic acidophiles that are capable of assimilating organic and inorganic carbon (30, 34, 42, 51). Similar to the published Sulfobacillus genomes (30, 31), a full suite of genes involved in the Calvin-Benson-Bassham cycle were predicted in all strains (Fig. 3 and Table S4), e.g., form I ribulose 1,5-bisphosphate carboxylase (RuBisCO), composed of eight large and eight small subunits (27). In chemolithoautotrophic acidophiles, such as Acidithiobacillus and Ferrovum spp., a gene cluster potentially encoding several copies of carboxysome shell proteins, carboxysome-associated carbonic anhydrase, and RuBisCO were identified (9, 27, 43, 44). These microorganisms utilize carboxysome-associated carbonic anhydrase to elevate the concentration of carbon dioxide near the RuBisCO via conversion of accumulated cytoplasmic bicarbonate into CO₂ (52), thereby enabling efficient utilization of the limited available carbon in the environment. However, no evidence supported the existence of a corresponding gene/gene cluster in Sulfobacillus strains. Instead, genes encoding putative carbonic anhydrases were observed to be dispersed in other genomic regions (Table S4). With the exception of autotrophic growth, Sulfobacillus isolates have the ability to grow mixotrophically and heterotrophically on various organic carbon substrates, such as glucose, fructose, and glycerol (30). Although S. thermosulfidooxidans strains and Sulfobacillus sp. CBAR-13 were predicted to lack 6-phosphofructokinase (Table S4), an enzyme commonly present in the Embden-Meyerhof pathway for glycolysis, enzyme assays using cell extracts showed that these cells can transform glucose under mixotrophic conditions (53). Metabolic enzymes associated with the oxidative portion of the pentose phosphate pathway, including glucose-6-phosphate 1-dehydrogenase, 6-phosphogluconolactonase, and 6-phosphogluconate dehydrogenase, were identified in all organisms. Pyruvate that accumulates in other pathways is catalyzed by pyruvate dehydrogenase to generate acetyl-coenzyme A via decarboxylation, and then it enters the tricarboxylic acid (TCA) cycle and certain macromolecular biosynthetic pathways (Fig. 3). Additionally, genes related to the TCA cycle were complete in all the genomes. An earlier study revealed that aerobic CO oxidation performed by the carbon monoxide dehydrogenase (CODH) complex is ubiquitously found in bacteria (54). This complex is composed of a medium subunit, a large/catalytic subunit, and a small subunit. Similarly, several copies of multigene clusters that are potentially related to CODH were identified in Sulfobacillus genomes (Table S4).

FIG 3.

Schematic representation of the predicted metabolic potentials and adaptive strategies for environmental stresses within Sulfobacillus strains. The potential metabolic traits were focused on carbon metabolism, nitrogen uptake, sulfur metabolism, iron oxidation, as well as hydrogen utilization. Stress management mechanism was discussed, including cell mobility, acidic stress tolerance, and heavy metal resistance. The corresponding genes are listed in Table S4. CBB, Calvin-Benson-Bassham cycle; CoA, coenzyme A; cyt, cytochrome; Hyd Grp., hydrogenase group.

The comparison of metabolic profiles in Sulfobacillus strains was performed to reveal variations in nitrogen metabolism. All genomes harbor the set of genes required for the utilization of urea via urease (Fig. 3). Ammonium generated by the catalysis of urea is used as a nitrogen source in all Sulfobacillus strains, and the released bicarbonate (another metabolite derived from the hydrolysis of urea) is predicted to be catalyzed by carbonic anhydrases. All strains shared the potential for ammonium uptake via ammonium transporters and the assimilation of ammonia into central metabolic pathways via a series of enzymes, including glutamate dehydrogenase, glutamine synthetase, and glutamate synthase (Table S4). Additionally, further analysis showed that S. thermosulfidooxidans strains and Sulfobacillus sp. CBAR-13 harbored genes for assimilatory nitrate reduction, as indicated by the nasA encoding assimilatory nitrate reductase catalytic subunit and nirA, which encodes ferredoxin-nitrite reductase. However, the electron transfer subunit NasB that is required for electron transfer from NADH to nitrate (55) was absent in the nitrate reductases, thereby rendering an unidentified electron donor (30). No other components for nitrogen fixation and dissimilatory nitrate reduction were found. In contrast, S. acidophilus strains shared a copper-containing NO-forming nitrite reductase and a nitric oxide dioxygenase that potentially oxidizes nitric oxide to nitrate (56).

