A Novel Adaptation Mechanism Underpinning Algal Colonization of a Nuclear Fuel Storage Pond

ABSTRACT Geochemical analyses alongside molecular techniques were used to characterize the microbial ecology and biogeochemistry of an outdoor spent nuclear fuel storage pond at Sellafield, United Kingdom, that is susceptible to seasonal algal blooms that cause plant downtime. 18S rRNA gene profiling of the filtered biomass samples showed the increasing dominance of a species closely related to the alga Haematococcus pluvialis, alongside 16S rRNA genes affiliated with a diversity of freshwater bacteria, including Proteobacteria and Cyanobacteria. High retention of 137Cs and 90Sr on pond water filters coincided with high levels of microbial biomass in the pond, suggesting that microbial colonization may have an important control on radionuclide fate in the pond. To interpret the unexpected dominance of Haematococcus species during bloom events in this extreme environment, the physiological response of H. pluvialis to environmentally relevant ionizing radiation doses was assessed. Irradiated laboratory cultures produced significant quantities of the antioxidant astaxanthin, consistent with pigmentation observed in pond samples. Fourier transform infrared (FT-IR) spectroscopy suggested that radiation did not have a widespread impact on the metabolic fingerprint of H. pluvialis in laboratory experiments, despite the 80-Gy dose. This study suggests that the production of astaxanthin-rich encysted cells may be related to the preservation of the Haematococcus phenotype, potentially allowing it to survive oxidative stress arising from radiation doses associated with the spent nuclear fuel. The oligotrophic and radiologically extreme conditions in this environment do not prevent extensive colonization by microbial communities, which play a defining role in controlling the biogeochemical fate of major radioactive species present.

IMPORTANCE Spent nuclear fuel is stored underwater in large ponds prior to processing and disposal. Such environments are intensively radioactive but can be colonized by microorganisms. Colonization of such inhospitable radioactive ponds is surprising, and the survival mechanisms that microbes use is of fundamental interest. It is also important to study these unusual ecosystems, as microbes growing in the pond waters may accumulate radionuclides present in the waters (for bioremediation applications), while high cell loads can hamper management of the ponds due to poor visibility. In this study, an outdoor pond at the U.K. Sellafield facility was colonized by a seasonal bloom of microorganisms, able to accumulate high levels of 137 Cs and 90 Sr and dominated by the alga Haematococcus. This organism is not normally associated with deep water bodies, but it can adapt to radioactive environments via the production of the pigment astaxanthin, which protects the cells from radiation damage. KEYWORDS Haematococcus, microbial ecology, nuclear waste, spent nuclear fuel T he generation of nuclear power and the resultant radioactive wastes, which have the potential to contaminate the environment with radionuclides, have led to intense public and academic interest in environmental radioactivity. This has most recently been highlighted by the events at the Fukushima-Daiichi nuclear power complex in Japan, where reports estimate that~10 16 Bq of radioactive cesium and 10 14 Bq of radioactive strontium were released accidentally into the environment following the To hoku earthquake and tsunami on 11 March 2011 (1). Undoubtedly, there will be contaminated equipment, fuel rods, and storage ponds that contain elevated concentrations of these and other radionuclides as a result of the accident. Thus, the development of safe management processes for radioactive wastes includes consideration of the large and complex global legacy materials and facilities which have resulted from more than 50 years of nuclear power generation and clearly affects any development of new nuclear power stations.
Microorganisms have been shown to survive and colonize a broad range of extreme habitats (2,3), and there have been sporadic reports suggesting microbial colonization of radioactive environments at nuclear facilities (4)(5)(6)(7)(8)(9)(10)(11). These include nuclear fuel waste disposal containers (12), contaminated soils, including the Department of Energy Hanford Site in Washington (4) and the Sellafield nuclear facility in the United Kingdom (13), water surrounding the damaged core reactor at the Three Mile Island Nuclear Power Plant (14,15), the walls of the damaged number four reactor of the Chernobyl Nuclear Power Plant in Ukraine (16), nuclear reactor cooling pool waters (17,18), and spent fuel ponds (5-8, 10, 19).
The fate of radionuclides in the environment is controlled mainly by the interaction between the background matrix of the radioactive material, the often complex chemistry of the radionuclides in question, and a broad range of biogeochemical factors associated with the extant microbial communities in the environment that has been contaminated. Indeed, in recent years, it has become clear that microorganisms exhibit significant control on radionuclide mobility in these environments via processes, including biosorption, bioaccumulation, biotransformation, biomineralization, and microbially enhanced chemisorption (20,21). These processes must be studied in more detail if we are to understand and control biological processes occurring in nuclear fuel cycle facilities, understand their impact on the long-term fate of stored radioactive materials, and harness their potential for the cost-effective bioremediation of sediments and waters influenced by radionuclides (22)(23)(24)(25). This is an extreme challenge, as nuclear facilities, including fuel storage ponds, are some of the most inhospitable facilities that environmental scientists have worked in, and are subject to extremely tight regulatory control. Thus, as a consequence of both safety and access issues, they remain largely uncharacterized.
