Bacterial Diversity in the Hyperalkaline Allas Springs (Cyprus), a Natural Analogue for Cementitious Radioactive Waste Repository

The biogeochemical gradients that will develop across the interface between a highly alkaline cementitious geological disposal facility for intermediate level radioactive waste and the geosphere are poorly understood. In addition, there is a paucity of information about the microorganisms that may populate these environments and their role in biomineralization, gas consumption and generation, metal cycling, and on radionuclide speciation and solubility. In this study, we investigated the phylogenetic diversity of indigenous microbial communities and their potential for alkaline metal reduction in samples collected from a natural analogue for cementitious radioactive waste repositories, the hyperalkaline Allas Springs (pH up to 11.9), Troodos Mountains, Cyprus. The site is situated within an ophiolitic complex of ultrabasic rocks that are undergoing active low-temperature serpentinization, which results in hyperalkaline conditions. 16S rRNA cloning and sequencing showed that phylogenetically diverse microbial communities exist in this natural high pH environment, including Hydrogenophaga species. This indicates that alkali-tolerant hydrogen-oxidizing microorganisms could potentially colonize an alkaline geological repository, which is predicted to be rich in molecular H2, as a result of processes including steel corrosion and cellulose biodegradation within the wastes. Moreover, microbial metal reduction was confirmed at alkaline pH in this study by enrichment microcosms and by pure cultures of bacterial isolates affiliated to the Paenibacillus and Alkaliphilus genera. Overall, these data show that a diverse range of microbiological processes can occur in high pH environments, consistent with those expected during the geodisposal of intermediate level waste. Many of these, including gas metabolism and metal reduction, have clear implications for the long-term geological disposal of radioactive waste.


Introduction
The United Kingdom has an extensive legacy of radioactive waste from more than 60 years of civil and military nuclear technology, and this inventory will only increase with the decommissioning of old facilities and the development of new nuclear power options. In 2008, deep geological disposal of intermediate-and high-level radioactive wastes, which are the most hazardous components of the UK waste legacy, was adopted as UK Government policy (Department for Environment, Food & Rural Affairs (DEFRA) 2008) and this situation is echoed throughout Europe and globally. Currently, in the United Kingdom, the proposed concept for intermediate level radioactive waste (ILW) disposal is based on a multibarrier system. The current generic model for disposal is that ILW will be grouted in steel containers emplaced in the geological disposal facility and eventually the waste will be sealed in the deep subsurface with a cementitious backfill (DEFRA 2008;NDA 2010a;Nirex 2003). When the host environment becomes saturated with groundwater, the highly alkaline conditions that will develop are intended to minimize radionuclide solubility and thus the risk of radionuclide transport to the biosphere (NDA 2010a;Nirex 2003). However, the biogeochemical gradients that will develop across the interface between the highly alkaline deep cementitious geological disposal facility and the geosphere are poorly understood in terms of their impact on the long-term performance of the geological disposal facility. It is now recognized that the wastes, which can include nitrate, iron, metal oxides, radionuclides, H 2 gas (produced by the corrosion of steel waste) and organic carbon (e.g., cellulose-derived compounds) are likely to create conditions favorable for microbial growth. Microbial transformations of radionuclides under reducing conditions are well reported at circumneutral pH (e.g., Anderson et al. 2003;Bernier-Latmani et al. 2010;Boyanov et al. 2011;Newsome et al. 2014). However, very little is known about microorganisms that can potentially reduce metals and radionuclides under alkaline conditions, although they may potentially control the speciation and solubility of several key radionuclides via complexation with ligands produced by microbial metabolism, by reduction or by mineralization processes. Therefore, understanding their activities will be critical in underpinning any safety case for a cementitious geodisposal facility.
To extrapolate results from laboratory and field experiments to the repository, the study of natural analogues of cementitious-based geological repositories is of high importance (Alexander and Milodowski 2011). Several sites with active hyperalkaline (pH > 10) groundwater systems have been studied as natural analogues for cementitious geological repositories. These include the Maqarin site in northern Jordan (Nagra 1992), the Semail ophiolitic complex in Oman (Bath et al. 1987), the Troodos ophiolite in Cyprus (Alexander and Milodowski 2011), and the Zambales ophiolite in the Philippines (Alexander et al. 2008). Ophiolites are sequences of mafic and ultramafic rocks representing ancient oceanic crust and upper mantle rocks that have been tectonically emplaced onto a passive continental margin (e.g., Troodos in Cyprus and Semail in Oman) or have uplifted within subduction zone accretionary complexes or subduction complexes, such as the Josephine and Coast Range ophiolites of California (Harper et al. 1994;Shervais et al. 2004).
