Pseudoalteromonas piratica strain OCN003 is a coral pathogen that causes a switch from chronic to acute Montipora white syndrome in Montipora capitata

Reports of mass coral mortality from disease have increased over the last two decades. Montipora white syndrome (MWS) is a tissue loss disease that has negatively impacted populations of the coral Montipora capitata in Kāne‘ohe Bay, Hawai‘i. Two types of MWS have been documented; a progressive disease termed chronic MWS (cMWS), that can be caused by Vibrio owensii strain OCN002, and a comparatively faster disease termed acute MWS (aMWS), that can be caused by Vibrio coralliilyticus strain OCN008. M. capitata colonies exhibiting cMWS can spontaneously switch to aMWS in the field. In this study, a novel Pseudoalteromonas species, P. piratica strain OCN003, fulfilled Koch’s postulates of disease causation as another etiological agent of aMWS. Additionally, OCN003 induced a switch from cMWS to aMWS on M. capitata in laboratory infection trials. A comparison of OCN003 and Vibrio coralliilyticus strain OCN008, showed that OCN003 was more effective at inducing the cMWS to aMWS switch in M. capitata than OCN008. This study is the first to demonstrate that similar disease signs on one coral species (aMWS on M. capitata) can be caused by multiple pathogens, and describes the first Pseudoalteromonas species that infects coral.


Introduction
Coral disease is an environmental and economic concern that threatens the continued survival of coral reef ecosystems and intensifies the impacts of other recognized detriments to coral health. Following the first report of coral disease in 1973 [1], over 40 coral diseases from at least 75 countries have been described and characterized [2]. More than 200 species of reef-a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 building corals are affected by different diseases, and reports of mass mortality events have been increasing over the last two decades [3][4][5][6][7]. For many of these diseases, a causative agent remains to be found.
Koch's postulates of disease causation are a set of guidelines that are commonly used to infer the causality of a microorganism as a disease agent [8]: (I) The microorganism must be present in all organisms that have the disease and should be absent in healthy organisms, (II) the microorganism should be isolated from a diseased organism and can be grown in pure culture, (III) the cultured microorganism should cause disease signs if inoculated into healthy hosts, and (IV) the same microorganism must be re-isolated from the inoculated, diseased hosts. These postulates have been utilized by researchers to describe causative agents of coral diseases, although only seven causal relationships linking pathogens to a coral disease have been successfully demonstrated to date [9][10][11][12][13][14][15].
The fulfillment of Koch's postulates has proven useful in organismal studies, but the advent of genetic analysis has prompted the repurposing of these postulates for use in molecular pathogenesis research. Using an analogous line of reasoning to organismal investigations, studies must fulfill Koch's molecular postulates to assess the role of a gene in a pathogen's virulence [16]. Like Koch's postulates, for a gene to be considered a virulence factor, mutations abrogating gene function lead to decreased virulence of the pathogen. Complementation of the mutation via reintroduction of the gene to restore function results in a concurrent return to wild type virulence levels. To date, only a handful of genes have been implicated as virulence determinants in coral pathogens. A toxin and a superoxide dismutase have been shown to be virulence factors produced by the pathogen Vibrio shiloi, which causes coral bleaching of Oculina patagonica [17,18]. Another study demonstrated that a metalloprotease produced by V. coralliilyticus strains BAA-450 and P1-4 caused tissue lesions in Pocillopora damicornis and Acropora millepora coral colonies, respectively [19,20]. In a recent study, mutation of a toxin regulator (toxR) and a gene within the type IV pilus operon (mshA) in V. coralliilyticus strains OCN008 and OCN014 significantly reduced infection rates in the coral M. capitata and Acropora cytherea, respectively [15]. While such studies on the molecular basis of pathogenesis have been conducted extensively in human pathogens, similar work is almost entirely lacking from coral disease research.
Montipora white syndrome (MWS) is a tissue loss disease that has negatively impacted populations of Montipora capitata, a major reef-building coral in Kāne'ohe Bay, Hawai'i [21]. Two types of MWS have been documented; a chronic, progressive disease that displays diffuse tissue loss termed chronic MWS (cMWS) [13], and a comparatively faster progressive tissue loss disease termed acute MWS (aMWS) [7,14]. Montipora capitata colonies exhibiting cMWS are observed at consistent levels throughout the year [21], while outbreaks of aMWS occur sporadically usually during the rainy, winter months [7]. Colonies with cMWS can survive months to years because of the comparably slower rate of tissue loss [21]. In contrast, aMWS-affected colonies can die within days to weeks during disease outbreaks [7]. During aMWS outbreaks, disease transmission has been observed between neighboring M. capitata colonies in the field, and cMWS-affected colonies were also observed to spontaneously switch to aMWS, which can lead to complete colony mortality [7]. While this switch from cMWS to aMWS has been observed both in laboratory infection trials and on coral colonies in the field [13,22,23], the mechanism underlying the change in disease signs remains unknown.
To date, two pathogenic Vibrio species have been shown to induce disease signs consistent with MWS in M. capitata; Vibrio owensii strain OCN002 induces cMWS and Vibrio coralliilyticus strain OCN008 induces aMWS (hereafter referred to as OCN002 and OCN008) [13,14]. During controlled laboratory infections of M. capitata, OCN002 caused cMWS in 53% of inoculated coral fragments in an average of 28 days post-inoculation [13], while OCN008 infected 80 to 100% of fragments resulting in aMWS within four days post-inoculation [14,15]. A comparison of the culturable bacterial communities from healthy and aMWS-affected M. capitata from Kāne'ohe Bay found that Pseudoalteromonas was the second most abundant bacterial genus in infected tissue (second to Vibrio) but was absent from healthy mucus [24]. Based on these findings, a Pseudoalteromonas strain isolated from aMWS-affected M. capitata was investigated as another potential pathogen. The Pseudoalteromonas strain originally described by Smith (2008) has recently been identified as a novel Pseudoalteromonas species, P. piratica strain OCN003 [25], which was assessed as another etiological agent of acute tissue loss disease in M. capitata (aMWS).
In this study, Koch's postulates of disease causation were fulfilled for P. piratica strain OCN003, which establishes it as another etiological agent of aMWS in M. capitata. In addition, when inoculated onto coral fragments exhibiting cMWS signs, OCN003 also induced the switch from cMWS to aMWS in M. capitata and again fulfilled Koch's postulates as an etiological agent for this type of infection. To show that the switch from cMWS to aMWS was due to inoculation with wild type OCN003, and demonstrate that motility is a critical component of the infection process, a non-motile ΔfliF::bla mutant of OCN003 was constructed and found to be incapable of inducing the switch from cMWS to aMWS. Complementation of the nonmotile OCN003 mutant restored the bacterium's ability to cause the switch from cMWS to aMWS, fulfilling Koch's molecular postulates of disease causation [16]. This work describes both the first strain of Pseudoalteromonas that infects coral and the first coral pathogen that can induce a switch in disease signs from a chronic (cMWS) to an acute (aMWS) type of tissue loss disease.