This study focuses on an outdoor spent nuclear fuel storage pond at Sellafield, in Cumbria, United Kingdom where seasonal microbial productivity significantly impacts plant operations at the site. The pond contains a dynamic inventory of spent nuclear fuel and has been operating since the early 1960s. This fuel has typically been derived from a fleet of power-generating light water reactors and is ultimately destined for reprocessing within the on-site reprocessing facility, THORP (thermal oxide reprocessing plant). The pond is open to the atmosphere and has steel-lined concrete walls, and as part of the pond management strategy, the demineralized water volume of 14,000 m 3 is purged (replaced) at an average rate of~400 m 3 a day. Most of the fuel in the pond is zirconium clad and has typically been stored in specialist metallic fuel storage containers known as multielement bottles (MEBs), which ideally isolate the spent fuel from bulk exchange with the pond water. As well as MEB storage, it is noteworthy that some of the spent fuel has been held in open-topped "skips" (large metal-bodied containers, typically used for refuse) within the pond, and the zirconium-clad fuel in these skips will have exchanged with bulk water, with presumably a higher radionuclide release than from the fuels stored in MEBs.
The outdoor pond is subject to sodium and chloride incursions due to Sellafield's close proximity to the coast, e.g., sea spray. Furthermore, the pond has a known history of seasonal generation of biomass typically referred to on-site as "algal blooms." This has led to visibility issues and resultant plant downtime during the summer months, affecting the forward management of the spent nuclear fuel inventory. As would be expected in a facility where the spent fuel inventory is being moved according to site operations, there is large variability in the measured radiation dose at the site, with dose rates at the MEB side panels up to~14 Gy h Ϫ1 . Gamma radiation dose rates from the MEBs are higher at the side panels of the containers, as there is less shielding compared to the top and bottom of the MEBs. Additionally, due to both attenuation of the radioactivity by water and to the inverse square law, dose rates drop off rapidly with distance from the MEBs. In day-to-day pond operations, MEBs are moved through the water column causing the pond contents to be disturbed, and localized high radiation fluxes can occur. Thus, the system is dynamic in terms of the dose rate that biota receive, and it is clear that any biological materials that are entrained within the pond circulation will need to withstand dose rates from ambient to extreme.
In addition to the radiation dose from the MEBs, the ponds also contain radionuclides from the stored fuels. The radionuclide inventory is dominated by the high-yield fission products 137 Cs and 90 Sr, which have intermediate half-lives (30 years and 28.8 years, respectively), and are present in high quantities in the spent fuel stored within the pond (26). Thus, pond waters and any resultant solids where 137 Cs and 90 Sr are present at elevated levels will need to be managed for several hundred years for radioactive decay to occur.
Understanding the microbial ecology and biogeochemistry of the Sellafield spent fuel storage ponds is an important first step required for minimizing plant downtime due to reduced visibility and will also inform decommissioning activities for fuel storage ponds worldwide. In addition, the evolving nuclear accident at Fukushima-Daiichi has generated a renewed interest in the management of radioactivity from accidental releases. It is clear that as a result of the series of accidents at the site, significant quantities of highly active effluents containing 137 Cs and 90 Sr have been generated and will continue to be held in storage during the ongoing efforts to stabilize the site (27). Indeed, there are likely to be parallels between extreme accident scenarios and controlled spent nuclear fuel pond environments that may assist in the longer-term management of both planned and accidental nuclear releases. The aim of this study was to characterize, in detail, the unique microbial ecology of a spent fuel pond throughout a microbial bloom and to investigate the impact of this increase in microbial biomass on the cycling of key radionuclides in the pond.

RESULTS
Water samples were collected from Sellafield's outdoor storage pond approximately every 7 days between May 2010 and October 2010 (n ϭ 16). The sampling campaign was aimed at characterizing changes in microbial ecology, biogeochemistry, and radiochemistry during a microbial bloom event affecting plant operations. Identification of key prokaryotic and eukaryotic species and their contribution to the bacterial/algal bloom and water chemistry was undertaken to further understand the radionuclide-microorganism-biosphere interactions in this extreme environment. Significant shifts in the microbial ecology of the ponds were noted over the 6-month sample period and included a photosynthetic bloom that occurred with changes in the inventory of the major radionuclides in the pond waters.