Oxidation of various inorganic sulfur compounds, including sulfur, tetrathionate, and sulfide, has been well documented in Sulfobacillus strains (35, 37–39). Sulfur oxygenase reductase (SOR) catalyzes the disproportionation reaction of linear polysulfide in the presence of molecular oxygen to generate hydrogen sulfide, sulfite, and thiosulfate (57, 58). SOR was identified in all Sulfobacillus spp., with S. thermosulfidooxidans strains and Sulfobacillus sp. CBAR-13 showing two copies (Table S4). Other metabolic enzymes potentially involved in sulfur oxidation of Sulfobacillus strains include sulfide:quinone oxidoreductase, thiosulfate:quinone oxidoreductase that is encoded by doxDA, and thiosulfate sulfurtransferase. Sulfobacillus genomes were predicted to possess tetrathionate hydrolase, which directs the disproportionation of tetrathionate to generate sulfate, thiosulfate, and other sulfur compounds (57). An additional route for tetrathionate metabolism in S. acidophilus strains was predicted to be driven by tetrathionate reductase, which is a three-subunit enzyme that catalyzes the reduction of tetrathionate to produce thiosulfate (59). Another important enzyme complex associated with sulfur compound oxidation in Sulfobacillus strains is heterodisulfide reductase-like protein, which is also frequently observed in other acidophilic sulfur oxidizers, such as Acidithiobacillus (57, 60, 61). All Sulfobacillus strains exhibited the predicted sulfate reduction, although only the S. acidophilus strains harbored the sulfite reductase (ferredoxin) as an additional enzyme.

Ferrous ion [Fe(II)] is rapidly oxidized to ferric iron in circumneutral environments, whereas it is stable under acidic conditions even in the presence of molecular oxygen (62). Thus, Fe(II) in acidic environments is available to iron-oxidizing acidophiles living in these habitats. Several Sulfobacillus isolates have been recognized as iron oxidizers (33, 37–39). In other acidophiles, such as Acidithiobacillus ferrooxidans (27), electrons derived from iron oxidation are ultimately transferred to either terminal electron acceptor NADH dehydrogenase (“uphill”) or to molecular oxygen (“downhill”). Similarly, S. thermosulfidooxidans and Sulfobacillus sp. CBAR-13 strains were predicted to harbor homologous redox proteins that are related to iron oxidation and the electron transfer chain. Membrane-associated c-type cytochromes potentially involved in iron oxidation were examined in all Sulfobacillus species. Electrons from membrane c-type cytochromes might be transferred to aa₃-type cytochrome oxidase via a “downhill electron pathway” (Fig. 3). In Leptospirillum spp. and A. ferrooxidans, a bc complex has been implicated in reverse electron flow (62). However, no homologous genes were identified in Sulfobacillus genomes. Two putative sulfocyanins, the blue copper protein having the conserved protein domain family (cd04230), were predicted to transfer electrons during iron oxidation in the S. thermosulfidooxidans strains and Sulfobacillus sp. CBAR-13 (Table S4), while no candidate gene encoding a sulfocyanin-like protein was identified in S. acidophilus. More details about iron oxidation in Sulfobacillus need to be further validated.

Under both oxic and anoxic environmental conditions, molecular hydrogen is widely used as an electron donor in bacteria and archaea to support growth (63). An earlier study has extensively described five distinct types of nickel-iron [NiFe]hydrogenases within the Sulfobacillus genomes (30). Accordingly, a conserved gene cluster encoding hydrogenase-like proteins was detected in all Sulfobacillus species, and the catalytic subunit was clustered with group 1 respiratory-uptake [NiFe]hydrogenase, which might oxidize H₂, transferring protons to the quinone pool via a membrane-integral cytochrome b subunit (64). Additionally, Sulfobacillus genomes contained the second multisubunit gene cluster related to hydrogenase, but the operon structure and catalytic subunits differed (30). The catalytic subunits in the S. acidophilus strains showed high sequence homology with subgroup A of group 2 hydrogenase, whereas the hydrogenases of the S. thermosulfidooxidans strains and Sulfobacillus sp. CBAR-13 were most similar to group 5 hydrogenases (Table S4). Other homologous proteins were assigned to group 4 hydrogenases, despite the absence of binding-site motifs in their catalytic subunits (30). Taken together, the presence of hydrogenase complexes in Sulfobacillus species indicated that molecular hydrogen might be a key source of low-potential electrons, although no experimental evidence for hydrogen utilization was observed in these microorganisms (30).