Highly alkaline groundwaters are often encountered within these systems and are associated with the reaction of percolating water with olivine and pyroxene within the mafic and ultramafic rocks to form serpentine minerals (serpentinization). Serpentinization occurs along several pathways (Moody 1976), although it is the low-temperature serpentinization (e.g., Barnes and O'Neil 1969;Barnes et al. 1972) that is particularly relevant to the active serpentinization sites in Cyprus, Oman and the Philippines. In this case, Mg(HCO 3 ) 2type meteoric groundwaters react with the ultramafic rocks of the ophiolite in an essentially open system and produce highly alkaline Ca-rich (spring) waters (generally between pH 10 and 11) rich in K C , Na C , Ca 2C , Mg 2C , and results in the production of serpentine minerals, magnetite, hydroxide, hydrogen (H 2 ) and methane (CH 4 ) gas, depending on the mineralogy of the site (Blank et al. 2009;Schulte et al. 2006).
Active serpentinization systems in ophiolites are sometimes studied as analogues for potential early ecosystems on Earth and Mars (Russell et al. 2010;Schulte et al. 2006;Sleep et al. 2011). However, most studies on serpentinization and the affiliated microbial communities have focused on deep-sea hydrothermal fields (Brazelton et al. 2006;Brazelton et al. 2012;Kelley et al. 2005), where it has been shown that serpentinization can generate large volumes of hydrogen gas (H 2 ), variable quantities of methane (CH 4 ) and low molecular weight compounds (McCollom and Seewald 2007;Proskurowski et al. 2008), and that microbial communities are dominated by methane-and sulfur-metabolizing Bacteria and Archaea (Brazelton et al. 2006).
Microbial studies in serpentinization-driven (terrestrial) ophiolitic environments have been carried out only recently, at the Cabeço de Vide aquifer in Portugal (Tiago et al. 2004;Tiago and Veríssimo 2013), the Del Puerto ophiolite in California (Blank et al. 2009), the Tablelands ophiolite in Canada (Brazelton et al. 2012;Brazelton et al. 2013), and the Leka ophiolite in Norway (Daae et al. 2013). Nevertheless, no comprehensive microbiological studies have been carried out yet on the ophiolitic sites that have been studied as natural analogues to geological repositories for radioactive wastes, despite the potential significance of these sites in informing radioactive waste disposal options. Furthermore, the potential for microbial metal reduction in terrestrial alkaline serpentinization-associated systems has not been explored to date and our knowledge on alkaline microbial metal reduction is restricted to only a few isolated microorganisms, such as Alkaliphilus metalliredigens QYMF (Roh et al. 2007), Bacillus sp. strain SFB (Pollock et al. 2007), and two Natronincola strains (Zhilina et al. 2009b).
The aim of this study was to investigate, for the first time, the microbial ecology of samples from the hyperalkaline (pH up to 11.9) Allas Springs (Troodos Mountains, Cyprus), a site of active low-temperature serpentinization within the Troodos ophiolite that has been studied as a natural analogue for cementitious radioactive waste repositories (Alexander and Milodowski 2011). The objectives were: i) to describe the phylogenetic diversity of natural microbial communities from an analogue of cementitious geological radwaste disposal site using molecular microbiology techniques; ii) to investigate the potential of indigenous microbial communities to catalyze metal reduction at alkaline pH; and iii) to obtain pure cultures capable for Fe(III) reduction at alkaline pH, for future studies on the microbial transformation of metals and radionuclides. Our findings are discussed in the context of the potential for microorganisms to colonize and influence the evolution of (alkaline) cementitious-based geological repositories for radioactive wastes.

