Microbial enrichment and gene functional categories revealed on the walls of a spent fuel pool of a nuclear power plant

Microorganisms developing in the liner of the spent fuel pool (SFP) and the fuel transfer channel (FTC) of a Nuclear Power Plant (NPP) can form high radiation resistant biofilms and cause corrosion. Due to difficulties and limitations to obtain large samples from SFP and FTC, cotton swabs were used to collect the biofilm from the wall of these installations. Molecular characterization was performed using massively parallel sequencing to obtain a taxonomic and functional gene classification. Also, samples from the drainage system were evaluated because microorganisms may travel over the 12-meter column of the pool water of the Brazilian Nuclear Power Plant (Angra1), which has been functioning since 1985. Regardless of the treatment of the pool water, our data reveal the unexpected presence of Fungi (Basidiomycota and Ascomycota) as the main contaminators of the SFP and FTC. Ustilaginomycetes (Basidiomycota) was the major class contributor (70%) in the SFP and FTC reflecting the little diversity in these sites; nevertheless, Proteobacteria, Actinobacteria, Firmicutes (Bacilli) were present in small proportions. Mapping total reads against six fungal reference genomes indicate that there is, in fact, a high abundance of fungal sequences in samples collected from SFP and FTC. Analysis of the ribosomal internal transcribed spacer (ITS) 1 and 2 regions and the protein found in the mitochondria of eukaryotic cells, cytochrome b (cytb) grouped our sample fungi in the clade 7 as Ustilago and Pseudozyma. In contrast, in the drainage system, Alphaproteobacteria were present in high abundances (55%). The presence of Sphingopyxis, Mesorhizobium, Erythrobacter, Sphingomonas, Novosphingobium, Sphingobium, Chelativorans, Oceanicaulis, Acidovorax, and Cyanobacteria was observed. Based on genomic annotation data, the assessment of the biological function found a higher proportion of protein-coding sequences related to respiration and protein metabolism in SFP and FTC samples. The knowledge of this biological inventory present in the system may contribute to further studies of potential microorganisms that might be useful for bioremediation of nuclear waste.


Introduction
There is consideration of extending the use of many Nuclear Power Plants (NPP) from 40 to 60 years of operation. The nuclear power plant Angra 1 in Rio de Janeiro, Brazil has a pressurized water reactor and after 12 months, part of the fuel (Uranium-235) in the core of the reactor is replaced. The spent fuel remains thermal active and radioactive and is transferred through a fuel transfer channel (FTC) to racks at the bottom of a spent fuel pool (SFP). The pool water contains Boron, a neutron-absorbing element to avoid critical heating of the pool. FTC and SFP are two of the locations to be studied for integrity and secure storage capacity of the fuel elements when considering extension of the duration of operation of the NPP. To ascertain security, operators search for any small leaks through micro-cracks in the liner and the drainage system by periodic inspection and through the SFP alarm level in the control room. Detecting microorganisms in the liner is important because their growth may cause deterioration of the NPP facility through microbial-induced corrosion (MIC) [1]. The water that eventually passes through the liner cracks is directed to the drainage system, and the volume is measured periodically to evaluate the waste. The pool water is oligotrophic and radioactive and does not provide a favorable environment to support life. However, there is a possibility of proliferation of algae, fungi, and bacteria, which together may form a biofilm [2]. Microorganisms that can grow under the harsh conditions of the NPP potentially can accumulate radionuclides and thus may be useful for bioremediation of radioactive environments [3][4][5][6]. Therefore, an in-depth analysis of the water quality also should include periodic assessment of possible growth of microorganisms in the pool and the inner coating of the drainer tubes. However, few studies have been conducted on microorganisms in NPP [3,7,8]. One study using analysis of 16S rRNA identified a diverse group of bacteria [8] and another isolated six bacterial taxa in the SFP pool water samples with radio-tolerance and biofilm-forming capabilities, but no species characterization was provided [3]. Also, an autotrophic freshwater green microalga was isolated by culture methods from a cooling pool of a nuclear reactor containing spent nuclear fuels in France [9]. This extremophile of the genus Coccomyxa is resistant to UV and gamma radiation. To estimate diversity and resolve taxa, improved tools to describe microbial diversity were described [10,11]. Also, the culture approach favors the microorganisms that were able to grow in a particular nutrient. Therefore, better resolving, unbiased methods are needed to characterize and describe biodiversity, especially in hazardous environments. Given the difficulties and limitations to collect samples from the SFP, small samples from the wall of this installation were collected. Subsequently, massively parallel sequencing (MPS) was applied for metagenomic sampling to analyze the microorganisms present in the walls of the SFP and FTC. Also, microorganisms in the drainage system may have traveled over the 12-meter column of the water pool where irradiated fuel elements with high activity (0.416 Gy/h) are stored. In addition to the inventory of microorganisms, biological function present in the system was assessed.

