MAG binning and general features
From the four hot springs, we assembled four associated metagenomes and then binned a total of 42 MAGs. We recovered 12 MAGs from RB10 hot spring, 13 from RB13, 14 from RB32 and 3 from RB108. Out of these 42 MAGs, 7 were of high-quality, 25 of nearly-high quality, 9 of medium quality and 1 of low quality (Table 1) from metagenomics standards [25]. The GC% was quite variable, ranging from 25.76% to 70.35% among all MAGs and between 32.15% and 69.21% only among the high- and near high-quality MAGs. With the exception of RB108 from which we only recovered bacterial MAGs, we retrieved both bacterial and archaeal MAGs in the other hot springs. Two thirds of the MAGs (26/42) were assigned to the domain Bacteria and the rest to the domain Archaea (Table 2). Accession numbers of the 4 metagenomes and 42 MAGs are given in Table 3.
Taxonomic and phylogenomic analyses of MAGs
The taxonomic affiliation of the MAGs was investigated in details through GTDB-Tk (release 95) (Table 2) and through phylogenomic analyses (Figure S1 A-I). We also tried to classify MAGs on the basis of overall genome relatedness indices (OGRI), which is detailed in supplementary material (Text S1, Table S2, Figure S2).
For Bacteria, GTDB-Tk analyses allowed us to place the MAGs in the following clades: six in the phylum Aquificota from the four different springs, comprising four MAGs belonging to the genus Hydrogenivirga (family Aquificaceae) (RB10-MAG07, RB13-MAG10, RB32-MAG07, RB108-MAG02), and two belonging to the family ‘Hydrogenobaculaceae’ (RB10-MAG12, RB32-MAG11) (Table 2, Figure S1A). Their closest cultured relatives originated either from hot springs or from deep-sea hydrothermal vents [26]. Three MAGs from three geothermal springs belonged to the phylum Armatimonadota (RB10-MAG03, RB13-MAG04, RB32-MAG03) and had no close cultured relatives. Seven MAGs have been classified into the phylum Chloroflexota: three MAGs belonging to the genus Thermoflexus from three different springs (RB10-MAG04, RB13-MAG05, RB32-MAG02), one affiliating with the genus Thermomicrobium (RB32-MAG08), one falling into the family Ktedonobacteraceae (RB108-MAG03) and one belonging to the class Dehalococcoidia (RB32-MAG04). Six MAGs from four various hot springs belonged to the phylum Deinococcota, and to the genera Thermus (RB10-MAG08, RB10-MAG11, RB13-MAG09, RB32-MAG10, RB108-MAG01) and Meiothermus (RB13-MAG13). One MAG belonged to the family ‘Sulfurifustaceae’ (RB13-MAG01), in the phylum Proteobacteria (Gamma- class). The MAG referenced as RB32-MAG13 was classified into the phylum ‘Patescibacteria’, in the class ‘Paceibacteria’, and was distantly related to MAGs originating from groundwater and from hot springs. Finally, two MAGs from two different springs belonged to the phylum WOR-3, in the Candidatus genus ‘Caldipriscus’ (RB32-MAG12, RB10-MAG09).
For Archaea, almost all the MAGs reconstructed in this study, e.g. 15 of the 16 archaeal MAGs, belonged to the phylum Thermoproteota. Among them, four belonged to the genus Ignisphaera (RB10-MAG05, RB13-MAG08, RB13-MAG11, RB32-MAG05), three to the genus Thermofilum (RB10-MAG06, RB13-MAG03, RB32-MAG09), two to the genus Zestosphaera (RB10-MAG02, RB13-MAG06), three to the family Acidilobaceae (RB10-MAG01, RB13-MAG02, RB32-MAG01) and two to the class Thermoprotea (RB10-MAG10, RB32-MAG06). Additionally, one belonged to the order Thermoproteales (RB13-MAG07). Lastly, the MAG belonging to another phylum (RB13-MAG12) was affiliated with the ‘Aenigmatarchaeota’, class ‘Aenigmatarchaeia’, and was distantly related to MAGs from hot springs and from deep-sea hydrothermal vent sediments [27,28].
