Vitamin B12 Auxotrophy in Isolates from the Deep Subsurface of the Iberian Pyrite Belt

Vitamin B12 is an enzymatic cofactor that is essential for both eukaryotes and prokaryotes. The development of life in extreme environments depends on cofactors such as vitamin B12 as well. The genomes of twelve microorganisms isolated from the deep subsurface of the Iberian Pyrite Belt have been analyzed in search of enzymatic activities that require vitamin B12 or are involved in its synthesis and import. Results have revealed that vitamin B12 is needed by these microorganisms for several essential enzymes such as ribonucleotide reductase, methionine synthase and epoxyqueosine reductase. Isolate Desulfosporosinus sp. DEEP is the only analyzed genome that holds a set core of proteins that could lead to the production of vitamin B12. The rest are dependent on obtaining it from the subsurface oligotrophic environment in which they grow. Sought proteins involved in the import of vitamin B12 are not widespread in the sample. The dependence found in the genomes of these microorganisms is supported by the production of vitamin B12 by microorganisms such as Desulfosporosinus sp. DEEP, showing that the operation of deep subsurface biogeochemical cycles is dependent on cofactors such as vitamin B12.


Introduction
The deep subsurface is a poly-extreme environment inhabited by microorganisms, [1,2] where they have to withstand several extreme factors, such as high pressure, low nutrient concentration, high temperatures and low water activity [3]. Additionally, and in contrast with extreme environments on the surface, biogeochemical cycles operate without photosynthesis as a source of organic matter [2]. Thus, biogeochemical cycles depend on chemosynthetic organisms as primary producers, geochemically and biologically produced H 2 [4,5] and the presence of electron acceptors for anaerobic respiration due to the existing anoxic conditions. To colonize these very extreme niches, microorganisms must cooperate in order to be able to use the few available resources to thrive [1,4,6]. The deep subsurface of the Iberian Pyrite Belt (IPB) has been thoroughly studied via several drilling projects [7][8][9], which have revealed the existence of an underground bioreactor. This bioreactor employs inorganic compounds such as Fe 3+ to dissolve metal sulfides such as pyrite (FeS 2 ) and increase nutrient availability in the deep subsurface. Nitrogen, carbon, iron, hydrogen and sulfur biogeochemical cycles have been confirmed throughout 600 m of rock [10]. These studies also allowed the isolation and sequencing of several microorganisms native to the subsurface and the study of their genomes and metabolic capabilities in silico [11][12][13][14][15] and in vitro [10,11].
Enzymes are often in need of cofactors such as vitamin B12 (Figure 1) that allow chemical reactions to flow. Although these are needed in very small quantities, they are essential for the organism. Cobalamin, also known as vitamin B 12 , is an essential cofactor for life, involved in processes such as DNA methylation and methionine or ribonucleotide for life, involved in processes such as DNA methylation and methionine or ribonucleotide synthesis. Vitamin B12 is a tetrapyrrole with a covalently linked cobalt atom and belongs to the cobamide group, also known as corrinoids. Additionally, these molecules have an upper ligand and a nucleotide loop with a lower ligand. These two positions differ between the different types of cobamides, and vitamin B12 is characterized by a 5,6-dimethylbenzimidazole (DMB) as the lower ligand. However, vitamin B12 can have different upper ligands, such as a methyl group (methylcobalamin), a hydroxyl group (hydroxocobalamin), a cyano group (cyanocobalamin) or a 5′-deoxyadenosyl molecule (adenosylcobalamin; Figure 1). In addition, its synthesis is only feasible in a limited variety of archaea and bacteria. As such, both prokaryotes and eukaryotes are heavily dependent on its synthesis by microbial producers for their survival [16]. Prokaryotes produce B12 with either a methyl group or with 5′-desoxyadenosyl, while the hydroxyl group is present when the natural forms are exposed to light and the cyano group in synthetic form is produced as a pharmaceutical complement [17]. The complete pathway for the synthesis of vitamin B12 ranges between 21 and 30 different reactions, making it a very demanding pathway in terms of energy and evolutionary cost. Its synthesis can take place in the presence or in the absence of oxygen [18]. Therefore, most microorganisms depend on the retrieval of this cofactor from the medium for survival [19]. This is known as the salvage pathway, and it requires a significantly smaller number of proteins and allows microorganisms that cannot synthesize vitamin B12 to capture intermediaries from the extracellular medium and modify them into adenosylcobalamin [17].