Seasonal variations in physical and chemical conditions. During the sampling campaign, pond water chemistry showed low overall variability (as expected, as the operational mode of the facility is designed to maintain highly controlled and constant conditions) with conductivity averaging 3.9 Ϯ 0.6 S cm Ϫ1 for all sample points and bulk pH changing between 6.4 and 8.0 (see Table S2 in the supplemental material).
Additional data for the wider pond visibility and temperature were also made available by Sellafield Ltd. for the relevant sampling period, and there was a clear inverse relationship between temperature and visibility in the pond with the coldest water temperatures in January coinciding with a maximum visibility of~5 m (Fig. 1). Pond water temperatures then increased from January to June as a result of increased solar radiation, with pond visibility at a minimum from May to July (Ͻ3 m). This drop in visibility was coincident with site microbial blooms hampering pond operations and affecting spent fuel management. Visibility in the pond then remained below 4 m for the remainder of the sampling campaign.
Biomass and chlorophyll concentrations. The chlorophyll a concentration, an indicator of trophic state (28), revealed that the system contained a large amount of biomass, typical of that of a biologically rich eutrophic water body during May 2010 to October 2010 and consistent with the visibility measurements. The water column visibility ( Fig. 1) in addition to algal cell counts and chlorophyll a content (Fig. 2) indicated that the most significant biomass bloom occurred around July 2010. The filter from the prebloom June 2010 sample (when visibility was already poor) was pale yellow in color and had a relatively low chlorophyll a concentration ( Fig. 1 and 2). In contrast, during the bloom period (mid-June to August), the filters were a red/orange color, and displayed consistently elevated chlorophyll a concentrations (Fig. 2). Additionally, the summer bloom also coincided with significantly higher concentrations of phosphate, nitrate, sulfate, calcium, and suspended solids in the pond (Table S2; data supplied by Sellafield Ltd.).
Microscopic algal cell counts ( Fig. 2) showed that the number of cells over the sampling period ranged from 850 Ϯ 244 cells per ml (on 19 May 2010) to a maximum of 4,550 Ϯ 495 cells per ml (on 7 July 2010). The peak number of algal cells detected coincided with the time of the high pond turbidity and increased chlorophyll a concentrations in late June to early July. However, a slight fall in chlorophyll a concentrations through August, which does not directly correlate with high cell counts, may suggest a shift to production of red pigments that coincides with the filter color. There was then a general decline in algal cell counts through late summer and autumn to values approaching those at the start of the sampling in May 2010.
Pond water radiochemistry. For the radionuclide inventory, there was a general trend upwards in concentrations of both 90 Sr and 137 Cs in unfiltered waters over the sampling campaign. In unfiltered waters, the 90 Sr concentration increased from 2.66 Ϯ 0.07 Bq ml Ϫ1 in May 2010 to a maximum concentration of 5.94 Ϯ 0.15 Bq ml Ϫ1 in late September 2010. The 137 Cs concentration in the unfiltered water samples also increased gradually from 1.22 Ϯ 0.03 Bq ml Ϫ1 in early June 2010 to a maximum concentration of 2.81 Ϯ 0.07 Bq ml Ϫ1 in late September 2010. These increases in 90 Sr and 137 Cs concentrations throughout sampling are apparently unrelated to the broad peak in bulk biomass in late June to early July (Fig. 2).
The influence of biomass on radionuclide behavior in the system was further examined by considering radionuclide retention on the filtered biomass (Fig. 3). For 90 Sr, the retention on the filters was highly variable throughout the year with between 5.6% Ϯ 0.9% and 71.6% Ϯ 7.9% retention observed at different time points. Interest- Microbial ecology and phylogenetic diversity of the pond. PCR amplifications using broad-specificity prokaryotic and eukaryotic primers produced a series of clone libraries showing a surprisingly rich microbial ecology, consisting of aquatic freshwater bacteria, freshwater algae, plankton, and to a lesser extent fungi and presumably reflecting the uncovered nature of the pond ( Fig. 4 and Fig. S1 [Phyla] in the supplemental material). The neutral pH of the pond with associated marine salt incursions, ingress of windblown debris from the coast, and nutrient additions from sources, including guano and rainfall, undoubtedly play an important role in the development of such a complex ecosystem. Indeed, while there is no bulk exchange of seawater with the purged pond water, it is possible that there may be minor introduction of microbial biomass and associated DNA from the atmospheric sources mentioned above. However, the high pond biomass, evidenced through several bloom events with significant chlorophyll concentrations, suggests that the microbial community detected is a result of high net productivity of the pond community that developed in situ.