Bacterial growth conditions
All bacterial strains used in this study are listed in Table 1. All marine bacterial strains were grown at 27˚C in glycerol artificial seawater (GASW) broth or on plates solidified with 1.5% (w/v) agar as previously described [13]. Escherichia coli strains were grown in LB-Miller medium and incubated at 37˚C. Antibiotics for plasmid selection in E. coli were used at the following concentrations: ampicillin, 100 μg/ml; kanamycin, 50 μg/ml; streptomycin, 50 μg/ml; spectinomycin, 100 μg/ml; and chloramphenicol, 30 μg/ml. Auxotrophic E. coli strains π3813 and β3914 were grown on media supplemented with deoxythymidine (DT) or diaminopimelate (DAP) at a final concentration of 0.3 mM each, respectively [26]. Bacterial sacB-mediated counterselection was achieved by supplementing GASW with 5% (w/v) sucrose [27].

Coral collection and infection trials
Healthy and cMWS-infected fragments of M. capitata were collected from the fringing reef surrounding Moku o Lo'e island in south Kāne'ohe Bay, Hawai'i, under Special Activities Permits SAP#2013-47, SAP#2015-17, and SAP#2015-48 granted by the State of Hawai'i, Department of Land and Natural Resources, Division of Aquatic Resources. Fragments with cMWS were identified by their characteristic lesions. Histology on the coral fragments was not performed, as the mucus associated bacteria are lost during sample processing, thus the specific causative agent of disease was not determined prior to experimentation. M. capitata fragments were allowed to recover from collection in a flow-through water table at ambient temperature for at least two days prior to experimental infection trials. Infection trials employed a block design in which all coral fragments used within an experimental block were collected from the same coral colony to control for intraspecific variability in disease susceptibility. Each experimental block consisted of one seawater control kept in a water table, one FSW control kept in its own tank, one bacterial control kept in its own tank, and any pathogen candidates tested.
Infection trials were conducted as previously described [13][14][15] with minor modifications (S1 Fig). Briefly, each M. capitata fragment with a cMWS lesion was individually housed in a nine-liter tank filled with seven liters of filtered seawater (FSW). Healthy M. capitata fragments were individually housed in four-liter tanks filled with three liters of FSW. All tanks were maintained at a constant 25˚C water temperature. For coral inoculation, overnight liquid cultures of the various bacteria tested were diluted 1:1000 in GASW broth, and then grown to an OD 600 (optical density measured at 600 nm) of 0.8 before being washed once and resuspended in autoclaved FSW. OCN003, the ΔfliF::bla mutant, the complemented ΔfliF::bla mutant, OCN004, OCN008, and OCN050 were inoculated to a final concentration of 10 8 CFU/ml of tank water unless otherwise stated during infectious dose determinations. Dilutions were prepared as previously described in Ushijima et al. (2014). OCN004 was used as a negative control organism to show that the presence of a high concentration of bacteria in tank water was not sufficient to induce disease on M. capitata as previously described [13,14].

