Biostimulation and microbial community profiling reveal insights on RDX transformation in groundwater

Abstract Hexahydro‐1,3,5‐trinitro‐1,3,5‐triazine (RDX) is a high explosive released to the environment as a result of weapons manufacturing and testing worldwide. At Los Alamos National Laboratory, the Technical Area (TA) 16 260 Outfall discharged high‐explosives‐bearing water from a high‐explosives‐machining facility to Cañon de Valle during 1951 through 1996. These discharges served as a primary source of high‐explosives and inorganic‐element contamination in the area. Data indicate that springs, surface water, alluvial groundwater, and perched‐intermediate groundwater contain explosive compounds, including RDX (hexahydro‐1,3,5‐trinitro‐1,3,5‐triazine); HMX (octahydro‐1,3,5,7‐tetranitro‐1,3,5,7‐tetrazocine); and TNT (2,4,6‐trinitrotoluene). RDX has been detected in the regional aquifer in several wells, and a corrective measures evaluation is planned to identify remedial alternatives to protect the regional aquifer. Perched‐intermediate groundwater at Technical Area 16 is present at depths from 650 ft to 1200 ft bgs. In this study, we examined the microbial diversity in a monitoring well completed in perched‐intermediate groundwater contaminated by RDX, and examined the response of the microbial population to biostimulation under varying geochemical conditions. Results show that the groundwater microbiome was dominated by Actinobacteria and Proteobacteria. A total of 1,605 operational taxonomic units (OTUs) in 96 bacterial genera were identified. Rhodococcus was the most abundant genus (30.6%) and a total of 46 OTUs were annotated as Rhodococcus. One OTU comprising 25.2% of total sequences was closely related to a RDX ‐degrading strain R. erythropolis HS4. A less abundant OTU from the Pseudomonas family closely related to RDX‐degrading strain P. putida II‐B was also present. Biostimulation significantly enriched Proteobacteria but decreased/eliminated the population of Actinobacteria. Consistent with RDX degradation, the OTU closely related to the RDX‐degrading P. putida strain II‐B was specifically enriched in the RDX‐degrading samples. Analysis of the accumulation of RDX‐degradation products reveals that during active RDX degradation, there is a transient increase in the concentration of the degradation products MNX, DNX, TNX, and NDAB. The accumulation of these degradation products suggests that RDX is degraded via sequential reduction of the nitro functional groups followed by abiotic ring‐cleavage. The results suggest that strict anaerobic conditions are needed to stimulate RDX degradation under the TA‐16 site‐specific conditions.

