Convergent and divergent evolution of metabolism in sulfur-oxidizing symbionts and the role of horizontal gene transfer

https://doi.org/10.1016/j.mib.2012.09.003Get rights and content

Symbioses between marine invertebrates and autotrophic sulfur-oxidizing bacteria have evolved from multiple lineages within the Gammaproteobacteria in a striking example of convergent evolution. These GammaSOX symbionts all perform the same basic function: they provide their hosts with nutrition through the fixation of CO2 into biomass using reduced sulfur compounds as an energy source. However, our review of recent –omics based studies and genome mining for this study revealed that the GammaSOX symbionts diverge in many other metabolic capabilities and functions, and we show how these divergences could reflect adaptations to different hosts and habitat conditions. Our phylogenetic analyses of key metabolic genes in GammaSOX symbionts revealed that these differed markedly from 16S rRNA phylogenies. We hypothesize that horizontal gene transfer (HGT) would explain many of these incongruencies, and conclude that HGT may have played a significant role in shaping the metabolic evolution of GammaSOX symbionts.

Highlights

► GammaSOX symbionts converge in their basic metabolism, oxidation of sulfur and CO2 fixation. ► GammaSOX symbionts differ greatly in additional metabolic capabilities involved in energy and carbon gain. ► Horizontal gene transfer may play a major role in the metabolic evolution of these symbionts. ► Phylogenies of metabolic genes provide insights into the ecology of symbionts. ► GammaSOX symbionts are excellent models for studying horizontal gene transfer.

Introduction

A vast array of animals has acquired the ability to live from inorganic energy and carbon sources by establishing symbioses with chemolithoautotrophic bacteria. In these chemosynthetic associations, the symbiotic bacteria provide their hosts with organic compounds via autotrophic CO2 fixation using reduced compounds such as hydrogen sulfide and hydrogen as energy sources [1, 2]. Chemosynthetic symbioses have evolved independently numerous times in convergent evolution. This is evident from the phylogenetic diversity of the hosts, with at least six animal phyla known to host such associations [1, 3•], and their bacterial symbionts, that belong to numerous lineages within the Alphaproteobacteria [4], Deltaproteobacteria [5, 6], Gammaproteobacteria [1, 7, 8] and Epsilonproteobacteria [9, 10].

Chemosynthetic symbioses are found in a wide range of habitats from shallow water sediments to deep-sea hydrothermal vents and cold seeps [1, 3•]. Correspondingly, symbionts from such different environments can experience very different conditions, such as (i) lower or higher pH or temperature, (ii) differences in the concentrations of electron donors, electron acceptors, and CO2, or (iii) the availability of nutrients such as nitrogen and phosphate. Biotic factors also have a strong effect on symbionts, such as their location in the host (extracellular or intracellular) or on the host (epibiotic), the way in which they are transmitted from one generation to the next, or the manner in which the host interacts with them (Table 1). Recent genomic, transcriptomic and proteomic studies now enable us to compare the metabolism of phylogenetically distinct chemosynthetic symbionts from different biotic and abiotic environments and with different modes of symbiont transmission to better understand how these affect their evolutionary ecology.

Our review will focus on sulfur-oxidizing symbionts from the Gammaproteobacteria (GammaSOX symbionts), because some of these symbionts have been extensively studied using –omic approaches (Table 1). The GammaSOX symbionts share a similar basic metabolism – the oxidation of reduced sulfur compounds for energy gain (thiotrophy) and autotrophic fixation of CO2 into biomass. Despite this similarity, GammaSOX symbionts did not originate from a single common ancestor. At least nine phylogenetically distinct clades of chemosynthetic symbionts exist within the Gammaproteobacteria, and these are interspersed with free-living bacteria, indicating that symbioses with GammaSOX bacteria have evolved multiple times in independent evolutionary events [1, 8].

In this review, we will highlight examples of convergent and divergent evolution of symbiont metabolism and show that despite the fact that all GammaSOX symbionts fulfill the same basic function for their hosts, they can differ greatly in many other metabolic capabilities. Based on our observation that the phylogeny of many key metabolic genes from GammaSOX symbionts are incongruent with their 16S rRNA phylogeny, and the recent recognition that horizontal gene transfer (HGT) is very frequent in bacterial populations and is a major driver in ecological differentiation [11, 12], we will argue that HGT may have enabled the acquisition of genes by GammaSOX symbionts that led to both the convergence and divergence of their metabolic pathways. In this review we give numerous examples for genes that may have been gained through HGT and highlight features of the GammaSOX symbionts that make them an excellent model for studying HGT.