To the best of our knowledge, the importance of vitamin B12 in the deep subsurface has not yet been described. The purpose of this study is to analyze 12 genomes which had been previously sequenced to study the ability of these microorganisms to synthesize vitamin B12 and to try to infer from this data the relevance of vitamin B12 in the deep subsurface of the IPB. The complete pathway for the synthesis of vitamin B 12 ranges between 21 and 30 different reactions, making it a very demanding pathway in terms of energy and evolutionary cost. Its synthesis can take place in the presence or in the absence of oxygen [18]. Therefore, most microorganisms depend on the retrieval of this cofactor from the medium for survival [19]. This is known as the salvage pathway, and it requires a significantly smaller number of proteins and allows microorganisms that cannot synthesize vitamin B 12 to capture intermediaries from the extracellular medium and modify them into adenosylcobalamin [17].
To the best of our knowledge, the importance of vitamin B 12 in the deep subsurface has not yet been described. The purpose of this study is to analyze 12 genomes which had been previously sequenced to study the ability of these microorganisms to synthesize vitamin B 12 and to try to infer from this data the relevance of vitamin B 12 in the deep subsurface of the IPB.

Genomes for In Silico Analysis
The genomes used in this study belong to twelve microorganisms isolated from the deep subsurface of the IPB (Table 1). These microorganisms were isolated using selective anaerobic media with rocks from the IPB subsurface as inoculum. The taxonomic identification of each isolate was achieved through the sequencing of the 16S rRNA gene [10,20]. Depth detection of each isolate is displayed in Supplementary Table S7. Isolation and subsequent growth of isolates were performed in an anaerobic atmosphere and in the absence of light.

Detection of Cobamide Biosynthesis and Dependence Genes in Genomes
To identify proteins and enzymes which are involved in each step of cobamide biosynthesis or that are cobamide-dependent or independent (Figure 2), we first built a database with several amino acid sequences for each of these proteins retrieved from UniProt (Supplementary Table S1). Then we queried (blastp) this database against the proteins predicted for the twelve aforementioned microorganisms (Table 1). We considered as positive those results which had a percent identity higher than 35%, a query cover higher than 50% and an E-value smaller than 10 −5 . Genes that had either the query cover or the percent identity above the threshold but did not meet the requirement on the other were compared with blastp against the UniProt Reference proteomes + Swiss-Prot database with the UniProt Blast suite [21].

Domain Detection and Structure Comparison between BtuB Transporters
Evaluated proteins are the putative BtuB transporters detected in this article and BtuB annotated proteins from Shewanella putrefaciens T2.3D-1.1 For the predictions of domains in each protein, we used the InterPro tool [22] with standard parameters (Supplementary Table S2). For the prediction of the structure of the putative BtuB transporters we used the algorithm AlphaFold 2 available inside the Galaxy European server [23] with standard parameters and the full database. The predicted structures were compared to the crystal structure of the BtuB transporter 1NQE and to the crystal structure of the FecA transporter (1KMO) from the PDB database [24]. Pairwise alignment was made through the FATCAT (Flexible structure AlignmenT by Chaining Aligned fragment pairs allowing Twists) online server using the flexible alignment mode with standard parameters [25].

Domain Detection and Structure Comparison between BtuB Transporters
Evaluated proteins are the putative BtuB transporters detected in this article and BtuB annotated proteins from Shewanella putrefaciens T2.3D-1.1 For the predictions of domains in each protein, we used the InterPro tool [22] with standard parameters (Supplementary Table S2). For the prediction of the structure of the putative BtuB transporters we used the algorithm AlphaFold 2 available inside the Galaxy European server [23] with standard parameters and the full database. The predicted structures were compared to the crystal structure of the BtuB transporter 1NQE and to the crystal structure of the FecA transporter (1KMO) from the PDB database [24]. Pairwise alignment was made through the FATCAT (Flexible structure AlignmenT by Chaining Aligned fragment pairs allowing Twists) online server using the flexible alignment mode with standard parameters [25].