Study location
The Almirante Á lvaro Alberto Nuclear Power Plant (Angra 1) is located south of Rio de Janeiro city, in Rio de Janeiro State (23˚00 0 31.42@ S, 44˚27 0 26.87@ W), Brazil. SFP and FTC are diagramed in S1 Fig. Physicochemical parameters of the water, such as temperature, conductivity, pH, and concentration of boron, chloride, fluoride, silica, and sulfate were performed by routine analysis of the SFP and did not vary in the period of the sample collection.

Sample collection
Samples were collected under Radiological Working License No. 543. Samples were collected with the aid of sterile cotton swabs with plastic rod tubes on 12/09/2015 between 09:45 h and 11:35 h. The stem of the swab was held at its lower half of the swab and rubbed firmly, from top to bottom and vice versa, then rotated approximately 180 degrees and the procedure was repeated. After collection, cotton swabs were inserted into the tube. Fourteen samples were collected in four different locations: Four samples were taken from the surface of the liner of the FTC and four others from the SFP, both at 15 cm below the water level. Six samples were collected from the drainage system, three from dried drainer (Dry Drainers-DD) and three from drainer tubes that were wet at the time of the collection (Wet Drainers-WD). Radionuclide measurements were performed using the 40 FH GL Series 22676 Radiometer equipment.

DNA extraction
Collected cotton swabs (N = 3 or 4) were pooled to be representative of the four distinct locations. These locations were (1) the liner of the spent fuel pool (SFP); (2) the liner of the fuel transfer channel (FTC); and the drainage system identified as (3) dry (DD) and (4) wet drainers (WD). Nucleic acid extraction was performed for all cotton swabs. Briefly, 500 μl of TE buffer (TRIS-HCl 10 mM, EDTA 1 mM, pH 7.5 containing freshly added 0.2 g/mL proteinase K and 0.1% SDS) was added to the cotton swabs and incubated for 16 h at 37˚C. An equal volume of the solution of phenol-chloroform equilibrated with TRIS buffer (10 mM, pH 8.0) was added to the mixture and vortexed for 30 seconds. DNA present in the aqueous phase was precipitated by the addition of 0.3 M (pH 5.5) sodium acetate and 2.5 volumes of ethanol followed by centrifugation at 13,000 rpm in a microfuge. The DNA was suspended in TE-4 solution (TRIS-HCl 10 mM, EDTA 0.1 mM, pH 7.5) quantified by spectrophotometry (NanoDrop 2000, ThermoScientific, USA) and stored at -20˚C.