The degradation of RDX through anaerobic biodegradation has been extensively investigated (Scheme 1) (Beller, 2002;Bernstein & Ronen, 2012;Fournier, Halasz, Spain, Fiurasek, & Hawari, 2002;Hawari et al., 2000;Jackson, Raylot, Fournier, Hawari, & Bruce, 2007). The main degradation pathways described involve either a sequential biotic reduction of the nitro functional groups followed by abiotic ring-cleavage  or a direct denitration followed by hydration and subsequent ring-cleavage (Jackson et al., 2007). The intermediates that accumulate in solution as a result of the sequential biotic reduction pathway include MNX, DNX, TNX, NDAB, and MEDINA. In contrast, when the breakdown pathway involves denitration and ring-cleavage, the breakdown products include MEDINA, NDAB, nitrate, and formaldehyde.
Stepwise denitration of RDX involves a nitrate reductase which is a ubiquitous enzyme possessed by a diverse group of bacteria, especially denitrifying bacteria (Bhushan et al., 2002). Degradation of RDX through denitration and ring-cleavage involves the microbial P450 system which was shown to be able to degrade RDX under both aerobic and anaerobic conditions (Jackson et al., 2007). The cytochrome P450 system (XplA and XplB) was originally cloned from Rhodococcus rhodochrous (Rylott, Jackson, Sabbadin, Seth-Smith, & Edwards, 2011;Rylott, Budarina, et al., 2011;Seth-Smith, Rosser, Basran, Travis, & Dabbs, 2002). Expression of xplA and xplB is highly induced in the presence of RDX (Indest, Hancock, Jung, Eberly, & Mohn, 2013;Indest, Jung, Chen, Hancock, & Florizone, 2010). Recent studies have shown that production of xplA in Arabidopsis plants confers both the ability to remove RDX from liquid culture and resistance to the toxic effects of RDX (Rylott et al., 2006;Rylott, Jackson, et al, 2011;Rylott, Budarina, et al., 2011). xplA and xplB exist in various genera including Rhodococcus, Gordonia, and Williamsia are commonly found in soil and groundwater (Halasz, Manno, Perreault, Sabbadin, & Bruce, 2012). The global distribution of RDX-degrading bacteria containing xplA and xplB gene homologs suggests that denitration may represent a key RDX degradation pathway in nature (Andeer, Stahl, Bruce, & Strand, 2009). Besides P450 enzymes, two flavin mononucleotidecontaining oxidoreductase genes xenA and xenB, have been cloned from Pseudomonas (Blehert, Fox, & Chambliss, 1999). Monoculture of the Pseudomonas strains harboring these two enzymes demonstrated groups followed by abiotic ring-cleavage. The results suggest that strict anaerobic conditions are needed to stimulate RDX degradation under the TA-16 site-specific conditions.

K E Y W O R D S
biodegradation, bioremediation, microbial structure, pseudomonas, water F I G U R E 1 Location of TA-16 and other Laboratory technical areas at Los Alamos National Laboratory that both XenA and XenB were able to degrade RDX (Fuller et al., 2009).
Interestingly, XenB exhibited a broader substrate specificity than XenA.
The activities of both enzymes are significantly high when degrading RDX for anaerobic conditions compared with aerobic conditions. Biostimulation has been examined as a remediation approach to treat HE contamination including RDX. Various nutrients including acetate and edible vegetable oil are known to promote bacterial growth and RDX degradation (Livermore, Oh, LePuil, Arnseth, & Mattes, 2013;Schaefer et al., 2007). Acetate is a widely applied carbon source which enriches Fe (III)-reducing bacteria such as Pseudomonas. Multiple studies have shown that Fe (III)-reducing bacteria degrade RDX by direct reduction or indirectly by electron shuttling (Hawari et al., 2000). On the other hand, emulsified edible oils have been successfully used to enhance biodegradation of RDX. The procedures and applications of vegetable oils for the bioremediation of RDX are applicable to numerous other biodegradable contaminants like nitrates, chlorinated solvents, and perchlorates.
Biostimulation using acetate and vegetable oil is normally carried out under anaerobic conditions, when RDX can act as an electron acceptor to support microbial respiration (Beller, 2002;Bernstein & Ronen, 2012).
Degradation of RDX in the presence of oxygen has also been reported and where microorganisms utilize RDX as a carbon source or a nitrogen source (Fuller, McClay, et al, 2010;Fuller, Perreault, & Hawari, 2010).
In this study, we surveyed the microbial profile of RDX-containing groundwater to determine if microorganisms are playing any active role in RDX degradation, examined the potential existence of RDX biodegradation signatures, and evaluated the response of endogenous microbes to biostimulation. This work focused on water samples col- The samples were immediately stored on ice in the field and during transportation, and then kept at 4°C in the laboratory until further analysis.

| Analytical techniques
RDX and its degradation products were analyzed on a Dionex Summit HPLC (Thermo Scientific, USA) system using the EPA method (METHOD 8330A). The HPLC was equipped with an Acclaim Explosives E1 column 25 cm × 4.6 mm E1 (Thermo Scientific, Waltham, MA). The flow rate used was 0.80 ml/min and the mobile phase composition was 52% MeOH and 48% DI Water. Absorbance detection wavelength was set at 254 nm. RDX certified standards (Ultra Scientific, North Kingstown, RI) were used for sample quantification. Degradation products TNX, DNX, and MNX were obtained from SRI International, Menlo Park, CA and were used to establish calibration curves for quantitative analysis.

