Differential Production and Secretion of Potentially Toxigenic Extracellular Proteins From A Hypervirulent Strain of Aeromonas Hydrophila

Background: Hypervirulent Aeromonas hydrophila (vAh) is a pathogen in freshwater aquaculture that results in the loss of over 3 million pounds of marketable channel catsh, Ictalurus punctatus, and catsh hybrids (I. punctatus, (cid:0) x blue catsh, I. furcatus, (cid:0) ) each year from freshwater catsh production systems in Alabama, U.S.A. vAh isolates are clonal in nature and are genetically unique from, and signicantly more virulent than, traditional A. hydrophila isolates from sh. Even with the increased virulence, natural infections cannot be reproduced in aquaria challenges making it dicult to determine modes of infection and the pathophysiology behind the devastating mortalities that are commonly observed. Despite the intimate connection between environmental adaptation and plastic response, the role of environmental adaption on vAh pathogenicity and virulence has not been previously explored. In this study, secreted proteins of vAh cultured as free-living planktonic cells and within a biolm were compared to elucidate the role of biolm growth on virulence. Results: Functional proteolytic assays found signicantly increased degradative activity in biolm secretomes; in contrast, planktonic secretomes had signicantly increased hemolytic activity, suggesting higher toxigenic potential. Intramuscular injection challenges in a channel catsh model showed that in vitro degradative activity translated into in vivo tissue destruction. Identication of secreted proteins by HPLC-MS/MS revealed the presence of many putative virulence proteins under both growth conditions. Biolm grown vAh produced higher levels of proteolytic enzymes and adhesins, whereas planktonically grown cells secreted higher levels of toxins, porins, and mbrial proteins. Conclusions: This study is the rst comparison of the secreted proteomes of vAh when grown in two distinct ecological niches. These data on the adaptive physiological response of vAh based on growth condition increase our understanding of how environmental niche partitioning could affect vAh pathogenicity and virulence. Increased secretion of colonization factors and degradative enzymes during biolm growth and residency may increase bacterial attachment and host invasiveness,


Background
Aeromonas hydrophila is a wide-spread and diverse species of Gram-negative bacterium ubiquitous in freshwater aquatic ecosystems. As a rapidly growing and metabolically diverse generalist (1)(2)(3)(4)(5), A. hydrophila is capable of exploiting a variety of ecological habitats and a broad range of hosts. A. hydrophila has been isolated from almost every freshwater aquatic environment and from diseased mammals, reptiles, amphibians, insects, and sh (1,(6)(7)(8). A. hydrophila has been found in association with processed poultry, meats, sh, and even bottled water. It is capable of withstanding chlorination and is resistant to multiple antibiotics (1,9).
In aquaculture, A. hydrophila is an important cause of disease in most freshwater production systems. Historically, A. hydrophila has been an important secondary pathogen in cat sh production systems, commonly responsible for cutaneous ulceration and muscle necrosis. Occasionally following sh stress (low oxygen, poor water quality, etc) the bacterium can cause a septicemia (motile aeromonad septicemia [MAS]), resulting in high mortalities (10)(11)(12)(13)(14). In 2009, a new, highly virulent strain of A. hydrophila was isolated from a diseased channel cat sh, Ictalurus punctatus, within a production pond in West Alabama. This strain, referred to as hypervirulent Aeromonas hydrophila, or vAh, was responsible for outbreaks of peracute motile aeromonad septicemia of epidemic proportions (11,(15)(16)(17)(18)(19). vAh apparently acts as a primary pathogen, and may not be preceded by immune insult (11). To date, vAh has been responsible for the loss of 30 million pounds of marketable channel cat sh from production farms in West Alabama. In 2017, A. hydrophila infections were responsible for the loss of 3.4 million pounds of farm-raised cat sh in Alabama alone, more than twice as much as the second leading cause of loss, Flavobacterium columnare. vAh has been the primary or secondary cause of cat sh loss in Alabama since the primary outbreak in 2009 (Hemstreet, AL Fish Farming Center).
