ISOLATION, MOLECULAR & PHYSIOLOGICAL CHARACTERIZATION OF SULFATE-REDUCING, HETEROTROPHIC, DIAZOTROPHS

Nitrogen (N2) fixation is the process by which N2 gas is converted to biologically reactive ammonia, and is a cellular capability widely distributed amongst prokaryotes. This process is essential for the input of new, reactive N in a variety of environments. Heterotrophic bacterial N fixers residing in estuarine sediments have only recently been acknowledged as important contributors to the overall N budget of these ecosystems and many specifics about their role in estuarine N cycling remain unknown, partly due to a lack of knowledge about their autecology and a lack of cultivated representatives. Nitrogenase reductase (nifH) gene composition and prevalence in Narragansett Bay sediments has revealed that two distinct phylogenetic groups dominate N-fixation. Analysis of nifH transcripts has revealed one active group to be the Desulfovibrionaceae, belonging to the Deltaproteobacteria. We see nifH expression from this group across sampling sites and times, despite the fact that Narragansett Bay sediments are replete with combined N, which is thought to inhibit N fixation in the environment. Here were present the genomic and physiological data in relation to N2 fixation by two heterotrophic members of the Desulfovibrionaceae, isolated from sediments of the Narragansett Bay estuary in 2010 and 2011, respectively: Desulfovibrio sp. NAR1 and Desulfovibrio sp. NAR2.To elucidate how nitrogenase activity in these organisms responds to the presence of different sources of combined N, and to link observed physiology with genomic potential (i.e. gene content), we performed a two-part study that coupled high-throughput genome sequencing and analysis with physiological investigations of growth on different N sources with N fixation rate measurements. The genomes of the two diazotrophic Narragansett Bay Desulfovibrio isolates (NAR1 and NAR2) were sequenced using a high throughput platform, subsequently assembled, annotated, and investigated for genes related to N fixation and overall N metabolism which were then compared across 34 additional Desulfovibrio genomes which were publicly available. To link findings at the molecular level with observations at the physiological level, N fixation rates were measured using the acetylene reduction assay (ARA) under conditions free of reactive N, and under the following combined N conditions: 12 mM urea, 12 mM NO3, and 12 mM NH3. Both isolates can sustain growth by N2 fixation in the absence of biologically available N and our data indicate that nitrogenase activity is completely inhibited by the presence of ammonia, yet uninhibited by nitrate and urea, which are other forms of combined N found in Narragansett Bay. This agrees with observations made at the genome level, as neither our isolates nor the other Desulfovibrio examined in comparison appeared to have the genetic capability to use NO3 or urea catabolically to meet cellular N demands. This study indicates that the Desulfovibrionaceae are restricted in terms of the N sources they are capable of using, and that this may be a factor contributing to the observed N fixation by this group in sediments that are not limited for sources of combined N. Genome sequencing also reveals both isolates to be metabolically versatile and unique. The NAR1 isolate possesses genes involved in bacterial mercury methylation, and displays near obligate biofilm formation. Genes were also found in the NAR1 isolate which suggest the involvement of c-di-GMP in cell-to-cell communication and biofilm formation. This is particularly interesting since biofilm formation and quorum sensing is not well characterized among the Desulfovibrio, despite biofilm formation being displayed by many members of this genus. While investigating the role of these organisms as important contributors of fixed N in Narragansett Bay, it was critical that we examine these additional aspects of their metabolism in order to gain a better understanding of controls on growth that may also impact biomass and the ability of these organisms to achieve significant rates of N fixation in the environment.

Narragansett Bay Desulfovibrio isolates (NAR1 and NAR2) were sequenced using a high throughput platform, subsequently assembled, annotated, and investigated for genes related to N fixation and overall N metabolism which were then compared across 34 additional Desulfovibrio genomes which were publicly available. To link findings at the molecular level with observations at the physiological level, N fixation rates were measured using the acetylene reduction assay (ARA) under conditions free of reactive N, and under the following combined N conditions: 12 mM urea, 12 mM NO 3 -, and 12 mM NH 3 . Both isolates can sustain growth by N 2 fixation in the absence of biologically available N and our data indicate that nitrogenase activity is completely inhibited by the presence of ammonia, yet uninhibited by nitrate and urea, which are other forms of combined N found in Narragansett Bay. This agrees with observations made at the genome level, as neither our isolates nor the other Desulfovibrio examined in comparison appeared to have the genetic capability to use NO 3 -or urea catabolically to meet cellular N demands. This study indicates that the Desulfovibrionaceae are restricted in terms of the N sources they are capable of using, and that this may be a factor contributing to the observed N fixation by this group in sediments that are not limited for sources of combined N. Genome sequencing also reveals both isolates to be metabolically versatile and unique. The NAR1 isolate possesses genes involved in bacterial mercury methylation, and displays near obligate biofilm formation. Genes were also found in the NAR1 isolate which suggest the involvement of c-di-GMP in cell-to-cell communication and biofilm formation. This is particularly interesting since biofilm formation and quorum sensing is not well characterized among the Desulfovibrio, despite biofilm formation being displayed by many members of this genus. While investigating the role of these organisms as important contributors of fixed N in Narragansett Bay, it was critical that we examine these additional aspects of their metabolism in order to gain a better understanding of controls on growth that may also impact biomass and the ability of these organisms to achieve significant rates of N fixation in the environment.