| Microbial profiling analysis
The groundwater samples were processed immediately after reception from the field. Processing consisted of filtering 100-500 ml of the water samples using 47 mm, 0.2-μm pore size polycarbonate filters (Thermo Scientific) to concentrate the microbial biomass. Total DNA was extracted from bacterial cells collected on each filter membrane following the method of The UltraClean ® Microbial DNA Isolation kit (MO BIO). DNA extracts were used to amplify the V4 region of bacterial 16S rRNA genes using bacterial barcoded primers (515F-806R [GTGCCAGCMGCCGCGGTAA and GGACTACHVGGGTWTCTAAT, respectively]) (Hugerth et al., 2014). Amplicons (equivalent to library fragments) were purified and size selected using AMPure XP beads, quantitated by picogreen assay, normalized, pooled, and sequenced on an Illumina MiSeq instrument.
The sequencing run resulted in ~100,000 paired reads per sample with an average read length of 295 bp. Paired reads were combined to produce 290 bp sequences corresponding to the V4 region and filtered to retain those with an average quality score of greater than or equal to 30. Sequence data were processed using the QIIME software package v1.9.1 (Caporaso et al., 2010). Operational taxonomic units (OTUs) were clustered at the 96% similarity level. The most abundant sequence in each cluster was chosen as a representative. Alpha diversity analysis was performed using QIIME script (alpha_rarefaction.py). To determine how bacterial community compositions varied across samples, principal coordinate analysis (PCoA) was performed by comparing unweighted UniFrac profiles for each sample in QIIME. NCBI Blast was used to assign representative sequences to genus or species levels. All sequences obtained in this study were deposited at the NCBI Sequence Read Archive (SRA) and are available under the accession number PRJNA318785 (https://submit. ncbi.nlm.nih.gov/subs/bioproject/SUB1472271/overview).

| Phylogenetic analysis
A comparative analysis of nucleotide sequences was performed by Basic Local Alignment Search Tool (BLAST) at the National Center for Biotechnology Information (NCBI; http://www.ncbi.nlm.nih.gov/ Blast.cgi/) to obtain sequences of 16S rDNA from species closely related to Pseudomonas and Rhodococcus for phylogenetic analysis.
16S rDNA sequences were aligned using CLUSTAL X (Jeanmougin, Thompson, Gouy, Higgins, & Gibson, 1998) and adjusted manually, as necessary. The resulting data matrix was first analyzed using equally weighted maximum parsimony in PAUP* (Swofford, 2002). Maximum parsimonious trees were sought using the heuristic search strategies of PAUP*. A neighbor-joining analysis was also performed using the uncorrected pairwise nucleotide differences ("p") in PAUP*. The confidence level of branches was evaluated by bootstrap analysis of 10,000 replicates (Felsenstein, 1985).