Of particular note was the unexpected dominance within the eukaryotic clone libraries of a close relative of Haematococcus pluvialis, a freshwater species of green alga, Chlorophyta (29), with a maximum dominance of the clone library occurring in October (~77%) (Fig. 4). Pyrosequencing technologies became available after the initial clone libraries had been generated in this study and were used to confirm our initial results from Sanger sequencing. The higher-throughput pyrosequencing data generated were in agreement with those of the clone libraries, confirming the dominance of a close H. pluvialis relative (H. pluvialis strain SAG 34-1b; 99.5% match [427/429] [30]) alongside a diverse prokaryotic community (Fig. S2).
Haematococcus pluvialis is well-known for its high content of the red ketocarotenoid pigment and strong antioxidant astaxanthin (3,3=-dihydroxy-␤,␤-carotene-4,4=-dione) which is important in aquaculture and various pharmaceuticals and cosmetics (31,32). Astaxanthin is normally present at low concentrations within the cell, but it is accumulated rapidly when environmental conditions become unfavorable for normal cell growth (33). Microscopic analysis of pond samples illustrated a change in the putative Haematococcus cell morphology from small (5-to 10-m), mobile, green cells with two flagella in the June samples to larger (10-to 30-m) vegetative cells with no flagella and clearly with cysts with a red pigment, putatively identified as astaxanthin (34), in the September and October samples (Fig. S3). Interestingly, this change in the putative Haematococcus cell morphology to the moderately encysted vegetative stage coincides with a peak in 90 Sr and 137 Cs levels in the unfiltered pond waters, and is thus possibly related to radiation stress.
Viability and encystation of experimentally irradiated H. pluvialis. To investigate whether the dominance of Haematococcus in the clone library, alongside morphological changes recorded in the samples, may be the result of tolerance to radiation stress, pure cultures of H. pluvialis were irradiated with daily acute doses of 16-Gy X-ray radiation to a total absorbed dose of 80 Gy. Due to the constraints associated with working on a regulated nuclear site, it was not feasible to isolate Haematococcus from FIG 4 Phylogenetic affiliations (phyla, with closest distinguishable phylogenetic ranks in parentheses) of organisms detected in Sellafield pond samples using PCR with broad-specificity primers for prokaryotic 16S rRNA genes (top) and eukaryotic 18S rRNA genes (bottom). fam, family; sub. div. subdivision.
Microbial Ecology of a Nuclear Fuel Storage Pond ® pond samples for further experimental work, so a pure culture of a very close relative was obtained from the Culture Collection of Algae and Protozoa (Scottish Marine Institute, Oban, United Kingdom). The dose regime chosen was also representative of radiation fluxes delivered to organisms in a flushed spent nuclear fuel pond, such as at Sellafield, where dose rates are dynamic, ranging from ambient to~14 Gy h Ϫ1 at MEB surfaces. However, localized high radiation fluxes can occur as the spent nuclear fuel is moved through the water column.
Viability remained similar for both control and irradiated cultures throughout the incubation period (Fig. 5A), suggesting that these doses did not have a significant impact on cellular physiology. However, by day 5, there was a slight drop in viability in irradiated cultures compared to controls. Although this was not significant (80% Ϯ 11% and 90% Ϯ 5%, respectively), it was accompanied by a noticeable increase in cell debris and "ghost" cells (dead cells that have lost their cytoplasmic contents and appear as a cell wall with clear interior) observed in irradiated cultures (Fig. 5E).
Analysis of cell morphology revealed an increase in encystation during the incubation period of the irradiated culture compared to the control culture ( Fig. 5B and D). By day 5, the percentage of encysted cells in irradiated cultures was 50% Ϯ 16% compared to 17% Ϯ 8% in control cultures. Microscopy images also showed that by day 3, irradiated cultures were dominated by green immotile palmelloid (Palmella-like aggregates) cells (Fig. 5C), and by day 5, the majority of cells appeared larger, had thicker cell walls, and displayed increased astaxanthin accumulation, visible as red pigmentation (Fig. 5D). This was largely absent in nonirradiated control cultures (Fig. 5F).

Analysis of H. pluvialis metabolic fingerprints by FT-IR spectroscopy.