Samples Collected from the Hyperalkaline Allas Springs
A suite of samples were collected from the Allas Springs site in the Argaki tou Karvouna valley, near Platania in the Troodos Mountains in October 2010. The sampling site corresponds to location A1-1, as described by Alexander and Milodowski (2011), where Ca-rich hyperalkaline groundwater (up to pH 11.9) discharges under artesian flow through a steeply inclined fracture in a large outcrop of harzbergite up to 4 m high (supplemental materials, Figure S1). As the water discharges from the fracture and flows over the outcrop it reacts with atmospheric CO 2 , resulting in the precipitation of calcite, aragonite and dolomite to form travertine (tufa) 74 Rizoulis et al.
deposits on the bedrock surface. At the foot of the outcrop, the springwater also pecolates through a dense covering of forest litter and the underlying highly fractured and altered bedrock. Samples Cyp1, Cyp2, Cyp3, Cyp5, CypR were collected from a broken stalactite/flowstone "rib" on the underside of the travertine-coated harzbergite outcrop (Figures S1 and S2). They consisted of: dripping hyperalkaline groundwater (Cyp1); a suspension of brown flowstone in groundwater (Cyp2); fragments of brown-stained flowstone in groundwater (Cyp3); a suspension of green microbial mats in groundwater (Cyp5); and a brown/green-stained flowstone-coated rock sample (CypR). A further sample (Cyp4) was taken from beneath the surface of the forest litter at the base of the harzbergite outcrop where the other samples in this study were collected.
This corresponds to Site A1-3 (Alexander and Milodowski 2011), and the sample taken consisted of a pink gelatinous layer, that was observed to occur at a depth of about 10 cm immediately beneath the buff/brown unconsolidated tufa ( Figure S3) that impregnates the base of the forest litter. The water seeping through this material was still hyperalkaline (pH 10) at the point of sampling. The water chemistry of the collected samples is shown in Table 1. More details about the location of the site and the collected samples are included in the supplemental materials. The tufa-coated rock sample CypR was carefully chiselled from the flowstone surface and wrapped in a plastic Ziplock bag that was rinsed (5 times) with the same hyperalkaline groundwater discharging over the rock surface at this point. All other samples were collected into clean, sterile 30-ml plastic bottles, which had been thoroughly rinsed (5 times) with the hyperalkaline groundwater associated with each respective sampling point prior to collecting the samples, and topped up so that no head space was left in the bottle. All samples were stored at 4 C prior to further analysis.

Microbial Community Analyses
DNA was isolated from 0.25 g of the rock sample and 0.5 ml of the Cyp1a, Cyp2, Cyp3, Cyp4, and Cyp5 water suspensions using the MoBio PowerSoil DNA Isolation Kit (MoBio Laboratories, Inc., Carlsbad, CA, USA), following the manufacturer's instructions. For the profiling of the bacterial communities present, PCR amplification was performed using universal bacterial 16S rRNA gene primers 8F (Edwards et al. 1989) and 1492R (Lane 1991). PCR products were purified using a Qiagen PCR purification kit (Qiagen, Inc., Valencia, CA, USA) and then ligated into the pGEM-T Easy Vector system (Promega, Madison, WI, USA) and transformed into One Shot TOP10 chemically competent Escherichia coli cells (Invitrogen, Inc., Carlsbad, CA, USA).
Positive clones were screened by PCR using primers SP6 and T7, and sequenced using the ABI Prism BigDye Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems, Life Technologies Corporation, USA) and forward primer 8F (Edwards et al. 1989). The obtained 16S rRNA gene sequences were checked for chimera formation using Mallard (Ashelford et al. 2006). All nonchimeric bacterial 16S rRNA gene sequences from this study were clustered into OTUs (Operational Taxonomic Units) at a level of similarity of 97% using Mothur v.1.24.1 (Schloss et al. 2009).
Mothur was also used to calculate and compute alpha diversity indices, rarefaction curves, and unweighted pair group method with arithmetic mean (UPGMA) clustering based on Bray-Curtis dissimilarity values. The phylogenetic classification for all obtained nonchimeric 16S rRNA gene sequences of this study was performed using the RDP classifier (at 80% confidence threshold) of the Ribosomal Database Project (Release 10, update 31; (Cole et al. 2009). In addition, the closest phylogenetic relatives (environmental sequence, cultured organism or bacterial type strain) for the OTUs with the highest number of reads were identified by nucleotide Table 1. pH, Eh measurements and concentration of major anions and cations in samples taken from the hyperalkaline Allas Springs in Cyprus A complete list of the water chemistry analysis is shown in Table S1. ND stands for not detected.