B 12 Dependence in the IPB Subsurface
Enzymes that have been reported to depend on vitamin B 12 to process their substrates are compared against the database that contains all the proteins from the 12 isolates from the IPB. In similar fashion, those vitamin B 12 -dependent proteins that have vitamin B 12 -independent counterparts or alternative enzymes are also compared to our database ( Table 2). In this way we seek to determine which microorganisms are auxotrophs for cobalamin. Expanded results for proteins dependent on B 12 can be found in Supplementary  Table S3. Additionally, expanded results for B 12 -independent counterparts proteins can be found in Supplementary Table S4. The vitamin B 12 -dependent methionine synthase MetH ( Figure 2A) is present in 11 out of the 12 isolates under study. Nevertheless, 6 out of the 12 also hold a vitamin B 12independent methionine synthase MetE ( Figure 2B). Only Desulfosporosinus sp. DEEP has not returned a match for this enzyme. Hence, Aestuariimicrobium sp. T2.26MG-19.2, Ciceribacter sp. T2.26MG-112.2, Ciceribacter sp. T2.30D-1.1, Rhodoplanes sp. T2.26MG-98 and Tessaracoccus sp. T2.5-30 would be dependent on vitamin B 12 for the synthesis of methionine, an essential amino acid.
In the case of the vitamin B 12 -dependent ribonucleotide reductase NrdJ (Figure 2A), 10 out of the 12 isolates have an annotated copy in their genomes (Table 2). We also searched for B 12 -independent ribonucleotide reductase NrdAB enzymes ( Figure 2B) annotated in their genomes and found that 8 out of the 12 isolates had this enzyme ( Table 2).
The vitamin B 12 -dependent methylmalonyl-CoA mutase BhbA enzyme ( Figure 2A) is also present in 6 out of the 12 genomes (Table 2). To the best of our knowledge there is no vitamin B 12 -independent ortholog for this enzyme.
The vitamin B 12 -dependent epoxyqueosine reductase QueG (Figure 2A) was found in 8 of our isolates (Table 2). Additionally, we have not obtained positive results for the vitamin B 12 -independent epoxyqueosine reductase QueH ( Figure 2B; Table 2).
The vitamin B 12 -dependent glutamate mutase GlmS (Figure 2A) was only found in Desulfosporosinus sp. DEEP (Table 2). To the best of our knowledge, there is no vitamin B 12independent glutamate mutase, but there are several enzymes involved in the degradation of glutamate that can also use glutamate as substrate. We have looked for two different B 12 -independent proteins that can use glutamate as a substrate, GdhA and AspC. The glutamate dehydrogenase GdhA ( Figure 2B) has been detected in all of the genomes except for those of Desulfosporosinus sp. DEEP and Microbacterium sp. T2.11-28 ( Table 2). The aspartate transaminase AspC ( Figure 2B) is common to all the assayed genomes except for that of Aestuariimicrobium sp. T2.26MG-19.2 (Table 2). Therefore, none of our isolates would depend entirely on vitamin B 12 for the degradation of glutamate.
Another vitamin B 12 -dependent enzyme that has returned a positive match in 3 isolates (Table 2) is the ethanolamine ammonia lyase EutBC (Figure 2A). There is an alternative pathway for ethanolamine consumption that is not dependent on vitamin B 12 , composed of 7 different genes (Csal_0675 to Csal_0681) [26]. The two isolates from the genus Ciceribacter present the complete set of proteins involved in the utilization of ethanolamine without the need for vitamin B 12 (Table 2). For AcsCDE, a vitamin B 12 -dependent methyltransferase (Figure 2A), we have not found a B 12 -independent counterpart. Nevertheless, this enzyme is present in 4 of the isolate's genomes ( Table 2).
The last vitamin B 12 -dependent enzyme to return a positive match is BchE (Figure 2A). It is involved in the synthesis of chlorophyll and bacteriochlorophyll, which only returned a positive match in Rhodoplanes sp. T2.26MG-98 ( Table 2). The vitamin B 12 -independent counterpart AcsF ( Figure 2B) also returned a positive match in Rhodoplanes sp. T2.26MG-98 (Table 2).