The pathways used by GammaSOX symbionts to oxidize reduced sulfur compounds are strikingly similar despite their high phylogenetic diversity. This indicates that there is a strong selective advantage for a symbiont to use one particular convergently evolved pathway. Until recently, the genomes of GammaSOX symbionts were thought to encode different sulfur oxidation pathways, the main difference being the presence or absence of the sox genes for thiosulfate use [13, 14•, 15••]. This appeared to explain results of early incubation experiments showing that different GammaSOX symbionts prefer different sulfur compounds [16, 17]. However, recent studies and our own analysis of the currently available genomic and transcriptomic data show that all GammaSOX symbionts have the potential to use both thiosulfate and sulfide as an energy source and all use similar pathways to do so [5, 13, 14•, 18••, 19] (Figure 1, Table S1). This is surprising because the sulfur oxidation pathways used by free-living sulfur oxidizers within the Gammaproteobacteria vary (SI Text) [20•, 21•].

The main function of chemoautotrophic symbionts is to fix inorganic carbon into organic compounds to provide for their host's nutrition [3]. All GammaSOX symbionts studied so far use the Calvin–Benson-Bassham (CBB) cycle for CO2 fixation [3] (Figure 1). The CBB cycle in the GammaSOX symbionts for which genome information is available does not correspond to the classical ‘textbook’ version, because the genes for the CBB cycle enzyme sedoheptulose-1,7-bisphosphatase are missing in all symbionts and the fructose-1,6-bisphosphatase in all except the Olavius algarvensis Gamma3 symbiont [5, 19, 22] (the Bathymodiolus sp. symbiont metagenome [2] was analyzed for this review). Instead, we recently proposed that the symbionts posses a trifunctional enzyme – a pyrophosphate-dependent 6-phosphofructokinase (PPi-PFK) [5, 19], which could not only replace the two missing enzymes but possibly also a third CBB cycle enzyme – the phosphoribulokinase [23]. We proposed that PPi-PFK acts in concert with a membrane-bound proton-translocating pyrophosphatase (HPPase) allowing energy gain from CBB cycle reactions that in the classical version of the CBB cycle do not yield energy [5] (Figure 1). The close metabolic relationship between PPi-PFK and HPPase is supported by the fact that the two genes form an operon in all GammaSOX symbionts [5] except for the Bathymodiolus sp. symbiont in which no HPPase gene was found in the immediate vicinity of the PPi-PFK gene (this review, Figure S2). The PPi-PFK phylogeny is incongruent with 16S rRNA phylogeny (Figure 2 and Figure S2) suggesting that HGT may have been involved in the acquisition of the more energy efficient CBB cycle in GammaSOX symbionts (SI Text).

Although all GammaSOX symbionts use the CBB cycle to fix CO2, they use different types of the key enzyme RubisCO [24] (Figure 3). Some use form I and some form II. These forms differ markedly in their subunit composition, affinity for the substrate CO2 and discrimination against the alternative substrate O2, which hinders CO2 fixation [25]. One explanation for the variability in the presence of form I and II RubisCO in GammaSOX symbionts is the different environments their hosts inhabit. CO2 and O2 concentrations vary greatly between chemosynthetic habitats [3•, 26, 27, 28], and it has been proposed that the symbionts posses the RubisCO form that functions best under the given environmental conditions they live in (reviewed in [29] and [30]).

More than half of the Proteobacteria known to use the CBB cycle have multiple RubisCO forms and can even have multiple distinct genes encoding a single form, enabling them to adapt to varying CO2 and O2 concentrations [25, 31] (Figure 3). By contrast, GammaSOX symbionts appear to have only one copy of either form I or form II RubisCO in their genomes, although so far, only the genomes of the vesicomyid clam symbionts have been closed. Bacteria with only one copy of RubisCO have been hypothesized to be specialists that fix CO2 at predictable and fairly constant CO2 and O2 levels [25]. Such stable conditions would be consistent with a buffered environment inside the host and might explain the presence of only one RubisCO in GammaSOX symbionts.