Identification and phylogenetic analysis of Pseudoalteromonas isolates
Pseudoalteromonas piratica strain OCN003 was originally isolated as described in Smith (2008). To determine whether OCN003, or similar P. piratica strains, may have been involved in the March 2010 aMWS outbreak samples, bacteria were cultured from the mucus of infected colonies. Coral mucus was homogenized in FSW and serial dilutions were plated on GASW plates as previously described [13]. A total of 265 bacterial isolates were cultured from four aMWS-infected coral fragments collected during the outbreak. These isolates were screened by PCR using primers that selectively amplify a unique region of the OCN003 genome (003unique-F and 003unique-R). The unique region used for OCN003 identification was found by searching the genome [28] for a large intragenic region between convergent genes that was not similar to any sequences present in the NCBI database. All oligonucleotides used in this study are listed in Table 2. Nucleotide sequences for the 16S rRNA gene and the genes for the multilocus sequence analysis (MLSA) for 14 Pseudoalteromonas species were obtained from whole genome sequences from NCBI (Table 3). For MLSA, the ftsZ, gapA, and recA genes from P. piratica strains OCN003, OCN050, OCN051, and OCN052 were amplified by PCR using their respective primers, then sequenced with the same primers (Table 2), and then aligned and analyzed as previously described [25].

Re-isolation of OCN003 and OCN050 from infected coral fragments
The re-isolation of OCN003 and OCN050 tagged with the non-self-transmissible vector pRL1383a [29] was conducted as previously described [13][14][15] with some modifications. Infected coral fragments were crushed and plated in dilutions on GASW plates supplemented with spectinomycin and streptomycin. Isolates that grew were screened using pRL1383a-MCS primers and the P. piratica-specific primers, 003unique-F and 003unique-R ( Table 2). The P. piratica-specific primers were designed to amplify a 511 bp intergenic region in the OCN003 genome between the divergently expressed coding sequences of a putative response regulator [AIY65323], with 83% identity to chemotaxis protein CheY from Pseudoalteromonas sp. P1-9, and a Cys regulon transcriptional activator [AIY65324], with 99% identity to the transcriptional regulator CysB from Pseudoalteromonas sp. P1-9. According to BLAST analysis, the intergenic region did not share significant nucleotide similarity with any other sequence in the NCBI database. Positive colonies displayed amplified bands following amplification with both the pRL1383a-and P. piratica-specific primer pairs. In addition, the 16S rRNA gene was amplified and sequenced to ensure that recovered P. piratica isolates were all identical to the OCN003 and OCN050 stock cultures.