From our set of B 12 -independent proteins, 5 have not returned positive matches in any of the genomes. These include the 2-hydroxybutanoil-CoA mutase HcmB, the 2methyleneglutarate mutase Mgm, the glycerol/diol dehydratase PduCDE, the reductive dehalogenase PceA and the mercury methyltransferase HgcAB (Table 2; Figure 2A).

B 12 Production in the IPB Subsurface
According to the last section, there are several subsurface isolates that depend on vitamin B 12 for very important enzymatic reactions. We looked for the complete array of proteins needed for the synthesis of adenosylcobalamin [16,17] across all our isolates (Supplementary Table S5). The first step involves the synthesis of uroporphyrinogen III starting from glycine and succinyl-CoA or glutamate ( Figure 2C). From the sample, 4 genomes have all the necessary proteins for the synthesis of this intermediary from either glutamate or glycine and succinyl-CoA. While Ciceribacter sp. T2.30D-1.1 has the necessary proteins to start from succinyl-CoA and glycine, it lacks those needed to start from glutamate (Supplementary Table S5).
For the aerobic/anaerobic pathways for the synthesis of adenosylcobyrinic acid ( Figure 2C), our results have not found any of the genomes to contain all the proteins ascribed to either of the options (Supplementary Table S5). The best results in the aerobic pathway are 10 proteins out of the total 12 proteins for Tessaracoccus sp. T2.5-30. The maximum detected for the anaerobic pathway is 10 of the 11 proteins of the anaerobic pathway, for both Desulfosporosinus sp. DEEP and Tessaracoccus sp. T2.5-30. Even so, a recurring pattern that emerged across this study is that protein IDs from our genomes would appear in both the aerobic and anaerobic pathways for different proteins, hence WP_154722398.1 7 of 14 from Ciceribacter sp. T2.26MG-112.2 appears as CobA and CysG (Supplementary Table S5). As a result, we have been unable to map some protein IDs to enzymes involved in either the aerobic or anaerobic pathways of synthesis of adenosylcobrynic acid.
As for the central pathway, Stutzerimonas sp. T2.31D-1 is the only isolate to have returned positive results from all the involved proteins (Supplementary Table S5).
Since most of our sampled genomes lack many of the proteins for the synthesis of vitamin B 12 , it was deemed important to compare our genomes against known importers for the salvage of vitamin B 12 . The uptake system of BtuBCDF to salvage vitamin B 12 is only complete in the S. putrefaciens T2.3D-1.1 genome. Aside from the BtuCDF and BtuB system, we also looked for alternative vitamin B 12 importers BtuM [27], Rv1819c [28] and CbrT [29]. BtuM and Rv1819c returned positive matches in Stutzerimonas sp. T2.31D-1 and Rhodoplanes sp. T2.26MG-98 respectively. On the other hand, CbrT was not found in any of the assayed genomes (Supplementary Table S5).
Although it has been stated previously that the S. putrefaciens T2.3D-1.1 genome contains 13 putative BtuB transporters [15], in this work, just 2 different BtuB transporters were detected (CAD6364477.1 and CAD6365988.1). The domain structure of these 13 putative BtuB copies and putative BtuB proteins found in this work (CAD6364477. To better understand similarities and differences between annotated BtuB proteins and BtuB proteins found in this article, we predicted the structure of the 13 BtuB annotated copies of S. putrefaciens T2.3D-1.1 [15], plus all the BtuB proteins that were been found in this work (Supplementary Table S6). Using the predicted structures, we have compared them through pairwise alignment with the Escherichia coli BtuB structure from the PDB (1NQE) and with another TonB-dependent receptor (1KMO) as a control (Supplementary  Table S6). BtuB proteins from this work (WP_048328632.1, WP_008262044.1, CAD6365988.1 and CAD6364477.1) have low p-values, attributed to significantly similar proteins, and high FATCAT scores and no twists when compared to 1NQE. When compared to 1KMO the FATCAT scores are lower but are also considered significant similar proteins. Out of these proteins, CAD6365988.1 from S. putrefaciens T2.3D-1.1 and WP_008262044.1 from Brevundimonas sp. T2.26MG-9 hold the best results with 1NQE ( Figure 3) and their comparison with the control 1KMO yielded lower scores (Supplementary Table S6). Even though some of the 13 annotated proteins from S. putrefaciens T2.3D-1.1 (CAD6367072.1 and CAD6364935.1) have high FATCAT scores with 1NQE, they still have lower scores than the BtuB proteins found in this work. Additionally, some of these annotated proteins, such as CAD6364935.1 and CAD6366340.1, show higher FATCAT scores when compared with the control 1KMO such as CAD6366340.1 and CAD6364935.1 (Supplementary Table S6). Table S6). Even though some of the 13 annotated proteins from S. putrefaciens T2.3D-1.1 (CAD6367072.1 and CAD6364935.1) have high FATCAT scores with 1NQE, they still have lower scores than the BtuB proteins found in this work. Additionally, some of these annotated proteins, such as CAD6364935.1 and CAD6366340.1, show higher FATCAT scores when compared with the control 1KMO such as CAD6366340.1 and CAD6364935.1 (Supplementary Table S6).