The RubisCO large subunit phylogeny of GammaSOX symbionts and their free-living relatives is another example for the incongruence between a metabolic gene and 16S rRNA phylogeny (Figure 2, Figure 3). This incongruence was previously observed for many free-living bacteria and was explained by multiple HGTs, gene duplications and losses of RubisCO genes (reviewed in [32]). All three factors may have played a role in shaping GammaSOX symbiont RubisCO phylogeny. For example: (1) The form II RubisCO of the Riftia/Tevnia symbiont falls into a phylogenetically distant clade from the form II RubisCOs of the Lamellibrachia spp. symbionts, although these symbionts are closely related based on their 16S rRNA genes. This suggests that one of the two symbiont groups has acquired a form II RubisCO from distantly related bacteria via HGT. Alternatively, initial possession of two form II RubisCOs and selective loss could also have led to a similar pattern. (2) The Bathmodiolus spp. and vesicomyid clam GammaSOX symbionts, which based on 16S rRNA form a monophyletic clade with their free-living relatives SUP05, and ARCTIC96BD-19, possess different forms of RubisCO. While the Bathymodiolus sp. GammaSOX symbiont has the form I RubisCO, the vesicomyid clam symbionts have form II. It was suggested that this pattern could result from the ancestor of the Bathymodiolus spp. and vesicomyid clam symbionts having both forms of RubisCO, which were later selectively lost [30]. This is a reasonable explanation considering that several free-living chemoautotrophs possess both RubisCO forms (Figure 3). However, acquisition of the respective RubisCO genes via HGT could also have played a role. Another interesting observation is the form I RubisCO from the symbionts of Bathymodiolus spp., Solemya velum, Lucinoma aff. kazani, and Oligobrachia haakonmosbiensis. Although these symbionts fall into separate phylogenetic groups based on their 16S rRNA genes, they have closely related RubisCOs that group together (Figure 2, Figure 3). How can we explain this degree of incongruence? The possession of multiple RubisCO copies in a very distant ancestor of these symbionts followed by differential gene loss is hard to imagine, while HGT would provide a much more parsimonious explanation. Given that ecological similarity shapes HGT [33••], we postulate that the symbionts of the hosts from our example above experience similar O2 and CO2 concentrations. The symbionts will not only be influenced by the concentrations of these gases in the environment but also within the host, where carriers for O2 and CO2 like blood, and enzymes that enhance CO2 fixation like carbonic anhydrase, will have a significant effect. To our knowledge, a comprehensive analysis of abiotic and biotic factors that contribute to the actual O2 and CO2 concentrations the symbionts experience has not yet been done, but would be critical for understanding the factors that drive metabolic evolution.

The Riftia/Tevnia symbiont uses an additional autotrophic pathway, the reverse TCA (rTCA) cycle, in addition to the CBB cycle. Both cycles are expressed in symbionts from the same host individual [15••, 18••, 19, 27], but it is currently unknown if both cycles are expressed in parallel in the same individual symbiont cell (SI Text).

The apparent absence of the rTCA cycle in all other sequenced GammaSOX symbionts and their free-living relatives [34] indicates that the genes for this pathway were acquired by the Riftia/Tevnia symbiont through HGT. Our BLAST search of two of the key rTCA genes (aclA and aclB) from the Riftia/Tevnia symbiont revealed best hits to genes from Alphaproteobacteria, Betaproteobacteria and Deltaproteobacteria but not to the Gammaproteobacteria. This could indicate their acquisition through HGT, although a more comprehensive phylogenetic study is needed to fully understand the origin of the rTCA cycle in the Riftia/Tevnia symbiont.

All GammaSOX symbionts have the same basic metabolism, but they diverge in many other metabolic pathways, such as those involved in nitrogen acquisition, waste recycling, carbon and energy storage and use of additional energy sources.

Thirty years after their discovery, reduced sulfur compounds were still the only known inorganic energy source for GammaSOX symbionts [1, 3•]. Recently, we discovered two additional energy sources, hydrogen and carbon monoxide, which are used by some GammaSOX symbionts (Figure 1). Hydrogen can be used by the symbionts of Bathymodiolus spp., Riftia/Tevnia and the vent shrimp Rimicaris exoculata, and carbon monoxide by the Gamma3 symbiont of O. algarvensis [2, 5]. The phylogeny of the [NiFe] hydrogenase large subunit, which is used by the GammaSOX symbionts for hydrogen uptake, is incongruent with their 16S rRNA phylogeny. For example, the hydrogenase sequences from Bathymodiolus GammaSOX symbionts fall within a group containing mostly sequences from Alphaproteobacteria and Betaproteobacteria, and this group is clearly separate to the sequences from free-living and symbiotic gammaproteobacterial sulfur oxidizers such as the Riftia/Tevnia symbiont (Figure 4). Furthermore, the hydrogenase sequences from Bathymodiolus GammaSOX symbionts and the Rimicaris exoculata symbiont are closely related, although their 16S rRNA genes are not (Figure 2, Figure 4). This provides yet another example for the possibility that HGT from phylogenetically distant species played a role in the acquisition of metabolic genes in some GammaSOX symbionts.