INTRODUCTION The Role of Heterotrophic Diazotrophs in Estuarine Sediments
Estuarine sediments, such as those in Narragansett Bay, have historically been shown to be major regions of nitrogen (N) removal via the activity of denitrifying bacteria (Nixon et al. 1996). Denitrification results in the release of nitrogen (N 2 ) gas and the loss of biologically reactive N. The opposing process of N fixation converts N 2 gas to biologically reactive ammonia, and has been demonstrated to occur in Narragansett Bay sediments by anaerobic diazotrophs (Fulweiler et  In Narragansett Bay the most active of these heterotrophic diazotrophs, based on the number of nifH (dinitrogenase reductase) transcripts, belong to two distinct phylogenetic groups; members of the Desulfovibrionaceae and the Geobacteraceae ). In order to gain a better understanding of the ecology of these active N fixers, we attempted to isolate representatives of both phylogenetic groups from Narragansett Bay sediments. We were initially successful in isolating two In other words, shifting from questions regarding "who, what, and where" to questions regarding "why and how" these organisms fix N in their respective habitats.
Anaerobic heterotrophic diazotrophs, specifically members of the Desulfovibrionaceae, present a unique challenge because currently there is no model or set of supporting physiological data that exist which explain why N-fixation activity by these organisms is observed in environments which are not limited for combined N.
The paradigm for aerobic N fixing bacteria is that combined N represses N fixation Howarth et al. 1988), and this is something that still remains to be investigated in anaerobic diazotrophs. Similarly, there has been very little investigation at the genomic level into the N metabolism of these organisms, their potential ability as a group to assimilate different combined sources of N, or a specific description of controls on their overall N fixation. We hypothesize that members of this genus may not be capable of using certain forms of N in an assimilatory manner and thus would need to continue to fix N under otherwise N replete conditions. This may be an important contributing factor to what has been seen regarding their activity in the environment (

Thesis Motivation and Outline
Members of the Desulfovibrio were found to be one of the primary bacterial groups responsible for driving N-fixation activity in Narragansett Bay ). Previously, sulfate-reducing bacteria (SRB) as a group have been found to fix N in culture   still much yet to be discerned regarding their metabolism, nutrient cycling properties, and responses or adaptations to different environmental stressors, particularly from a genomic perspective and more specifically as those factors pertain to N fixation.
There is a similar lack of information regarding N fixation rate data among the Desulfovibrio. Although there are previous studies examining N fixation rates for heterotrophic diazotrophs in sediments using the acetylene reduction assay (McGlathery et al. 1998;Welsh et al. 1996b), these studies have used environmental samples and thus are examining the overall N fixation ability of a mixed microbial consortia and are not able to establish cell-specific rates of N 2 fixation, or a connection between those rates and the responsible cells. This lack of evidence is most likely due to the fact many of these organisms remain uncultured, and where there are cultivated representatives the potential contributions to heterotrophic N fixation by those specific organisms remain largely disregarded.
The lack of cultivated Desulfovibrio representatives and corresponding analysis of their genomic capabilities in regards to N fixation and metabolism, combined with a lack of organism specific physiological data regarding N fixation rates and rate responses to environmental N conditions, have provided motivation for the current study.