Discussion
In our array of genomes, we have found 13 enzymes dependent on vitamin B12 (Table  2) with different degrees of importance. Of these, 8 have returned positive matches in some microorganisms from our sample ( Table 2). These are the methionine synthase, glutamate mutase ribonucleotide reductase, epoxyqueosine reductase, methylmalonyl-CoA mutase, ethanolamine ammonia lyase, methyltransferase (anaerobic magnesium-protoporphyrin IX monomethyl ester cyclase) and the bacteriochlorophyll cyclase ( Figure 2A). The methionine synthase, methylmalonyl-CoA mutase, epoxyqueosine reductase and ribonucleotide reductase are highly conserved among bacteria [16].
The methylmalonyl-CoA mutase enzyme catalyzes the reversible conversion of (R)methylmalonyl-CoA to succinyl-CoA [30]. It is a very important reaction involved in many pathways such as the fermentation of lactate to propionate, H2 and acetate; for the

Discussion
In our array of genomes, we have found 13 enzymes dependent on vitamin B 12 (Table 2) with different degrees of importance. Of these, 8 have returned positive matches in some microorganisms from our sample ( Table 2). These are the methionine synthase, glutamate mutase ribonucleotide reductase, epoxyqueosine reductase, methylmalonyl-CoA mutase, ethanolamine ammonia lyase, methyltransferase (anaerobic magnesium-protoporphyrin IX monomethyl ester cyclase) and the bacteriochlorophyll cyclase ( Figure 2A). The methionine synthase, methylmalonyl-CoA mutase, epoxyqueosine reductase and ribonucleotide reductase are highly conserved among bacteria [16].
The methylmalonyl-CoA mutase enzyme catalyzes the reversible conversion of (R)methylmalonyl-CoA to succinyl-CoA [30]. It is a very important reaction involved in many pathways such as the fermentation of lactate to propionate, H 2 and acetate; for the conversion of succinate to propanoate; or in the 3-hydroxypropanotae cycle for the fixation of CO 2 [31]. The epoxyqueosine reductase is responsible for the production of queosine from epoxyqueosine, which is a modified base present in the wobble position for the synthesis of amino acids such as histidine, aspartic acid, asparagine and tyrosine [32]. The ribonucleotide reductase synthesizes the deoxyribonucleotides essential for DNA synthesis [33], and the methionine synthase is responsible for the synthesis of the essential amino acid methionine [34]. Although conserved, most of these deeply rooted enzymes have vitamin B 12 -independent counterparts, except for the methylmalonyl-CoA mutase. Even so, our isolates depend on vitamin B 12 for at least one of these four conserved reactions. In particular, Ciceribacter sp. T2.26MG-112.2 and Ciceribacter sp. T2.30D-1.1 depend on vitamin B 12 for three highly conserved enzymatic activities among bacteria: methionine synthase, ribonucleotide reductase and the epoxyqueosine reductase (Table 2).