Plasmid construction
All plasmids used in this study are listed in Table 1. Plasmid pBU187 is a suicide plasmid based on pSW4426T [26]. A fragment containing araC-PBAD-ccdB was removed from pSW4426T by amplifying it with PCR using the primers pSW4426T-up-AraC-SacI and pSW4426T-down-ccdB, digesting the product with SacI and DpnI, and then self-ligating it to create pBU187. Plasmid pSB123 is a suicide plasmid based on pBU187 used to create genetic mutations in OCN003. The R6K oriV, oriT, and the chloramphenicol acetyltransferase gene (cat) were excised from pBU187 as a~1.7 kb SacI/XbaI fragment and cloned into the same sites of pBluescript SK+ (Stratagene) to create pSB122. The R6K oriV, oriT, cat, and multiple cloning site were amplified from pSB122 with PCR using the primers M13-F and M13-R, and the product was cloned into the EcoRV sites of pRL277 [30] to replace the existing oriV and oriT and create pSB123.
Plasmid pSB126 is a suicide vector based on pSB123 used to delete all but the first and last 18 nucleotides of the fliF coding region in OCN003. Regions up-and downstream of the fliF gene were amplified by PCR from OCN003 chromosomal DNA using the primers pairs 003-fliF-up-F and 003-fliF-OEX-R and 003-fliF-down-OEX-F and 003-fliF-down-R, respectively. The up-and downstream fragments were fused together by overlap extension PCR [31] and the product was cloned into the EcoRV site of the plasmid pEVS78 [32] to create pSB124. A fragment containing P bla -bla was amplified by PCR from pBBR1MCS-4 [33] with the primers bla-F and bla-R and cloned into the SmaI site in pSB124 to create pSB125. A fragment harboring regions up-and downstream of the fliF gene and P bla -bla was amplified from pSB125 by PCR using the primers pEVS78-MCS-F and pEVS78-MCS-R. The resulting product was digested with SpeI and cloned into the same site in pSB123 to create the vector pSB126. Plasmid pSB129 is a replicative plasmid based on pRL1383a used to complement the OCN003 ΔfliF::bla mutant. The fliF gene was amplified by PCR from OCN003 chromosomal DNA using the primers 003-fliF-F and 003-fliF-R and ligated into the SmaI site of pBlue-scriptSK+ to create pSB127. The gentamicin resistance cassette was amplified by PCR from pUC18-mini-Tn7T-Gm-lacZ [34] using the primers Gm R -F and Gm R -R and cloned as a XhoI-ClaI fragment into the same sites of pSB127 to create pSB128. The gentamicin resistance cassette and fliF gene were excised from pSB128 as a SacI and XhoI fragment and cloned into the same sites in pRL1383a to create pSB129.

Strain creation and bacterial conjugation
The fliF gene was deleted from the OCN003 genome by allelic exchange with the suicide vector pSB126. The suicide vector pSB126 was introduced into OCN003 using tri-parental conjugation with E. coli. The suicide vector was maintained in E. coli strain β3914 [26] and the selftransmissible vector pRK2013 [35] was maintained in strain π3813 [26]. For tri-parental conjugation, the donor and recipient strains were grown overnight under antibiotic selection for plasmid maintenance and with DAP or DT as required. Overnight cultures were diluted 1:1000 in fresh culture medium, grown to an OD 600 of 0.7, and washed three times with either GASW or LB-Miller for Pseudoalteromonas or E. coli strains, respectively. The strains were then resuspended in GASW to a total volume of 30 μl, combined, and spotted onto GASW plates supplemented with DAP and DT. Conjugation spots were incubated at 28˚C for 24 h before being resuspended in GASW, washed three times with fresh GASW, and then dilutions were plated onto GASW agar supplemented with chloramphenicol, but lacking DAP or DT for counterselection against the auxotrophic E. coli donor strains. Chloramphenicol resistant colonies, which consisted of bacteria with the suicide vector introduced through a single recombination event, were used to inoculate GASW broth and incubated for 15 h. After incubation, cultures were washed with GASW three times, and then dilutions were plated onto GASW agar supplemented with sucrose and ampicillin to isolate double-recombinant mutants. The resulting strain, OCN003 ΔfliF::bla, was verified by PCR with the primer pair 003-fliF-outside-F and 003-fliF-outside-R, which anneal outside the region of DNA used to make the mutations and sensitivity to chloramphenicol. The complementation plasmid was introduced into the OCN003 ΔfliF::bla mutant via triparental conjugation as described above. Colonies that grew on GASW plates supplemented with gentamicin were confirmed by PCR using the primers pRL1383a-MCS-F and pRL1383a-MCS-R and then screened for motility using light microscopy.

Microscopy
Motility of the OCN003 ΔfliF::bla mutant, the complemented OCN003 ΔfliF::bla mutant, and wild type OCN003 were determined using both light microscopy and a semi-solid assay (GASW broth and 0.15% agar) in poured plates for the differentiation of motile and nonmotile colonies. Cell morphology of wild type OCN003, OCN003 ΔfliF::bla, and the complementation of OCN003 ΔfliF::bla were examined using transmission electron microscopy (TEM). Overnight cultures were deposited on a Formvar-coated copper grid, contrasted with 1% uranyl acetate, and viewed on a Hitachi HT7700 TEM at 100kV. The presence of a flagellum was examined and photographed with an AMT XR-41B 2k x 2k CCD camera.

