Characterization of Biofilm Formation, Chemotaxis, and the Genome of Aliiroseovarius crassostreae

Aliiroseovarius crassostreae, the causative agent of Roseovarius Oyster Disease, is a marine α-Proteobacterium and a member of the Roseobacter clade. Roseovarius Oyster Disease, formerly known as Juvenile Oyster Disease, has been the reason for high mortality rates in hatchery-raised eastern oysters (Crassostreae virginica) since the late 1980s. Juvenile oysters less than 25 mm in shell length are more heavily impacted by the disease than large adult oysters. Because mortality rates can exceed 90%, the disease is responsible for large economic losses to the New England aquaculture industry. The probiotic organism, Phaeobacter gallaeciensis is a marine α-Proteobacterium and like A. crassostreae a member of the Roseobacter clade. Oyster larvae that are pretreated with P. gallaeciensis show significantly reduced mortality rates in challenge experiments with A. crassostreae. The goal of this study was to elucidate the physiological responses of A. crassostreae and P. gallaeciensis to oyster pallial fluid and identify putative virulence factors via genome analysis of the pathogen. A. crassostreae and P. gallaeciensis grew rapidly in oyster pallial fluid. When growth medium was supplemented with oyster pallial fluid biofilm formation by A. crassostreae was significantly increased. Both organisms were chemoattracted to pallial fluid and a molecule >10 kDa seems to be responsible for this positive chemotactic response. These results suggest that oyster pallial fluid is likely to contribute to the initial colonization of A. crassostreae in the oyster in three ways – by promoting positive chemotaxis, growth, and biofilm formation. Genome analysis of A. crassostreae revealed multiple putative virulence genes including cytolysins, RTX toxin and related Ca-binding proteins, and a serralysin peptidase. In addition, the genome encodes pilus/fimbriae biogenesis machinery and other proteins that appear to facilitate surface attachment.

v Thank you to Robert Deering for his help with analyzing oyster pallial fluid for free amino acids. I would also like to thank Jiadong Sun for his help with the carbohydrate analysis of pallial fluid. The availability of these data sets was invaluable to much of my research.
I would also like to thank Paul Johnson and Janet Atoyan at the URI Genomics and Sequencing Center for their tremendous help with sequencing.
I would like to thank the entire faculty as well as all the graduate students of the Cell and Molecular Biology department. It was an honor to get to know every one of you, to learn from you, and to work with you over these last couple of years.
vi DEDICATION I dedicate this thesis to my parents Luise and Ulli Keßner. Thank you for always believing in me and encouraging me to further my education and professional career. I am unendingly grateful for your support and continuous confidence in me.
I would also like to dedicate this thesis to Jason LaPorte. Thank you for being my rock and continuously supporting me. I am incredibly grateful that we share a passion for microbiology and that I was able to undertake this journey through graduate school together with you. vii

PREFACE
The following thesis has been prepared in Manuscript format according to the guidelines of the Graduate School of the University of Rhode Island. This thesis contains a literature review and two manuscripts.
The first manuscript "Physiological Effect of Oyster Pallial Fluid on Aliiroseovarius crassostreae and Phaeobacter gallaeciensis S4" has not been published.
A short version of the second manuscript "Draft genome sequence of Aliiroseovarius crassostreae CV919-312 T Sm, causative agent of Roseovarius Oyster Disease (formerly Juvenile Oyster Disease)" will be submitted to Genome Announcements.

Introduction
Roseovarius Oyster Disease (ROD), formerly known as Juvenile Oyster Disease (JOD), was first reported in the late 1980s and has since been the cause of high mortality rates in hatchery raised eastern oysters (1)(2)(3)(4). High mortality rates result in significant economic loss for the shellfish industry (3). Oysters are one of the most important aquaculture products in the U.S. and worldwide. The United States consumes almost 60% of the world's total supply of oysters (5). Approximately 66% of the U.S. aquaculture production value lies in bivalve mollusks like oysters, clams, and mussels (6).
In 2005, Aliiroseovarius crassostreae, formerly known as Roseovarius crassostreae, was reported to be the causative agent of ROD (7). A. crassostreae is an aerobic, Gram-negative α-Proteobacterium and a member of the Roseobacter clade (7). Phaeobacter gallaeciensis is a marine organism that has been isolated from the inner shell surface of a healthy oyster (8). Like A. crassostreae, P. gallaeciensis is an α-Proteobacterium and a member of the Roseobacter clade. P. gallaeciensis is a probiotic organism that has been shown to increase survival of oyster larvae when challenged with A. crassostreae or Vibrio coralliilyticus (8). The probiotic mechanisms of P. gallaeciensis include the ability to form biofilms, the production of tropodithietic acid (TDA), and quorum quenching with N-acyl-homoserine lactones (9).

Roseobacter clade overview
Roseobacter clade members share ≥ 89% identity of their 16S rRNA gene sequence and are heterotrophic organisms (10)(11)(12). Members of the Roseobacter clade represent a large part of the marine microbiota and many of its species live in symbiotic relationships with marine phytoplankton, invertebrates, as well as vertebrates (11). This clade is found exclusively in marine or hypersaline environments (11,12). Depending on location, more than 20% of the coastal bacterioplankton communities in the upper mixed ocean waters may consist of the Roseobacter clade alone (11,12). In addition to forming symbiotic relationships with other marine organisms, members of the Roseobacter clade inhabit diverse niches in ocean waters including costal and polar regions, sea ice, hydrothermal vents, macroalgae, and sediments (13). Most often however, members of this clade are found to be attached to organic particles or near phytoplankton or macroalgae blooms (14).
Some of these microorganisms are capable of chemotaxis which provides an advantage to locate nutrients or hosts in heterogeneous and nutrient poor ocean waters (14,15).
While a great metabolic diversity can be observed in the Roseobacter clade, certain physiological traits appear to be characteristic for the clade. These traits include aerobic anoxygenic phototrophy (11,13), carbon monoxide oxidation (11,13), aromatic compound degradation (11), sulfur metabolism (11), production of secondary metabolites (11), and quorum sensing (13). Many Roseobacters have the ability to utilize two key pathways for the degradation of the sulfur based compound dimethylsulfoniopropionate (DMSP) (9,11,16). Marine algae and coastal vascular plants produce this compound as an osmolyte, which is an important sulfur source in the oceans and plays a significant role in marine biogeochemical processes (11,16,17). One of the two pathways, the so called cleavage pathway, plays a role in the release of the volatile gas dimethylsulfide (DMS) while the other pathway is involved in demethylation to 3-(methylmercapto) propionic acid (MMPA), which can then be converted into methanethiol (MeSH) and incorporated into the sulfur containing amino acid methionine (16,18). P. gallaeciensis is known to have this ability to degrade DMSP (9). However, no studies have yet been performed on the ability of A.
Roseobacter clade members are the most dominant primary surface colonizing group of bacteria and it is common for them to form rosettes (8,11,19). The ability of Roseobacters to form pili is most likely an important feature for these characteristics (14). Slightom et al. (14) selected 28 members of the clade to perform a comparative genome analysis and found that all of the 28 members contain gene clusters that encode the machinery for flp (fimbrial low-molecular-weight protein) pili formation.
This widespread colonization island has been implicated with biofilm formation, colonization of surfaces, and pathogenesis in various genera (20). The presence of flppili has been confirmed in Phaeobacter gallaeciensis strain 2.10 and strain BS107 (14). Components of the type 4 secretion system can also be found in many Roseobacter clade genomes (14).