Aside from highly conserved enzymes such as methionine synthase, methylmalonyl-CoA mutase, epoxyqueosine reductase and ribonucleotide reductase, we have also looked for enzymes that are less conserved in bacteria (Table 2). To start with, the vitamin B 12 -dependent glutamate mutase allows the conversion of L-glutamate to L-threo-3-methylaspartate and is involved in the degradation of glutamate [35]. The vitamin B 12 -independent enzymes glutamate dehydrogenase and aspartate transaminase convert L-glutamate into 2-oxoglutarate [36] and catalyze a reversible reaction from L-glutamate to L-aspartate, respectively [37]. Only Desulfosporosinus sp. DEEP has returned a positive match against the vitamin B 12 -dependent glutamate mutase, but all the isolates do have at least one vitamin B 12 -independent pathway for glutamate degradation.
The ethanolamine lyase catalyzes the degradation of ethanolamine to acetaldehyde and ammonia [38]. This nitrogenated compound can be used by some bacteria as a source of N and C. Acetaldehyde is then transformed by other enzymes into acetate and released to the extracellular medium [39]. Alternatively, there is a vitamin B 12 -independent pathway in Chromobacter salexigens that would yield glycine instead of acetate similar to the vitamin B 12 -dependent pathway [26]. Thus, for Stutzerimonas sp. T2.31D-1, Tessaracoccus sp. T2.5-30 and Desulfosporosinus sp. DEEP, the final product from ethanolamine consumption would be acetate and ammonia, and for both Ciceribacter strains it would be glycine. Nevertheless, NH 4 + and acetate have been detected in the subsurface of the IPB [10]; hence, its abundance could be tied to the production of B 12 .
The methyltransferase AcsCDE catalyzes the transfer of a methyl group from methyltetrahydrofolate to the vitamin B 12 cofactor of the protein in the fixation of CO 2 in the Wood-Ljungdahl pathway [40]. Both Ciceribacter strains and Microbacterium sp. T2.11-28 have returned positive matches for at least one of the three subunits, and only Desulfosporosinus sp. DEEP has matched the three subunits of this transferase (Supplementary Table S4). Different strains of the Desulfosporosinus genus have been reported to fixate CO 2 through the reductive acetyl-CoA pathway [41][42][43], and our isolate does possess all the necessary proteins except for FdhB [10]. Thus, Desulfosporosinus sp. DEEP is a good candidate for CO 2 fixation in the subsurface of the IPB [10].
The bacteriochlorophyll cyclase enzyme, or BchE, requires vitamin B 12 for the synthesis of 3,8-divinylprotochlorophyllide, an intermediary in the formation of the isocyclic ring of chlorophylls [44]. Its B 12 -independent counterpart, AcsF, can act under both aerobic and anaerobic conditions [45]. Due to the anoxic conditions and the absence of light in the subsurface of the IPB, it is as yet unknown what role the synthesis of chlorophylls could have. Regardless, this is not the first report of microorganisms that are deemed photosynthetic living in the continental subsurface [6,10]. In our sample, only Rhodoplanes sp. T2.26MG-98 returned positive results for both the vitamin B 12 -dependent and the vitamin B 12 -independent enzymes. Although not entirely dependent, vitamin B 12 would still be needed for the synthesis of chlorophyll.
Of the analyzed genomes, Ciceribacter sp. T2.26MG-112.2 and Ciceribacter sp. T2.30D-1.1 would depend on vitamin B 12 for three highly conserved enzymatic activities among bacteria, namely methionine synthase, ribonucleotide reductase and the epoxyqueosine reductase. Other isolates depend on vitamin B 12 for less conserved reactions, such as Stutzerimonas sp. T2.31D-1 and Tessaracoccus sp. T2.5-30 for the consumption of ethanolamine or the fixation of CO 2 in the case of Desulfosporosinus sp. DEEP. Overall, all our isolates depend on vitamin B 12 in at least one of the cobalamin-dependent activities that have been studied. These results shed light on the requirements for life in the deep subsurface. Vitamin B 12 may play a relevant role as a cofactor and as a key component to the development of its communities and for some of its metabolisms.