Introduction:
Estuarine sediments typically exhibit a nitrogen (N) cycle that is dominated by processes of N removal, such as coupled nitrification and denitrification (1). The opposing process of N 2 fixation by prokaryotic organisms, known as diazotrophs, is the primary source of reactive N to the world's oceans and consequently acts as a control on both the N budget and primary production in many marine ecosystems Historically, N fixation has been a process primarily attributed to cyanobacterial species residing in the water column, although genetic potential to fix N is widely distributed amongst prokaryotes, including members of bacteria and archaea (4). Due to the energetically unfavorable nature of biological N fixation and typically abundant concentrations of combined N found in estuarine systems, heterotrophic N fixation was previously thought to be an inconsequential process in these environments (5). However, recent research involving direct observations of N 2 flux across the sediment-water interface in a variety of marine, salt marsh, and sea grass systems support that heterotrophic N fixation is an important source of reactive N in these systems, and has begun to alter the historically held conclusions regarding the role of heterotrophic diazotrophs (6-10). Additionally, significant N fixation has been documented in waters where cyanobacteria are not believed to be present or active (11)(12)(13). reducers and have been shown to fix N in culture (15). Members of this genus are also noted for their ability to perform a wide variety of metabolic functions, including the reduction of sulfate to sulfur and sulfide species, the ability to utilize recalcitrant carbon sources, the ability to fix N, and the ability to transform certain metal species (15)(16)(17)(18) Additionally, it has been known for some time that many sulfate reducing bacteria (SRB), particularly members of the Desulfovibrio, have the genetic potential to fix N (23). These organisms have been shown to fix N in a laboratory setting (15,24) and in a variety of habitats including coral reefs, photosynthetic microbial mats, mangrove sediments, sea grass rhizospheres (25-27), shallow estuarine sediments (6,14,28), bioturbated sediments (29), and salt marshes (9). There is further evidence supporting that these organisms play a critical role in supplying fixed N to their environment, particularly in anaerobic or anoxic sediments (14,30) and benthic sediments which are N deficient (31)(32)(33)

Acetylene Reduction Assay
The NA of Desulfovibrio sp. NAR1 was measured under various combined N conditions. 10 mL cultures of NAR1 containing 2 g of 1 mm diameter glass beads in carbonate buffered NBSO, NBSO + 12 mM NH 3 , NBSO + 12 mM NO 3 -, NBSO + 12 mM urea were prepared in either duplicate or triplicate and NA was assessed using the acetylene reduction assay (ARA) and methods described previously by Capone (44).
Acetylene was generated in house by reacting calcium carbide with water and was collected in a 1 L Supel-Inert film gas sampling bag. Acetylene was added to the culture tubes to a final concentration of 10 to 20% of the total headspace. A Shimadzu GC8 gas chromatograph (Shimadzu Corporation, Kyoto, Japan), with a 2.5 m long stainless steel column containing Haysep T packing 80/100 mesh, was used to measure ethylene production in all ARAs. The injector and column were set to 130°C and 100°C, respectively. Gas samples of 100 uL were taken from the tube's headspace with a gas tight syringe and immediately injected into the gas chromatograph.
Ethylene production was measured over the course of 24 hours. Samples were usually measured 1, 3, 6 and occasionally 24 hours after acetylene was added. Cultures that were part of the same assay were inoculated at the same time using the same parent culture. A set of multiple potential parent cultures was established for each set of experimental tubes; NA was measured in the parent tubes to ensure that the inoculum for the experimental cultures was actively fixing N at the time of inoculation, once NA was established in one member of the parent culture series the next un-tampered parent culture was used to inoculate the experimental culture series. A series of potential parent cultures was necessary, rather than measuring the NA of one culture and using that same culture as a parent inoculum, to mitigate the negative effects of long term acetylene exposure on the growth of NAR1. Cultures were also no longer sterile after being used in the acetylene reduction assay (ARA), and so could not themselves be used as parents. Additionally, measurement of NA in the parent culture was a critical means of determining when the culture was ready to be used as inoculum, since cell enumeration of the NAR1 isolate remains challenging, and as of yet no hard correlation between culture age and NA has been established.
After establishing a baseline NA using N limited cultures (Fig. 7 Biomass estimates were ultimately made by extracting DNA from pure cultures grown under N limited conditions, and subsequently using the DNA yield to calculate the total number of genomes present in the culture volume, which was used as representation of cell abundance. Figure 6 shows the log growth of NAR1 under N limited conditions. This organism exhibits very slow growth and very little change in overall biomass when grown under N limited conditions. However, slow growth is not unusual for environmental bacteria such as members of the Desulfovibrio (58), and it is likely that there is some loss of DNA during the extraction process, which would result in a lower reported cell count.
Ethylene production by NAR1 (Fig. 7) was first measured on N limited cultures over a period of ~14 days prior to examining the effects of combined nitrogen on NA in the NAR1 isolate. Previous pilot experiments had shown that no NA occurred before the third day after inoculation, so these time points were not included in the study. Figure 7 shows NA for the NAR1 isolate for the same time period in which growth was measured (Fig. 6). These measurements indicate that peak NA in N limited cultures of NAR1 occurs early on in the growth of the organism, late lag or early log phase, which is a trend that had been observed in previous assays (data not shown). Peak NA is within range of that described for other bacterial isolates (59).
The observed high activity on Day 5, when the culture is young and the cell count is lower, may be due to the N demands of the organisms as they prepare to enter log phase. It is unclear at this time whether biofilm formation plays a role in increased NA or overall N demand. Since Desulfovibrio biofilms are known to be composed primarily of protein (60,61), this remains a possibility and should be considered in future investigations. Further studies would be needed in order to elucidate what specific factors contribute to timing of peak NA in this isolate.