Biofilm assays
Biofilm formation of the wild type OCN003, OCN003 ΔfliF::bla mutant, and complemented OCN003 ΔfliF::bla mutant was tested using a microtiter dish biofilm formation assay as previously described [36] with minor modifications. Bacterial cells were grown overnight in GASW broth, 10 μl of an overnight culture was used to inoculate each microtiter dish well (24-well dish) containing two ml of sterile GASW broth. The microtiter dish was incubated for 24 to 96 h at 28˚C. After incubation, the culture was aspirated and the wells were rinsed twice with deionized water. To stain the biofilm, two ml of a 0.1% crystal violet solution was added to each well and then incubated at room temperature for 15 min. The crystal violet solution was aspirated and then the plate was rinsed with water and dried overnight at room temperature. For the quantification of the biofilm, two ml of 30% acetic acid was added to each well to solubilize the crystal violet, and then the solution was transferred to a new microtiter plate and the absorbance was measured at 550 nm. The data was analyzed by one-way ANOVA, followed by Tukey's multiple comparison with no significantly different results.

Pseudoalteromonas piratica strain OCN003 is another etiological agent of acute tissue loss lesions (aMWS) in Montipora capitata
The novel bacterial species Pseudoalteromonas piratica strain OCN003 [25]; hereafter OCN003) was originally isolated from an M. capitata colony displaying aMWS [24]. Due to the abundance of members from the Pseudoalteromonas genus in aMWS-affected M. capitata tissue, OCN003 was assessed for virulence against M. capitata. When healthy fragments of M. capitata were exposed to OCN003, 33% of the coral fragments developed acute tissue loss (aMWS) in an average of 22 days post-inoculation (McNemar's test, n = 24, p = 0.01; Fig 1). The lesions induced by OCN003 in laboratory fragments were visually similar to aMWS observed on M. capitata both in the field and during previous laboratory experiments [7,14]. As controls, healthy M. capitata fragments were exposed to filtered seawater (FSW) or inoculated with the negative-control bacterium, Alteromonas sp. strain OCN004 [13], neither of which induced tissue loss (n = 24 per treatment). Using these two controls indicates that neither the microflora harbored by coral from the field nor the addition of a high concentration of control bacteria were sufficient to induce tissue loss lesions. Prior studies defining coral pathogens as etiological agents of disease demonstrated the successful re-isolation of the pathogenic strain from experimentally infected coral fragments [13][14][15]19]. To facilitate re-isolation and ensure that the re-isolated bacteria were derived from the laboratory stock culture, OCN003 was genetically tagged with a non-self-transmissible plasmid, pRL1383a, as previously described [13]. The infection rates between the wild type OCN003 and tagged OCN003 were not significantly different (conducted simultaneously; Mantel-Cox test, n = 20, p = 0.98), with each bacterium inducing aMWS in 20% of the fragments after an average of 19 days post-inoculation. The tagged OCN003 was re-isolated from all experimentally infected fragments that developed aMWS (n = 20), using previously described methods [13]. Collectively, these results demonstrated that OCN003 was isolated from diseased coral, grown in pure culture, used to experimentally infect laboratory specimens, and was re-isolated from infected specimens. This fulfillment of Koch's postulates of disease causation indicates that OCN003 is an etiological agent of aMWS.

P. piratica strain OCN003 induces the switch from cMWS to aMWS in M. capitata
In the field, M. capitata colonies with cMWS can suddenly display acute tissue loss that resembles aMWS [7]. This switch from a cMWS to aMWS could be due to many factors, one of which is the onset of a new or additional infection. To investigate whether OCN003 could induce a switch from a cMWS to aMWS, M. capitata fragments displaying cMWS were collected from the field and inoculated with OCN003. Following inoculation with OCN003, 56% of fragments with pre-existing cMWS switched to aMWS in an average of eight days postinoculation (McNemar's test: n = 16, p < 0.01; Fig 2). Following the OCN003-induced switch from cMWS to aMWS in the laboratory, the newly formed tissue loss lesions appeared visually similar to aMWS observed in the field and in the laboratory infections described above [7]; S2  Fig). The genetically-tagged OCN003 strain used above was also able to cause a switch from cMWS to aMWS at levels similar to the wild type (Mantel-Cox test: n = 6, p = 0.92), and could be re-isolated from all M. capitata fragments that switched from cMWS to aMWS. The FSW or OCN004 negative controls did not induce a switch from cMWS to aMWS (n = 16 per treatment; S3 Fig). The minimum dose of OCN003 required to induce the cMWS to aMWS switch was then determined by conducting infection trials on fragments with cMWS with inoculum concentrations ranging from 10 4 to 10 8 CFU/ml of tank water. After inoculation with 10 4 and 10 5 CFU of OCN003 per ml of tank water, 0% and 18% of fragments with cMWS developed aMWS, respectively. Therefore, the minimum dose required to induce the switch from cMWS to aMWS was found to be between 10 4 and 10 5 CFU/ml (n = 11). These results collectively fulfill Koch's postulates to confirm that OCN003 acts as a pathogen to induce the switch from cMWS to aMWS.