Library construction and Next Generation Sequencing
The DNA libraries were prepared using the Ion Xpress Plus Fragment Library Kit as recommended by the manufacturer (Thermo Fisher, San Francisco, CA). The pooled samples from SFP, FTC, DD and WD were used to generate four libraries. A total of 100 ng of DNA from each sample was enzymatically fragmented for two minutes using Ion Shear kit according to the manufacturer's instructions (Life Technologies, Carlsbad, CA). Fragmentation patterns were evaluated by electrophoresis microfluidics using DNA 1000 Kit in a Bioanalyzer (Agilent Technologies, Santa Clara, CA) and quantified using the ds DNA HS assay Kit Qubit fluorometer (Invitrogen Co. Carlsbad, CA). Sequencing was performed using a 318 chip (Life Technologies, Carlsbad, CA) in the Ion Torrent PGM platform (Life Technologies, Carlsbad, CA).

Data analysis
Sequence data were exported in FASTQ format by the Torrent Suite software and uploaded to the MG-RAST server (http://metagenomics.anl.gov/) on May 2016. Processing of shotgun reads was performed using the metagenomics pipeline [12] with default options for initial quality control (QC) filtering of raw reads. Low quality (Phred<15) and duplicated sequences were removed. The maximum e-value threshold of 1e-5, minimum alignment length of 50 bp and minimum of identity cutoff of 60% were used for functional annotation and classification. Venn diagrams were prepared at http://bioinfogp.cnb.csic.es/tools/venny/index.html. GC-pattern for each sequenced sample was evaluated through GC distribution among the reads to determine the heterogeneity of the DNA sequences on the online server MG-RAST. The search for fungal and bacterial genera was performed on the online server MG-RAST, using the M5NR database. This database is an integrated database that comprises sequences from NCBI, KEGG, EBI and other databases. Functional category identification was performed using reads matching to the M5NR database [13], which is based on genomic annotation data that are curated in a set of functional roles (SEED system) [14]. Functional category analysis of the genes found in the metagenomic samples was performed to investigate the metabolic capabilities of the microorganisms present in the SFP, FTC and drainage system. The taxonomic analysis was performed to determine the distribution of prokaryotic and eukaryotic organisms in the sampled areas. All the reads assigned to a taxon were counted, and their relative abundance was established. The potential of the metagenomic sequencing reads encompassing partial or entire genomes was tested by mapping these reads to reference genomes. Mapping was performed using CLC Genomics Workbench v8.5.1 with the following parameters: no global alignment; 0.8 similarity fraction; 2 of mismatch cost; 0.5 of length fraction. Assembling reads to a contig was performed using the parameters: reads were mapped back to contigs, using mismatch cost = 2; insertion cost = 3; deletion cost = 3; length fraction = 0.5; similarity fraction = 0.8. Subset of the variants calls, or ploidy variant detection was performed using the following parameters: fixed ploidy = 2; required variant probability = 90%; minimum coverage = 10; minimum count = 2: minimum frequency (%) = 20%. To characterize the fungal species, the barcode analysis was performed using ITS. The search for ITS consensus sequences was performed using Blastn against our sequence database of contigs using ITS sequences from different fungi as query sequences. An alignment of ITS sequences from 25 members of Ustilaginales selected from the NCBI database was performed using clustalW progressive alignment methods. The algorithm computes a rough distance matrix between each pair of sequences using pairwise sequence alignment scores. Progressively more distant groups of sequences are aligned until a global alignment is obtained. MEGA 6 was used to generate a phylogenetic tree by Maximum Likelihood, using Kimura 2-parameter, gamma distribution and evolutionarily invariable sites (K2+G+I) as nucleotide substitution model, and bootstrap of 500 replicates. The cytb phylogenetic tree was built using amino acid data, applying three different algorithms. Mega 6 was used to find the best-fit substitution model. For the cytb analysis, the best one, with smaller BIC (Bayesian Information Criterion) was mtREV24, described as General Reversible Mitochondrial [15], and it was used for building a Maximum Likelihood tree. The JTT best-fit model [16], was used for generating a Neighbor-Joining tree. No substitution model was used to generate a Maximum Parsimony tree.