Blastn search. All partial bacterial 16S rRNA gene sequences for this study were deposited to GenBank, under accession numbers JQ766531-JQ766937. In addition to the profiling of the bacterial communities, 16S rRNA gene PCR amplifications were carried out to investigate the presence of Archaea and methanogens in our samples, using Archaeal-specific primers Arch21F and Arch958R (DeLong 1992), and methanogen-specific primers 1AF and 1100AR (Hales et al. 1996), respectively.

Setup and Sampling of Anaerobic Enrichment Cultures
Enrichment cultures were set up anaerobically by supplementing each of samples Cyp1, Cyp2, Cyp3, Cyp4, Cyp5 with an equal volume of a ferric-citrate containing medium that was largely based on a medium that has been used previously to isolate metal-reducing alkaliphilic bacteria (Ye et al. 2004). The medium used in this study contained 9.4 mM NH 4 Cl, 4.3 mM K 2 HPO 4 , 4 mM NaHCO 3 , 6.1 mM Na 2 SeO 4 , 17.1 mM NaCl, 10 ml L ¡1 mineral stock solution (Lovley et al. 1984), 7 mM sodium lactate, 7 mM sodium acetate, 0.025 g L ¡1 yeast extract, and 15 mM Fe(III)-citrate. The pH of the medium was adjusted to 10 with the addition of NaOH. Following sparging with N 2 gas and sterilization by autoclaving, 20 ml of the medium was mixed with approximately 20 ml of either of the Cyp1-Cyp5 samples in sterile 100 ml serum bottles, under a N 2 :H 2 (98% : 2%) atmosphere in an anaerobic glove box. The serum bottles were sealed with butyl rubber stoppers and aluminium crimps, incubated at 20 C and sampled aseptically after 7 and 18 days, after which pH and Eh measurements were taken as described above, and the concentration of reduced Fe(II) was determined with the ferrozine assay (Lovley and Phillips 1987).

Isolation of Pure Cultures and Identification by 16S rRNA Gene Sequencing
For the isolation of pure cultures, the enrichment medium was solidified with 1.8% gellan gum, plated under a N 2 :H 2 (98% : 2%) atmosphere in an anaerobic glove box, and inoculated with 30 ml of the Cyp4 and Cyp5 enrichment cultures. In addition, 30 ml of the Cyp4 and Cyp5 enrichment cultures were plated onto solidified medium that contained 15 mM Na 2 SO 4 instead of Fe(III)-citrate as the electron acceptor to assess the potential for sulfate reduction in the experiments. After incubation at 20 C for 7 days under anaerobic conditions, single colonies were picked randomly and used to streak new gellan gum plates that contained either ferric citrate or sodium sulfate. One week later, 135 single colonies (65 from the ferric-citrate and 70 from the sodium sulfatecontaining plates respectively) were used to inoculate 10 ml liquid enrichment medium containing either Fe(III) (as citrate) or sulfate (as the NaSO 4 salt) as the electron acceptor.
After incubation at 20 C for 7 days, a 250-ml subsample was used for DNA extraction with the MoBio PowerSoil DNA isolation kit. A total of 84 of the isolated bacteria were identified by 16S rRNA gene sequencing, after PCR amplification of the 16S rRNA gene using primers 8F and 1492R, PCR purification using a Qiagen PCR purification kit, and 16S rRNA gene sequencing using the ABI Prism BigDye Terminator v3.1 Cycle Sequencing Kit and primer 8F (as described here for the clone libraries). The closest phylogenetic relatives of the isolated bacteria were found by nucleotide Blastn search. Then, phylogenetic tree showing their phylogenetic associations was constructed with MEGA version 5.10 (Tamura et al. 2011), after alignment with Clus-talW v.1.4 (Thompson et al. 1994). Evolutionary relationships were inferred by the neighbor-joining method (Saitou and Nei 1987) and the evolutionary distances were computed using the Jukes-Cantor model (Jukes and Cantor 1969). All positions containing alignment gaps and missing data were eliminated in pairwise sequence comparisons (pairwise deletion option) and bootstrapping was performed for 1,000 replicates.

Fe(III)-Reduction by Isolated Bacteria
The potential of the isolated microorganisms of this study to reduce Fe(III) was tested in 10-ml liquid cultures, containing the same medium that was used for the enrichment cultures, i.e., with 15 mM ferric citrate as the electron acceptor and 7 mM lactate and 7 mM acetate as the electron donor, but with the pH adjusted to 9. The pure cultures were incubated at 20 C for 7 days, before pH, Eh and Fe(II) measurements were taken, as described above.