Aliiroseovarius crassostreae CV919-312 T
Previously isolated from Crassostrea virginica tissue during the 1997 epizootics at the Damariscotta River in Maine, A. crassostreae measures approximately 0.25 x 1.0 μm and is motile via one or two polar flagella (4,7). Cells are rod to ovoid in shape (7). The organism can also possess a polar tuft of fimbriae that appear to facilitate the attachment of the organism to the inner shell surface of oysters ( Figure 1) (7,21,22). Colonies take approximately two days to grow ~1 mm in diameter at 27˚C and are grey in color. According to Boettcher et al. (2005), A.
crassostreae grows optimally at temperatures ranging from 34 to 37˚C and prefers pH levels between 6.5 and 8.0. Optimal growth has been observed at salinities between 1% and 1.5% (7). These observations suggest the adaption of this organism to shallow coastal waters (7).
A. crassostreae is an aerobic, non-fermentative organism, produces a weak catalase reaction and is oxidase positive (7). The organism's whole cell fatty acid content is predominantly made up of C18:1ω7c (7). In addition, Boettcher et al. (2005) showed that A. crassostreae is capable of utilizing glycerol and β-hydroxybuturate and has the ability to perform denitrification by reducing nitrate to nitrite and further reducing nitrite to nitrogen gas (7). Esterase, esterase lipase, acid and alkaline phosphatase, leucine and valine acylamidase, as well as lipase activity have been observed (7).
Aliioseovarius crassostreae CV919-312 T (Roseovarius crassostreae at the time) was first described in 1999 in association with oysters affected by Roseovarius   (3,29). Mortality rates in excess of 90% have previously been reported (3,29). Mortality rate seems to be related to animal size and age since small juvenile oysters <25mm in shell length are more heavily impacted by the disease than fully grown adults (3,24,29). Typical symptoms of ROD can be observed as   excessive cupping of the left valve, uneven valve margins, lesions and retraction of the   mantle, emaciated tissue, as well as conchiolin deposits (proteinaceous insoluble organic matrix (4, 30)) on the inner shell surface and overall retarded growth of juvenile oysters (1,3,21,26,29,31,32). Conchiolin deposits can range from mild to severe depending on size of the animal and severity of infection ( Figure 2) (26).
Myoepithelial and muscle degradation was observed in some affected oysters resulting in detachment of soft tissues from the shell (3). In addition, dense coccoid bodies can be observed in the epithelial tissue of affected oysters (3,24,33). Hemocyte infiltration of affected epithelial tissue has also been observed (1,3,24). This hemocyte infiltration is an indicator of microbial infections, as phagocytosis by hemocytes is one of the major immune responses of oysters to bacterial invasion (34)(35)(36)(37). Signs of the disease are often not apparent until one week prior to mortality (24,26,31).
Conchiolin deposits may not be evident in very young oysters that are less than 10 mm in shell length (26). It has been proposed that very young oysters may not have the metabolic resources for rapid conchiolin formation (26). However, large adult oysters have been observed to grow large conchiolin depositions without any evidence of growth cessation or tissue damage (26) (Figure 2). Conchiolin formation may serve as a barrier to the invading microorganism. This barrier can isolate the organism and thus prevent it from reaching the soft tissue (3,21,24,38). .

Pathogen-Host interactions
A. crassostreae-shaped organisms can be observed via A. crassostreae-specific immunofluorescent labeling within conchiolin deposits and on the mantle surface of juvenile oysters challenged with A. crassostreae (22,28). A. crassostreae cells attached to and embedded in conchiolin deposits can also be visualized in infected animals by electron microscopy (21,22,28). SEM images show that A. crassostreae attaches to the oyster's inner shell surface by its polar end (21,22) (Figure 3). In affected oysters that showed visible signs of ROD, the organism can be recovered from the inner shell in 89% and 95% of the left and right valves, respectively (21).
However, in affected animals, A. crassostreae was isolated by culture from only 21% of pallial fluid samples and 47% of tissue surface samples (21). Of animals that tested positive for A. crassostreae by culture, ~ 66% of total CFUs cultured from the inner shell surface were identified as A. crassostreae, whereas 13.1% and 9% of total CFUs were identified as A. crassostreae in tissue surface samples and pallial fluid, respectively (21). These observations suggest that A. crassostreae prefers to colonize the oyster's inner shell surface over oyster tissues. Pallial fluid (fluid in pallial cavity) and extrapallial fluid (fluid in mantle cavity) are always in contact with the inner shell surface of oysters and conchiolin deposits form at the pallial line (edge of mantle) on the shell surface (1,3,7,21,29). Thus, pallial fluid may facilitate or even induce colonization of A. crassostreae to the shell surface.
Since oysters obtain food sources and nutrients directly from seawater, it is suspected that an exchange of oyster pallial and extrapallial fluid and ocean water occurs when oysters are feeding. The filter-feeding behavior of oysters may result in the accumulation of bacteria from ocean waters inside the oyster (36,39). Besides the shell, pallial fluid is one of the first parts of the oyster that marine organisms come in contact with when encountering the host. Pallial fluid may offer a nutrient source for A. crassostreae as well as P. gallaeciensis in the generally nutrient poor ocean water.
As a result, pallial and extrapallial fluid may play an important role in disease onset and progression.
Previous research shows that virulence as well as growth of certain marine pathogens can be induced when the organism is exposed to mucus or mucus-like substances that can be found in the natural host. For example, growth rates of Perkinsus marinus, the causative agent of 'dermo' disease in oysters, were significantly increased when growth medium was supplemented with oyster mucus (40). Cultures of this organism that were supplemented with oyster pallial mucus appeared to be significantly more virulent in challenge experiments (40). Exposure to oyster pallial mucus was found to significantly increase the expression of putative virulence genes in P. marinus (41). Furthermore, previous research shows that the fish pathogen V. anguillarum grows rapidly in Nine Salt Solution supplemented with salmon gastrointestinal mucus (42)(43)(44). In addition, expression of the extracellular metalloprotease EmpA, a virulence factor of V. anguillarum, is induced by salmon intestinal mucus (42). Studies suggest that the alimentary tract containing the mucus is the site of amplification for V. anguillarum (43). Just as fish intestinal mucus is an excellent growth medium for V. anguillarum and induces expression of virulence factors, pallial fluid may represent a similar growth medium for A. crassostreae and P.
gallaeciensis and might induce expression of virulence genes in the pathogen.
Metalloproteases and/or cytolysins are virulence factors that are secreted by many marine pathogens. The genomes of the two marine pathogens V. anguillarum and V. coralliilyticus encode multiple metalloprotease genes and both organisms produce hemolysins as virulence factors (42,45,46). Vibrio cholerae, Vibrio vulnificus, and Pseudomonas aeruginosa are known to secrete proteases as virulence factors (47)(48)(49). P. marinus secretes a serine-protease that is involved in virulence of 'dermo' in oysters (41,50). Studies by Gomez-Leon et al. (2008) show that oyster hemocyte viability decreased when hemocytes were co-incubated with A. crassostreae (28). A similarly low viability of oyster hemocytes can be observed when hemocytes are treated only with extracellular products of the pathogen (28). These results infer that A. crassostreae produces an extracellular protein that kills the oyster hemocytes.
Pore-forming cytolysins have been described in many Gram-negative bacteria (51) and might be the cause for this significant drop in hemocyte viability. Oyster pallial fluid may mimic the host environment and induce expression of these potential virulence factors. The expression of virulence genes can also be controlled by the stringent response. For example, virulence processes of Pseudomonas aeruginosa are tightly regulated by the organism's ability to enter the stringent response because synthesis of virulence factors was impaired in a relA and spoT double mutant (52).
Thus, a starvation induced stringent response may elicit the upregulation of virulence factors as can be observed in other organisms (53).
Substances that are naturally secreted by a host can also promote positive chemotaxis of pathogens. Bordas et al. (54) have shown that V. anguillarum demonstrates a strong chemotactic response towards gilt-head sea bream mucus. In addition, chemotaxis is required for virulence and host colonization of the soilborne plant pathogen Ralstonia solanacearum (55). Since oyster pallial fluid is one of the first parts of the oyster that microorganisms encounter, it may stimulate a chemotaxis response in A. crassostreae and P. gallaeciensis.