| Enumeration of bacterial population
Culturable microbial populations were enumerated by dilution plate count technique (Wang, Korban, Pusey, & Zhao, 2012;Wang et al., 2015). For dilution plating, groundwater samples were serially diluted in normal sterile saline (0.9%) and 100 μl of suspension from each dilution was plated on Luria-Bertani (LB) agar plate in triplicate and incubated at room temperature for 8 days. Total microbial counts were determined by direct count using a Hemocytometer plate (Cambridge Instruments, Inc) according to manufacturer's protocol. added to the sealed vial at the start of the experiment using a syringe equipped with a needle. About 100 ml air was sealed within each vial to create microoxic treatments. For strict anoxic stimulation, vials were purged with nitrogen gas for 10 min to remove oxygen. Samples were incubated in the dark at room temperature (22°C) under constant shaking at 50 rpm. Samples were prepared in triplicates for the following four conditions: AC1: CdV-16-4ip cultivated with acetate + initial oxygen; AC2: CdV-16-4ip cultivated with acetate-oxygen; OIL1: CdV-16-4ip cultivated with safflower oil-oxygen; OIL2: CdV-16-4ip cultivated with safflower oil + initial oxygen. Controls with no amendments and sterile controls were also setup in triplicates to account for RDX degradation under unstimulated biotic and abiotic conditions. For abiotic controls, the vials were autoclaved at 121°C to kill indigenous microbes present in the groundwater. Triplicate control cultures were also prepared under aerobic conditions in conical flasks and amended with acetate and safflower oil. All vials were sampled routinely every 3 to 7 days by removing a sample through the rubber stopper using a 1.0 ml syringe. The samples were analyzed for RDX degradation and production of RDX degradation products, and acetate concentration. Samples were collected from each cultivation condition after 5 weeks (i.e., AC1, AC2, OIL1, OIL2, controls) and processed for microbial profiling.

| Physical and chemical characteristics of groundwater
The groundwater samples collected in our studies were obtained from well CDV-16-4ip which has a screened interval between 815 and 879 ft interrogating the Puye Formation. This formation is primarily made up of poorly sorted, unconsolidated, dacitic boulders, cobbles, and gravels that are either clast-supported or matrix-supported. Sand, silty sand, and silt are common matrix materials (LANL, 2011). The groundwater is well oxygenated with oxygen concentrations varying from 7.5 to 8.0 mg/L, the ORP measurements varied between 200 and 275 mV, and the pH is neutral typically varying between 7.33 and 7.67. The concentrations of the major anions in one of the CdV-16-4ip samples were: chloride = 3.5 mg/L, fluoride = 0.1 mg/L, nitrate as nitrogen = 0.89 mg/L, sulfate = 3.5 mg/L, and Na + = 9.7 mg/L. The alkalinity CO 3 +HCO 3 = 50 mg/L and its Ca 2+ content is 10 mg/L, Mg 2+ = 3.1 mg/L. The total organic carbon concentration is 0.57 mg/L and total dissolved solids = 126 mg/L. The concentration of RDX is about 160 ppb and the concentrations of the degradation products TNX, MNX, and DNX are typically less than 1 ppb. Culturable cell counts as enumerated using LB agar medium were about 3.6 × 10 2 CFU/ml. In contrast, total bacteria counted using hemocytometer was 8.7 × 10 4 cells/ml.

| Bacterial community analysis
A total of 98,405 bacterial 16S rRNA gene sequences were recovered for the CdV-16-4ip sample and used for community analyses by QIIME. OTUs were assigned by clustering sequences with over 96% sequence identity. A number of 1,605 OTUs distributed in 15 phyla were identified indicating high microbial diversity in the sample.  (Figure 2b). Table S1 lists dominant OTUs (>1%) and their closest relatives found in the GenBank. The two most dominant OTUs (>20%) from CdV-16-4ip are identified as Rhodococcus erythropolis HS4 (NR_074622) with 100% sequence identity and is known to degrade RDX (Chong, 2011) and Nocardia ignorata DSM 44496 also with 100% sequence identity and is a human nocardiosis pathogen isolated from respiratory specimens in Europe (Rodriguez-Nava, Couble, & Khan, 2005). Other relative abundant OTU sequences were related to F I G U R E 2 Microbial community analysis of CdV-16-4ip groundwater sample. (a) Relative abundances of major bacterial phyla. Data were analyzed using QIIME (Caporaso et al., 2010