Motility is required for OCN003 to induce the cMWS to aMWS switch
It is possible that the addition of an infectious strain, rather than the action of the strain itself, is able to induce the switch from cMWS to aMWS, thus, assessment of a mechanistic basis of virulence underlying the switch between cMWS and aMWS is pertinent. To assess a potential virulence factor of OCN003 contributing to its ability to induce the switch from cMWS to aMWS, a non-motile OCN003 mutant was created. Previous studies analyzing motility as a requirement for infection have demonstrated that a flhA mutant in V. coralliilyticus strain YB2 failed to form a flagellum and was incapable of infecting the coral Pocillopora damicornis [37]. In contrast, non-motile mutant strains of the human pathogen Vibrio cholerae were shown to remain infectious [38]. Rather than remove the flagellum as was previously done in V. coralliilyticus strain YB2 [37], a mutation abolishing flagellar movement was constructed in OCN003 to preserve the cell's morphology but not propel the organism. Previous work has shown that flagella can be quite immunogenic [39][40][41], so the introduction of two simultaneous phenotypic changes stemming from one mutation, the lack of a flagellum and abrogation of motility, would introduce additional difficulty into the data interpretation. Therefore, the fliF homolog in OCN003, which encodes a flagellar motor protein, was deleted to create a non-motile mutant that remained flagellated. Deletion of fliF in other bacteria results in the production of a non-functional flagellum, which abolishes bacterial motility and potentially impairs adhesion to the host [42,43]. The non-motile strain of OCN003 was created by replacing all but the first and last 18 nucleotides of the coding region of the fliF homolog in OCN003 [AIY66328] with the bla gene, which confers resistance to the antibiotic ampicillin. The resulting strain, OCN003 ΔfliF::bla, displayed wild type physiology, growth rate, and retained the ability to form a biofilm (S4 Fig). Though OCN003 ΔfliF::bla formed a flagellum (S5 Fig), it was non-functional and the mutant could no longer swim as confirmed by light microscopy and semi-solid agar assays (Fig 3). When the OCN003 ΔfliF::bla strain was used to infect cMWS-affected M. capitata fragments, the mutant was incapable of inducing the switch from cMWS to aMWS (Mantel-Cox test: n = 8, p < 0.01). The introduction of a plasmid carrying a functional copy of fliF into the OCN003 ΔfliF::bla mutant complemented the mutation and restored its ability swim and cause the switch from cMWS to aMWS at rates that were similar to wild type (Fig 4; Mantel-Cox test: n = 8, p = 0.90). The complemented OCN003 ΔfliF:: bla mutant was also successfully re-isolated from all of the experimentally infected fragments. These results fulfill Koch's molecular postulates for the requirement of fliF, and motility by extension, for proper virulence of OCN003 in the cMWS to aMWS switch. This indicates that  Pseudoalteromonas piratica, a coral pathogen wild type OCN003, not just the presence of viable OCN003 cells, is required to induce the switch from cMWS to aMWS.
Pseudoalteromonas piratica strains were isolated from diseased M. capitata fragments collected during the 2010 aMWS outbreak Based on the finding that OCN003 can cause aMWS, it is possible that coral colonies displaying aMWS in the field during outbreaks were infected by OCN003. To determine whether OCN003, or similar P. piratica strains, may have been involved in the March 2010 aMWS outbreak that affected several hundred colonies of M. capitata [7], bacteria were cultured from the mucus of infected colonies and identified. A total of 265 bacterial isolates were cultured from four aMWS-infected coral fragments collected during the outbreak. These isolates were screened  Pseudoalteromonas piratica, a coral pathogen by PCR using previously described primers (OCN008-42310-F and OCN008-43080-R) specific to V. coralliilyticus strain OCN008, a pathogen known to cause aMWS in M. capitata in Kāne'ohe Bay [14], or primers that selectively amplify a unique region of the OCN003 genome (003unique-F and 003unique-R). The unique region used for OCN003 identification was found by searching the genome [28] for a large intragenic region between convergent genes that was not similar to any sequences present in the NCBI database (see Materials and Methods for a description). Five isolates from the 2010 outbreak grew on Vibrio-selective TCBS agar, but none yielded a PCR product using the OCN008-specific primers. In contrast, three of the 265 isolates yielded a PCR product when screened with the OCN003-specific primers. The 16S rRNA gene sequences from each of the three OCN003-like isolates, referred to as OCN050, OCN051, and OCN052, were identical to the sequence from OCN003 and all three isolates clustered with OCN003 during MLSA comparison, which supports their classification as P. piratica strains OCN050, OCN051, and OCN052 (Fig 5). Based on the similarity of these isolates, the remainder of the study utilized OCN050 as a representative isolate from the outbreak.
While sequence identity indicated that strains OCN003 and OCN050 are genetically similar, the next step was to determine whether OCN050 could also induce aMWS in healthy M. capitata fragments and induce the cMWS to aMWS switch in a manner similar to OCN003. During infection trials conducted like those above, OCN050 infected healthy fragments of M. capitata at levels that were similar to those of OCN003, 25% and 33%, respectively (Mantel-Cox test: n = 12, p = 0.71; Fig 6). None of the negative control fragments displayed any signs of tissue loss following exposure to FSW or the negative control bacterium, OCN004 (n = 12 per treatment). In addition, OCN050 induced a switch from cMWS to aMWS in 38% of the exposed cMWS-affected fragments (McNemar's test: n = 16, p = 0.04), which was not significantly different from the OCN003-mediated switch from cMWS to aMWS (Mantel-Cox test: n = 16, p = 0.50; Fig 6). These data indicate that OCN003 and OCN050 infect M. capitata fragments similarly and can induce the switch from cMWS to aMWS in laboratory conditions. Taken together, OCN003 and OCN003-like strains are etiological agents of acute tissue loss disease in M. capitata and may have been involved in the 2010 aMWS outbreak. P. piratica strains and V. coralliilyticus strain OCN008 induce the cMWS to aMWS switch at different rates Like OCN003, Vibrio coralliilyticus strain OCN008 has fulfilled Koch's postulates as an etiological agent of aMWS for M. capitata [14]. It is possible that any bacterial agent that can cause Pseudoalteromonas piratica, a coral pathogen aMWS can induce a switch from cMWS to aMWS, and that this is not an OCN003-specific interaction with M. capitata. To assess the specificity of two aMWS-causing pathogens to induce the switch from cMWS to aMWS, cMWS-affected M. capitata fragments were exposed to OCN008. When healthy coral fragments were exposed to OCN008 as a primary pathogen, 80% of the fragments developed aMWS in an average of two days post-inoculation (McNemar's test: n = 22, p < 0.01). However, when aliquots of the same OCN008 cultures were used to inoculate M. capitata with cMWS, only 46% of the fragments developed aMWS in an average of 10 days post-inoculation (McNemar's test: n = 13, p = 0.04; Fig 7). In contrast, cMWSaffected fragments from the same diseased colonies were more susceptible to OCN003, which was able to induce the switch to aMWS in 62% of specimens in an average six days post-inoculation during concurrent infection trials. Analysis of the infection rates of OCN003 on healthy and cMWS-affected M. capitata shows that its ability to infect fragments with pre-existing cMWS is significantly higher than its infection of healthy fragments (Mantel-Cox test: n 1 = 16, n 2 = 12, p = 0.03). The opposite result was found for OCN008, in which its ability to infect cMWS-affected M. capitata fragments was statistically lower than its infection of healthy fragments (Mantel-Cox test: n 1 = 13, n 2 = 22, p < 0.01). When healthy and cMWS-affected fragments were exposed to the control bacterium, OCN004, no fragments switched from cMWS to aMWS. Taken together, these results demonstrate that OCN008 infects healthy M. capitata more effectively while OCN003 is more effective at infecting cMWS-affected colonies.