Out of these 42 MAGs, 23 corresponded to different taxa at the taxonomic rank of species or higher. Eighteen of them belonged to lineages with several cultivated representatives and were distributed respectively, into 1 known species called Thermus thermophilus, 10 new genomic species within the genera Zestosphaera, Thermoflexus, Ignisphaera (× 2), Thermofilum, Hydrogenivirga, Thermus, Meiothermus, Caldipriscus and Thermomicrobium, 5 putative new genera belonging to the families Acidilobaceae, ‘Thermocladiaceae’, ‘Sulfurifustaceae’, ‘Hydrogenobaculaceae’ and Ktedonobacteraceae, and 2 putative new orders within the classes Dehalococcoidia and Thermoprotea. In addition, five MAGs belonged to lineages that are predominantly or exclusively known through environmental DNA sequences. They were classified as 1 new genomic species in the phylum Armatimonadota, 2 putative new genera in the classes ‘Aenigmatarcheia’ and ‘Paceibacteria’, and 1 putative new family in the phylum Chloroflexota. Thus, these 42 MAGs comprised a broad phylogenetic range of Bacteria and Archaea at different levels of taxonomic organization, of which a large majority were new.
The approaches implemented here were not intended to describe the microbial diversity present in these sources in an exhaustive way and to compare them in a fine way, and do not allow it because of storage. However, they do provide an overview of the microbial diversity effectively present. If we compare the phylogenetic diversity of the MAGs found in the 4 hot springs, we can observe that 3 shared phyla (Deinococcota, Aquificota and Chloroflexota: phyla names according to GTDB), 2 shared families (Thermaceae and Aquificaceae), and 2 shared genera (Hydrogenivirga and Thermus) were found among the four sources (Figure 2). In addition, hot springs RB10, RB13 and RB32, that are geographically close (< 60 m), also share 2 other phyla (Thermoproteota and Armatimonadota) and 5 other families in common (Acidilobaceae, Ignisphaeraceae, Thermofilaceae, Thermoflexaceae, and HRBIN17) (Figure 2). These phyla and families that are shared between sources are widespread lineages in terrestrial geothermal habitats (e.g. [4-6,12]). Phyla and families detected in the hot environments of Antarctica are also found here, such as Patescibacteria for example [15].
In summary, this metagenomic analysis highlighted the presence of bacterial and archaeal lineages commonly found in hot springs, and lineages found in hot habitats from polar areas (e.g. [4-6,15,29]). The microbial communities in these Kerguelen Islands hot springs were diverse, particularly in RB10, RB13, and RB32 hot springs. However, within these lineages that have been previously reported to occur in geothermal environments, a majority of the genomic taxa detected here were new, sometimes at a high taxonomic rank.
Metabolic potential of MAGs
An extensive genomic characterization of the 42 MAGs has been performed to explore the metabolic pathways and the possible adaptations of the microbial populations from which these MAGs originate.
KEGG Decoder visualization highlighted various pathways associated with carbohydrate degradation, oxidative phosphorylation and sulfur, nitrogen, and amino-acid metabolisms, among others (Figure 3).
To confirm these initial metabolic predictions, further annotation was performed by combining data generated by Prokka with the MetaCyc database. Efforts have been directed at studying catabolic pathways, particularly those involving inorganic electron donors and acceptors. These results are not representative of the metabolic diversity of all the hot spring ecosystems studied, but they do reflect some of the microbial catabolisms likely to be used in situ to produce energy and, by assumption, the most abundant ones. Metabolic predictions are presented hereafter, at different taxonomic ranks.