| Pseudomonas and Rhodococcus
Among known RDX-degrading genera, only Pseudomonas and Rhodococcus were detected in the CdV-16-4ip sample. Many Rhodococcus species are enriched by polycyclic aromatic hydrocarbon and related contaminants (Yu, Ke, Wong, & Tam, 2005). Rhodococcus was the most abundant genus in CdV-16-4ip (30.6%). A total of 46 OTUs were annotated as Rhodococcus. Phylogenetic trees of partial 16S rDNA sequences reconstructed using neighbor-joining methods revealed evolutionary positions of the 46 OTUs in relation to 27 known Rhodococcus species (Figure 3). Four strains in related genera Agreia, Herbiconiux, Tessaracoccus, and Propionibacterium were selected as outgroups. All of the 46 OTUs were clustered together with Rhodococcus strains and separated from outgroups confirming the QIIME annotation. Based on their distance to known Rhodococcus species, OTUs here were grouped into two categories. One comprising those distantly related to known Rhodococcus strains: Clade 1 (23 OTUs), Clade 2 (10 OTUs), and Clade 3 (4 OTUs); while the other constituting 36 OTUs closely related to known species. One OTU (573976 CBN. 40 73) comprising 25.2% of total sequences was closely related to a RDX-degrading strain R. erythropolis HS4 (Figure 3) (Chong, 2011). In contrast, Pseudomonas appeared to be less diverse and abundant than Rhodococcus. Most OTUs were closely clustered with reported strains except for two OTUs (boxed) in Clade 1 (Figure 4). An OTU (1108886 CBN. 39 944) was placed in the same clade with two RDX-degrading strains P. putida II-B and P. fluorecens I-C ( Figure 4).
Together, these data suggest that the CdV-16-4ip groundwater harbors a diverse group of Rhodococcus and Pseudomonas including some related to RDX-degrading strains.
F I G U R E 4 Phylogeny of Pseudomonas in CdV-16-4ip groundwater sample. Phylogenetic tree of 16S rRNA gene sequences showing the phylogenetic affiliation of the operational taxonomic units (OTUs) from the groundwater samples. The neighbor-joining tree was constructed from the 16S rRNA V4 hypervariable sequences of representative clones of each OTU and sequences retrieved from the GenBank database. The branch indicated by "*" contains outgroups. Numbers on branches represent bootstrap estimates from 10,000 replicate analysis; values <50% are not indicated. Strain names, if any, and GenBank accession numbers are given following the species names. Known RDX-degrading Pseudomonas strains were underlined F I G U R E 3 Phylogeny of Rhodococcus in CdV-16-4ip groundwater sample. Phylogenetic tree of 16S rRNA gene sequences showing the phylogenetic affiliation of the operational taxonomic units (OTUs) from the groundwater samples. The neighbor-joining tree was constructed from the 16S rRNA V4 hypervariable sequences of representative clones of each OTU and sequences retrieved from the GenBank database. The branch indicated by "*" contains outgroups. Numbers on branches represent bootstrap estimates from 10,000 replicate analysis; values <50% are not indicated. Strain names, if any, and GenBank accession numbers are given following the species names. Known RDX-degrading Rhodococcus strains were underlined