While much current research is focused on MAS disease prevention, there are many important unanswered areas of research to understanding the mechanisms of vAh pathogenesis, bacterial-host interactions, and bacterial adaptive responses under different environmental conditions. Current studies of vAh pathogenesis and virulence are performed almost exclusively with planktonicallycultured bacteria despite the fact that most free-living generalist bacteria in aquatic systems reside primarily within bio lm (20)(21)(22)(23)(24)(25). One study using only planktonically-cultured vAh reported the presence of 228 extracellular proteins (ECPs) in the supernatant of vAh broth cultures, at least 23 of which were putative virulence factors (18). No studies have evaluated the secretome of bio lm grown vAh. A recent study by Cai et al. (2018) found no vAh present in the water column through the survey period, (July-October), while vAh resident in bio lm and pond sediment was detected at an increasing rate in the same sampling period, suggesting that bio lms serve as a stable reservoir for vAh survival when planktonic conditions are less favorable. Bio lm-associated bacteria generally have increased adhesive properties (20,21,(26)(27)(28) and may have increased production of proteolytic enzymes, both of which could increase virulence (12,(29)(30)(31). Red eld (2002) suggested that extracellular proteases are expressed when diffusion and/or mixing is reduced. vAh residing within a bio lm may have an advantage in attaching to and invading sh tissues due to increased secretion of proteolytic enzymes and adhesins. Given the data supporting the presence of vAh within pond bio lms, it is important to identify virulence factors secreted during bio lm residence that could impact host attachment and invasion. In this study, we compared the secreted protein pro les (secretomes) of bio lm-or planktonically cultured vAh strain ML09-119 to determine if niche occupation could in uence vAh pathogenicity in natural environments.

Results
Bio lm grown vAh express higher protease activity but less hemolytic activity than planktonic cultures Protease activity in bio lm samples was observed to be more than 2 times higher than in planktonic samples, and 1.2 times higher than the trypsin positive control (p < 0.05; Figure 1). Similarly, elastase activity was signi cantly higher for bio lm grown vAh, which expressed 2.7 greater elastase activity than observed for planktonically grown cultures (p < 0.05; Figure 2). In contrast, hemolytic activity was greatly increased in planktonic cultures, with more than 6 times higher hemolytic activity compared to bio lm grown vAh (p < 0.05; Figure 3).

Severe sh tissue necrosis was induced by the secretome of bio lm-grown vAh
To determine if the increased proteinase and elastase activities observed from vAh when grown as an in vitro bio lm would result in tissue damage indicative of MAS disease, 10 µg of secreted proteins from each growth condition was injected intra-muscularly into channel cat sh. Two hours post-injection, loss of dermal pigment was noted at the injection site in bio lm-injected sh, but no changes were observed for the injected planktonic secretome. After 24 hours, substantial tissue necrosis was observed grossly in all bio lminjected sh ( Figure 4). Fish injected with planktonic-associated ECPs developed no gross lesions even after 7 days. No control sh developed any gross lesion at the injection site after 7 days.
Histopathology was performed on skin and subcutaneous tissues collected from injection sites of channel cat sh. Bio lm-injected sh tissue was edematous, hemorrhagic, and there was extensive tissue necrosis at the site of injection ( Figure 4). Despite substantial tissue damage, few in ammatory cells were present. In contrast, sh injected with planktonic ECPs were identical to the control sh, with no perceptible damage to skin, subcutaneous adipose tissue, or muscle.
Bio lm and planktonically grown vAh have distinct secretomes The differences in enzyme activities and tissue damage observed for bio lm versus planktonically-cultured vAh supported the hypothesis that niche occupancy has a signi cant in uence on vAh exoprotein expression. To further test this hypothesis, a secretome analysis was performed to identify differentially secreted proteins present under the two culture conditions. A total of 272 proteins were identi ed in the secretomes of bio lm and planktonically-cultured vAh. Eighty-two proteins were identi ed that were present in both secretomes, while 98 were identi ed only in bio lm secretomes and 92 were unique to planktonic secretomes. ROTS and T-test analyses identi ed 53 proteins that signi cantly (FDR < 0.05, p = 0.01) varied in abundance. The protein abundance ratios of 52 ROTS-identi ed proteins were above the signi cant fold change threshold of ≥1.5 (Table 1) Figure 5) revealed extensive secretion of degradative enzymes and toxins in both bio lm and planktonic secretomes, with degradative enzymes, such as elastase, metalloprotease, chitinase, and endochitinase, dominating bio lm secretomes and cytotoxic and cytotonic toxins, such as ahh1-type hemolysin and extracellular lipase enriched in planktonic secretomes. In both planktonic and bio lm secretomes, degradative enzymes and toxins made up the majority of signi cant proteins, representing 79.8% of planktonic proteins and 55.7% of bio lm proteins. Proteins involved in transport (16.5%), carbohydrate metabolism (8.5%), and pilus and agellin (3.6%) contributed signi cantly to the bio lm secretome, while pilus and agellin proteins (5.8%), outer membrane proteins (4.0%), and proteins involved in transcriptional regulation and electron transport (3.5%) were other signi cant contributors to planktonic secretomes ( Figure 5). Of particular interest were the presence of polar agellar proteins (AHML_09345 and _09350) present in higher quantities in the bio lm secretome and type I pili proteins (AHML_2665 and _2690) that were present in planktonic secretomes, but absent from bio lm secretomes. Polar agella, typically considered motility agella, are important in adhesion and invasion in A. hydrophila that lack lateral agella, such as vAh (32), while type I pili are thought to contribute to host colonization, but not host invasion (33).