Water Chemistry of Collected Samples
Six samples were collected from the hyperalkaline Allas Springs, including groundwater only (Cyp1), brown-stained flowstone fragments in groundwater (Cyp2, Cyp3), a greenstained microbial mat in groundwater (Cyp5), a sample from the base underneath the artesian flowstone (Cyp4), and a rock sample (CypR). The chemical compositions of the collected samples are presented in Table 1 and are similar to compositions previously reported for groundwaters from this site (Alexander and Milodowski 2011).
The pH measurements indicated that samples Cyp1-3 were highly alkaline (pH between 11.52 and 11.71; Table 1), reflecting the alkalinity of the water that discharges from the harzbergite. The pH in samples Cyp4 and Cyp5 was lower (9.24 and 9.40, respectively), presumably as a result of atmospheric CO 2 reaction with sample Cyp5 and by reaction with the forest litter in the case of Cyp4. The redox potential was reducing in Cyp5 (¡158 mV), while the Eh in the other samples was mildly oxic, between C68 and C158 mV. All samples were characterized by very high concentrations of Na C (1348-1635 mg L ¡1 ), Cl ¡ (1880 to 2230 mg L ¡1 ), and elevated concentrations of K C (69-82 mg L ¡1 ) and sulfate (54-130 mg L ¡1 ).
The concentration of Ca 2C was elevated in samples Cyp1 and Cyp2 (38 and 14 mg L ¡1 respectively), while in the remaining samples Ca 2C ranged from 2.2 to 4 mg L ¡1 . This variation in Ca 2C concentrations may also reflect the 76 Rizoulis et al.
interaction of the hyperalkaline water with CO 2 , which results in the removal of Ca 2C from solution through the precipitation of the extensive calcite and aragonite tufa on site. The concentration of HCO 3 ¡ also varied, from below detection limits in Cyp5, 60-130 mg L ¡1 in Cyp1-3, and 480 mg L ¡1 in Cyp4. The water chemistry is not untypical for ophiolites, but the Na C and Cl ¡ levels are somewhat high compared to most ophiolite systems, possibly reflecting the presence of relict seawater in the host rock (see Alexander and Milodowski 2011). All other measured concentrations were either below 1 mg L ¡1 (Mg 2C , Al, Si, Ba) or negligible/ nondetectable (nitrate, nitrite, total Fe, Mn, Sr). The complete set of measurements is shown in Table S1.
The Cyp5 clone library was dominated by unclassified a-Proteobacteria (63.5%) and to a lesser degree by GpIV Cyanobacteria (17.6%), while the rock sample CypR contained group GpIV Cyanobacteria (40.8%) and Rhodobacter (16.3%), Pseudomonas (2%), Silanimonas (2%) genera. The microbial community in the sample that was taken from beneath forest litter at the base of the harzbergite outcrop (Cyp4) was significantly different to the microbial communities of the other samples as indicated by the UPGMA cluster dendrogram ( Figure S5), with most of the sequences grouping within unclassified a-Proteobacteria (46%) and g-Proteobacteria (22.4%), as well as in the Spirochaeta (7.9%) genus. The number of OTUs (at 97% similarity level) identified in each sample varied between 14 and 36 (total number of identified OTUs D 104), indicating moderately diverse microbial communities, in agreement with the calculated diversity indices (Table S2) and rarefaction curves ( Figure  S4). The OTUs that contained the highest number of sequences (more than 1.9% of the total number of sequences of this study) and their closest phylogenetic affiliations are shown in Table 2.
In addition to the bacterial diversity detected, PCR amplification using Archaeal-or methanogen-specific primers indicated that these microorganisms were present in samples Cyp4, Cyp5 and CypR, but probably in low abundances, as indicated by the low PCR yields (faint bands in the agarose gels; Figure S6). Further description of the phylogenetic diversity in these communities was beyond the scope of this study and it remains to be investigated by future sequencing efforts.