| RDX degradation in biostimulated cultures
Both acetate and vegetable oil are known to stimulate microbial activity and enhance RDX degradation. We amended groundwater samples with acetate and safflower oil and examined RDX degradation under variable geochemical conditions to determine which factors are most relevant to RDX degradation. As shown in Table 1, little bacterial growth and RDX transformation were observed in the control samples in which no amendments were added. RDX also remained unchanged in the abiotic controls (data not shown). Acetate amendment stimulated microbial growth, with a stronger effect under microoxic conditions. However, RDX concentration were not significantly reduced in these samples over 5 weeks. A similar effect was also observed for safflower oil under microoxic environment.
Even though the microbes grew to densities of 7.9 × 10 9 cells/ml, no significant RDX degradation was observed. In contrast, safflower oil enhanced bacterial growth to the same level and promoted RDX degradation under more strict anaerobic conditions. Quantitative analysis of RDX and its degradation products by HPLC (Figure 5a, b) showed a transient increase in the concentrations of DNX and MNX during the active degradation of RDX, but their concentrations decrease rapidly when RDX degradation reaches a plateau. These data are consistent with the literature data which identified DNX, MNX, and TNX as intermediate products in the anaerobic degradation of RDX (Bernstein & Ronen, 2012). It also suggests that RDX degradation proceeds by the stepwise reduction of the nitroso groups followed by ring-cleavage and is consistent with the mechanisms reported in the literature .
The concentration of 4-NDAB remained fairly constant throughout the experiment and TNX was barely detectible until the 5th week.
The concentration of NDAB was elevated in all samples including controls with no noticeable RDX degradation and did not change over time (Fig. S2). This is likely due to the lack of separation of NDAB from other small molecular weight metabolites present in the media which are not well resolved by the HPLC method. These results show that RDX degradation is enhanced under strict anaerobic conditions, but very negligible in the presence of oxygen. The availability of oxygen seems to inhibit RDX degradation which indicates that RDX under our current experimental conditions is degraded mainly through anaerobic respiration, which is also supported by the detection of degradation products identified as intermediates of RDX anaerobic respiration.
RDX remained stable in the anaerobic reactors amended with acetate under both anoxic and microoxic conditions, although the cell numbers were significantly increased. This means that the type of nutrients provided is critical for RDX degradation possibly by differentially regulating the assemblage of groundwater microbial community.
T A B L E 1 Effect of acetate, safflower oil, and oxygen on microbial growth and RDX degradation. Initial groundwater bacterial population was 8.7 × 10 4 cells/ml. Final cell numbers and RDX concentrations were measured after 5 weeks

| Evolution of the microbial population after biostimulation
To determine how nutrients and oxygen content affect the microbial community in the stimulated groundwater, we performed a microbial profiling of the bacterial communities before and 7 weeks after the biostimulation under different conditions. In the two nonstimulated control samples, a slight increase was found in Actinobacteria and a decrease was observed for Proteobacteria. In contrast, Proteobacteria populations were significantly increased in all stimulated samples comprising over 90% (Figure 6a). The microbes representing the phyla Bacteroidetes, Nitrospirae, Planctomycetes, Chlamydiae, Chloroflexi, and Verrucomicrobia were decreased or completely eliminated. Rarefaction curve analysis show a significant reduction of species richness after biostimulation (Figure 6b). The original 1,605 OTUs were reduced to 440, 539, 512, and 577 in the AC1, AC2, OIL1, and OIL2, respectively. Dominant OTUs representing over 67.5% of the total population were all significantly decreased, with a stronger effect under anaerobic conditions (Table 2). Three OTUs related to Frateuria, Hydrogenophaga, and Pedobacter were completely eliminated suggesting that biostimulation is detrimental to these species.
Rhodococcus are well known as aerobic bacterial species and therefore are not able to survive under anoxic conditions. As expected, they were barely detected under anaerobic growing conditions for both acetate and vegetable oil (Figure 7a). Species in Pseudomonas are capable of using a range of nutrients including vegetable oil (Song, Jeon, Choi, Yoon, & Park, 2008). Therefore, they are favored when cultivated with safflower oil, especially under microoxic conditions ( Figure 7b).
Since OIL1 was the only sample showing RDX degradation ( Figure 5; Table 1), we were particularly interested in the bacterial OTUs enriched in this sample. found in the OIL2-, AC1-, and AC2-stimulated samples. Instead of OIL2 which shares the same nutrient type with OIL1, it is AC2 that showed a more similar pattern of dominant OTUs to OIL1 (Table 3). We also performed a principal coordinate analysis (PCoA) of the OTUs in the five samples ( Figure S1). PC1, PC2, and PC3 accounted for 62.97%, 23.35%, and 12.53% of total variance, respectively. The score plot PC1 clearly showed that the four stimulated samples were separated from the original CdV-16-4ip sample. This indicates that biostimulation significantly shifts the composition of the microbial community. The two anaerobic samples AC2 and OIL1 were grouped together; while, AC1 and OIL2 were relatively far from each other in the plot. Altogether these results suggest that nutrient type and oxygen are interconnected and both are critical in shaping the microbiome during biostimulation.