(ii) Management strategies for environmental stresses.

Each of the Sulfobacillus genomes analyzed in this study harbored a large number of genes assigned to COG categories T (signal transduction mechanisms; Table S2) and N (cell mobility), many of which were predicted to be associated with chemotaxis signal transduction systems and flagellar formation (Table S4). These special traits, such as flagella and chemotaxis, enable Sulfobacillus strains to actively move across environmental gradients, thereby providing a competitive advantage in aquatic environments.

A high number of strain-specific genes related to the COG category L (replication, recombination, and repair) were observed compared to that in other COG categories (Fig. 2C). This finding might be explained by a previous notion that protein and DNA repair systems in acidophiles play a critical role in dealing with acid stresses (65), given that low pH in these particular habitats could cause DNA and protein injury. For a long time, the potential mechanisms for pH homeostasis of acidophiles have attracted considerable interest of researchers. Baker-Austin and Dopson (65) described that distinctive structural and functional characteristics, mainly including highly impermeable cell membranes, reversed membrane potential, and the predominance of secondary transporters enable acidophiles to survive and grow under these extreme conditions. Similar to various acidophiles, Sulfobacillus spp. might generate a reversed membrane potential via the Kdp-type potassium transport system (Table S4), thereby partially deflecting the inward flow of protons (9, 65, 66). Additionally, genes encoding the Kef-type potassium efflux system and voltage-gated potassium channel proteins were identified in Sulfobacillus genomes. Monovalent cation/proton antiporters, i.e., Na⁺(K⁺)/H⁺ antiporters, may facilitate the active efflux of cytoplasmic sodium and/or potassium from the extracellular matrix in exchange for hydrogen (67). Accordingly, Sulfobacillus strains observed in this study were predicted to exclude intracellular redundant protons by Na⁺/H⁺ antiporters.

Sulfobacillus genomes were identified to harbor several genes/gene clusters that are potentially involved in stress tolerance strategies to cope with high metal loads (Table S4). A gene cluster related to the reduction of arsenate was identified in the S. thermosulfidooxidans genomes and Sulfobacillus sp. CBAR-13, whereas S. acidophilus strains were predicted to employ the arsenical efflux pump for arsenical resistance. Some transporting ATPases for the resistance to various heavy metal ions, including lead, cadmium, zinc, mercury, and copper, were identified in all species. We also found that Sulfobacillus organisms harbored genes that encode mercuric ion reductase and the cation diffusion facilitator family member CzcD.

Potential driving forces of genome evolution.

A comparison of the predicted metabolic profiles suggested that Sulfobacillus spp. harbor a core of genes that were essential to metabolic functions (Table S4). Further inspection also revealed distinguishing metabolic traits among these species. Apart from the core genome, bacterial genomes have been reported to have a number of accessory genes that were probably acquired by horizontal gene transfer (HGT) and were beneficial under environmental conditions (68). Linking the differences in the inferred metabolic profiles to their own genome architectures allows a detailed comparison of potential driving forces that contribute to genome evolution and diversification of Sulfobacillus strains.

(i) Detection of mobile genetic elements.

As an evolutionary force shaping the content of microbial genomes, HGT events frequently occur in microbes and are regarded as an effective means of rapid adaptation to changing environmental demands (69). A significant part of HGT is facilitated by mobile genetic elements (MGE) and is often characterized by certain signatures, e.g., integration sites usually associated with tRNA genes, varied codon usage, or abnormal G+C contents (68, 70, 71). Gain and/or loss of MGE, such as integrative conjugative and mobile elements, genomic islands, and insertion sequences (IS), may play a crucial role in adaptation to environmental stresses (71). Transposases and integrases were identified and classified in all Sulfobacillus genomes by using the online platform ISfinder (Table S5). Generally, a higher number of transposable elements was predicted in relatively large genomes (Cutipay, 272 transposable elements; CBAR-13, 160; TPY, 298; and DSM 10332, 299) than small genomes (DX, 76 transposable elements; ZBY, 79; ZJ, 79; ST, 90), indicating that gene turnover in larger genomes was relatively frequent. IS classes IS3, Tn3, and ISL3 were abundant in all genomes, whereas several IS classes were only identified in the genomes of individual strains (Table S5). Further inspection revealed the potential correlation between genome size and the number of insertion sequences among S. thermosulfidooxidans strains. For instance, S. thermosulfidooxidans Cutipay, which has a larger genome (3.86 Mb), harbored more transposable elements (272) than the others (Table S5). We thus proposed that HGT might be more frequent in larger genomes, thereby contributing to intraspecific divergence.