Discussion
While vAh has established itself as a primary pathogen in natural settings (11), laboratory-cultured vAh appears to mimic its opportunistic relatives during immersion challenges. Planktonically-cultured vAh is extremely virulent, causing death in a matter of hours in intraperitoneal injection challenges. However, models meant to mimic more natural infections including submersion and gavage have been unreliable, even when challenged with arti cially high colony forming units (CFUs) of the bacterium (2, 34). Zhang et al. reported consistent MAS mortality was attainable in channel cat sh immersion trials only following scari cation and challenge with 2 × 10 7 CFU/ml of planktonically-cultured vAh (35). This suggests that some environmental stimuli are not present in arti cial broth culture, which, in pond systems, could be responsible for inducing bacterial virulence and resulting in large scale MAS epidemic outbreaks. Since most environmental bacteria spend much of their time in bio lm, either attached to a substrate or oating as bacterial ocs (20,22,23,25,36), bio lm-associated vAh may produce proteins that increase invasiveness and allow initial colonization in vivo (37). The ability to form a bio lm is commonly considered a virulence factor, particularly in human disease conditions (38). Likewise, for A. hydrophila bio lm formation and residency may induce global changes in gene expression resulting in increased production and secretion of degradative enzymes and other factors that increase pathogenicity or invasiveness. Aeromonas spp. produce extracellular enzymes that facilitate nutrient acquisition in aquatic environments and produce adhesins that aid in the attachment and colonization of benthic surfaces (5). In aquatic environments, these enzymes provide nutrients by degrading the organic compounds including suspended detritus and benthic substrates. These enzymes may also be important in the pathophysiology of disease by enabling degradation of animal tissues (5,20,39).
Previous research reported the presence of 23 potentially toxigenic extracellular proteins in the supernatant of planktonically-cultured vAh (18). Because many opportunistic bacteria like vAh reside largely in bio lms and not as sustained planktonic populations (40), it was important to evaluate the in uence of bio lm growth on vAh exoprotein expression. This study found that degradative activities were signi cantly increased in the supernatant of bio lm-associated vAh ( Figs. 1 and 2). Furthermore, when bio lm-grown vAh ECPs were injected into the muscle of channel cat sh, signi cant necrosis and cytolysis occurred within 24 hours, while secreted proteins of planktonically-cultured vAh failed to produce necrotic lesions after seven days.
A secretome analysis was conducted to examine in more detail how bio lm growth in uenced vAh exoprotein expression, which revealed signi cant differences in the secretomes of the two cultures, both in complexity and quantity. The bio lm secretome contained 248 proteins, including 183 unique proteins, while planktonic secretomes contained 183 total proteins, including 101 unique proteins. Of the 82 proteins secreted under both culture conditions, at least 36 had previously been identi ed as putative virulence factors (18,41,42). Under both growth conditions, vAh secreted an abundance of potential virulence proteins, the majority of which were not statistically signi cant in differential secretion analyses. However, secretomes of vAh cultured in bio lm were signi cantly more varied and, in general, relative protein abundance was increased.