Growth and acetylene reduction of NAR1 under differing combined N conditions
After establishing a baseline NA using N limited cultures (see previous section), ethylene production for cultures of NAR1 grown under different combined N treatments was measured using the ARA (Fig. 9) Keeping in mind that replicates in this experiment are separate cultures and that Fig. 8 shows the total number of cells present in these cultures measured at the respective time-point, instead of the same culture with change in cell abundance quantified over time, it is plausible that the lower cell count on day 10 for all treatments except nitrate is due to fewer cells being present in the inoculum used for those cultures compared to the day 8 cultures. Possibly inflated ethylene production rates due to discrepancies in culture density are corrected for by normalizing ethylene production rates to cell abundance.

Genome sequencing outputs, assembly and annotation
Genome sequencing using the Illumina MiSeq platform was carried out for Desulfovibrio sp. NAR1 and Desulfovibrio sp. NAR2. The number of paired-end, 250bp reads obtained for each isolate was more than 16 million. Phred per-base quality scores of ≥ 10 were reported for NAR1 and NAR2 raw reads, which represent inferred base call accuracy of at least 90% (64,65), for all reads. Ambiguous N bases were not detected in either raw data set. Although these values indicate that the probability of an incorrect base call was minimal for the raw sequences, both data sets were trimmed prior to being used in de novo assemblies. The number of paired-end reads remaining for each isolate after trimming was more than 13 million, which represents a range between 700x and 900x expected genome coverage for each isolate.
Phred per-base quality scores of ≥ 30 were reported for NAR1 and NAR2 trimmed reads, which in turn represent inferred base call accuracy of at least 99.9% for all reads. Ambiguous N bases were not detected in either trimmed data set. The average read length after trimming was 200bp and 210bp for NAR1 and NAR2, respectively.
These resulting quality values represent an improvement over the raw data sets and indicate that the probability of an incorrect base call is minimal for these sequences, at this point both trimmed data sets were considered to be appropriate for generating de novo assemblies.
Statistics for the de novo assemblies of D. sp. NAR1 and D. sp. NAR2 are shown in Table 1 ( Table 1). The NAR1 genome data was assembled first, and used to establish a pipeline for working with the NAR2 data. Both isolate data sets were trimmed and assembled using identical parameters, using all reads in their respective sets. Individual contig coverage values were averaged for both assemblies, and are reported in Table 1. Fold coverage for individual contigs for NAR1 and NAR2 ranged from 339x-909x, and 705x-1,876x, respectively (Supplemental Tables S1 and S2).
Comparisons of assembly statistics such as N 50 , N 75 , number of contigs, and longest contigs for these isolates compared to other published draft genomes indicate that good-quality assemblies have been achieved for both isolates (66,67 were ultimately reordered using Mauve alignments (57), as shown in Figure 2.

Comparison of isolate genomes using IMG/ER and RAST
A combination of IMG/ER, RAST, and manual annotation using BLAST tools (50, 69, 70) was used to annotate both isolate draft genomes. IMG/ER was used as the primary annotator and comparison tool, RAST was used sparingly, and manual annotation was used primarily where automatic annotations seemed questionable, were missing, or when genes were of special interest to this study.
Total coding bases were split into PCGs, RCGs, and hypothetical PCGs. This data, along with the breakdown of RCGs into those accounting for rRNA, tRNA, and other RNA genes, is shown in Table 2 along with the same data for two closely related Desulfovibrio species for each isolate; D. desulfuricans str. ND132 and D. piezophilus for NAR1, and D. desulfuricans subsp. aestuarii and D. alaskensis str. G20 for NAR2.
The more phylogenetically distant representative, D. vulgaris str. Miyazaki F, was also included as an additional point of comparison. Tabulated attributes are similar overall between isolates and the previously sequenced genomes listed. However, the NAR1 genome is note-worthy with respect to the two annotated clustered regularly interspaced short palindromic repeat regions (CRISPRs), where 3 of the other genomes included in Table 2 have only one CRISPR region and 3 do not have any.
CRISPR-Cas systems play a role in adaptive immunity against phages and other invading genetic elements and are present in approximately 40% of sequenced eubacteria genomes and 90% of archaea genomes (71,72). It is interesting to note that NAR2 lacks a CRISPR-Cas system, particularly since these organisms were isolated from the same site and would presumably have been exposed to similar phage attacks and the same foreign DNA in the environment.
Isolate PCGs with KEGG, COG, Pfam and TIGRfam (73-76) annotations in IMG/ER were compared across subcategories in terms of number of genes contained in each subcategory and the percentage of total PCGs with pathway association that the gene number represents. The percentage is not particularly meaningful on its own because it is taken from the total number of genes with pathway association, which can vary depending on which database is being considered, and it is not a representation of total PCGs. Even so, the percentage of genes represented in each of the various subcategories remained relatively consistent between the two isolates. So from a broad, overall standpoint, the two isolates look similar in terms of gene content with function assignments. However, not all possible metabolic subcategories are represented, and details regarding gene content rather than overall gene number cannot be assessed in this manner.
The RAST database uses SEED (69)  This difference is in agreement with the observed exudate and biofilm production in NAR1, which is not observed at all in the NAR2 isolate. The number of genes assigned to motility and chemotaxis subsystems is also of interest; the NAR1 isolate has 136 genes assigned to this subcategory, with 100 associated with flagellar motility and 36 being assigned to bacterial chemotaxis. NAR2, comparatively, has 94 genes assigned to motility and chemotaxis, with all of these genes being associated with flagellar motility. This difference between the two isolates in terms of genes involved in chemotaxis could also be related to the observed biofilm formation in NAR1, and could be part of the pathway that signals biofilm formation. As this pathway is not yet described among the Desulfovibrio, these genes should be considered as targets for future studies involving biofilm formation in this genus.  (84). So although the presence of predicted betalactamases in these isolates may not be surprising in regards to environment or genus assignment, it should still be noted and considered as a possible target for future physiological studies, especially since the potential human pathogenicity of these isolates remains unknown. The NAR1 isolate in particular should be considered, as it contains additional putative genes for colicin V production and fosfomycin resistance.
The presence of these genes has been observed and annotated in other Desulfovibrio, but there has yet to be physiological confirmation of colicin V production in this genus, while fosfomycin resistance has already been observed in some Desulfovibrio species with clinical relevance (85).