To quantify the impact of ionizing radiation on the metabolic fingerprints of H. pluvialis, infrared spectra were collected from both control and irradiated cultures immediately after each daily dose for 5 days (Fig. S4). The spectra from each treatment were qualitatively similar, making it difficult to discriminate between spectral features by visual inspection alone. Therefore, in order to detect the possible development of an irradiated phenotype, cluster analysis was performed using principal-component discriminant function analysis (PC-DFA). Figure 6A shows that the replicates of each treatment form tight clusters in the score plot of discriminant function 1 (DF1) versus DF2, suggesting that the metabolic fingerprints of cells from each treatment class were reproducible. In addition, the samples from both control and irradiated cultures taken on each treatment day cluster tightly together, such that day five control samples (C5), for example, appear most closely related to day five irradiated samples (X5) in the hierarchical cluster analysis (HCA) dendrogram (Fig. 6C). Likewise, there were no strong patterns of separation between clusters of control and treated samples in the score plot of DF1 versus DF3 (Fig. 6B). While there is some evidence of discrimination between respective irradiated and control clusters, Euclidean distances were not quantifiably related to dose (Fig. 6C).
Collectively, these data indicate that the metabolic fingerprints of cells from both the control and irradiated cultures on each day are quantitatively very similar. Hence, it appears that total absorbed doses of up to 80 Gy had no discernible effect on the metabolic profiles detected by FT-IR spectroscopy, despite significant astaxanthin production as a result of irradiation (evidenced via cyst formation and pigmentation; Fig. 5).

DISCUSSION
Despite the obvious challenges of working with materials from highly radioactive nuclear fuel storage facilities, it is becoming clear that microorganisms are uniquely able to colonize a range of nuclear sites worldwide (5-8, 10). This study demonstrates the rich microbial ecology of an outdoor nuclear fuel pond environment, and in Microbial Ecology of a Nuclear Fuel Storage Pond ® particular defines the species involved in microbial blooms at this site and explores their influence on the solubility of the radioactive inventory. Of particular interest is the dominance of a close relative to the freshwater chlorophyte Haematococcus pluvialis in the eukaryote clone libraries at the end of the summer bloom period when the levels of the fission products 90 Sr and 137 Cs in the pond water are at a maximum, and presumably when radiation fluxes within the biomass are also relatively high. Although this species is widely distributed in most parts of the world, it is typically associated with small ephemeral pools which are often subjected to high UV radiation fluxes (35), rather than the large, flushed body of water with its high radionuclide inventory studied here. The dominance of this species in the clone library throughout the algal bloom period, alongside morphological changes recorded in the samples, suggest that this organism may have an adaptive advantage which allows it to thrive in the outdoor spent fuel pond environment.
Experimental irradiation of pure cultures of H. pluvialis indicated that ionizing radiation had little impact on cell viability or the metabolic profile over the time frame tested. However, this coincided with the development of an astaxanthin-accumulating phenotype, which we suggest afforded protection against reactive oxygen species (ROS) generated by ionizing radiation (36,37). Intriguingly, this could imply that this species may be able to withstand the highly radioactive conditions in areas of the outdoor pond as a result of its ability to produce astaxanthin, and thus thrive as a significant primary producer within the surprisingly complex microbial community in this unique environment. It is perhaps surprising that the development of this phenotype was not revealed by FT-IR spectroscopy, as this technique has been used previously to demonstrate molecular diversity within species (38) and to discriminate between different algal species (39,40). It may be that the metabolic fingerprint of H. pluvialis is dominated by biomolecules such as proteins or lipids, which may mask any subtle changes to other biochemical components such as carotenoids. Indeed, the FT-IR spectrum of astaxanthin exhibits absorbance at IR regions similar to those of lipids and proteins (41).
While direct spectroscopic determination of astaxanthin in laboratory cultures was hindered by the difficulty in homogenizing cysts for astaxanthin extraction, light microscopy illustrated a greater percentage of astaxanthin-rich nonmotile encysted cells with increased radiation dose. Carotenoids, such as astaxanthin, protect the photosynthetic apparatus from light-mediated stress by quenching triplet state chlorophyll molecules, singlet oxygen molecules, and other ROS formed within the chloroplast via photooxidation (42). Astaxanthin production in H. pluvialis has been observed under a range of environmental and metabolic stressors, including high UV fluxes, which generate ROS that can damage DNA, proteins, and membranes (31,32). Hence, the production of this molecule in irradiated cultures is likely to quench and scavenge ROS generated by ionizing radiation within cultures of H. pluvialis and is likely to have contributed to the survival and eventual dominance of this organism in the algal bloom noted in this nuclear fuel storage pond.