Assays to measure general and speci c proteolytic potential of the secreted proteins revealed signi cant increases in both caseinolytic and elastinolytic activity in bio lm secretomes when compared to planktonic ECPs (Figs. 1 and 2). A signi cant difference in proteolytic potential was also seen upon inspection of the secretome analysis. ROTS analysis revealed at least seven degradative proteins were present in the bio lm secretomes at signi cantly higher observed abundance relative to planktonic secretomes. There was a 5-fold increase in elastase abundance in bio lm secretomes, with an average quantitative protein value (QPV) of 122, compared to an average QPV of 23 in planktonic secretomes, which agreed with the results obtained from elastase enzyme activity measurements. There was a 3-fold increase in the M66 -family metalloprotease AHML_05230 in bio lm secretomes, with average QPVs of 103 and 30 in bio lm and planktonic secretomes, respectively. Both elastase and the M66 zinc metalloprotease are considered signi cant virulence factors of A. hydrophila as well as other pathogens, such as Vibrio cholerae, and enterohemorrhagic Escherichia coli (43,44). Five other proteolytic enzymes were secreted in statistically signi cant quantities in bio lm secretomes but were not detected in the planktonic secretomes and likely increase the overall proteolytic potential of bio lm ECPs (Table 1).
While the majority of the differentially secreted degradative enzymes present in the bio lm secretome were proteolytic, two important glycolytic proteins, chitinase and chitin binding protein (CBP) were found in signi cantly higher amounts in bio lm secretomes. While chitinase and CBP are integral in the breakdown of environmental chitin, these proteins may also play integral roles in virulence. Though vAh can use chitin as a sole carbon source (45), the lack of chitin in the TSB growth medium would make it unlikely that chitinase and CBP production would be energetically favorable. Therefore, it is hypothesized that these chitin-associated proteins play other roles in vAh tness or pathogenicity. In other pathogens, chitinases and CBPs are considered virulence factors not because they target chitin but because of their interactions with substrates other than chitin. In some virulent E. coli and V. cholerae, chitinases and CBPs target host glycoproteins and glycolipids that contain N-acetylglucosamine (GlcNAc), the monomer present in mucus (46,47). Outer membraneexpressed chitinases and CBPs have also been indicated as accessory molecules responsible for initiating host cell adhesion and invasion (46)(47)(48). In a murine model, E. coli chitin-binding domain interacts with intestinal epithelial cells, increasing invasiveness and pathogenicity (47). In V. cholerae, Bhowmick et al. (2008) found chitinases function to break down the GlcNAc of mucin and reported upregulation of chitinases resulting from exposure to exogenous mucin. Furthermore, the V. cholerae chitin binding protein GbpA was shown to bind to the protective mucus layer of mammalian intestinal epithelium, resulting in bacterial colonization and disease initiation. Likewise, chitinases and CBPs produced by clinical Pseudomonas aeruginosa strains isolated from patients with cystic brosis (CF) were also upregulated in response to mucin-containing sputum and likely play an integral role in primary adhesion to lung epithelium in the initiation of CF (46). In sh, the mucosal barrier covering the gills, skin, and intestinal surfaces are considered the rst line of defense against invading pathogens (49,50). The presence of chitinases and CBP may act to degrade not only the cat sh slime coat, but also to bind to and degrade the epithelial mucins in the digestive tract, increasing vAh invasiveness. Peatman et al (2018) reported a direct link between feed consumption and vAh-induced MAS, with survival in vAh-challenged cat sh decreasing signi cantly when sh were fed to satiation 4 hours prior to challenge. The mucus coating of the intestinal epithelium may decrease after eating, as ingesta moves through the digestive tract and takes mucus with it. Chitinases and CBPs may then be capable of breaking down the remaining mucus, gaining access to the underlying epithelium and, eventually, the bloodstream (51). The presence of chitinase and CBP could help explain the intestinal epithelial damage found on necropsy in sh naturally infected with vMAS (52). Although signi cantly higher in bio lm secretomes, chitinase and CBP was prominent in both planktonic and bio lm secretomes, suggesting they play an important role in bacterial tness regardless of growth condition.