Phaeobacter gallaeciensis S4
P. gallaeciensis S4 has been isolated from the inner shell surface of a healthy oyster (8). P. gallaeciensis is a short Gram-negative rod that can elongate and form rosettes in stationary phase (8). The organism is a strict aerobe and heterotrophic. It has one or two polar flagella and is an excellent biofilm former (9). P. gallaeciensis grows optimally at temperatures between 18 to 30 ˚C and prefers salinities of either 2 or 3%. P. gallaeciensis is a probiotic organism that has been shown to protect oyster larvae from bacterial pathogens (8). Oyster larvae that are pretreated with P.
gallaeciensis show significantly increased survival rates when challenged with A.
crassostreae CV919-312 T or Vibrio coralliilyticus RE22 (8). While oyster larvae that were challenged with A. crassostreae or V. coralliilyticus resulted in only 14-31% survival, survival rates of oyster larvae that were first pretreated for 24 hr with S4 were significantly increased ( Figure 4). Furthermore, P. gallaeciensis can protect cod and turbot larvae infected with Vibrio anguillarum, the causative agent of vibriosis (56,57). This antagonistic effect of Phaeobacter against bacterial pathogens has been confirmed by surface attached Phaeobacter cells that are capable of killing Vibrio anguillarum in in vitro assays (58). These probiotic properties can be attributed to the organism's ability to produce secondary metabolites like the broad spectrum antibiotic  tropodithietic acid (TDA) as well as N-acyl homoserine lactones (AHLs), but also the organism's superior ability to form biofilms (9). The ability to form a thick biofilm is an essential part of the probiotic mechanism of P. gallaeciensis, not just by competing for adhesion sites and nutrients, but also to by preventing contact of pathogenic organisms with the host (9). Previous studies show that P. gallaeciensis S4 forms thick biofilms on glass surfaces under static culture conditions (9). In comparison, A.
crassostreae forms much weaker biofilms with only 13.4% the amount of biofilm formed by P. gallaeciensis S4 (9) ( Table 1). Similar results have been obtained for two other aquaculture pathogens, V. coralliilyticus, a pathogen of oyster larvae, and V.
anguillarum, a fish pathogen (9). Additionally, previous studies have shown that TDA production by P. gallaeciensis plays a key role in the organism's probiotic activity because a TDA null mutant did not protect cod larvae challenged with Vibrio anguillarum as well as wild type did (56). Furthermore, the TDA null mutant did not reduce cell densities of V. anguillarum while wild type Phaeobacter significantly reduced or completely killed V. anguillarum in culture based systems (58).
In addition to biofilm formation and TDA production, AHLs have been shown to contribute to the probiotic effect of P. gallaeciensis against V. coralliilyticus by reducing transcription of vtpB and vtpR, two genes involved in protease activity of the pathogen (9). These AHLs are thought to disrupt the quorum-sensing pathway in V.
coralliilyticus that is responsible for activating transcription of genes involved in protease activity (9).

Goals of this study
The overall goal of this study was to elucidate the physiological responses of A. crassostreae and P. gallaeciensis when exposed to oyster pallial fluid. In order to obtain nutrients it is suspected that an exchange of oyster pallial fluid and ocean water occurs when oyster are open (34,59). When encountering the host, both organisms come in contact with the oyster shell and pallial fluid. Thus, pallial fluid may play an important role in colonization of oysters with P. gallaeciensis or A. crassostreae.
Furthermore, pallial fluid may affect ROD onset and progression by A. crassostreae.
The first aim of this investigation was to determine the effects of pallial fluid on growth of A. crassostreae and P. gallaeciensis and establish a chemotactic response pattern of these organisms to oyster pallial fluid. A. crassostreae and P.
gallaeciensis were both grown in pallial fluid and growth rates were compared to the traditional growth medium. Chemotaxis assays were performed to study chemotactic responses.
The second aim of this study was to investigate the effects of pallial fluid on biofilm formation of A. crassostreae. Biofilm assays with A. crassostreae and P.
gallaeciensis were performed on various materials with and without supplementing the growth medium with pallial fluid.  Phaeobacter gallaeciensis is also a Gram-negative α-Proteobacterium and member of the Roseobacter clade (17). P. gallaeciensis is a probiotic organism and can protect oyster larvae from pathogens. Oyster larvae that are pre-treated with P.
gallaeciensis have significantly higher survival rates when challenged with A.
gallaeciensis, are excellent colonizers of marine surfaces (19). P. gallaeciensis forms a thick biofilm when grown under static culture conditions (18). Furthermore, Zhao et al. (2014) have shown that biofilm formation is an important characteristics for the probiotic activity of P. gallaeciensis (20).
Previous research shows that natural host secretions may serve as a nutrient source for marine pathogens (10,21,22). Pales Espinosa et al. (2013) showed that growth rates of oyster pathogen Perkinsus marinus, the causative agent of 'dermo' disease, were significantly increased when growth medium was supplemented with mantle mucus (21). Additionally, virulence of P. marinus significantly increased in challenge experiments when cultures were supplemented with oyster pallial mucus (21). As a result, pallial fluid may represent an excellent growth medium for A.
crassostreae and P. gallaeciensis and might induce expression of virulence genes in the pathogen.
Oyster pallial fluid may not only affect virulence, but also initial colonization of oysters by A. crassostreae. Because oyster pallial fluid is one of the first parts of the oyster that A. crassostreae and P. gallaeciensis come in contact with, it may serve as a chemotactic agent. Chemotaxis towards the host or molecules secreted by the host has been demonstrated in various pathogenic bacteria (23,24). As a result, oyster pallial fluid may elicit a chemoattractive response in A. crassostreae and P. gallaeciensis.
In this study, we determine whether oyster pallial fluid is a nutrient source and growth medium for A. crassostreae and P. gallaeciensis and if it can serve as a chemoattractant. A. crassostreae was tested for potential virulence factors like hemolysins as well as extracellular protease activity. Additionally, we assess the effects of oyster pallial fluid on biofilm formation by A. crassostreae and P. gallaeciensis.

Bacterial strains and growth conditions
A. crassostreae CV919-312 T was originally isolated from a ROD-affected oyster during the 1997 epizootic in the Damariscotta River in Maine (25). P.
gallaeciensis S4 was isolated from the inner shell surface of a healthy oyster (18).

Extracellular protease activity
Protease activity in A. crassostreae culture supernatant was determined by the azocasein protease activity assay as described by Denkin and Nelson (1999) (10).