Some GammaSOX symbionts appear to be obligate autotrophs. This was suggested by Newton et al. for the vesicomyid clam symbionts because transporters for organic substrates are absent in their genomes and their TCA cycle is incomplete [22, 35, 36]. In particular, the lack of the key TCA cycle enzyme α-ketoglutarate dehydrogenase in the clam symbionts indicates their obligate autotrophy [37••]. By contrast, other GammaSOX symbionts have the potential to grow heterotrophically or mixotrophically. The Riftia/Tevnia symbiont has genes for a complete oxidative TCA cycle in a separate gene cluster from the rTCA genes, in addition to genes for glycolysis, fructose degradation, uptake transporters for organic substrates, and potentially for cellulose degradation [18••, 27, 38]. The Gamma3 symbiont of O. algarvensis has genes for a complete oxidative TCA cycle, a variety of uptake transporters for amino acids, peptides, urea, glycine betaine, and dicarboxylates, and for the breakdown of transported organic compounds [5, 6]. The Gamma1 symbiont of O. algarvensis has genes for a few uptake transporters for organic substrates including sugar and dicarboxylates and also for a novel pathway, a modified version of the 3-hydroxypropionate bi-cycle that allows it to assimilate and recycle waste products of anaerobic host fermentation like acetate, succinate, malate and propionate [5] (Figure 1). The discovery of a partial 3-hydroxypropionate bi-cycle in the Gamma1 symbiont was surprising, because this pathway is not present in its close free-living Chromatiaceae relatives or other GammaSOX symbionts. A BLAST search of the genes for this pathway yielded hits with Chloroflexaceae for some genes and Clostridia for others, providing circumstantial evidence that this pathway might have been acquired by the Gamma1 symbiont through HGT from several unrelated groups of bacteria.

The GammaSOX symbionts also differ in their use of electron acceptors. Most GammaSOX symbionts use oxygen as the primary electron acceptor [3], but some symbionts may also use nitrate as a terminal electron acceptor (reviewed in [3]). Nitrate is readily available in the deep-sea as well as in some coastal sediments, and some siboglinid hosts store nitrate in their blood [39]. In recent years, sequencing of GammaSOX symbiont genomes has revealed genes for nitrate respiration in several but not all GammaSOX symbionts (Figure 1).

A clear case for the recent loss of respiratory nitrate genes was found in vesicomyid clam symbionts. While Candidatus Vesicomyosocius okutanii has a membrane-bound nitrate reductase to respire nitrate to nitrite [35], genes for this enzyme are not present in Candidatus Ruthia magnifica [36] (Figure 1). The loss of the respiratory nitrate reductase in Ca. R. magnifica cannot be easily explained by their environmental conditions. Nitrate concentrations measured in the habitat of Ca. R. magnifica would be sufficient for nitrate respiration by free-living bacteria, although the symbionts may need higher concentrations given their intracellular location. Nitrate concentrations in Ca. V. okutanii's habitat have not yet been measured. [28, 36]. Furthermore, oxygen availability may be more important in determining whether the selective advantage of keeping genes for nitrate respiration outweighs their cost; if oxygen is readily available in the environment, then competition with the host for oxygen may not provide a strong selective pressure to keep respiratory nitrate reduction genes.

Intriguingly, while Ca. V. okutanii has a respiratory nitrate reductase, it lacks an assimilatory nitrate reductase (nasA). This suggests that Ca. V. okutanii uses the respiratory nitrate reductase for both respiration and assimilation. By contrast, Ca. R. magnifica, which lacks respiratory nitrate reductase genes has the nasA gene. Both symbionts can thus reduce nitrate to nitrite, which can then be further reduced by the assimilatory nitrite reductase (nasD), which both symbionts possess (Figure 1). The presence of exactly one pathway for ammonium production from nitrate in each symbiont leads us to hypothesize that the superfluous nitrate reductase genes were differentially lost due to the ongoing genome reduction in the vesicomyid clam symbionts.