Finally, there are clearly multiple controls on the radionuclide flux in the dynamic pond environment and on the radionuclide retention mechanisms operating throughout the sampling period. Nonetheless, it is apparent that the microbial biomass present in the pond, dominated by the Haematococcus species, can have a significant impact on radionuclide partitioning within the pond environment. The controls on 137 Cs retention, which peaked during the peak bloom period and 90 Sr, which peaked at the end of the summer are different but are likely linked to photosynthetic activity and biocycling in this complex environment. The photosynthetic activity of the biomass is likely to affect the behavior of radionuclides in several ways. First, during photosynthesis, alkaline conditions up to pH~10.5 can be generated at the localized cell/water interface, due to consumption of CO 2 (43). Under these conditions, sorption of cations is likely to be enhanced, as localized pH values rise, resulting in the biomass surfaces becoming negatively charged and thus more attractive for Sr 2ϩ and Cs ϩ sorption (44)(45)(46)(47)(48). Additionally, high pH favors biomineralization of carbonate phases, which may affect 90 Sr behavior via incorporation into the biomineral matrix (44,47,49). Indeed, when we modeled pond water chemistries, open to the atmosphere and across a pH gradient between 7 and 10 (reflecting credible pH variations within the local biomass environment under photosynthetic conditions [28]), calcite (CaCO 3 ) oversaturation was predicted at pH values greater than 9.2 (see Table S1 in the supplemental material). Importantly, Sr 2ϩ is well-known to incorporate into carbonate mineral phases (47,(49)(50)(51)(52)(53), and thus it appears that in this system bioprecipitation of calcium carbonates due to pH elevation during photosynthesis may contribute to a localized effect where enhanced 90 Sr retention may occur (54,55). In addition, it is possible that biological uptake of these radionuclides may also have occurred (e.g., Cs ϩ substitution for K ϩ [56,57] and Sr 2ϩ substitution for Ca 2ϩ [58]), with higher uptake and retention possibly resulting from the increased algal cell numbers during the bloom period.
In summary, the initial primary aim of this work was to gain fundamental insight into the microbial ecology of a nuclear fuel storage pond, identifying the key organisms that were able to colonize this highly radioactive environment and the cellular adaptation mechanisms at play. These are clearly key steps in understanding how to control microbial biomass production in this large facility, which complicates pond management. However, the bioaccumulation of key radionuclides by the robust microbial community studied here suggests that there could be an important role for such organisms in the sustainable bioremediation of highly radioactive waters in such facilities worldwide, and also in natural aquatic environments contaminated with radionuclides, including 90 Sr and 137 Cs.

MATERIALS AND METHODS
Safety. Samples from a nuclear licensed site are subject to strict protocols during sampling and extensive monitoring procedures prior to shipping to any external facility. Samples containing radionuclides must be handled only by suitably qualified and experienced personnel in a properly equipped radiochemistry laboratory. The possession and use of radioactive materials are subject to statutory controls.
Water sample collection, storage, and characterization. For this study, a custom-made monitoring program was employed, with grab samples of water from the outdoor spent fuel storage pond collected on an approximately weekly basis from May 2010 to October 2010 (n ϭ 16). Samples were collected from approximately 10 cm below the pond surface in sterile 500-ml bottles. Gross alpha and beta measurements were taken at Sellafield, and after appropriate checks, samples were then transported to The University of Manchester within 2 days upon receipt. The samples were stored in a refrigerator at 4°C in the dark. The samples were then filtered (0.2-m Supor 47-mm filter; VWR, Chicago, IL, USA), typically within 24 h of receipt, and the filtrate and filters were stored immediately at 4°C in sterile containers prior to molecular ecology and radiometric and chemical characterization. Samples were analyzed as received ("unfiltered"), and a select series of biomass samples were also analyzed after filtration for molecular ecology and radiochemical and chemical characterization.
Extraction, amplification, and sequencing of 16S and 18S rRNA genes. DNA was extracted from filtered biomass using a PowerSoil DNA isolation kit (Mo Bio Laboratories, Inc., Solana Beach, CA, USA). Extracted 16S and 18S rRNA genes (typically 1,500 bp in length) were selectively amplified by PCR, using oligodeoxynucleotide primers. Purified DNA (2 l) and 1 l of 25 M primer stocks were added to the reaction mix to a final volume of 50 l. Primers used for bacterial 16S rRNA gene amplification were the broad-specificity 8F forward primer and the reverse primer 1492R (59). Primers used for the eukaryote 18S rRNA gene amplification were Euk F and Euk R (60). Samples were amplified using an iCycler (Bio-Rad) thermal cycler. Here the PCR mixtures contained 5 l PCR buffer, 5 l of 25 mM MgCl solution, 1 l of 10 mM deoxynucleoside triphosphate (dNTP) solution, 0.5 l of 25 M primer, and 0.5 l Sigma Taq DNA polymerase, which was made up to a final volume of 50 l with sterile water. Thermal cycling was performed as follows for the bacterial 8F and 1492R primers: 35 cycles of PCR, with one cycle consisting of initial denaturation at 94°C for 4 min, melting at 94°C for 30 s, annealing at 55°C for 30 s, and extension at 72°C for 1 min with the final extension step of 72°C for 5 min (59). For eukaryotic 18S rRNA gene amplification, the following temperature cycle was used: 30 cycles with 1 cycle consisting of denaturation at 95°C for 1.5 min, annealing at 55°C for 1.5 min, and extension at 72°C for 1.5 min (60). The purity of the amplified PCR products was determined by electrophoresis in Tris-acetic acid-EDTA (TAE) gel 1% (wt/vol). DNA was stained with ethidium bromide and viewed under short-wavelength UV light using a Bio-Rad Geldoc 2000 system (Bio-Rad, Hemel Hempstead, Herts, UK).