Chemotaxis assay
Chemotaxis was determined by using a modification of the method described by Adler (32). Stationary phase A. crassostreae or P. gallaeciensis cells were diluted into artificial sea water (30 g L -1 of ocean salt, Instant Ocean) (1000-fold dilution).
Through a small hole in the lid of the petri dish, the cells were exposed to a microcapillary tube containing 10 µl of pallial fluid that had been diluted to a protein concentration of 3 mg/ml (opf #1 and 2). As a control, a microcapillary tube filled was also tested by this assay. The retentate containing molecules ≥10kDa was further diluted (7.5-fold) to obtain their approximate natural concentrations in pallial fluid.
The flow through containing molecules <10kDa was tested undiluted.

Biofilm formation
Biofilm formation was determined using a modified version of the crystal violet staining method described by Belas  Cell densities of planktonic A. crassostreae in each tube were determined from culture medium after the 24 h incubation step. Culture medium was gently withdrawn with a 2 ml serological pipette and placed into a sterile 2 ml microcentrifuge tube.
This culture medium was then serially diluted with artificial seawater and spot plated (10 µl/spot) in triplicate on YP30 agar plates containing 200 µg/ml streptomycin.
After a 2 d incubation, colonies were counted and CFU/ml determined.

Statistical analysis
Statistical data analysis between the individual treatment groups and the control was performed using the Student's T-test. Data with p<0.05 was considered to be statistically significant.

Analysis of pallial fluid
Bradford assay results show that the protein concentration of opf #1 was 4 mg/ml while opf #2 contained 3.5 mg/ml protein ( Table 1).
Analysis of both batches of untreated pallial fluid by HPLC revealed the presence of 12 different amino acids in concentrations ranging from 4 to 2200 µM.
These are free amino acids that were detected in pallial fluid and not amino acids that resulted from the hydrolysis of proteins. The four most abundant amino acids were found to be serine, glycine, alanine, and proline (Table 2).
Phenol-sulfuric acid assay results revealed that opf #1 contained 1.1 mg/ml of carbohydrates and opf #2 contained 0.6 mg/ml of carbohydrates (Table 1). Ion Chromatography of opf #2 only detected 31.9 µg/L glucose to be present; no other mono-or oligosaccharides were identified by this method. Ion chromatography was not performed with opf #1.  Histidine unknown unknown

Growth of A. crassostreae in oyster pallial fluid
Our results show that oyster pallial fluid is a good growth medium for both organisms with growth rates that are comparable to growth in YP30 (Figure 1 and 2).
When A. crassostreae was grown in YP30 the shortest generation time observed was ~71 min while the average generation time was ~79 min. The shortest generation time when grown in pallial fluid (opf #1) was ~66 min whereas the average generation time was determined to be 94 min (Table 3). Additionally, when grown in pallial fluid, A.
A. crassostreae grown in opf #2 grew as well as in opf #1. The fastest generation time in opf #2 was ~68 min and the average generation time was ~102 min ( Table 3).

Growth of P. gallaeciensis in oyster pallial fluid
Like A. crassostreae, P. gallaeciensis grew well in oyster pallial fluid ( Figure   2). The shortest generation time in YP30 was determined to be 73 min while the shortest generation time in opf #1 was ~96 min ( Table 4). The average generation time in YP30 was 120 min and 138 min in opf #1.

Hemolytic activity
No hemolytic activity against fish erythrocytes by A. crassostreae was detected in any of the four conditions tested (Figure 3). Dark zones around inoculum represent colony growth after 2 d.      C. D.

Extracellular protease activity
The highest amount of extracellular protease activity for the positive control organism Vibrio coralliilyticus was observed at 6 h with 100 protease activity units.
While a low protease activity for A. crassostreae under both conditions was observed, these results are negligible and cannot be attributed to any significant extracellular protease activity ( Table 5).

Chemotaxis of A. crassostreae and P. gallaeciensis
Chemotaxis assay results showed that pallial fluid (opf #2, diluted to 3.0 mg/ml protein) is a chemoattractant for both A. crassostreae and P. gallaeciensis.
Microcapillary tubes filled with pallial fluid contained almost 8 times more A.
crassostreae cells than tubes filled with artificial sea water (p=0.001) (Figure 4) while microcapillary tubes filled with pallial fluid attracted almost 5 times more P.
gallaeciensis cells than control tubes filled with artificial sea water (p=0.001) ( Figure   5). HPLC analysis of pallial fluid revealed that numerous free amino acids were present in pallial fluid. Early chemotaxis experiments performed by Mesibov and Adler (1972) (34) showed that Escherichia coli exhibited positive chemotaxis to multiple amino acids including alanine, serine, and glycine. Since these and other amino acids have been identified in pallial fluid, we tested all 12 amino acids detected in pallial fluid for their chemottractive properties using A. crassostreae and P.
gallaeciensis. Neither A. crassostreae nor P. gallaeciensis showed any attraction towards the amino acids tested (Figures 4 and 5). An amino acid cocktail containing   gallaeciensis was also attracted to this fraction, showing 5-fold more cells in the microcapillary tube containing the ≥10 kDa retentate than the artificial sea water tube.
The fraction containing molecules <10 kDa did not reveal any significant chemoattractive properties (Figures 4 and 5). The previously determined optimal addition of 0.5 mg/ml protein from pallial fluid for increased biofilm formation was used to test and compare biofilm formation of both A. crassostreae and P. gallaeciensis in polypropylene, polystyrene, and glass tubes. Again, a significant increase in biofilm formation was observed for A.
crassostreae grown in glass tubes (p= 0.0027) ( Figure 9). However, no significant increase in biofilm was observed in polypropylene, nor polystyrene tubes when compared to glass tubes. Additionally, no significant difference in biofilm formation in polypropylene and polystyrene tubes was detected when growth medium was supplemented with pallial fluid versus un-supplemented YP30 ( Figure 9).
P. gallaeciensis did not form an increased amount of biofilm when the growth medium was supplemented with pallial fluid (Figure 10). The amount of biofilm produced by P. gallaeciensis was not affected by the addition of pallial fluid.
However, biofilm formation is enhanced when the organism is grown in polypropylene (p=0.035) or polystyrene (p=0.039) tubes compared to biofilm formation in glass tubes ( Figure 10).