Considering the need for vitamin B 12 for the given set of very ubiquitous microorganisms of the IPB [10], we have also looked for the set of proteins required for the synthesis of vitamin B 12 . The synthesis of vitamin B 12 involves a very complex and extensive list of proteins ( Figure 2C) [17]. The first step involves the synthesis of the precursor uroporphyrinogen III, which is shared among different tetrapyrrole-based molecules such as the heme group, chlorophylls or the coenzyme F430 [18]. There are two different starting points; one depends on glutamate and uses the proteins GtlX and HemAL to produce 5-aminolevulinic acid, and the other path starts from glycine and succinyl-CoA to also yield 5-aminolevulinic catalyzed by HemA. Afterwards, the proteins HemBCD produces uroporphyrinogen III in both cases. For the construction of the coring ring, uroporphyrinogen III can then be processed through an aerobic or anaerobic pathway. Shelton and colleagues (2018) [19] conducted a larger analysis of the production of vitamin B 12 and obtained similar results, reporting that some proteins involved in the aerobic/anaerobic pathways CobI:CbiL, CobJ:CbiH, CobM:CbiF, CobK:CbiJ, CobL:CbiT, CobL:CbiE, CobM:CbiC and CobB:CbiA are orthologous ( Figure 2C). Therefore, differentiating between aerobic and anaerobic pathways based on the sequence alone was untenable. Finally, the central pathway, where the nucleotide loop is assembled, is shared between the aerobic and anaerobic pathways and produces adenosylcobalamin (vitamin B 12 ) from adenosylcobyrinic acid [18].
All in all, although none of our genomes have shown a complete vitamin B 12 Table S5). Additionally, Shelton and colleagues (2018) [19] found a core set of eight proteins for the synthesis of this cofactor that are shared between the genomes of in vitro tested vitamin B 12 producers (Table 2). According to this criterion, only Desulfosporosinus sp. DEEP holds the core set of genes necessary to synthesize vitamin B 12 . Therefore, Desulfosporosinus sp. DEEP could play a central role in the ecosystem of the IPB not only as a vitamin B 12 producer but possibly as a primary producer as well through the fixation of CO 2 via the Wood-Ljungdahl pathway. Nevertheless, Ciceribacter sp. T2.26MG-112.2, Ciceribacter sp. T2.30D-1.1, Rhodoplanes sp. T2.26MG-98 and Tessaracoccus sp. T2.5-30 contain all the proteins except CobC, making them good candidates for vitamin B 12 production as well.
Even so, intermediaries of the vitamin B 12 biosynthesis or the vitamin itself can also be obtained from the extracellular medium and constitute a very relevant interaction between producers and vitamin B 12 auxotrophs at different ecosystemic levels [46][47][48]. The proteins involved in the salvage of vitamin B 12 belong to the central pathway (CobU/CobP, CobS/CobV and CobC) and different transporters for its uptake [17]. The most studied protein complex for the capture of vitamin B 12 is the ABC system BtuCDF with the outer membrane receptor BtuB [49,50]. With the exception of S. putrefaciens T2.3D-1.1, none of our isolates, although dependent, would have the means to capture extracellular vitamin B 12 through this system. Reference BtuB proteins CAD6020855.1 and RIH46231.1 have 7 different domains (the plug domain (IPR012910 and IPR037066), the interaction with TonB and β-barrel domains (IPR000531, IPR036942), the TonB Box (IPR010916), the TonBdependent receptor conserved site (IPR010917 and IPR039426) and the BtuB specific domain (IPR010101)). When compared with the annotated BtuB transporters from S. putrefaciens T2.3D-1.1 on the domain level, only CAD6364477.1 from S. putrefaciens T2.3D-1.1 shows 6 out of the 7 domains (Supplementary Table S2). Since the BtuB proteins found in S. putrefaciens T2.3D-1.1 in this work were different from those annotated previously [11], a more in-depth study was performed to reinforce the obtained results.
Structure prediction and comparison with the model 1NQE confirms that our putative BtuB are closer to BtuB than the control structure 1KMO. The control structure 1KMO belongs to the outer membrane transporter FecA [51], which was also a result in the Blastp search for BtuB transporters. We used 1KMO as a control TonB-dependent receptor to see if we could discern it from 1NQE using both structure and sequence to study putative TonB-dependent transporters. Structural differences between TonB-dependent transporters, such as FecA and BtuB, are very small, and substrate specificity depends on the outer loops between β-barrell strands [49], accounting for a low number of amino acids out of the total sequence that conform the protein. Therefore, the information that can be obtained from these structures is limited. When looking for BtuCDFB proteins, most of the results are ABC transporters involved in the transport of siderophores, hemin, sulphate/thiosulphate, and, to a lesser extent, cobalamin. Hence, we have found many ABC transporters in the genomes of our isolates, but we remain unable to assign specific functions to many of these transporters based only on their sequence. As stated above, based only on sequence, it is very difficult to ascertain what type of ligands and the specificity of such transporters might be. Nevertheless, the combination of sequencebased screening and protein structure prediction shows that the BtuB transporters detected in this work resemble the reference model more closely than the annotated vitamin B 12 transporters from S. putrefaciens T2.3D-1.1.