Other characteristic functions related to genomic islands were also investigated using the online platform IslandViewer3 (72). A higher number of putative genomic islands was identified in larger genomes than in smaller genomes (Table S6), further suggesting highly frequent genetic exchanges. Several CDSs located in genomic islands were annotated as hypothetical proteins. Further inspection identified numerous mobile element proteins, indicating that various genomic islands might have been acquired by HGT. Because HGT was critical to the expansion of the gene repertoires of prokaryotes (69, 73, 74), we deduced that the genomic islands in these species might play a predominant role in functional recruitment, thereby enabling them to adapt to specific environmental niches.

(ii) Linking differences in metabolic profiles of Sulfobacillus spp. to genome architectures.

The Sulfobacillus strains were further classified into four types of genomes based on their genome sizes (Fig. 4). Each type of genome was then used as reference for BLASTN-based whole-genome comparisons to unravel the presence/absence of genomic regions and to some extent reflect the relationship of Sulfobacillus strains based on the sequence identity.

FIG 4.

BLASTN-based whole-genome comparison of Sulfobacillus strains using Circos. The representative genomes from S. thermosulfidooxidans DX (A), Cutipay (B), Sulfobacillus sp. CBAR-13 (C), and S. acidophilus TPY (D) were used as references. Matches to each reference genome are displayed on rings 1 to 8 with different colors. Additionally, insertion sequences, tRNA, and G+C content are shown on rings 9 to 11. Cross-references link to other figures and tables showing more details of special regions.

Figure 4 shows that a number of genomic regions related to metabolic functions were only distributed over each type of Sulfobacillus genome. Apparently, some genomic regions potentially involved in metabolic pathways were only found in certain larger genomes, such as those of S. thermosulfidooxidans Cutipay (sections 1 to 7 in Fig. 4B) and Sulfobacillus sp. CBAR-13 (sections 8 to 13 in Fig. 4C). Several genes located in these genomic regions were predicted to encode putative proteins with unknown functions (Table S7). Further investigation to identify HGT signatures was conducted to infer the possible origin of these putative genes. Several transposases, integrases, and phage-associated proteins were predicted to disperse in the corresponding genomic locus (Table S7), thereby suggesting that these clusters might have been acquired by HGT. Also, several ABC transporter-related proteins were predicted to be acquired by HGT, probably indicating a high exchange rate for substances.

The genomic regions of the three new Sulfobacillus strains harbored genes involved in the utilization of urea (Table S4 and Fig. 4A and D). The urease gene clusters in the three Sulfobacillus species were distributed in distinct genomic regions, which were further visualized by the presence of colinear blocks (Fig. 5). Although several homologous genes were identified in these two urease gene clusters, these differed in terms of gene content and nucleotide sequence identity, thereby suggesting the divergent evolution of the urease gene cluster in the Sulfobacillus species. Two transposases were identified in the genomic neighborhood of the urease gene cluster of the S. acidophilus strains (Fig. 5B), whereas no signature for a recent HGT was identified in the S. thermosulfidooxidans strains and Sulfobacillus sp. CBAR-13, thereby suggesting that the urease gene cluster in S. acidophilus genomes was introduced via HGT. A gene encoding carbonic anhydrase (CA) was predicted upstream of the urease gene cluster in the S. thermosulfidooxidans strains and Sulfobacillus sp. CBAR-13 (Fig. 5A). Unlike other recognized acidophiles, such as Acidithiobacillus spp. (44) and “Ferrovum” spp. (9), in which CA gene was located in a carboxysome gene cluster, no carboxysome-associated genes were found in the Sulfobacillus strains. Additionally, two gene clusters involved in carbon monoxide dehydrogenase (CODH), which were potentially acquired by HGT, are discussed here. Form I CODH was identified in certain Sulfobacillus strains with larger genomes (Table S4 and section 24 in Fig. 4D), and form III CODH was only found in S. acidophilus strains (Table S4 and section 19 in Fig. 4D). One or several transposases identified in the genomic neighborhoods suggested that these genomic regions underwent several HGT events. In the S. thermosulfidooxidans strains and Sulfobacillus sp. CBAR-13, 15 genes located in a genomic region were predicted to encode transport-associated enzymes, such as sugar transporter and glycosyltransferases that are potentially related to the synthesis of the cell envelope polysaccharides (9), and signatures of a recent HGT event, i.e., phage-associated genes, were found in the genomic neighborhood (Fig. S1). We also identified some transposases around the hydrogenase group II gene cluster, which was only present in S. acidophilus strains (section 25 in Table S7). HGT events contribute to the acquisition of novel genes that were originally derived from apparently taxonomically unrelated species (70) and play an important role in functional recruitment as a number of accessory genes encode adaptive traits that might be beneficial to inhabiting various econiches (68, 71). We thus inferred that gene acquisition via HGT might be an efficient way of rapidly adapting to changes in the environment, thereby providing the advantage for bacterial survival, growth, and reproduction.