DISCUSSION
Our study shows that oyster pallial fluid is likely to contribute to the initial colonization of A. crassostreae or P. gallaeciensis in the oyster in two waysby promoting positive chemotaxis and growth. Additionally, pallial fluid seems to promote colonization of A. crassostreae by stimulating biofilm formation. O'Toole et al. (1996) showed that chemotaxis is required for V. anguillarum to infect fish (35).
Furthermore, the soilborne plant pathogen Ralstonia solanacearum requires chemotactic abilities to invade host plants (24). We demonstrate that pallial fluid serves as a chemoattractant, able to attract these microorganisms towards or into the pallial cavity of oysters. Filter-feeding may enhance colonization once these microorganisms are in the proximity of oysters. We have shown that the molecule/s of interest that elicits this chemotactic response is >10 kDa in mass. Further studies will be necessary to identify the molecule/s.
In order to capture nutrients from ocean waters, one oyster is capable of processing over 100 L of seawater per day (36). This filter-feeding behavior exposes the animal to many microorganisms that are present in the ocean including the pathogen A. crassostreae (36,37). During this filter-feeding process and during ejection of pseudofeces microorganisms come in contact with pallial fluid which could promote colonization of the oyster. However, the exact route of infection by A.
crassostreae still remains to be discovered.
Virulence factors in other marine pathogens are up-regulated when exposed to mucus found in the natural host (10,13,21). For instance, genes that affect apoptosis and immunity in invertebrates, proteases and other genes are up-regulated in the oyster pathogen P. marinus when exposed to oyster pallial mucus (13). Additionally, growth rates of P. marinus were faster when media was supplemented with mantle mucus from C. virginica (21). However, our data show that A. crassostreae remains non- P. gallaeciensis is a probiotic organism that protects oyster larvae and increases their survival rates when challenged with A. crassostreae (18). Like A.
Pallial fluid also provides an excellent growth medium for this organism. Colonization by P. gallaeciensis appears to benefit larval oysters by reducing mortality caused by pathogens including A. crassostreae or V. coralliilyticus RE22 (18). However, our study shows that pallial fluid does not promote surface attachment of P. gallaeciensis as it does with A. crassostreae ( Figure 10). This study and previous research on Phaeobacters (17,45,46) demonstrate that these bacteria are excellent biofilm formers and readily attach to various surfaces. Despite an increase in biofilm by A.
crassostreae when supplemented with pallial fluid, P. gallaeciensis still grows a thicker biofilm. For example, the highest average OD580 value for P. gallaeciensis in this study was 1.2 (observed in polypropylene tubes treated with pallial fluid), whereas the highest average OD580 value for A. crassostreae was observed to be 0.6 (detected in glass tubes when YP30 was supplemented with 0.5, 1.0, and 4 mg/ml protein from pallial fluid) (Figures 6 and 10). Thus, at a minimum, P. gallaeciensis biofilm formation is 2-fold greater than biofilm formation by A. crassostreae. The same was observed between A. crassostreae biofilm in glass tubes supplemented with pallial fluid and P. gallaeciensis biofilm in glass tubes; P. gallaeciensis formed >2-fold biofilm on glass than A. crassostreae (Figures 9 and 10). This correlates with the results of biofilm competition assays performed by Zhao et al. (2014), which show that glass coverslips pre-colonized with P. gallaeciensis significantly inhibit colonization of V. coralliilyticus RE22 (20). Thus, inhibition of colonization of pathogens is most effective, when surfaces are pre-colonized with P. gallaeciensis (20).
In conclusion, oyster pallial fluid is rich in free amino acids and other nutrients and serves as an excellent growth medium for both A. crassostreae and P.   Selected genes or gene clusters will be discussed that are putative virulence genes involved in either oyster colonization or pathogenicity.

Growth Conditions and DNA Extraction
Prior to sequencing, a spontaneous streptomycin resistant mutant was isolated

Sequence Trimming, Genome Assembly, and Annotation
Sequence trimming and de novo assembly was performed using the CLC Genomics Workbench (v8.0.1). Sequences from paired end and mate pair sequencing were trimmed and sequences with an average coverage above 100 were used for de novo assembly with the CLC Genomics Workbench. Trimmed reads were also assembled with SPAdes Genomic Assembler (v3.1.1). Resulting contigs from both CLC Genomics workbench and SPAdes Genomic Assembler were joined using the CLC Microbial Genome Finishing module.

Bioinformatics Tools for Genomic Analysis
Gene annotation was performed with Rapid Annotations using Subsystems Technology (RAST) (12)(13)(14). In addition, the draft genome was submitted to and annotated by Integrated Microbial Genomes/Expert Review (IMG/ER) (15). When these annotation pipelines identified open reading frames without any functional annotation, NCBI's Basic Local Alignment Search Tool (BLAST) (16)(17)(18)(19) search tool was used to help identify homologous proteins in other organisms and to compare amino acid sequences to sequence databases. BLAST was also used to detect conserved regions within sequences that could help identify the functional role of a hypothetical protein and/or characterize members of protein families.

Overview and Statistics
The draft genome consists of 26 contigs with a total sequence length of 3,706,831 bp and G+C content of 57.4% plus one complete plasmid of 18,548 bp with G+C content of 58.3%.
According to IMG/ER the draft genome contains 3763 total ORFs including 3694 protein encoding genes (15). RAST detected 3812 protein encoding ORFs and 59 features encoding tRNA or rRNA (12)(13)(14). IMG/ER assigned a predicted function to almost 74% of protein encoding genes (15). The number of genes identified as encoding tRNAs is 50 and while minimally one 16S, one 5S, and one 23S rRNA gene is required, three 16S, three 5S, and three 23S rRNA genes have been annotated (15).
In order to simplify references of gene loci to certain contigs in this study, the 26 A.
crassostreae contigs were numbered and matched to individual GenBank sequence accession numbers. Contig numbers and associated GenBank sequence accessions are displayed in Table 1.