Isolated Bacterial Strains from the Hyperalkaline Allas Springs
Eighty-four of the bacterial isolates were identified by 16S rRNA gene sequencing. The phylogenetic analysis indicated that the isolates were grouped within 5 OTUs (at OTU similarity level of 99%). Most of the isolates (82 isolates, OTUs/ strains P1, P2, P3) belonged to the Paenibacillus genus (Figure 3), and they shared 98% ID similarity to various facultative anaerobic Paenibacillus type strains, such as P. odorifer TOD45 (NR_028887), P. borealis KK19 (NR_025299), and P. wynnii LMG 22176 (NR_042244). The remaining two isolates (OTU/strains A1 and A2) were most closely related (99% ID similarity) to the uncultured bacterial clone Alchi-chica_AQ2_2_1B_147 (JN825558) from the alkaline lake Alchichica in Mexico (Couradeau et al. 2011) and the uncultured bacterial clone TX2_2O07 (JN178047) from an extreme saline-alkaline soil of the former lake Texcoco in Mexico (Figure 3).

Alkaline Fe(III)Reduction in Enrichment Cultures and by Isolated Bacteria
The potential for Fe(III) reduction was tested by supplementing the Cyp1-Cyp5 samples from the hyperalkaline Allas Springs with an equal volume of an enrichment medium containing 15 mM Fe(III)-citrate as the electron acceptor, and 7 mM lactate and 7 mM acetate as the electron donors. Thus, the initial concentration of soluble Fe(III) in the enrichment cultures was approximately 7.5 mM. During the 18-day incubation, no Fe(III) reduction was observed in the Cyp1-3 enrichment cultures (Figure 4), while the pH decreased slightly to approximately 9.40 and the redox potential decreased from 78-168 mV to 56-112 mV. In contrast, in the Cyp4 and Cyp5 enrichment cultures, up to 40% and 32% of the Fe(III) was reduced, respectively, during the same incubation period (Figure 4). The reduction of Fe(III) was accompanied by a notable decrease in the redox potential to around -240 mV, and a drop in pH to 8.76 and 8.34 for the Cyp4 and Cyp5 enrichment cultures, respectively (Figure 4).
Furthermore, preliminary experiments with five isolated microorganisms (Paenibacillus affiliated strains P1, P2, P3, and Alkaliphilus related strains A1 and A2) showed that they can all reduce Fe(III)-citrate at alkaline pH 9 but to a different extent. After 7 days of incubation, the concentration of Fe(II) in solution ranged from 1.7 mM (A1 strain) to 7.6 mM (A2 strain), and it was between 2.6 and 3.7 mM in the cultures of the Paenibacillus strains ( Figure 5). During incubation, the pH dropped significantly from pH 9 to 6.54 in the medium inoculated with strain P1, while in the remaining cultures the pH remained alkaline, between 8.3 and 8.8 ( Figure 5).

Microbial Diversity in Samples from the Hyperalkaline Allas Springs, Cyprus
Despite the potential role of microbial metabolism on the long-term performance of geological repositories for radioactive wastes, comparatively few studies have examined the microbial ecology of high pH natural environments, analogues to these systems. In this study the bacterial diversity in samples from a natural analogue for a cementitious based geological repository, the hyperalkaline Allas Springs (Troodos Mountains ophiolite) in Cyprus, was investigated by 16S rRNA gene cloning and sequencing. The results showed diverse bacterial communities present, but many of the sequences were not closely related (less than 95% sequence similarity) to isolated microorganisms (Table 2), an indication that these systems have not been studied extensively yet and/or that isolation of these highly alkaliphilic microorganisms is challenging.
Of the sequences that were closely affiliated to cultured genera, only a few were related to known alkaliphilic genera (for example Silanimonas genus; Table 2), and the majority were related to genera with type strains that are not known as alkaliphilic (Hydrogenophaga, Pseudomonas, Acidovorax, Acinetobacter, Paracoccus, Propionibacterium; Figure 2). Some of these genera have also been detected in samples from other highly alkaline sites, such as the Maqarin site (Pedersen et al. 2004) and the Cabeço de Vide aquifer (Tiago and Veríssimo 2013), which could indicate that alkali-tolerant microorganisms, adapted to environmental niches, may persist in such highly alkaline environments. Interestingly, the presence of a large number of sequences related to the Hydrogenophaga genus is observed among most of the studies on terrestrial ophiolitic environments.