Aside from the BtuCDF and BtuB system, we have also looked for alternative vitamin B 12 importers. BtuM is a cobalamin transporter that can transport vitamin B 12 to the periplasm in cooperation with BtuB without the need for BtuCDF. Additionally, BtuM can also remove the CN − group from cyanocobalamin, allowing its conversion to physiological forms [27]. BtuM has only been found in Stutzerimonas sp. T2.31D-1. It also has been suggested that proteins such as BtuR could further transport vitamin B 12 in collaboration with BtuM [51], which is more widespread in our sample and is present in Stutzerimonas sp. T2.31D-1 as well. Another alternative is Rv1819c, which has been described as a hydrophilic compound transporter that could also transport vitamin B 12 [28] Table 2). Despite being unable to find the putative vitamin B 12 transporter for many of our isolates, other ways to introduce this cofactor to the cytoplasm must exist to ensure its development in the deep subsurface.
In this study, we conclude that some of the isolates depend on vitamin B 12 to carry out essential functions for their survival. Additionally, some of them, such as Desulfosporosinus sp. DEEP, might be able to synthesize vitamin B 12 . These factors have great relevance in the context of the IPB. During the IPBSL project, different techniques were used to identify the microorganisms present along the length the borehole. Moreover, the results of this project allowed the reconstruction of C, N, Fe, H and S biogeochemical cycles that take place in the subsurface of the IPB [10]. Thanks to all this information, we know that the isolates studied in this project are distributed throughout the subsurface and that they play a role in the aforementioned cycles. As a potential vitamin B 12 producer, Desulfosporosinus sp DEEP is distributed throughout the column and has been detected in 14 out of the 43 sampled depths. Particularly, at depths of 416 mbs (meters below surface) and 594 mbs, 8 out of 12 microorgansms have been detected, including Desulfosporosinus sp DEEP. Thus vitamin B 12 sharing seems plausible for most of our sample. This could most certainly enhance the flow of the biogechemical cycles that take place in the deep subsurface of the IPB, meaning that the synthesis of vitamin B 12 in the subsurface is essential for the operation of the bioreactor. At these depths, we hypothesize that there must be close interaction between these isolates, in which some of them would produce vitamin B 12 which then would be used by the others. Additionally, microorganisms, such as S. putrefaciens T2.3D-1.1 could be sharing other important molecules such as H 2 [11] as an electron donor, and that has been reported as a substrate for some Desulfosporosinus strains [42]. As such, our results would suggest that these microorganisms coexist and develop together in the deep subsurface through cooperation and sharing of essential molecules such as vitamin B 12 .

Conclusions
In spite of the limitations inherent to the given methodology, we have detected several vitamin B 12 -dependent enzymatic reactions with no vitamin B 12 -independent counterparts in the genomes of our twelve isolates from the deep subsurface. However, we have not been able to ascertain the entry pathways for vitamin B 12 in all the isolates, so questions regarding this are still open. Although not complete sensu stricto, we have found a complete core set of proteins [24] that could lead to the synthesis of cobalamin in the genome of Desulfosporosinus sp. DEEP and an almost complete set in the genomes of Ciceribacter sp. T2.30D-1.1, Ciceribacter sp. T2.26MG-112.2 and Rhodoplanes sp. T2.26MG-98. All in all, we have found 12 isolates with different cobalamin needs, and some of them are promising vitamin B 12 -producing candidates which may possibly meet those needs. The deep subsurface, although vastly unexplored, seems to depend on very similar factors and micronutrients such as vitamin B 12 when compared with above ground ecosystems. Hence cooperative interactions between microorganisms in poly-extreme environments such as the deep subsurface [10] could be responsible for the colonization and ecopoiesis of the IPB.