FIG 5.

Comparison of genomic regions in Sulfobacillus strains involved in urease (A), urea carboxylase, and carbon monoxide dehydrogenase (B). More details about genes distributed in selected regions are shown in Table S4.

Interestingly, three clustered regularly short palindromic repeat (CRISPR)/CRISPR-associated protein (Cas) systems were also identified in these genomic regions (sections 3, 5, and 27 in Table S7). CRISPR/Cas modules are adaptive immunity systems that widely occur in various bacterial species and almost all archaea (75, 76). Based on a comparison to systems in other bacteria (76), the CRISPR/Cas systems in S. thermosulfidooxidans Cutipay were classified as subtype I-C (Dvulg or CASS1; section 3 in Table S7) and subtype I-E (Ecoli or CASS2; section 5 in Table S7), and the CRISPR/Cas system in the S. acidophilus strains was assigned to subtype I-B (Tneap-Hmari or CASS7; section 27 in Table S7). The presence of the CRISPR/Cas systems in these species indicated that phage-host coevolution might be a means for HGT and play a key role in the genomic evolution in Sulfobacillus strains.

Comparative analysis of genome architecture revealed that several genomic regions were present in S. acidophilus strains and absent in S. thermosulfidooxidans strains and Sulfobacillus sp. CBAR-13 (Fig. 4 and Table S7). These regions were predicted to harbor genes involved in amino acid (section 16 in Table S7) and cofactor biosynthesis (section 17 in Table S7). However, no evidence for the presence of integrases, transposases, or phage-associated proteins was identified in the genomic neighborhoods, suggesting that the absence of these genomic regions in the S. thermosulfidooxidans strains might be the result of several gene loss events rather than HGT. We then inferred that the abandonment of conditionally essential genes contributed to the reduction in genome size, as well as shaped the metabolic profiles of the S. acidophilus genomes. Notably, the fourth gene cluster involved in carbon monoxide dehydrogenase (CODH) in S. acidophilus strains (Fig. 5B and section 28 in Table S7) was likely derived from a common ancestor, whereas the gene cluster in the S. thermosulfidooxidans strains and Sulfobacillus sp. CBAR-13 might have been lost during the evolution process, because no HGT signatures were identified. Compared to other CODH gene clusters, this cluster significantly differed in terms of gene order and orientation (Table S4). We thus proposed that alien genes might be integrated into the fourth CODH gene cluster and be caused by several rearrangements, thereby yielding different locations of the homologous genes in the respective genome. Additionally, S. acidophilus genomes harbor an 11-subunit NADH-quinone oxidoreductase, which lacks three genes (nuoEFG) encoding the “N-module” subunits that are associated with NADH binding (30), thereby rendering an unclear function (Table S4 and section 18 in Table S7). Accordingly, the results could provide a plausible adaptive explanation for the widespread loss of accessory genes with low contribution to cellular fitness.