Previous research by Boardman et al. (2008) revealed that A. crassostreae
prefers to colonize the inner shell surface of oysters (9). A. crassostreae was isolated from challenged culture-positive oysters from the left or right valves 89% and 95% of the time, respectively (9). In only 21% and 47% of culture positive oysters was the organism identified in pallial fluid or soft tissue samples, respectively (9). The percent   revealed that flp-1, rcpA, rcpB, tadB, tadD, tadE, and tadF are necessary for the expression of fimbriae in Aggregatibacter actinomycetemcomitans (25). A locus homologous to flp/tad has also been described in the non-pathogenic Caulobacter crescentus and is responsible for the formation of polar pili by that bacterium (22).
Instead of flp prepilin, C. crescentus encodes for pilA as a prepilin and cpaABCEF are homologous to the tadV-rcpCA-tadZA genes in A. actinomycetemcomitans (22). Many organisms contain two (sometimes three) distinct loci of the flp/tad gene cluster (22).
While some of the involved genes are homologous to genes found in type 2 or type 4 secretion systems, many of the genes associated with the flp/tad locus in other organisms like A. actinomycetemcomitans are entirely novel, including rcpC, rcpB, tadZ, tadD, and tadG (22,26).
With the exception of tadF and rcpB all of the genes associated with flpfimbriae biogenesis have been located in the draft genome of A. crassostreae CV919-312 T (Figures 2, 3, 4). Most of these genes are located in two distinct clusters on contig 6 and contig 19 (Figure 2, 3). Two more loci were detected that contain a gene encoding TadG and a gene encoding a TadE-like protein. TadG appears to be essential for complete fimbriae expression; it is thought to anchor the pili/fimbriae by a transmembrane domain to the inner (cytoplasmic) membrane (22,25,27). Tomich et al. (22) suggest that these genes might be paralogues because TadE is 22% identical to the amino acid sequence of TadF. TadE and TadF are pili-like proteins that are not assembled into the flp-fimbriae, but appear to form a pilus-like structure in the periplasm that is anchored to the inner (cytoplasmic) membrane and may assist in extrusion of fimbriae (22). TadE and TadF are thought to interact with each other in the assembly of this extrusion mechanism (22).
One cluster containing 15 genes involved in flp-fimbriae biogenesis is located on contig 6 ( Figure 2) (12-15). RcpB, tadE, tadF, and tadG are missing from this cluster; RcpB, along with RcpA, RcpC and TadD, are outer membrane components of the secretion apparatus (22,26). RcpA is a secretin that is thought to form a pore in the outer membrane through which the pilus structure is secreted (22,26) and RcpB may stabilize the secretion apparatus and maintain the integrity of the outer membrane (26).
The role of RcpC remains to be elucidated (26); however, it has been proposed that RcpC may post-translationally modify the pilus structure, or bind to peptidoglycan as a scaffolding protein (22).
TadD is a lipoprotein containing a tetratricopeptide repeat and due to its homology with lipoproteins in T4SS and T2SS is thought to assist in integration or polymerization of RcpA (22,26). Tetratricopeptide repeat motifs facilitate proteinprotein interactions and are often associated with the assembly of multiprotein complexes (28,29). Interestingly, rcpA, rcpB, and tadD are not found in Grampositive bacteria, suggesting that they localize to the outer membrane in Gramnegative organisms (22,26). The cluster of flp/tad genes on contig 6 encodes for three flp prepilins (12)(13)(14). In A. actinomycetemcomitans Flp-1 is the major pilin subunit while Flp-2 seems to be a homologue of Flp-1 (22,30). Figurski et al. (2013) showed that Flp-1 is a glycoprotein in A. actinomycetemcomitans (31). While most organisms contain either one or two flp genes, three or even more flp genes have been found in a few organisms (22). A. crassostreae is one of those exceptions as it carries three flp prepilin genes on contig 6 (12)(13)(14). These major structural units of the pili/fimbriae are usually 50 to 80 amino acids in size and are cleaved by TadV, the prepilin peptidase, into mature pili subunits (22,31). TadV also required for maturation of pseudopilins TadE and TadF (22). All three Flp prepilins located on contig 6 in A. crassostreae are 65 amino acids in length (pre-modification). The genes for RcpC and RcpA are found downstream of the Flp-prepilins followed by a gene encoding the outer membrane protein A (OmpA) (12)(13)(14)(15). The ompA gene is usually not part of the traditional flp/tad gene cluster and its role in this gene cluster has not been identified. However, studies in A. actinomycetemcomitans suggest that OmpA might be involved in pathogenicity or immune responses in host organisms (24,32). Further downstream from ompA the pilus assembly gene tadZ is found (12)(13)(14)(15). TadZ belongs to the parA/minD superfamily (33). ParA and MinD are proteins that localize to a specific region within the bacterial cell and facilitate localization of other proteins (33). MinD is required for proper cell division while ParA is required for accurate DNA segregation in bacteria (33). TadZ appears to localize the flp/tad secretion apparatus to the polar end of the cell (22,33). Immediately downstream of tadZ are tadABC ( Figure 1) (12)(13)(14)(15). TadA is an ATPase that powers pilus biogenesis (22,34). TadB and TadC are inner membrane components of the pilus assembly apparatus with high similarity in amino acid sequence and homology to proteins involved in type 2 secretion (22,26).
However, their exact role in pilus biogenesis has not been elucidated. It has been proposed that TadB and TadC are molecular pistons that transfer energy from TadA in order to utilize it in flp-polymerization or the two genes are merely involved in scaffolding of the apparatus in the inner membrane (22). The cluster encodes for two TadD molecules that immediately follow one another (12)(13)(14). As previously mentioned, TadD assists in assembly of RcpA into the outer membrane (22). TadV is located immediately downstream of tadD (12)(13)(14)(15) and responsible for modifying the Flp prepilin into mature pilin subunits (22). According to RAST, a predicted ATPase with chaperone activity that is associated with Flp pilus assembly is also located near the tad gene cluster on contig 6 ( Figure 2) (12)(13)(14). BLASTX results confirmed this gene to be an ATPase with multiple hits that have an E-value of 0 and a query cover of 99%. One of the hits aligned to a putative ATPase with chaperone activity, associated with Flp pilus assembly in Celeribacter marinus (99% query cover, E-value = 0, 73% identity). A hypothetical protein encoded by 1644 bp is located on the negative strand (opposite of the other genes in the tad cluster) between the tad gene cluster and this Flp pilus assembly associated ATPase (Figure 2) (12)(13)(14)(15). BLASTX results did not reveal any functional properties besides a conserved region (PHA03307) spanning the first 29-457 bp with an E-value of 5x10 -5 proposing the location of a transcriptional regulator (16)(17)(18)(19).
Additionally, a BLASTX search resulted in similarity to Flp pilus assembly protein TadG in the organism Labrenzia alba with an E-value of 1x10 -42 (99% query coverage and 32% identity, accession CTQ52072.1) (35). TadG is followed by an flp gene encoding a 73 amino acid Flp prepilin (15). RcpC and rcpA are also part of this cluster followed by a 65 amino acid long hypothetical protein (12)(13)(14)(15) This gene cluster on contig 19 also contains sequences for tadZ, tadA, tadB, and tadC (12)(13)(14)(15). However, a gene (558 bp) encoding a protein that is 185 amino acids long is located within the cluster and was annotated as 'L,D-transpeptidase catalytic domain' by IMG/ER (15) and as 'ErfK/YbiS/YcfS/YnhG' by RAST (12)(13)(14).   to NCBI, this type of transpeptidase is an alternative pathway for peptidoglycan crosslinking (16)(17)(18)(19). While all the genes described so far in this gene cluster on contig 19 are encoded on the same strand, one gene that is located immediately downstream of the cluster is encoded on the opposite strand. This gene has been annotated as 'Predicted ATPase with chaperone activity, associated with Flp pilus assembly' by RAST (12)(13)(14). This prediction was confirmed by BLASTX with multiple alignments to ATPases in other organisms including an 'ATPase with chaperone activity, associated with Flp pilus assembly Loktanella cinnabarina' (96% query cover, E-value = 9x10 -107 , 41% identity) (35).
In addition to these two tad gene clusters two loci that each contain a tadE-like gene and tadG have been detected (15) (Figure 4). One of them is localized to contig 23, while the other one is located on contig 19, approximately 315 kb downstream of the cluster of tad genes that are located on contig 19 ( Figure 4).
BLAST alignments of the two TadE-like proteins revealed some similarities with an E-value of 8×10 -7 while the two tadG loci also revealed minor similarity at the amino acid level (E-value = 4×10 -11 , 58% query cover) (35). These genes are most likely orthologs. As previously mentioned, TadE and TadF are homologous to each other (22). These TadE-like proteins might actually be TadF proteins or may have taken on the role of TadF in the pilus assembly machinery.
The flp/tad gene cluster has been described in many prokaryotes and seems to be a mobile genomic island, which has been labeled as a 'widespread colonization island' (22). While the G+C content of the tad/flp gene cluster on contig 6 is 58.2%,   (22).