MAGs belonging to the genus Thermoflexus (RB10-MAG04, RB13-MAG05, RB32-MAG02) encode pathways for carbon monoxide oxidation (via aerobic carbon monoxide dehydrogenase), hydrogen oxidation and nitrate respiration; However, the only cultivated known representative of this genus is a heterotrophic bacterium [30]. The Dehalococcoidia’s MAG (RB32-MAG04) encodes only a carbon monoxide oxidation pathway, whereas the genus consists of strict anaerobic hydrogenotrophic, organohalide-respiring bacteria [31]. In the MAG associated with the genus Thermomicrobium (RB32-MAG08), we predicted pathways for dimethylsulfide degradation, thiosulfate disproportionation and carbon monoxide oxidation, whereas only carboxydotrophic growth has been reported in this genus and demonstrated by culture [32]. In the Chloroflexota’s RB32-MAG14, carbon monoxide oxidation and thiosulfate disproportionation pathways are present but no CDS associated with phototrophy, which may suggest a chemoorganotrophy mode of energy production [33]. The Ktedonobacteraceae’s MAG (RB108-MAG03) encodes enzymes of four pathways of hydrogen oxidation (aerobic), carbon monoxide oxidation, dimethylsulfide degradation, selenate reduction, thiosulfate oxidation and disproportionation and finally tetrathionate oxidation; yet, the few taxa of this family isolated so far are mesophilic heterotrophic bacteria [34]. Within Hydrogenobaculaceae MAGs (RB10-MAG12, RB32-MAG11), we predicted a thiosulfate disproportionation pathway; most of the species within this family are capable of chemolithotrophic microaerophilic or anaerobic growth [35]. MAGs belonging to the genus Hydrogenivirga (RB10-MAG07, RB13-MAG10, RB32-MAG07, RB108-MAG02) possess genes encoding enzymes of aerobic respiration, thiosulfate oxidation, thiosulfate disproportionation, tetrathionate reduction, and hydrogen oxidation (aerobic and anaerobic), which is consistent with what is known about the genus (nitrate and molecular oxygen respiration combined to hydrogen, sulfur, or thiosulfate oxidation) [35]. In MAGs associated with the genus Thermus (RB10-MAG08, RB10-MAG11, RB13-MAG09, RB32-MAG10, RB108-MAG01), we predicted pathways for aerobic respiration, assimilatory sulfate reduction, hydrogen oxidation, selenate reduction, thiosulfate oxidation and thiosulfate disproportionation; cultivated species of this genus grow mainly chemoorganoheterotrophically by aerobic respiration, but some have their genome coding for chemolithotrophic and anaerobic respiration enzymes [36]. The MAG belonging to the genus Meiothermus (RB13-MAG13) encodes pathways for carbon monoxide oxidation, hydrogen oxidation, thiosulfate oxidation and thiosulfate disproportionation; in the current state of knowledge, the growth of Meiothermus strains is based on chemoorganotrophy and oxygen or nitrate respiration [37]. For the RB13-MAG01 belonging to the Sulfurifustaceae, we predicted the genetic potential for aerobic respiration, ammonia oxidation, dissimilatory sulfate reduction, sulfite oxidation, sulfide oxidation (to sulfur globules), tetrathionate reduction, thiosulfate oxidation and thiosulfate disproportionation; Sulfurifustaceae (referenced as Acidiferrobacteraceae in the LPSN taxonomy) are known to be able to oxidize sulfur and iron, and the microorganism corresponding to this MAG may possess a larger panel of chemolithotrophic abilities [38]. For Armatimonadota’s members (RB10-MAG03, RB13-MAG04, RB32-MAG03), we predicted pathways for assimilatory sulfate reduction, carbon monoxide oxidation, selenate reduction and thiosulfate disproportionation; the members of the phylum are known as aerobic heterotrophs [39]. In Zestosphaera’s and Ignisphaera’s MAGs (RB10-MAG02, RB13-MAG06) (RB10-MAG05, RB13-MAG08, RB13-MAG11, RB32-MAG05), we predicted sulfur reduction (sulfur and polysulfides) pathways; those MAGs classified in the Desulfurococcaceae in the LPSN taxonomy, are known as heterotrophs respiring sulfur species [40,41]. MAGs belonging to the class Thermoproteia (RB10-MAG10, RB13-MAG07, RB32-MAG06), encode pathway for dissimilatory sulfate reduction; various catabolic pathways are described in this class [42]. In MAGs related to the genus Caldipriscus (RB10-MAG09, RB32-MAG12), phylum Patescibacteria (RB32-MAG13), family Acidilobaceae (RB10-MAG01, RB13-MAG02, RB32-MAG01), family Thermofilaceae (RB10-MAG06, RB13-MAG03, RB32-MAG09) and class Aenigmatarchaeia (RB13-MAG12), we did not predict any catabolic pathway of inorganic nutrients among those reported in the MetaCyc database. This could be explained by the low completion of the MAGs and/or the fact that only well-known pathways are documented in this database. However, all these MAGs have pathways associated with carbohydrate and protein degradation. This may indicate that these taxa are chemoheterotrophs, which has already been reported in geothermal environments and already described for relatives of some of these taxa [43,44].