PCR products were then purified further using a QIAquick PCR purification kit (Qiagen) and ligated directly into the cloning vector pCR 2.1 (Invitrogen) before transformation into competent Escherichia coli cells. White transformants that grew on LB agar containing ampicillin (100 mg ml Ϫ1 ) and 40 ml of 40 mg ml ؊1 of 5-bromo-4-chloro-3-indolyl-␤-D-galactopyranoside (X-Gal) were screened for an insert using PCR (61). Primers were complementary to the flanking regions of the PCR insertion site of the pCR 2.1 cloning vector (ST1F and ST1R primers). The gene products of the clones were treated with shrimp alkaline phosphatase (Promega, UK) and exonuclease I (New England Biolabs, UK) and sequenced directly using an ABI Prism BigDye terminator cycle sequencing kit (v3.1) (Applied Biosystems, UK) following the manufacturer's instructions. Sequences were obtained using the reverse PCR primer 1492R for prokaryote sequences and the reverse Euk R primer for eukaryote sequences. DNA sequences were determined with a Life Technologies 3730XL sequencer at the Genome Analysis Centre (John Innes Genome Laboratory, Norwich, United Kingdom). Sequences (typically 500 bp in length) were analyzed against the NCBI (United States) database using BLAST program packages and matched to known 16S and 18S rRNA gene sequences. Gene sequences were aligned using ClustalX software package and corrected manually.
Radionuclide analysis. To determine 137 Cs activity and retention on biomass, analysis of unfiltered and filtered pond waters was undertaken using gamma ray spectroscopy on samples in a standard geometry. The detector was calibrated using a certified standard solution of 137 Cs, and analyses were performed using an Ortec High Purity germanium detector. For 90 Sr analysis, Cerenkov counting (liquid scintillation counting of Cerenkov radiation) was performed (Packard Tri-Carb model 1900). Again, samples were calibrated against a certified standard solution of 90 Sr and in a standard volume to obtain the counting efficiency (62). In order to assess the error on these samples, triplicate analysis of a filtered water sample was undertaken, and the standard deviations of the measurements were calculated.
PHREEQC analysis. PHREEQC, version 2 (63), using the WATEQ4F.dat data set (64) was used to model the chemical speciation and saturation index of the representative pond water chemistry in a range of scenarios relevant to algal photosynthetic activity. The saturation index [SI ϭ log(IAP/K T ), where K T is the solubility product constant and IAP is the corresponding observed activity product] provides a thermodynamic basis for indicating whether a mineral phase is over-or undersaturated, and in this case, it has been used to assess the potential for biomineralization in the ponds (64).
Chlorophyll a analysis. Chlorophyll a in the Sellafield pond samples was used as a proxy for total phototrophic biomass. A known volume of pond water was collected and filtered (0.2-m Supor 47-mm filter; VWR, Chicago, IL, USA), stored at Ϫ20°C until analysis of samples (typically after 2 weeks), then extracted with 96% ethanol for 20 h at 5°C before centrifugation for 10 min at 2,000 ϫ g (65). The UV absorbance of the supernatant at 665 nm (chlorophyll a) and 750 nm (correction for turbidity) was then determined on a Jenway 6700 UV-visible (UV-Vis) spectrophotometer.
Algal counts. Small subsamples from the spent nuclear fuel pond were preserved with Lugol's iodine within 2 h of sample collection, affording a unique opportunity to analyze the algal biomass at the species level using microscopic counts (29). After full radiological safety assessment, enumeration was conducted on 1-ml samples of pond water, giving sufficiently large count numbers for statistical validity.