Binding Proteins
Extracellular proteins produced by A. crassostreae CV919-312 T are capable of eliciting a similar mortality rate in oyster hemocytes as hemocytes treated with the organism itself (36). This indicates that extracellular proteins, potentially hemolysin/leukotoxin type toxins, play a role in pathogenicity (36). Furthermore, it has been suggested that the etiological agent of ROD most likely produces a toxin as a virulence factor (1,8,37).  (35). Repeats-in-Toxins (RTX) is a family of proteins that is secreted by Gram-negative bacteria (38,39). RTX proteins have characteristic glycine and aspartate rich nonapeptide repeats near the carboxyl terminal end of proteins (38,39). These repeats are known to bind Ca 2+ ions upon secretion to promote proper folding of the protein into its functional conformation (39).
All known RTX proteins are secreted via a type I secretion system (T1SS) and have a wide range of functions (38,39). Since the secretion signal is located at the carboxyl-terminal end of the protein, only fully translated proteins are secreted (38,39). The T1SS consists of an ATPase localized to the inner membrane (ABC transporter), a membrane fusion protein (MFP), and a TolC-like outer membrane protein (OMP) ( Figure 5) (38,39).
The α-hemolysin of E. coli (HlyA) and its secretion apparatus is one of the most extensively studied RTX toxins (40). Like many other RTX loci, this toxin along with the toxin activating protein (HlyC), the ABC transporter (HlyB), and the MFP (HlyD) are encoded within the same operon while the OMP (TolC) is located outside the hly operon (39,40). The toxin activating protein HlyC is an acyltransferase that post-translationally activates the protoxin by attaching a fatty acyl residue to the protein (39,40). Pore-forming RTX cytotoxins include hemolysins as well as leukotoxins, many of which are considered to be species and cell specific (39). While IMG/ER annotated most of the 11 ORFs in A. crassostreae as RTX toxins and related Ca 2+ -binding proteins or hemolysins, one ORF on contig 10 has been identified as a hypothetical protein (  conserved region associated with RTX toxin related Ca 2+ -binding proteins (16)(17)(18)(19).
This gene (contig 10) is encoded by 7914 bp and an acyltransferase is located directly adjacent to it ( Figure 6) (15). However, a T1SS that could potentially be involved in secretion of this protein has not been detected at this locus and could be located elsewhere on the genome.
A second RTX toxin related Ca 2+ -binding protein with a cytolysin-activating acyltransfease immediately following has been located on contig 3 (15). Again, a T1SS is absent. In addition to these two cytolysin-activating acyltransferases that are located immediately next to a RTX toxin related Ca 2+ -binding protein, one other putative cytolysin-activating acyltransferase was identified that is not located near a potential cytolysin gene (Table 4). While many of the known RTX proteins are mostly known for their poreforming cytotoxin activity as hemolysins and leukotoxins, this family of proteins also includes proteases, lipases, adenylate cyclases, bacteriocins, and S-layers (38)(39)(40). Out of the 11 RTX toxin related Ca 2+ -binding proteins in the draft genome 7 contain conserved domains that are linked to serralysin peptidases (pfam08548) ( Table 3) (16)(17)(18)(19). One of the 7 genes in particular has been annotated as sarralysin by IMG/ER and in addition to having a conserved domain found in serralysins (pfam08548), the gene Figure 6. Putative RTX toxin on contig 10 of A. crassostreae genome with a cytolysin activating acyltransferease located immediately upstream (15). Although IMG/ER annotated this putative RTX toxin as a hypothetical protein, BLASTX revealed that this gene contains a conserved region associated with RTX toxin related Ca 2+ -binding proteins (16)(17)(18)(19). carries a second conserved domain found in Zinc-dependent metalloproteases belonging to a serralysin-like subfamily (cd04277) (16)(17)(18)(19). According to NCBI, this family of proteins contains a calcium-binding carboxyl-terminal domain that forms a beta role potentially involved in translocation (16)(17)(18)(19). All these conserved domains within this gene confirm the annotation by IMG/ER. Serralysin is an important virulence factor of Serratia marcescens (41). According to Stocker et al. (1995), a Cterminal β-sandwich structure binds calcium ions in order to activate serralysin peptidases (42). Serralysin-type protease PrtA of the insect-pathogenic bacterium Photorhabdus luminescens has been shown to digest proteins in hemolymph of Manduca sexta that have immune-related functions like immune recognition and signaling (43). These results suggest that the substrate specificity of serralysins may be directed towards components of the innate immune system (43). Additionally, bacterial proteases have the ability to degrade numerous antimicrobial peptides (41,44,45). Since antimicrobial peptides are present in oysters as part of their innate immune system, a serralysin peptidase could help in the infectious process of ROD by hydrolyzing these innate immune antimicrobial peptides.
Members of a newly discovered subgroup of RTX proteins function as adhesins or biofilm associated proteins (38). Some of the genes with conserved RTX toxin Ca 2+ -binding protein domains in the A. crassostreae genome might be involved in surface attachment rather than cytolytic activities. RTX adhesins can promote interbacterial interaction or interaction between bacteria and their host (38). LapA for example is a RTX protein found in Pseudomonas putida and Pseudomonas fluorescens and is required for biofilm formation in these organisms (38). BapA (biofilm associated protein) is a protein found in Salmonella enterica that shows homology to RTX proteins and promotes cell-cell interaction (46). FrhA is a RTX protein in Vibrio cholera and associated with hemagglutination, adherence to epithelial cells, biofilm formation as well as chitin binding (38). Similarly, RtxA is thought to promote contact to phagocytic host cells in the intracellular pathogen Legionella pneumophila in order to facilitate entry and promote pathogenesis (38).
Furthermore, an RTX related protein in Shewanella oneidensis (BpfA) is known to be secreted by a T1SS and aids in the formation of biofilm (47). BpfA is positively regulated by the availability of calcium ions (47). Calcium concentration is known to influence biofilm formation in bacteria (48). Increased calcium concentrations also promote biofilm formation and affect protein expression in Pseudoalteromonas sp. It should be noted that while IMG/ER predicted most of the RTX toxin related Ca 2+ -binding proteins and hemolysin activating proteins, RAST annotated many of these genes as alkaline phosphatases.

Putative Hemolysins/Leukotoxins Other than RTX Related Proteins
In addition to all of the above-mentioned RTX related proteins, a putative 'hemolysin III' has been annotated by both IMG/ER and RAST (Table 5) (12)(13)(14)(15).

Conserved Hint Domains
While 11 genes have conserved regions related to RTX toxin related Ca 2+ -binding proteins, one gene on contig 23 has been annotated by IMG/ER (15) as 'Ca 2+ -binding protein, RTX toxin-related' without showing any of such conserved regions in BLASTX (35) ( Table 3). The only conserved region associated with this gene is a Hint 92 domain (pfam13403) that is usually found in inteins (16)(17)(18)(19). BLASTX results show that this protein is most likely connected to a T1SS (Table 3). Interestingly, three of the above mentioned genes that have been annotated as RTX toxins and related Ca 2+binding proteins by IMG/ER and contain RTX toxin conserved domains also contain the same type of conserved Hint domain. Inteins are genetic elements that are transcribed and translated with the host gene (50). After translation these self-splicing elements cleave themselves from the host protein and the host protein is joined again by a peptide bond leaving the protein completely intact and functional (50). This Hint domain has first been observed in Hedgehog proteins (belonging to the family of Hog proteins) and inteins (51,52). While inteins are found in archaea as well as in bacteria, plastids, and viruses, Hog proteins are usually found in multicellular eukaryotes (50,52). Traditional inteins are usually found in conserved proteins that are involved in nucleic acid metabolism and DNA replication (50). However, Amitai et al. (2003) have found that new bacterial intein-like domains are usually found in variable protein regions and are flanked by regions that are found in secreted proteins (52). These flanking domains include calcium binding RTX repeats (52). Since most bacterial intein-like domains are found in secreted proteins, it has been proposed that Hint domains in bacteria may enhance variability in secreted proteins (52).