Physicochemical and radiochemical parameters of the spent fuel pool and transfer channel
Measurements of physical and chemical parameters of the spent fuel pool and transfer channel SFP are described in Table 1. Radiation dosage at the collection points in the SFP and FTC was 2μSv/h. The concentration of radionuclides for each cotton swabs used to collect samples was recorded previously in a conservative manner by spectrometry containing 51 Cr, 58 Co, 60 Co, 137 Cs radionuclides. Total activity was 5.122 x 10 2 Bq/g.

Metagenomic sequencing and taxonomic analysis of microorganisms present in the spent fuel pool, transfer channel, and drainage systems
A total of 2,741,680 reads, obtained from sequencing, was uploaded to the online service MG-RAST for metagenomic analysis. Taxonomic classification results showed that parts of the microbial communities of fungi (class) and bacteria (phylum) are shared among the NPP sites, SFP, FTC, DD and WD (S2 Fig), but the relative abundance of microbial taxa distribution is different ( Table 2).
Bacteria comprised 85% of the microorganisms in these samples and Basidiomycota and Euryarchaeota in the drainage system were minor contributors (less than 0.7%). Proteobacteria represented 76.4% (DD) and 81.1% (WD). This group was followed by Actinobacteria (8.1% and 2.5%), Bacteroidetes (3.0% and 8.4%), and Firmicutes (3.9% and 1.9%), in DD and WD samples, respectively. Regarding the class of bacteria, Alphaproteobacteria (54.5% and 57.6%) was the most abundant in DD and WD sampled locations, respectively. The major genera of Alphaproteobacteria were Sphingopyxis (2.1% and 7.5%), Mesorhizobium (7.1% and 6.9%), Erythrobacter (4.2% and 3.9%), Sphingomonas (3.0% and 3.7%), Novosphingobium (3.6% and 3.4%), Sphingobium (2.4% and 2.4%), Chelativorans (2.6% and 2.1%), and Oceanicaulis (2.0% and < 0.1%), in DD and WD samples respectively. Pseudomonas, a genus of Gammaproteobacteria, Acidovorax a genus of Betaproteobacteria and fungal constituents, such as Sordaria, Ustilago, and Aspergillus, were found in minor proportions in DD and WD samples. Conversely, in SFP and FTC samples, the major contributors were eukaryotes (around 94%), in which the fungi Basidiomycota (79%), and Ascomycota (11%) were the major phyla. The dominant class was Ustilaginomycetes making up 70.9% and 70.2% of the SFP and FTC samples, respectively, followed by Sordariomycetes (4.6% and 4.7%), Tremellomycetes (3.3% and 3.9%), Agaricomycetes (2.4% and 2.3%), Exobasidiomycetes (2.3% and 2.5%), Saccharomycetes (1.9% and 1.7%) and Schizosaccharomycetes (1.9% and 1.7%). Bacteria constituted only 3.6% of the phyla in the SFP and FTC samples. Ustilago represented 72% of the fungal genera observed in the SFP and FTC. Other genera of fungi were found at lower percentages (less than 4%) in both samples. The detected fungi included Cryptococcus (Tremellomycetes), Schizosaccharomyces (Schizosaccharomycetes), Malassezia (Malasseziomycetes), Gibberella (Sordariomycetes), Neurospora (Ascomycetes) and Tilletia (Exobasidiomycetes). Bacteria constituted less than 3.5% of the taxa in SFP and FTC, and the genus Acidovorax was present in a very low percentage. Further analysis focused on the fungi as they were far more abundant than the bacteria in the FTC and the SFP. The presence of the dominant Ustilago (Ustilaginomycetes) in the SFP and FTC ( Table 2) was evaluated by mapping the total reads against available fungal genome sequences. Ustilago and closely related fungi genomes were compared to the total reads obtained from SFP and FTC. Total reads from SFP (80,270,543 bp) and FTC (78,047,151 bp) were mapped against the available Ustilaginales fungal sequences ( Table 3). The highest percentage of mapped reads (22% and 23%) from the SFP and FTC were obtained with Sporisorium reilianum when compared with the others fungal genomes Mycosarcoma maydis (previously known as Ustilago maydis) [17][18][19], Pseudozyma hubeiensis, Moesziomyces aphidis, Kalmozyma braziliensis, U. esculenta. On the other hand, only 0.81% and 0.38% of total reads from samples DD and WD mapped to the S. reilianum genome. Because environmental metagenome types usually have a distinct GC-pattern, we analyzed the GCcontent of our metagenomics reads (Fig 1). A similar GC distribution was observed for SFP and FTC sequencing reads peaking at 52 to 60%. In contrast, DD and WD samples peaked at 60 to 70% of GC-content. To further investigate the diversity of fungal and bacterial elements in our samples, the assembled reads into contigs from SFP, FTC, DD and WD were compared separately. Table 4 shows the contigs assembly data from SFP and FTC reads. They were, on average, bigger (772 bp and 647 bp, respectively) and smaller in number (19,986 and 25,688), when compared to DD and WD samples reads (482 bp and 436 bp; 43,391 and 46,188 contigs).