The streamlining hypothesis supporting genome reduction has been proposed in the context of increased cellular economization (77). Examples of genome streamlining have been documented in various microorganisms, such as Prochlorococcus marinus, in which genome reduction was proposed as an adaptive mechanism for the efficient utilization of limited nutrients (78). A strategy to engineer the Streptomyces avermitilis genome has been performed, in which nonessential segments are deleted (79, 80) to improve the expression of heterologous pathways (81). Because acidic habitats have nutrient-deficient substrates (27), genome reduction appears to confer adaptive fitness advantages that drive the abandonment of expensive genes, thereby enabling the effective acquisition of limited nutrients. For long-term evolution, the streamlining process has allowed bacteria to maintain essential genes, as well as facilitate in the loss of dispensable ones, thereby resulting in a simpler but still fully functional genome.

(iii) Identification of putative cross-kingdom HGT events.

Phylogenetic analysis of sulfur oxygenase reductase (a critical enzyme involved in the sulfur oxidation of many acidophiles) unraveled a close relationship between the Sulfobacillus genus and archaeal Ferroplasma acidarmanus (31). Accordingly, it is plausible that archaeal phyla acted as potential gene donors to Sulfobacillus strains. Therefore, the genes of the Sulfobacillus genomes were likely to have been acquired via cross-kingdom HGT events. To identify other genes that potentially underwent adaptive evolution, we analyzed the occurrence of cross-kingdom HGT events within various Sulfobacillus genomes.

The genomic sequences of the three novel S. thermosulfidooxidans strains were annotated against the truncated database, in which protein sequences belonging to the Sulfobacillus strains were excluded. Several proteins with apparent archaeal BLASTP first hits were then identified, despite the lower percentage than that in all examined sequences (1.49%; Fig. 6). About 53 genes in the S. thermosulfidooxidans genomes were of putative archaeal origin (Table S8), with the majority of these genes phylogenetically affiliated with phylum Euryarchaeota (Fig. S2). Archaea constitute a significant fraction of the microbial biomass on Earth, although details and bases for this taxonomic classification are limited (82). Most of the examined genes that were potentially derived from archaeal donors were annotated as hypothetical proteins, whereas a fraction of the genes were predicted to be involved in central metabolism, e.g., CoB-CoM heterodisulfide reductase, sulfur oxygenase reductase, and TQO small subunit DoxD in sulfur oxidation and glucose-6-phosphate isomerase in glycolysis. Accordingly, it appears that these HGT processes represent an evolutionary mechanism that contributed to enhancement of metabolic capacities and thereby confer rapid adaptation to changing environments. Additionally, our analysis also suggested that the archaeal phyla are potential donors of transport-related genes of S. thermosulfidooxidans species. These results indicate that the acquisition of genes via cross-kingdom HGT events might represent a potential mechanism that drives the adaptive evolution of S. thermosulfidooxidans strains.

FIG 6.

Proportion of genes with nonself BLASTP first hits in three newly sequenced S. thermosulfidooxidans genomes. Protein sequences within three novel strains in this study were aligned against the truncated NCBI-nr database, in which sequences belonging to Sulfobacillus genomes were excluded.

Concluding remarks.

Comparative genomics has improved our current knowledge of genome evolution and species-level identification of Sulfobacillus strains. HGT may be a key evolutionary force that drives the genome expansion of Sulfobacillus strains, thereby controlling the response and adaptation of microorganisms to environmental changes. Furthermore, gene acquisition via a cross-kingdom HGT process is likely to be an effective way to recruit novel functionalities in microorganisms. These genomes undergoing genome minimization events, although relatively small, have thus inherited a core of essential genes and biological functionalities that are necessary for survival and stress tolerance in specific environments. Collectively, bacterial genomes have undergone a variety of processes, such as HGT and the loss of gene/genome segments, thereby contributing to the diversification and adaptive evolution of microorganisms (68, 69, 83).

New insights were gained into the potential evolutionary mechanism of Sulfobacillus strains. The assessment of mobile elements was performed, suggesting that genetic exchanges might be highly frequent in relatively large genomes. We further focus on certain genomic regions mainly related to metabolic potentials which were predicted to potentially undergo HGT and/or gene loss events. However, genomic information shows us the possible HGT within Sulfobacillus genomes but does not quantify HGT. Thus, experimental quantification of HGT should be conducted in the future.

MATERIALS AND METHODS

Bacterial cultivation and genomic DNA extraction.