In fact, Hydrogenophaga related sequences were present in three of the clone libraries of this study (Cyp3, and to a lesser degree in Cyp1 and Cyp2; Figure 2), made up more than 35% of a 16S rRNA gene clone library and 20% of the pyrosequencing 16S rRNA reads in samples from the alkaline Cabeço de Vide aquifer (Tiago and Veríssimo 2013), and were the most dominant genus in samples from the Leka ophiolite (Daae et al. 2013). Hydrogenophaga related sequences also dominated the bacterial communities in two spring water samples from the Tablelands ophiolite (Brazelton et al. 2013), and were detected among the metagenomic reads and contigs of one of these samples (Brazelton et al. 2012).
Although Hydrogenophaga spp. are not known alkaliphilic microorganisms, one Hydrogenophaga sp. isolate has been shown previously to grow on benzene at pH up to 8.5, with optimum growth at pH 8 (Fahy et al. 2008). In general, members of the Hydrogenophaga genus are facultative anaerobic, chemoorganotrophic or chemolithoautotrophic, such as H.  Rizoulis et al. The closest type strain or cultured relative is given instead of the first GenBank match, in case sequence similarity is 95% or higher.
defluvii, which can use the oxidation of H 2 as an energy source and CO 2 as a carbon source, and their growth is not inhibited by high levels of O 2 in the atmosphere (K€ ampfer et al. 2005). Thus, it appears that abiotic liberation of H 2 gas during the serpentinization of ophiolites, potentially combined with the biotic generation of H 2 by fermentative microorganisms, may have led to the enrichment of Hydrogenophaga species in these systems, despite the presumptively adverse alkaline pH.   This is significant in the context of a geological repository for radioactive wastes, because alkali-tolerant hydrogen oxidizing bacteria (such as Hydrogenophaga species) in this setting could also potentially utilize excess H 2 that is expected to be produced during the abiotic corrosion of steel (NDA 2010b) and the biotic release of H 2 by fermentative microorganisms (particularly those degrading cellulose in low and intermediate-level wasteforms). This has significant implications as consumption of hydrogen by this previously poorly recognized biotic pathway has the potential to mitigate any over pressurization and transport effects in the geological disposal facility and/or host rock associated with hydrogen production from corrosion of steel (Libert et al. 2011). Oxidation of hydrogen could also be potentially linked to the reduction of a range of electron acceptors, including radionuclides (Lloyd 2003).
Another finding of this study was that samples Cyp5 and CypR, which contained visible green-colored microbial mats, yielded a relatively high number of sequences related to Cyanobacteria (17.6% and 40.8% respectively; Figure 1). Interestingly, these Cyanobacterial sequences were closely related to Leptolyngbya isolates from Antarctica (Taton et al. 2006), and a separate study has shown that Leptolyngbya species dominate the microbial communities on stromatolite benthic samples from the alkaline (pH 10.4) lake Untersee in Antarctica (Andersen et al. 2011). In addition, Cyanobacterial stromatolite-related sequences were also detected in the microbial mat sample (40% of the clones) from the Del Puerto Ophiolite, and have been linked to biological precipitation of carbonates (Blank et al. 2009). However, microbial precipitation or dissolution of calcium carbonates and other minerals is not limited to photosynthetic Cyanobacteria, as it may also be promoted by other microbial processes linked to sulfate-reducing, nitrate-reducing or fermenting bacteria, under alkaline anaerobic conditions (Dupraz and Visscher 2005), and within the microbial community of the Cyp4 sample of this study, siliceous stromatolite-related sequences affiliated to g-Proteobacteria were detected (Table 2). Thus, the potential significance of the microbial precipitation and dissolution of calcium carbonate (or other minerals, such as silicates) within the calcium-rich cement leachate that will be formed within the radwaste geological repository should not be overlooked, as it may influence the alkalinity and the evolution of the leachate plume during long-term radwaste disposal.
In regard to other microbial groups, although sulfatereducing bacteria (SRB) or dissimilatory sulfite reductase gene fragments have been detected in other terrestrial highly alkaline or ophiolitic environments (Blank et al. 2009;Pedersen et al. 2004), and high concentrations of sulfate were present in most of the Cyprus samples, no sequences closely related to known SRB were detected in this study. Furthermore, methanogenic and anaerobic methane-oxidizing Archaea have been found to dominate deep hydrothermal vents (Brazelton et al. 2006) and to be present in terrestrial environments too (Blank et al. 2009), and their presence was confirmed by PCR amplification in some samples from the hyperalkaline Allas Springs ( Figure S6). However their phylogenetic diversity remains to be investigated.