Lower sequence diversity was observed in SFP and FTC. In fact, 23% of the sequences organized in contigs from SFP were mapped to the 23 chromosomes of S. reilianum meaning i.e., 3,531,104 bp, from a total of 15,428,667 bp (data not shown). To investigate the diversity of the microorganisms present in the samples the nucleotide variation within all the assembled contigs was analyzed. S2 Table shows the consensus length, the total read counts and the average coverage of each of the 19,986; 25,688; 43,391 and 46,188 contigs assembled from reads obtained from SFP, FTC, DD, and WD, respectively. A subset of the variant calls was performed to verify the diversity of the consensus sequences for each contig considering a coverage of at least 10X for each nucleotide position with more than 10% of frequency (Fig 2).
We found 2,597 variants in SFP contigs, of which 1,455 were single nucleotide variation (SNV) and 1,142 other variants including deletions and insertions) and 2,420 variants in FTC contigs (1,388 SNV, and 1,037 variants including, deletions and insertions). In contrast DD  and WD contigs had 13,726 variants (10,827 SNV) and 16,917 variations (13,437 SNV), respectively. The number of the contigs in samples DD and WD were roughly twice as many as in the SFP and FTC, but the variants found in these contigs were 4 to 5 times more abundant in DD and WD samples. The diversity of microorganisms found in DD and WD was higher compared to SFP and FTC samples. To characterize the fungal species, present in SFP and FTC, the internal transcribed spacer (ITS) was analyzed in our metagenomics reads. One contig for each sampled area (SFP and FTC) was obtained that encompasses the ribosomal gene cluster. In the SFP contig archive, contig 111 was found which has a consensus length of 7,268 bp, 18,203 read counts and 28X of average coverage. In the FTC contig archive, contig 201 was found which has a consensus length of 7,720 bp, covered by 1,148 reads and 25X of average coverage (Fig 3). These two contigs containing the genes for 18S, 5.8S and 28S ribosomal rRNA and the respective ITS1 and ITS2 were aligned to closely related Ustilaginomycetes fungi (S3 Table).
No nucleotide variations were observed in the aligned region, containing ITS1 and ITS2 and 5.8S rRNA, of the two contigs (201 and 111). Alignment of an ML phylogenetic tree was based on a 630-bp region of the ITS1-5.8S rRNA-ITS2 sequences of 25 members of the order Ustilaginales selected from the NCBI ITS region from fungi strains and reference material. The ML phylogenetic tree (Fig 4) shows the relationship of the fungus in our sample. The contig consensus sequences of SFP and FTC were positioned among the clade of Ustilago austro-africana (Clade 7). The ML tree of our SFP and FTC samples using ITS barcoding were grouped in the clade of Ustilago austro-africana (Clade 7).
Also, the phylogenetic tree of cytochrome oxidase protein (cytb), with available amino acid sequences from related fungi, confirmed grouping into the clade 7 with Pseudozyma thailandica, in all three topology algorithms (S3 Fig).