Putative Adhesin
While no complete T1SS is located near any of the above described RTX toxin and related Ca 2+ -binging proteins, a surface protein that seems to be involved in adhesion is separated by 73 bp from a complete T1SS on contig 4 (12)(13)(14)(15). According to IMG/ER annotation, this T1SS may be responsible for the secretion of the adhesin that is located upstream (Figure 7). This surface protein has been annotated as 'Ig-like domain (group 3)' and 'T1SS secreted agglutinin RTX' by IMG/ER and RAST, respectively (12)(13)(14)(15). NCBI's conserved domain database confirms this annotation showing conserved domains containing an Ig-like fold. This family of proteins is found to be surface associated in other bacteria (pfam13754) (16)(17)(18)(19). While the majority of BLASTX hits are hypothetical proteins, one hit was aligned to "RTX family exoprotein (Tateyamaria sp. ANG-S1)" (90% query cover, E-value = 2x10 -75 , 35% identity, accession: WP_039682725.1) (35).
An outer membrane protein, an ATP-binding cassette, and a membrane fusion protein are all encoded in an operon-like structure (Figure 7). IMG/ER annotated the ATPase that is part of the T1SS as "ATP-binding cassette, subfamily C, LapB" while RAST annotated the outer membrane protein as "Type I secretion system, outer membrane component LapE" (Table 6) (12-15).
As mentioned earlier, LapA is a RTX protein found in Pseudomonas fluorescens and is required for biofilm formation in this organism (38,53). In this T1SS system, LapB is a cytoplasmic membrane-localized ATPase, LapC is a membrane fusion protein, and LapE is an outer membrane protein (53). These annotations confirm that this T1SS is associated with the secretion of a molecule involved in surface-attachment. The SiiE protein in Salmonella enterica has 53 copies of an immunoglobulin (Ig)-like repeat and is important for adhesion to epithelial cells (38). This suggests that this protein is indeed a surface associated protein that

Type 4 Secretion Systems
T4SSs are found in Gram-negative bacteria and are related to the bacterial conjugation machinery (54). However, T4SSs are also virulence factors in Gramnegative bacteria as they can translocate not only DNA, but also virulence effector molecules into host cells (55,56). The T4SS can transport substrate molecules by direct cell-to-cell contact (56). T4SSs can be divided into a T4ASS and T4BSS. The T4ASS is based on the structure and subunits found in the Agrobacterium tumefaciens VirB/D4 system consisting of 12 genes (virB1-vriB11and virD4) (55,57) while the T4BSS is based on the model found in the Legionella pneumophila Dot/Icm system (57). While almost all of the genes of a T4ASS are present in the draft genome of A. crassostreae, a partial T4BSS can also be located.
While VirB7, together with VirB9 and VirB10, is supposed to form the core complex of the T4ASS that spans the periplasm (56), identification of VirB7 is often challenging since VirB7 has not been described in all T4SSs (54,59). With 4.5 kDa (60, 61) VirB7 is the smallest of all the proteins that are part of the T4ASS and homologues are sometimes difficult to identify (54). The size for VirB7 ranges from 47 to 69 amino acid residues (54). Two hypothetical proteins are encoded within the virB gene cluster that fall within this size range of VirB7, one with 58 amino acids (177bp), and the other with 52 amino acids (159 bp) (12)(13)(14)(15). BLASTX results did not reveal any conserved domains (16)(17)(18)(19), however IMG/ER linked the 52 amino acid protein to a family of proteins associated with a 'Prokaryotic membrane lipoprotein lipid attachment site profile' that can be found in the Swiss Institute of Bioinformatics Resource Portal's Prosite database (62,63). Prosite is a database of protein domains, families and functional sites (62,63). This connection between the 52 amino acid protein encoded on the draft genome and this protein family was confirmed by submitting the 52 amino acid sequence to ScanProsite. ScanProsite matches similarities between the entered amino acid sequence and signature profiles in proteins in order to detect functional and structural domains within a protein (64,65).

99
The protein profile detected by Prosite was 'Prokaryotic membrane lipoprotein lipid attachment site profile' (accession PS51257) and matched to the first 17 amino acid residues (64,65). VirB7 is a membrane associated lipoprotein that is thought to anchor and stabilize the pilus complex to the outer membrane (54,56). This profile match would suggest that the 52 amino acid protein encoded within the virB gene cluster is a VirB7 homologue (Figure 9).
Interestingly, VirD2, a protein involved in T-DNA processing and transfer in Agrobacterium tumefaciens (66,67), is encoded just upstream of virD4 in the draft genome of A. crassostreae. In A. tumefaciens, oncogenic T-DNA is transferred into plant cells where it is integrated into the nuclear genome and able to modify plant hormonal balance, cell differentiation, transcription, and metabolism leading to tumors (66,68). VirD2 is the pilot protein that is directly involved in the transfer of T-DNA (68).
A gene located just downstream of virD4 has been annotated as 'AAA domaincontaining protein' by IMG/ER (15) while the GenBank annotation service identified this gene as a 'DNA repair protein' (35). These annotations correlate with the observation that RecN DNA repair proteins exhibit ATPase activity, which is stimulated by the addition of DNA (16)(17)(18)(19)69). In addition to the genes mentioned above, a second copy of a putative virB6 gene is located approximately 50 kb upstream of the virB gene cluster. This 597 bp gene has been annotated as 'TrbL/VirB6 plasmid conjugal transfer protein' by IMG/ER and 'Inner membrane protein of type IV secretion of T-DNA complex, VirB6' by RAST (12)(13)(14)(15).

Type 4B secretion system
In addition to the T4ASS, a partial T4BSS has been annotated which is often involved in pathogenesis (70) (Figure 10). The T4BSS was first discovered in the human pathogen Legionella pneumophila (70). L. pneumophila is an intracellular pathogen that multiplies within and kills macrophages (71,72). The T4BSS seems to contain components that are homologous to proteins found in type 2, type 3, type 4A, and type 6 secretion systems (70). Like the T4ASS, this T4BSS has some similarity to bacterial conjugation systems (70). In L. pneumophila, the dot/icm genes that encode this T4BSS are necessary for conjugal transfer of IncQ plasmids as well as translocation of effector molecules into host cells (70). Two groups have independently discovered clusters of genes that are responsible for intracellular replication and macrophage killing and while one group named the genes dot (defect in organelle trafficking), the other named them icm (intracellular multiplication) (70).
A typical T4BSS has multiple clusters of genes, one cluster containing dotD-dotC-dotB, a second cluster encodes dotM/icmP-dotL/icmO, and yet another cluster encodes dotI/icmL-dotH/icmK-dotG/icmE (70). Part of the T4BSS is a gene encoding dotA and dotO/icmB (70) (see Figure 11 for putative assembly and location of proteins associated with T4BSS assembly). Annotations revealed that many of these genes are present in the draft genome of A. crassostreae ( Figure 10).
A protein within this cluster of genes was annotated as 'Putative outer membrane core complex of type IVb secretion' by IMG/ER based on alignments to the protein family 'T4BSS_DotH_IcmK' (PF12293) (15). This alignment was confirmed by performing a sequence alignment search with Pfam (73). DotH/IcmK is a transporter protein localized to the outer membrane and is also part of the T4BSS (70).
RAST annotated the gene that is immediately downstream as IcmE/DotG whereas IMG/ER called it 'conjugation TrbI-like protein' (12)(13)(14)(15) the assignments of functional roles to the genes described in this study. Biofilm assays 106 of wild type and knockout mutants would reveal whether the putative genes associated with surface attachment that were discussed in this report are indeed involved in oyster shell colonization.
This Whole Genome Shotgun project has been deposited in DDBJ/ENA/GenBank under the accession LKBA00000000. The version described in this paper is version LKBA01000000.

FUNDING INFORMATION
This work was supported by an award from the Rhode Island Science and Technology Advisory Council to Dr. David Nelson and Dr. David Rowley.