Irradiation ). These cultures were irradiated with 16 Gy X-radiation at ambient room temperature using a Faxitron CP-160 Cabinet X-radiator (160 kV; 6 mA; tungsten target). The dose rate was determined as 0.8 Gy min Ϫ1 using Fricke dosimetry (66,67). A further three biological replicates formed the nonirradiated control cultures. These cultures were placed inside the irradiator to control temperature and were shielded by an appropriate thickness of lead. After irradiation, all cultures were incubated at 24°C Ϯ 1°C and shaken at 100 rpm in a light incubator with a photon flux density of 80 mol m Ϫ2 s Ϫ1 with a 16-h:8-h light-dark cycle. Irradiations, and subsequent analysis, were repeated each day for 5 days, giving a total absorbed dose of 80 Gy.
Algal cell number and viability in irradiated laboratory cultures. Cell concentration and viability were determined immediately after irradiation each day via five replicate cell counts using a Neubauer hemocytometer and a light microscope; Zeiss Axio Imager A1 (Carl Zeiss Microimaging GmbH, Germany). To determine cell viability, samples were stained using a fresh solution of Evans blue dye, an azo dye that penetrates nonviable cells only. Briefly, 200 mg of Evans blue (Sigma-Aldrich) in 10 ml phosphatebuffered saline (PBS) solution was added to the samples to give a final concentration of 0.1% (vol/vol) (40). The percentage of encysted cells was also determined by cell analysis under the light microscope. Encysted cells were identified by their larger size (Ͼ10 m), lack of flagella (nonmotile), increased cell wall thickness and orange-red color (increased astaxanthin accumulation [34]). Additionally, cultures were photographed daily using a Zeiss microscope camera connector and a digital camera (Olympus B071).
Analysis of metabolism by FT-IR spectroscopy. The metabolic fingerprints of control and irradiated cells were recorded by Fourier transform infrared (FT-IR) spectroscopy. Aliquots were taken from each biological replicate immediately after irradiation each day. Samples were then centrifuged at 3,000 ϫ g for 5 min; after centrifugation, the supernatant was removed, and the cell pellet was washed twice with sterile 0.9% NaCl solution prior to being centrifuged, flash frozen, and stored at Ϫ80°C. Upon analysis, samples were homogenized by freeze-thawing three times and then resuspended in sterile 0.9% NaCl solution to an OD 625 of 1. A 96-well Si sample plate was washed thoroughly with 2-propanol and deionized water and allowed to dry at room temperature prior to use. Then, 20 l of each algal sample was applied evenly in triplicate onto the plate at random locations (technical replicates) prior to drying at 55°C in an oven for 10 min. All FT-IR spectroscopic analyses were conducted using an Equinox 55 infrared spectrometer equipped with a high-throughput motorized microplate module, HTS-XT (Bruker Optics, Coventry, UK). A deuterated triglycine sulfate (DTGS) detector was employed for absorbance measurements of the samples to be acquired. Thus, three spectra from each biological replicate were collected over the wavenumber range of 4,000 to 600 cm Ϫ1 using the Opus software (Bruker Optics). Spectra were acquired at a resolution of 4 cm Ϫ1 with 64 spectra coadded and averaged to improve the signal-to-noise ratio. The collection time for each spectrum was approximately 1 min.
Multivariate statistical analyses. Prior to statistical analysis, FT-IR spectra were visually inspected and outlying spectra that deviated from the natural variability were removed from the data set. Features arising from CO 2 at 2,400 to 2,275 cm Ϫ1 and below 700 cm Ϫ1 were removed from the spectra and replaced with a smoothed trend. An extended multiplicative scatter correction (EMSC) method was used to normalize spectra. This method was originally developed to reduce the effects of light scattering by particles and is particularly effective at removing noise and unavoidable intensity shifts from the spectra (68,69). Prior to principal-component discriminant function analysis (PC-DFA) and hierarchical cluster analysis (HCA), spectra were scaled using an autoscaling process, such that for all spectra, the intensities at each wavenumber had a mean of zero and a standard deviation of one. Unsupervised principalcomponent analysis (PCA) was performed on the data to reduce data dimensionality (70). Principal components (PCs) were then extracted during this analysis, and discriminant function analysis (DFA) was applied to these PCs (i.e., PC-DFA). DFA is a supervised method, whereby discrimination between groups is based on a priori knowledge of experimental class structure. The algorithm acts to maximize between-group variance and minimize within-group variance. In this experiment, classes were assigned based on treatment and day of experiment. Discriminant function scores from PC-DFA were then used for HCA. The HCA algorithm uses the mean of the scores of each group to construct a dendrogram based on Euclidean distance between groups.
Accession number(s). DNA sequences from the Sellafield outdoor pond bacterial clone libraries have been submitted to GenBank under GenBank accession numbers KC009702 to KC009761.