Functional analysis of the microorganisms
The metabolic capabilities of the microorganisms present in the SFP, FTC and drainage system are shown in Fig 5. Relative abundance of functional gene categories from SFP, FTC, DD and WD indicate that protein coding sequences involved in the amino acids and derivatives pathway, carbohydrates, respiration, RNA metabolism and metabolism of aromatic compounds, were present.
Relative abundance indicate that the functions related to DNA metabolism, cell wall, membrane transport, virulence, and defense were more abundant in the WD and DD samples than in the FTC and SFP samples. In contrast, a higher number of protein sequences related to respiration and protein metabolism (Fig 5) were associated with SFP and FTC samples. Detailed analysis of eukaryote genes such as cytochrome c oxidase, NADH dehydrogenase, succinate dehydrogenase cytochrome b560 subunit, mitochondrial precursor, ATP synthase chains are displayed on S4 Table. These genes are present as major components of the eukaryote oxidative phosphorylation (S4 Fig).

Microbial diversity in spent fuel pool and fuel transfer channel in an acidic environment with boron
The data indicate that microorganisms from the wall of the SFP and FTC are part of a sophisticated, surface-attached community. The microorganisms found in these samples were collected in pH 4.0 to 4.7 at 35˚C. The group of bacteria and archaea described here can be found   [20,21]. Indeed, some microbial populations from acidic forest soils are more tolerant to acidic cultivation conditions [22]. Fungi are capable of adapting to environments with diverse pH values [23] and transcriptional response induced by changes in the pH of the medium from neutral to acidic or to alkaline was reported in Mycosarcoma maydis [24]. Conserved signal transduction pathways mediate ambient pH sensing and adaptation in yeast and fungi for alkaline pH [25], but the mechanisms helping to cope with acidic pH are still unknown. Fungi produce organic acids, and therefore are capable of contributing to MIC. Ustilaginomycetes, which were found in high content in the SFP and FTC samples, under certain environmental conditions may be a source of organic acids [26] to the aerobic and anaerobic bacterial types, modifying the environment creating a favorable situation for the corrosion of the stainless steel [27]. Although, the bio-corrosion in these NPP installation has not been an actual problem due to the constant monitoring of the drainage system, the presence of these microorganisms may pose a threat to the integrity of the stainless-steel liner. Sulphide or sulphate-reducing bacteria (SRB) were not observed in the stainless-steel biofilm. These microorganisms can alter the local physical/chemical conditions and lead MIC and may modify the environment with the presence of metabolites on the stainless steel creating fissures [27,28]. The observation that Proteobacteria (Alpha, Beta, Delta, and Gamma), Actinobacteria, Firmicutes (Bacilli), were present only in low proportions in the SFP and FTC may be due to the use of boron in the spent pool to capture neutrons. Boron is an essential element for human and animals [29] in which in a dose-dependent manner, between 0.77 and 1.545 mg/ml, has been known to have antibacterial effects [30]. The unexpected detection of Fungi (Basidiomycota and Ascomycota) as the major microbial contributor to the spent fuel pool and transfer channel suggests that they can grow in the presence of boron concentrations of 2,500 mg/L to 3,000

Microbial diversity in spent fuel pool, fuel transfer channel, dry drainer, wet drainer samples and exposure to radiation
Our results indicated Ustilago (Ustilaginomycetes) as the dominant genus (around 72%) in the SFP and FTC. The GC distribution observed for SFP and FTC sequencing reads peaked at 52 to 60% which are consistent with the high fungal content in the SFP and FTC and were similar to the fungal reference genomes. Mapping the metagenomic reads to the available genome sequences among the Ustilaginales was insufficient to identify the genus of the fungus in our sample. Members of this order are facultative biotrophic basidiomycetes and occur in a variety of environments, and an effort was made to obtain consistency in the fungal systematics of Ustilagomycetes [31][32][33][34][35][36][37]. ITS barcoding analysis grouping the SFP and FTC fungus in the clade 7 of Ustilago austroafricana suggests that this fungus in our SFP and FTC samples belongs to the Ustilaginales according to Wang et al. 2015 [36]. In their tree, using LSU rRNA D1/D2 domains and ITS1 and 2 regions including 5.8S rRNA gene, the group described that clade 7 is composed of Sporisorium veracruzianum, Pseudozyma thailandica, Macalpinomyces viridans, M. spermophora, Pseudozyma tsukubaensis and Ustilago austro-africana. The cytb tree grouped our sample in the same clade as Pseudozyma thailandica confirming the correct group of fungi (clade 7). Additional research may include incorporating morphological characters, parasite population data, physiology, radio resistance characteristics to clarify the species position in the Ustilago-Sporisorium clade 7. These closely related fungi might share the same properties that are seen in Ustilago as being capable of growing in a radionuclide environment up to 6000 Gy [38,39]. Black molds growing in and around the Chernobyl Nuclear Power Plant have been reported to be radiotrophic [40].