Discussion
Here we describe a novel Pseudoalteromonas species, P. piratica strain OCN003, which acts as another etiological agent of aMWS on healthy M. capitata coral and as a pathogen that can induce the switch from cMWS to aMWS. Additional strains of P. piratica, were isolated from M. capitata displaying acute tissue loss collected during a 2010 aMWS outbreak. These P. piratica strains also infected healthy M. capitata fragments in controlled laboratory experiments, suggesting that P. piratica may have been involved in the 2010 aMWS outbreak in Kāne'ohe Bay. Additionally, motility was required as a virulence mechanism influencing the ability of Kaplan-Meier survival curve of healthy or cMWS-affected M. capitata fragments exposed to OCN008 or the control bacterium, OCN004, at 25˚C. The solid line with open black rhombi represents healthy coral fragments exposed to OCN008 (n = 22); the dashed line with closed black rhombi represents cMWS-affected coral fragments exposed to OCN008 (n = 13); the dotted line with open circles represents healthy or cMWS-affected coral fragments exposed to OCN004 (n = 22 or n = 13, respectively). The concentration of bacteria used was 10 8 CFU/ml of seawater.
https://doi.org/10.1371/journal.pone.0188319.g007 OCN003 to induce the switch from cMWS to aMWS. This work also demonstrates that two different bacteria (P. piratica strain OCN003 and V. coralliilyticus strain OCN008) can elicit the same disease signs (aMWS) in a single coral species (M. capitata). Additional undiscovered pathogens may be capable of causing aMWS and other coral diseases with known etiologies.
The two pathogens, OCN003 and OCN008, displayed very different levels of virulence on healthy vs. cMWS-affected M. capitata. OCN003 infected 33% of healthy fragments roughly 22 days post-inoculation, but infected 56% of cMWS-affected fragments about eight days postinoculation. In contrast, OCN008 infected 80% of healthy fragments two days post-infection but only infected 46% of cMWS-affected fragments about ten days post-infection. The differences in rates of infection between these bacteria on healthy vs. cMWS-affected M. capitata suggests that some component of either the coral or the pathogen are modulating virulence under these conditions. The microflora present in healthy coral mucus confer some level of protection against bacterial infection either by competing for space or nutrients or by producing antimicrobial compounds to inhibit pathogen growth [44][45][46][47][48][49]. Gochfeld and Aeby (2008) [50] demonstrated that crude aqueous extracts from healthy M. capitata colonies from Hawaiian reefs including Kāne'ohe Bay, exhibited a significant amount of antibacterial activity with a high degree of selectivity; certain bacteria were inhibited whereas others were not. Bacteria, including OCN003 and OCN008, vary in their sensitivity to antimicrobial compounds. OCN003 is resistant to kanamycin (100 μg) whereas OCN008 is resistant to both kanamycin (25 μg) and ampicillin (800 μg) [14,25]. Perhaps OCN008 and OCN003 differ in their sensitivities to antibiotics produced by the microflora on M. capitata, which could affect their respective capacities to infect.
OCN008 and OCN003 differed significantly in their ability to infect a compromised host (coral with cMWS). Studies have shown that the bacterial communities in coral mucus change significantly during infection [51][52][53][54][55][56]. This pattern holds true for M. capitata; colonies with cMWS have different bacterial communities compared to uninfected colonies [24]. The shift in the coral-associated microflora could differentially modulate the infectivity of these pathogens, increasing the ability of OCN003 to infect and decreasing the infectivity of OCN008. OCN003 may be able to infect cMWS-affected M. capitata colonies more readily because some members of the bacterial microflora present in healthy mucus inhibited OCN003 infection, but not OCN008 infection, and were absent from diseased fragments. For example, Smith (2008) found that Streptomyces, a genus known for their capacity to produce antibiotics [57], was a dominant component of the culturable bacterial community in healthy M. capitata but were absent in cMWS-affected coral. Why OCN008 and OCN003 differ in their abilities to infect healthy and cMWS-affect M. capitata is currently unknown but this study raises some interesting hypotheses that warrant further investigation.