Whole genome comparison and alignments
The draft genomes of isolates Desulfovibrio sp. NAR1 and Desulfovibrio sp.
NAR2 were compared to all complete or well-annotated draft Desulfovibrio genomes available in GenBank or IMG databases (50, 86) at the time of this study (for a list of genomes see Supplemental Table S3), which was a total of 34 additional genomes. A multi-gene phylogenetic tree ( Figure 1) for all 36 Desulfovibrio representatives was constructed using 20 different vertically transferred genes (Table S4) alignment, suggest that this genome contains a significant amount of sequence variability compared with D. alaskensis str. G20, which is not surprising as G20 is only 89% similar to NAR2 at the 16s level. The more closely related organisms to NAR2 did not have closed genomes and so could not be used as references in this assessment. The NAR1 alignment contains less white space than the NAR2 alignment, which agrees with the higher level of sequence similarity, 95%, between NAR1 and D.
piezophilus at the 16s level. In both alignments note the "X" pattern formed by connected LCBs, this typically occurs at the origin of replication in aligned genomes.
The number, arrangement, and heights of similarity profiles in the whole genome alignments of both environmental isolates are indicative of organisms related at the genus level, but not at the species or strain level.

Gene network analysis
Evolutionary gene networks (87) were used to compare the genomes of the two environmental isolates, three of their closest relatives, and D. vulgaris Hildenborough, whose genome has been well studied and serves as an additional point of comparison.
Aspo-2 were included as part of the NAR1 cluster, and the genomes of D. acrylicus, D. desulfuricans susbsp. aestuarii and D. alaskensis str. G20 were included as part of the NAR2 cluster. The initial gene network was run using parameters discussed previously in the Methods section, and results were subsequently filtered to select networks that consisted of only NAR1 and NAR2 (Fig 3.2), only NAR2 ( Fig. 1, and only 87% 16s sequence similarity between the two isolates, these networks could indicate proteins that confer specific benefits for survival in Narragansett Bay sediments. The fact that some of these genes (e.g. beta-lactamases, DMTs) are involved in bacterial defense, bacterial detoxification (glyoxalase family proteins) or as of yet have an un-described function but are exclusively shared between these two isolates, support this hypothesis. Because of this potential connection these proteins should be considered as targets for future investigations, especially those involving transcriptomic or gene expression analysis.
The filtered network containing only NAR1 had approximately 650 connected components (see Appendix A for a list of corresponding proteins), the majority of these connected components consisted of proteins involved in signal transduction and amino acid transport, with a smaller portion of the overall networks consisting of proteins involved in bacterial defense. Components of particular interest are shown in Fig 3.3, which include a putative sensory box/GGDEF family protein network (Fig 3.3 A), a putative diguanylate cyclase and receptor proteins network (Fig 3.3 B), and a periplasmic binding and signal transduction proteins network (Fig 3.3 C), as they have implications in biofilm formation in NAR1. These proteins and their potential role in this isolate's biofilm production will be discussed further in a later section. Additional proteins found to be unique to NAR1 primarily had to do with signal transduction and amino acid transport, which is suggestive of involvement in biofilm formation and exudate production. The fact that these proteins network exclusively from any predicted proteins in NAR2 agrees with what we have observed at the physiological level, with biofilm and exudate production being restricted to NAR1 and not observed at all in NAR2.
There were only 2 connected components that consisted of just NAR2 (Fig   3.4), non-specific predicted membrane proteins (Fig. 3.4 A), and bacteriophage head to tail connecting proteins (Fig. 3.4 B). The two non-specific predicted membrane proteins (Dn2DRAFT 02896, Dn2DRAFT 02903) share the highest degree of BLAST homology using BLASTX with a hypothetical protein in Thiocapsa marina, a purple sulfur bacterium. Dn2DRAFT 02896 shares 59% amino acid identity across 97% of the query, and Dn2DRAFT 02903 shares 60% amino acid identity across 98% of the query with a hypothetical protein (Seq ID: ref|WP_007193013.1) in T. marina. All BLAST results with a high enough degree of amino acid similarity to be of interest (≥ 60% sequence similarity) were to other hypothetical proteins, primarily from betaproteobacteria, and so did not reveal any insight as to the possible function of this protein in NAR2.