Exposure to the ionizing radiation of Chernobyl may change the electronic properties of melanin and might enhance the hyphae growth towards sources of radiation [41]. Examination of fungi and bacteria at the same time allows the direct comparison of the abundance patterns of the different domains. Few numbers of members of Proteobacteria, Actinobacteria, Firmicutes, Bacteroidetes and Cyanobacteria were present in the SFP and FTC when compared to fungal abundance. These findings suggest that the microorganisms found in SFP and FTC, which are exposed to a uranium radiation and boron, may favor the growth of fungi. Analogous findings [42] associated the outgrowth of fungi over bacteria when uranium up to 1,500 ppm was added to culture media. Paenibacillus species were prevalent on LB plates containing 3 ppm of uranium but colony count decreased as concentration rose to 600 ppm. Some Bacillus and Paenibacillus were also isolated from high background radiation (2.9 μGy/h) from the Chernobyl nuclear power plant disaster area [43], suggesting that Firmicutes/Bacilli have abundant genera diversity, broad physiological characteristics and adaptation capabilities [44][45][46]. Similarly, regarding the drainage system, Sphingopyxis, Mesorhizobium, Erythrobacter, Sphingomonas, Novosphingobium, Sphingobium, Chelativorans. Oceanicaulis, Acidovorax, and Cyanobacteria were found in our metagenomic study. One can assume that the microorganisms identified in this area (DD and WD) have traveled over the 12-meter column of the water pool with 894 irradiated fuel elements stored with high activity 0.416 Gy/h are stored. It was reported that Actinobacteria, Flavobacteria, Firmicutes (Bacillus) and Proteobacteria as well as Deinococcus-Thermus and Cyanobacteria were identified in the areas after the Chernobyl accident, resisting an exposure of 4000 Gy/h [43,47]. One Microalgae isolated from a biofilm growing on a projector immersed 3 m below the surface water from the water pool used to store spent fuel elements in a research nuclear reactor in France was able to survive intense gamma-rays irradiation, up to 2,000 times the dose lethal to human [9].
Studies of the NPP of Cofrentes in Valencia, Spain, [7] identified microorganisms attached to the spent nuclear fuel pool wall using culture, PCR-DGGE, sequencing of 16S rRNA and LSU rRNA fungal identification system. Groups such as Betaproteobacteria, Actinomycetales, Bacillus, Staphylococcus and Trichocomaceae [7] were identified. Further studies in the same NPP [8,48,49], using standard culture methods, DGGE, and sequencing of 16S rRNA fragments found Ralstonia, Pseudomonas, Burkholderia, Stenotrophomonas, Sphingomonas, Methylobacterium, Afipia, Bradyrhizobium, Rhizobium, Streptococcus, Staphylococcus, Mycobacterium, Nocardia, and Chryseobacterium [40]. Subsequently, they were able to isolate Bacillus and Stenotrophomonas from the biofilm that was produced in 34 months by immersing stainless steel plates into the spent fuel pool, but on titanium plates, Ralstonia and Mycobacterium were the most present [8,49].