The bacterial strains used in this study (ZBY, DX, and ZJ) were originally isolated from various copper mine tailings from around the world. Detailed information on the geochemical conditions of two Chinese copper mines is also described in this report (84, 85). Unfortunately, the environmental attributes of the sampling sites in the Zambian copper mine at that time were not assessed. Strains were cultivated aerobically in 100 ml of liquid 9K basic medium [3.0 g/liter (NH₄)₂SO₄, 0.5 g/liter MgSO₄·7H₂O, 0.1 g/liter KCl, 0.01 g/liter Ca(NO₃)₂, and 0.5 g/liter K₂HPO₄ (initial pH 1.6)] with 4.5% (wt/vol) ferrous sulfate (membrane filtered) and 0.02% yeast, as previously described (31). The temperature and shaking speed of the bacterial cultures were 45°C and 170 rpm, respectively. At mid-exponential-growth phase, the bacterial cells were harvested by centrifugation at 12,000 × g for 10 min at 4°C. Total genomic DNA was extracted previously described (57) and then used for genome sequencing.

Genome sequencing and assembly.

Whole-genome sequencing of all three novel strains (ZBY, DX, and ZJ) was performed on an Illumina MiSeq sequencer (Illumina, Inc., USA). Shotgun libraries with an average 300-bp insert size were constructed and were then used for 2 × 150-bp paired-end sequencing. The generated read sequences were then assembled de novo, as previously described (43).

Bacterial phylogenetic reconstruction.

The 16S rRNA gene sequences in the three genome assemblies were extracted using RNAmmer (86). Phylogenetic reconstruction of the 16S rRNA genes was then performed by using MEGA version 5.05 using the maximum likelihood method. Nodal support was evaluated using 1,000 bootstrap replications. The resulting phylogenetic tree initially showed that these three new strains were affiliated with S. thermosulfidooxidans, thereby prompting us to download the genome sequences of five available Sulfobacillus species from the public database, the National Center for Biotechnology Information (NCBI), for subsequent analysis. Contig fragments from all assembled scaffolds of S. thermosulfidooxidans strains ST and Cutipay were extracted using an in-house Perl script. The completeness of all available genome assemblies was estimated using CheckM (87). JSpecies version 1.2.1 (88), with default parameters, was employed to further infer the phylogenetic relationship among the Sulfobacillus strains by comparing their average nucleotide identity (ANI) (89) by using BLAST (ANIb) and MUMmer (ANIm) (90), as well as tetranucleotide frequencies (Tetra) (91).

Pangenome analysis.

Genome functional annotation was performed using the Rapid Annotations using Subsystems Technology (RAST) platform (http://rast.nmpdr.org/) (92). Subsequently, protein sequences and corresponding annotation were acquired as previously described (43, 93). PanOCT version 3.18 (94), with an E value cutoff of 0.00001 and sequence identity cutoff of 65%, was applied for the identification of orthologs that were shared between species or strains. Protein sequences extracted by using a Perl script were then annotated against the extended Clusters of Orthologous Groups (COG) (95). For functional comparison, the KEGG Automatic Annotation Server (50) was employed to predict the putative metabolism-related genes in Sulfobacillus genomes.

Identification of putative HGT events.

Transposases or insertion sequences (IS) within the Sulfobacillus genomes were identified using ISfinder (96). A recently developed Web server, IslandViewer 3 (72), which integrates a single comparative genomic island prediction method, IslandPick (97), and two sequence composition genomic island prediction methods, SIGI-HMM (70) and IslandPath-DIMOB (98), were used to identify putative genomic islands that were dispersed across the Sulfobacillus genomes. To identify potential cross-kingdom HGT events, protein sequences belonging to Sulfobacillus species that were detected by NCBI-nr were excluded. The three novel genomes in this study were then aligned against the truncated database to identify genes that were potentially derived from archaeal donors. Genes assigned to the archaea were then visualized using MEGAN5 (available at http://ab.inf.uni-tuebingen.de/software/megan5/).

Accession number(s).

These whole-genome shotgun projects for three novel strains have been deposited at the DDBJ/ENA/GenBank under the accession numbers MDZD00000000 (DX), MDZE00000000 (ZBY), and MDZF00000000 (ZJ). The versions in these papers are versions MDZD01000000, MDZE01000000, and MDZF01000000, respectively.