Nitrogen fixation
Of the 34 representatives used in comparison to the environmental isolates, only 7 representatives lack the nif operon (Table 3). Only one representative, Desulfovibrio sp. U5L has genes for an alternate Fe-Fe nitrogenase. Both environmental isolates NAR1 and NAR2 have a full nif operon (Fig. 4), with the arrangement of their N fixation gene cluster being similar to those of their closest Nfixing relatives. There is also supporting physiological evidence that both isolates fix N. The presence of an iron-molybdenum nitrogenase appears to be a shared characteristic for this representative group of Desulfovibrio.
In addition to the examination of N fixation, an analysis of metabolism of other N substrates (ammonia, nitrate, and urea) was performed and discussed in the following sections. The analysis of additional aspects of N metabolism in these isolates is critical to improving our understanding of why D. sp. NAR1 and D. sp.
NAR2 exhibit the N fixation behavior we have observed in Narragansett Bay, as an inability to use other sources of N that are commonly found in the environment could account for a continued need to fix N. To do that it is important for us to be able to take the physiology discussed in the previous sections, and connect it with related functional gene content, which is discussed below.

Urea metabolism
Currently, the most attention regarding bacterial urea metabolism is given to organisms that make up mammalian gut consortia and intestinal human pathogens, and little focus has been placed on the urea metabolism of environmental representatives like the Desulfovibrio, or on sulfate-reducing bacteria in general. However, there have been some examinations of the uptake and metabolism of urea by environmental bacteria and phytoplankton (88,89). These studies have shown that rates of bacterial urea uptake in the environment are highly variable, genes for urea transport and catabolism are not wide-spread amongst bacteria, and that other forms of N are generally preferred to urea, which could be due to the fact that urea catabolism is an energetically expensive process. Although some members of the Desulfovibrio are known to have urea transporters and ureases, little is known about the fate of urea once it enters a Desulfovibrio cell.
There exist at least four families of transporters that facilitate selective permeation of urea: an ATP-dependent ABC type urea transporter (90), an ion motive force-dependent urea transporter (91), an acid-activated urea channel that belongs to the urea/amide channel family (92), and the urea transporter (UT) family, this last type being the most widely distributed family. UT members are found in bacteria, fungi, insects and vertebrates (91,(93)(94)(95)(96). In many bacteria and eukaryotes, urea in the cell can be broken down to ammonia and CO 2 by a urease. Some bacteria and eukaryotes also use urea amidolyases (UALase) to decompose urea (88). A crystalline structure for a UT family urea transporter from Desulfovibrio vulgaris Hildenborough is available and its activity and mechanism of action have been proven in vivo (97).
However, subsequent urea metabolism after transport has not been thoroughly investigated.
In this study, an examination using IMG/ER, with visual sequence confirmation in Geneious, and further confirmation using tblastn revealed that of the has an ability to use urea catabolically. These findings in the genome data agree with what has been observed at the physiological level for both isolates, to the extent that both isolates fix N even in the presence of urea, which supports the conclusion that they cannot catabolize urea. However, the NAR1 isolate has exhibited increased NA in the presence of urea, which would seem to indicate that the isolate has some means of sensing its presence. It would appear that any possible method NAR1 could be employing to sense urea and/or transport it into the cell is not a part of described urea metabolism or transport in bacteria. Since urea metabolism in environmental representatives is currently not very well described, further physiological and molecular investigations are needed to elucidate the mechanism and response seen in NAR1. The fact that neither isolate is predicted to be capable of catabolizing urea does not make them unique amongst the Desulfovibrio, or amongst the eubacteria in general.