| DISCUSSION
RDX degradation in groundwater is considered to be an outcome of complex interplay between the physicochemical factors and microbial activities. Although culture-independent studies have revealed F I G U R E 6 (a) Relative abundances of bacterial phyla in CdV-16-4ip before and after cultivation. Phyla with over 1% relative abundance were plotted. (b) Rarefaction curves for microbial diversity generated by QIIME (Caporaso et al., 2010). 200-1000 16S rRNA sequences were randomly selected and operational taxonomic units Also, the utilization of RDX as a carbon source seems to be very limited under our experimental conditions. Overall, the geochemical conditions of the groundwater represent a nutritionally limited medium that is not very conductive to the indigenous microorganism activity. This is consistent with the low RDX degradation activity via microbes.
Using culture-independent studies, we identified the presence of 96 genera in 15 phyla within the groundwater sample. A high abundance of OTUs was closely related to Rhodococcus strain HS4 (Figure 3) which exhibit low RDX biodegradation activities (Seth-Smith et al., 2008). In contrast, no OTUs are closely related to Rhodococcus strain 11Y ( Figure 3) which contain the highly effective RDX-degrading XplA-B system (Seth-Smith et al., 2002). It seems that OTUs related to Rhodococcus strain HS4 are enriched by RDX at the Los Alamos site, whereas Rhodococcus 11Y-related strains are not present. This explains the low RDX degradation activities as indicated by little RDX degradation products in the groundwater. Our study also establishes that in spite of poor nutrient content in groundwater, the bacterial diversity in the RDX-contaminated groundwater remains significantly high.
Although denitration and ring-cleavage via Rhodococcus genes xplA and xplB generate MEDINA, NDAB, nitrate, and formaldehyde, production of the nitroso-derivatives suggests that RDX is degraded via XenA and XenB sequential reduction by the enriched Pseudomonas OTU. Since the complete set of microbial species involved in RDX degradation is not well defined, it is equally possible that other bacterial genera enriched by safflower oil rather than acetate are involved in RDX degradation in anaerobic conditions. Ongoing work is being carried out to identify genes expressed under different conditions to help identify the specific microbes involved in RDX degradation.
Our microbial profiling shows that bacterial communities in the perched-intermediate groundwater environment at TA-16 are composed of diverse taxonomic groups. Furthermore, biostimulation analysis identified environmental conditions favoring RDX degradation.

F I G U R E 7 Relative abundances of Rhodococcus (a) and
Pseudomonas (b) in the groundwater samples before and after cultivation. Figures were generated using numbers from QIIME analysis (Caporaso et al., 2010). 4ip: CdV-16-4ip before cultivation; AC1: CdV-16-4ip cultivated with acetate + oxygen; AC2: CdV-16-4ip cultivated with acetate-oxygen; OIL1: CdV-16-4ip cultivated with safflower oil-oxygen; OIL2: CdV-16-4ip cultivated with safflower oil + oxygen These data show that strict anaerobic conditions are required to drive biotic RDX degradation. Such conditions were also seen in remediation of contaminated aquifers that contain uranium or other radionuclides (through coffinization, etc.), which favors the immobilization of uranium (Guo et al., 2014(Guo et al., , 2015. Our studies also show the importance of the initial availability of oxygen on the success of a bioremediation approach. Studies in an open system with continued flow are currently being carried out to specifically examine the effect of oxygen availability on microbial diversity and RDX degradation. The results will be reported in a separate paper.

ACKNOWLEDGMENT
We thank the office of environmental programs for facilitating groundwater sampling. Support for this work was provided by the DOE office of environmental management and Los Alamos National Laboratory.