Conclusions
This work describes the first non-Vibrio etiological agent of acute tissue loss disease in M. capitata, Pseudoalteromonas piratica strain OCN003, and is also the first member of this genus confirmed as a coral pathogen. We show that OCN003 can infect healthy M. capitata as an etiological agent of aMWS but was able to infect corals at a significantly higher rate when inoculated onto M. capitata displaying chronic MWS (cMWS). We contrast this with the known pathogen V. coralliilyticus strain OCN008, which infects compromised coral (cMWS) at a significantly lower rate than healthy coral. The different levels of virulence displayed by these two bacterial pathogens toward M. capitata when used to induce aMWS in either healthy or cMWS-affected fragments, suggests that some component of either the coral or the pathogen can modulate virulence in this system. This work provides the first demonstration of two different genera of bacterial pathogens that act as etiological agents producing identical lesions in corals.
Supporting information S1 Fig. Schematic of the coral infection protocol utilized in this work. The organisms used for infection are grown to the optical density specified, washed with artificial seawater (ASW), and inoculated into temperature-controlled aquaria housing healthy (right track) fragments of M. capitata or fragments displaying chronic Montipora white syndrome (cMWS; denoted by a dark spot on the fragment). Following inoculation, fragments were monitored for the onset of tissue loss similar to acute Montipora white syndrome (aMWS), which can result in exposure of the white coral skeletons. Prior to complete lysis, the remaining tissue from aMWS infectioned fragments was harvested, homogenized, and either plated on appropriate media to recover the pathogen and used to identify desired bacteria. showing the presence of a polar flagellum (arrow). Cells used for analysis were deposited on Formvar-coated copper grids and contrasted with 1% uranyl acetate for viewing on a Hitachi HT7700 TEM at 100 kV. Images were captured with an AMT XR-41B 2k x 2k CCD camera. Scale bar represents one μm. (TIFF)