Section snippets

Conclusions

Phylogenies of metabolic genes often do not yield useful information about organismal relationships as they are strongly influenced by HGT, gene loss and gene duplications; however, they can provide valuable information about the lifestyle, physiology and specific adaptations of symbionts to their habitats and hosts. We should therefore mine our metabolic gene trees to discern broader ecological and evolutionary patterns. This approach is not new [40], but the recent increase in genomic

References and recommended reading

Papers of particular interest, published within the period of review, have been highlighted as:

  • • of special interest

  • •• of outstanding interest

Acknowledgements

Funding was provided by the Max Planck Society, the German Research Foundation Cluster of Excellence at MARUM, Bremen, and a PhD scholarship from the Studienstiftung des deutschen Volkes to M. Kleiner.

References (44)

  • H. Kuwahara et al.

    Reduced genome of the thioautotrophic intracellular symbiont in a deep-sea clam, Calyptogena okutanii

    Curr Biol

    (2007)
  • K. Katoh et al.

    MAFFT: a novel method for rapid multiple sequence alignment based on fast Fourier transform

    Nucleic Acids Res

    (2002)
  • N. Dubilier et al.

    Symbiotic diversity in marine animals: the art of harnessing chemosynthesis

    Nat Rev Microbiol

    (2008)
  • J.M. Petersen et al.

    Hydrogen is an energy source for hydrothermal vent symbioses

    Nature

    (2011)
  • C.M. Cavanaugh et al.

    Marine chemosynthetic symbioses

  • H.R. Gruber-Vodicka et al.

    Paracatenula, an ancient symbiosis between thiotrophic Alphaproteobacteria and catenulid flatworms

    Proc Natl Acad Sci USA

    (2011)
  • M. Kleiner et al.

    Metaproteomics of a gutless marine worm and its symbiotic microbial community reveal unusual pathways for carbon and energy use

    Proc Natl Acad Sci USA

    (2012)
  • T. Woyke et al.

    Symbiosis insights through metagenomic analysis of a microbial consortium

    Nature

    (2006)
  • J.M. Petersen et al.

    Methanotrophic symbioses in marine invertebrates

    Environ Microbiol Rep

    (2009)
  • J.M. Petersen et al.

    Origins and evolutionary flexibility of chemosynthetic symbionts from deep-sea animals

    Biol Bull

    (2012)
  • J.J. Grzymski et al.

    Metagenome analysis of an extreme microbial symbiosis reveals eurythermal adaptation and metabolic flexibility

    Proc Natl Acad Sci USA

    (2008)
  • J.M. Petersen et al.

    Dual symbiosis of the vent shrimp Rimicaris exoculata with filamentous gamma- and epsilonproteobacteria at four Mid-Atlantic Ridge hydrothermal vent fields

    Environ Microbiol

    (2010)
  • R.T. Papke et al.

    How bacterial lineages emerge

    Science

    (2012)
  • B.J. Shapiro et al.

    Population genomics of early events in the ecological differentiation of bacteria

    Science

    (2012)
  • M. Harada et al.

    Expression of genes for sulfur oxidation in the intracellular chemoautotrophic symbiont of the deep-sea bivalve Calyptogena okutanii

    Extremophiles

    (2009)
  • F. Stewart et al.

    Metatranscriptomic analysis of sulfur oxidation genes in the endosymbiont of Solemya velum

    Front Microbiol

    (2011)
  • S. Markert et al.

    Physiological proteomics of the uncultured endosymbiont of Riftia pachyptila

    Science

    (2007)
  • D.B. Wilmot et al.

    The bacterial symbiont from the hydrothermal vent tubeworm Riftia pachyptila is a sulfide specialist

    Mar Biol

    (1990)
  • S. Belkin et al.

    Symbiotic assimilation of CO2 in two hydrothermal vent animals, the mussel Bathymodiolus thermophilus and the tube worm Riftia pachyptila

    Biol Bull

    (1986)
  • J.C. Robidart et al.

    Linking hydrothermal geochemistry to organismal physiology: physiological versatility in Riftia pachyptila from sedimented and basalt-hosted vents

    PLoS ONE

    (2011)
  • S. Markert et al.

    Status quo in physiological proteomics of the uncultured Riftia pachyptila endosymbiont

    Proteomics

    (2011)
  • C. Dahl et al.

    Sulfur oxidation in prokaryotes

    eLS

    (2008)
  • Cited by (43)

    View all citing articles on Scopus
    View full text