Sulfide oxidation may be a possible energy production pathway for 28 MAGs based on KEGG Decoder (Figure 3), since they code for a sulfide:quinone oxidoreductase (K17218) and a flavoprotein chain of sulfide dehydrogenase (K17229), but this hypothesis was not confirmed by MetaCyc except for RB13-MAG01. Due to high representations of sulfur metabolisms, genes encoded in MAGs were evaluated with DiSCo, which gave similar results to those obtained when analyzed with Pathway tools. DiSCo confirmed complete dissimilatory sulfate reduction pathways for two MAGs, predicted for sulfate reduction process (RB13-MAG07) or sulfide oxidation process by reverse sulfate reduction pathway (RB13-MAG01). The assimilatory sulfate reduction pathway is more represented in the overall dataset formed by all MAGs than the dissimilatory pathway, which is consistent with the low sulfate concentration measured in the four hot springs (Table S1). The thiosulfate disproportionation pathway predicted by MetaCyc in many MAGs simply refers to the detection of an enzyme, the rhodanese-type thiosulfate sulfurtransferase. However, in the current state of knowledge on the disproportionation pathways of inorganic sulfur compounds [45,46], this enzyme alone does not allow the implementation of this catabolic pathway. If we consider all the genes present in these MAGs, nothing indicates that the microorganisms from which these MAGs originate can achieve the disproportionation of inorganic sulfur compounds.
No enzymes clearly associated with photosystems I and II were found. Nevertheless, it cannot be ruled out that these energy production pathways are absent in microorganisms indigenous to these sources. On the other hand, our results show that these sources host chemolithoautotrophic taxa involved in the carbon and sulfur cycle, and to a lesser extent in the hydrogen and nitrogen cycles. Several taxa are likely to be involved in the primary production of these sources through chemolithoautotrophy, but in addition, heterotrophs appear to be very present and diverse in the collected samples. However, additional studies will be required to better apprehend the metabolic diversity, the trophic webs of these hot springs and their microbial actors, and to better understand the functioning of the microbial communities of the Kerguelen hot springs and their interactions with their biotic and abiotic environment.
In conclusion, this first metagenomic overview of the microbial diversity of Kerguelen hot springs allowed the assembly of 42 MAGs, from 4 hot springs, many of which correspond to putative new taxa, namely 11 new genomic species, 7 new putative genera, 1 new putative family and 2 new putative orders affiliated to Bacteria and Archaea. Based on their genetic potential, these taxa appear to be chemolithoautotrophs and chemoheterotrophs and thus probably involved in the carbon, sulfur, hydrogen and nitrogen cycle. As geographically isolated sites, the Kerguelen Islands are reservoirs of diversity and taxa of novel microorganisms that should be interesting to study the evolution of microbial life and speciation processes. It has been difficult to fully assess the microbial metabolic diversity in these geothermal pools due to the inherent limitations of MAG reconstruction and the state of knowledge of microbial pathways that remains limited. However, these geothermal ecosystems could be reservoirs of novel metabolic pathways, physiological properties and adaptive mechanisms and should be examined in detail through further and broader metagenomic studies and cultural approaches.