Ammonia metabolism
Ammonium is known to be the most universally utilized source of biologically available N, and is taken up preferentially by estuarine microbes (89). Accordingly, we would expect to find genes for ammonium uptake and incorporation in all Desulfovibrio representatives. An examination of environmental isolates and additional representatives revealed that all Desulfovibrio, with the exception of D.
cuneatus, possess at least one copy of the ammonium transporter (amt, TIGR accession: TIGR00836) ( Table 3). It is possible that because D. cuneatus is a draft, the ammonium transporter was either missed in annotation or is missing from the assembly, and that the organism may in fact have the transporter. The majority of representatives have multiple copies of the ammonium transporter, where both NAR1 and NAR2 have a single copy, making them slightly atypical in this regard. There are, however, 4 additional representatives that also have a single copy of the transporter. A BLAST search did not reveal any additional copies of the ammonium transporter in either isolate, but it is possible that both isolates could have an additional copy/copies of the transporter that are missing from the current assemblies, or that they have a different protein acting as ammonium transporter that has not yet had that function formally assigned to it. The ammonium transporter for NAR1 is located on DESnar1_contig6, at 122,261-123,619bp, in the forward direction. It is immediately followed by a copy of N regulatory protein P-II, two hypothetical proteins, and a copy of glutamate synthase ~3.5kb downstream, an arrangement which makes sense in terms of nitrogen regulation and activity which has been seen in other bacteria (34,36,38). In the NAR2 isolate, the ammonium transporter is located on DESnar2_contig2 at 904,931-905,257bp, in the forward direction. It is immediately preceded by a copy of N regulatory protein P-II, and immediately followed by an isocitrate dehydrogenase and a protein disulfide isomerase. NAR2 does have a copy of glutamate synthase, however it is located on a different contig.
Whether ammonia is used directly from the environment or is derived from other N sources, its assimilation involves metabolites. Some metabolites, such as 2oxoglutarate, signal N sufficiency or deficiency to the regulatory apparatus (98).
To confirm the potential for ammonia uptake and incorporation in both isolates, additional proteins involved in the N assimilatory pathway were examined in both isolates and the additional Desulfovibrio representatives ( Table 3). The N regulatory protein P-II is a 2-oxoglutarate (2OG) sensor, which is involved in the adenylation cascade that regulates the activity and concentration of glutamine synthetase (GS), in response to N source availability (99). The majority of the Desulfovibrio genomes examined here, including both environmental isolates, have between 2-4 copies of this protein. NAR1 has 4 copies of the protein, with two copies located side by side in between nifH and nifD on DESnar1_contig13, and a third copy located near the previously mentioned ammonium transporter. The fourth copy is located on DESnar1_contig1 at 248,250-248,591 in the forward direction. It is surrounded on either side by hypothetical proteins, further upstream are proteins involved in cellular respiration and downstream are proteins involved in the shikimate pathway. NAR2 has 3 copies of the N regulatory protein P-II, with two copies located side by side and preceded immediately by nifH and followed immediately by nifD, the same arrangement seen in NAR1. The third copy is located on DESnar2_contig2 and was previously mentioned in relation to the ammonium transporter, which immediately follows this copy of N regulatory protein P-II. The locations of all copies of the P-II protein in both isolates make sense in terms of transcription, regulation, and activity, given the proximity to other genes involved in N metabolism.
Because the majority of prokaryotes possess the glutamine synthetase (GS)/glutamate synthase (GOGAT) pathway of assimilation, both environmental isolates were assessed for components of this pathway. The isolates were also assessed for an alternate pathway involving the NADP-linked glutamate dehydrogenase, which catalyzes the amination of 2-oxoglutarate to form glutamate. NAR1 has two copies of glutamine synthetase, the first copy is annotated as being a type III glutamine synthetase, and the second is annotated as a type I glutamine synthetase, both of which have been found previously in prokaryotes (100,101). NAR1 also has all subunits for the NADPH type GOGAT and an NAD-specific glutamate dehydrogenase. NAR2 has a typical prokaryotic type 1 glutamine synthetase, all subunits of the NADPH GOGAT, and a glutamate/leucine dehydrogenase.
Both environmental isolates appear to have a complete GS/GOGAT system of ammonia assimilation, as well as a glutamate dehydrogenase. They both possess the critically important N regulatory protein P-II, with copies of this gene found at genomic locations that make sense in terms of N sensing and regulation. This provides evidence that both isolates have a predicted means of sensing ammonia in the cell, a means of assimilating it, and a means of signaling the regulation of other genes involved in N metabolism under differing N conditions. These findings support our physiological observations, in that neither isolate exhibited NA in the presence of ammonia.