Potential for Metal Reduction at Alkaline pH
Microbial metal reduction at alkaline pH is likely to be a significant factor in controlling the solubility and mobility of key radionuclides such as uranium and technetium, during disposal in cementitious-based geological repositories.
Recently it was shown that a clear succession of electron acceptor utilization existed in alkaline microcosm cultures up to pH 11, as rapid nitrate reduction was followed by slower soluble Fe(III)-citrate, insoluble Fe(III)-oxyhydroxide, and sulfate reduction (Rizoulis et al. 2012). Further sediment microcosm studies from our group have also shown that alkaline bioreduction of Fe(III)-oxyhydroxide can lead to the formation of magnetite (Williamson et al. 2013) and that microbially mediated reduction of U(VI) can occur at pH 10 (Williamson et al. 2014).
In this study, we have found that two out of the five microbial communities sampled from a natural analogue of cementitious-based geological repositories (the hyperalkaline Allas Springs) also exhibited the ability to reduce soluble Fe(III) (samples Cyp4 and Cyp5, Figure 4). Considering that the clone libraries of the Cyp4 and Cyp5 samples did not contain any sequences related to known Fe(III)-reducing bacteria, it is evident that complementary to molecular microbiology approaches, enrichment and isolation studies should be carried out when examining the potential for metal/radionuclide reduction by diverse natural microbial communities.
Furthermore, five bacterial strains isolated during this study, three of them belonging to the Paenibacillus genus and two of them associated with the Alkaliphilus genus, were also capable of Fe(III) reduction at alkaline pH. These metalreducing strains were not detected among the environmental sequences of the corresponding microbial communities prior to the establishment of the enrichment cultures (i.e., the Cyp4 and Cyp5 clone libraries), indicating either culturing selectivity or that microbial species can be enriched and drive biogeochemical processes when new electron acceptors become available, under specific geochemical conditions.
Noticeably, with the exception of the Serratia strain, all the other microorganisms that have been shown to reduce Fe (III) at alkaline pH (from this or other studies) are Gram-positive Bacilli (Bacillus or Paenibacillus genera) or Clostridial (Alkaliphilus and Natronincola genera) strains of the Firmicutes phylum (Figure 3). It is not clear whether this is due to culturing bias (Firmicutes are often obtained by culturing on solidified rich media) or because alkaline metal reduction is carried out preferentially by Gram-positive bacteria. To date, the mechanisms of microbial Fe(III)/metal reduction have been studied mostly at circumneutral pH using model Gramnegative bacteria, such as Shewanella oneidensis MR-1 and various Geobacter species. Extracellular electron transfer by Geobacter and Shewanella, implicated in the reduction of insoluble Fe(III) minerals, is mediated via c-type membrane cytochromes and/or conductive pili, as reviewed by Shi et al. (2009).
Shewanella species are also able to mediate extracellular reduction via the secretion of redox-active flavins, which act as electron shuttles (von Canstein et al. 2008), and may also play a role in Fe(III) reduction in high pH systems (Fuller et al. 2014). For Gram-positive bacteria, the mechanisms of extracellular electron transfer are largely unexplored and only recently has it been suggested that (at circumneutral pH) c-type cytochromes may be involved in the respiratory Fe (III) reduction by Thermincola potens strain JR (Carlson et al. 2012;Wrighton et al. 2011). Thus, more targeted studies are needed in order to determine the potential for and the fundamental mechanisms of microbial metal and radionuclide reduction at alkaline pH as these questions are highly pertinent to the management and disposal of the global legacy of radioactive waste materials.

Conclusions
The findings of this study indicate that phylogenetically diverse microbial communities can exist in a natural serpentinization-driven environment, analogous to a cementitious-based geological repository for radioactive waste. The presence of sequences affiliated to non-alkaliphilic genera may indicate that alkali-tolerant bacteria can persist at high pH. The phylogenetic results also indicate that microbial metabolism may have a significant role on important biogeochemical processes that have been largely overlooked to date in the context of geological repositories, such as (but not limited to), microbial oxidation of H 2 gas, reduction of metals and radionuclides, and precipitation or dissolution of calcium carbonates and other minerals. Therefore, this study highlights the potentially significant impact of microbial activity on long-term geological disposal of radioactive waste.