Functional capacity of microorganisms in spent fuel pool and fuel transfer channel
Function analysis presented in this work provided information about the microbial metabolism in the collected samples. The distribution of gene functional categories was similar between SFP and FTC and between DD and WD, suggesting similar microbial populations in these areas. Genes related to carbohydrate, amino acid, and vitamin metabolisms were similarly abundant in all sampled areas and probably reflect housekeeping functions. In contrast, genes related to respiration and protein metabolism were more abundant in SFP and FTC, and genes related to DNA metabolism, cell wall, membrane, and transport were more abundant in DD and WD. Interestingly, genes related to photosynthesis were almost exclusively found in WD, and those related to motility, chemotaxis, iron, and phosphorus metabolisms were mostly observed in DD and WD. These differences may reflect varying microbial populations, which in turn may be a result of divergent metabolic requirements. In addition, the detection of eukaryotic gene sequences related to respiration (~15%) suggests an abundance of fungi in SFP and FTC. Eukaryote cytochrome c oxidase, NADH dehydrogenase, succinate dehydrogenase cytochrome b560 subunit, mitochondrial precursor, ATP synthase chains are the major components of the eukaryote oxidative phosphorylation and indicate that the functional gene category respiration (S4 Fig) which are present in the SFP and FTC are likely due to the high abundance of the fungi.

Conclusion
This work contains relevant information about the microbial content of the water of the SFP and the monitoring drainage systems of a NPP. Metagenomic analyses using massively parallel sequencing can identify potential taxa at least to the genus level. This capability is a vast improvement over using a single biomarker such as 16S rRNA and allows better identification of microorganisms (both bacterial and archaea). Since the microorganisms were able to survive in a radioactive environment, they may be able to accumulate and concentrate radionuclides from water. Future work should focus on isolating these organisms, characterizing them better and determining their potential decontamination capabilities. The surprisingly high amount of sequences of fungal origin indicated that Ustilaginomycetes identified in the spent fuel pool and transfer channel suggests that these fungi are able to grow in a radioactive environment. Radiation levels were low in the samples collected from the pool wall, and no problems related to bio-corrosion materials were observed. Further genome and transcriptome analyses of the fungus may bring some understanding of what contributes to its persistence in the SFP and FTC. Our results were obtained without culture methods or PCR amplification of target genes. The use of cotton swabs was adequate for sampling material and the benefit of showing negligible levels of radionuclides (5.122 x 10 2 Bq/g). Therefore, collected samples could be combined and associated with success via a fast detection method to identify bacterial and fungal distribution in the SFP and FTC. This work also described functional genes related to respiration in SFP, contributing to understanding the NPP environment. Our results may be of value for further studies at other sites and to identify potential microorganisms that can be used for bioremediation of nuclear waste.
Supporting information S1 Table. General information of reads (after quality and trimming processes) uploaded to MG-RAST server. showing the robot arm, at the bottom, used to transfer nuclear fuel to the reactor or from the reactor to the SFP. SFP and FTC facilities are massive concrete structures with a corrosive resistant stainless-steel ASTM A240, type 304L liner material to guarantee structural quality. The nuclear power plant Angra 1 in Rio de Janeiro, Brazil has a pressurized water reactor (PWR) that uses 121 fuel elements (256 rods, 4 meters high each) of enriched Uranium-235. After 12 months, part of the fuel in the core of the reactor is replaced. The spent fuel remains thermal active and radioactive and is transferred through a fuel transfer tube to 1252 storage cells of super compact racks at the bottom of a spent fuel pool (SFP). The quality of the water must meet strict requirements of purity and clarity allowing operators to handle irradiated fuel elements.