Nitrate metabolism
No member of the Desulfovibrio examined here, including the environmental isolates, has a predicted means of assimilating nitrate (Nas-type nitrate reductases) ( However, the genes responsible for biofilm formation in these organisms have not been conclusively identified or well-studied, and cell-to-cell communication and quorum sensing pathways involved in biofilm formation for these organisms remains unclear. An examination of NAR1 gene annotations in IMG/ER and Geneious revealed no genes belonging to the well described lux family of quorum sensing genes (108,109), these results were then confirmed using a tblastn analysis with the assembled NAR1 contigs as the database and Lux proteins from Vibrio fischeri as the queries.
Putative genes involved in biofilm formation in NAR1 were ultimately discovered using a filtered evolutionary gene network (87) (Fig. 3.3). The majority of these genes belong to a family of diguanylate cyclases with GGDEF (110) domains, and c-di-GMP receptor domain proteins.
Cyclic di-GMP (c-di-GMP) is a bacterial second messenger that is widely utilized by bacteria, with more than 80% of sequenced bacteria predicted to use this signal (111,112). C-di-GMP controls a variety of phenotypes, including biofilm formation, motility, and virulence in multiple bacteria (111,113,114). The fact that the predicted diguanylate cyclases in NAR1 did not network with proteins from any other closely related Desulfovibrio representatives support the hypothesis that these proteins may serve a unique function in NAR1. Given the documented role of diguanylate cyclases and c-di-GMP in biofilm formation, and that NAR1 is unique in its near obligate biofilm lifestyle when compared with NAR2 and other representatives of the Desulfovibrio, it is possible that the unique function served by these proteins is coordination of biofilm formation in NAR1. These genes and the proteins they code for should be further investigated to confirm any role in biofilm formation in this isolate. A separate examination of c-di-GMP levels in NAR1 and other biofilm-forming Desulfovibrio during biofilm growth and during planktonic growth should be considered in order to elucidate the role of c-di-GMP as a signaling molecule for biofilm formation in these organisms.

Mercury methylation
Mercury (Hg) is a pervasive global pollutant known to found in Narragansett Bay sediments (79); in its methylated form (CH 3 Hg + ), it bioaccumulates and is highly toxic to humans and other organisms (115). Unlike inorganic forms of Hg, which originate from atmospheric deposition and point discharge, CH 3 Hg + is generated in the environment by microorganisms. Hg methylation is largely restricted to the proteobacteria and primarily to anaerobic organisms (116). Sulfate-reducing bacteria, such as the Desulfovibrio, are the main producers of CH 3 Hg + (117,118), although iron-reducing bacteria and methanogens can also be involved (119, 120).
The genetic basis for bacterial mercury methylation was recently described by Parks, et al. (18). Because some of the closest relatives to NAR1 are confirmed Hg methylators and because Narragansett Bay sediments are known to contain Hg, the draft genomes of both environmental isolates were searched for hgcA and hgcB, the genes required for bacterial Hg methylation (18). The amino acid sequences for both Hg methylation proteins from D. desulfuricans str. ND132 were used as queries to search the draft genomes of NAR1 and NAR2 using tblastn. No homologs for these proteins were found in the genome of NAR2, however homologs were found in the genome of NAR1 (Fig. 5)

Fig X. N fixation gene clusters of Narraganset Bay isolates (black arrows) and three closesly related Desulfovibrio representatives. Color-coded arrows indicate coding sequences (CDS) and their orientatio
Other/unknown     Nitrogenase activity of NAR1 grown under N limited conditions. Reported as nmols C 2 H 4 produced per cell per day, measured using the acetylene reduction assay. Error bars represent one standard deviation from sample mean.  Nitrogenase activity is reported as nmols C 2 H 4 produced per cell per day, measured using the acetylene reduction assay. Error bars represent one standard deviation from sample mean.
Cultures with added 12 mM urea were found to reduce acetylene to ethylene, again with peak ethylene production occurring 10 days after inoculation but with rates that were slightly higher than the reactive-N free control and nitrate sets. Increased NA for NAR1 in the presence of urea is something that was observed in earlier pilot experiments, and is a result for which there is currently no explanation or known cause. Cultures with added 12 mM ammonia were not found to reduce acetylene to ethylene, exhibiting no measureable NA, which agrees with what is widely accepted regarding the behavior of the nitrogenase enzyme . The genomes of both NAR1 and NAR2 isolates were further found to lack genes necessary for catabolism of either nitrate or urea, but were found to possess genes necessary for assimilating and catabolizing ammonia and deaminating amino acids, which agrees with our ARA based observations. Additionally, these gene-level observations held true for the majority of the Desulfovibrio genomes that were assessed in comparison to our environmental isolates, suggesting that an inability to utilize forms of reactive N other than ammonia may be characteristic of the Desulfovibrio. This inability could in turn be part of why we observe N fixation by this group even in environments that are not limited for sources of combined N, such as Narragansett Bay.