Whereas bio lm secretomes were ush with degradative exoenzymes, such as elastase, chitinases, and multiple Zn-dependent and metalloproteases, planktonic secretomes consistently produced more hemolytic and cytotoxic ECPs. Notably, both aerolysin-type and ahh1-type hemolysins were detected in much higher quantities in planktonic secretomes, as were two extracellular serine proteases (neither of which were identi ed in any bio lm sample) and extracellular lipases, all of which exert hemolytic activity against erythrocytes, and have been shown to be cytotoxic to cells (53,54). Interestingly, the alpha-hemolysin, phospholipid-cholesterol acyltransferase, which was present in planktonic secretomes but absent in bio lm, has been reported to produce signi cant lysis of salmon erythrocytes following activation by serine protease (55). The presence of substantial amounts of both proteins in the planktonic secretomes suggests that the production of these proteins could allow a multi-pronged approach to cell death, with each toxin acting independently, but increasing the collective virulence resulting from multiple exoproteins. Aerolysin-type hemolysin has been implicated as the main virulence factor of A. hydrophila (54), and was signi cantly higher in planktonic secretomes, with a three-fold increase compared to bio lm. However, ahh1-type hemolysin was present in planktonic secretomes at greater than three times the amount of aerolysin-type hemolysin. Ahh1 hemolysins are homologous to hlyA hemolysins of V. cholerae (56). The activity of this poreforming hemolysin is not erythrocyte-speci c, but targets erythrocytes, leukocytes, lymphocytes, and epithelial and endothelial cells in a multitude of eukaryotes (57) and, as such, are considered cytotoxins. This supports the in vitro hemolysis assay results that found 80% hemolysis of channel cat sh erythrocytes in one hour when exposed to planktonic supernatants, compared to less than 15% average hemolysis of erythrocytes that were incubated with bio lm supernatants (Fig. 3). The presence of these hemolysins and other cytotoxins in planktonically-cultured vAh may also help explain the rapid mortality seen in cat sh when challenged by intraperitoneal injection, as these bacteria may be primed to produce vast amounts of toxins in vivo.
Biological functions of secreted proteins as analyzed by gene ontology found carbohydrate utilization to be the dominant function of secreted proteins under both conditions. Proteins involved in hemolysis, lipid and nucleotide catabolism, arginine biosynthesis, protein folding and transport were dominant biological functions of planktonic secretomes. Signi cant bio lm proteins were largely involved in transmembrane transport, amino acid processing, and transport of ions, amino acids, and carbohydrates. Interestingly, agellar motility was also important in bio lms. This is likely due to A. hydrophila's use of agella in bio lm construction and not for bacterial motility (32). This increase in polar agella may also contribute to an increased host colonization in bio lm-associated vAh. While lateral agella are often considered imperative for bio lm production and adhesion (27,58), Aeromonads that lack lateral agella are capable of using polar agella for bio lm formation as well as cellular adhesion (28,32,59,60). The increased polar agella required for bio lm formation could act secondarily as adhesins when bio lm-derived bacteria come into contact with cat sh mucosal surfaces and could act in concert with other secreted invasins to colonize and destroy host mucosal barriers.

Conclusions
Most aquatic bacterial generalists, such as A, hydrophila, spend the majority of time resident in bio lms and host-microbe interactions are likely in uenced by niche-speci c microbial phenotype. Because bio lm-associated bacteria have emergent properties that cannot be elucidated by the study of free-living cells, it is imperative to study organisms within bio lms to understand how niche adaptations may in uence overall pathogenicity and virulence. This study is the rst comparison of the secreted proteomes of vAh when grown in two distinct ecological niches. These data on the adaptive physiological response of vAh based on growth condition increase our understanding of how environmental niche partitioning could affect vAh pathogenicity and virulence. Increased secretion of colonization factors and degradative enzymes during bio lm growth and residency may increase bacterial attachment and host invasiveness, while increased secretion of hemolysins, porins, and other potential toxins under planktonic growth (or after host invasion) could result in increased host mortality. These shifts in protein expression and secretion indicate that growth under bio lm and planktonic conditions results in massive changes in gene expression. Future research should explore the global regulatory factors that affect vAh gene expression under these growth conditions. Taken together, these data may help in our understanding of the unique aspects of this emerging pathogen that contribute to the devastating impact of MAS disease outbreaks.

Methods And Materials
Bacterial Strain vAh strain ML09-119 was isolated from a diseased channel cat sh from a MAS outbreak in a West Alabama aquaculture facility in 2009. Molecular characterization and genome sequencing of vAh ML09-119 have been performed (19) and the complete genome sequence deposited in GenBank (Accession CP005966). Aliquots of vAh ML09-119 were cryogenically stored in 10% glycerol freeze medium at -80°C.

Culture Media and Culture Conditions
Tryptic soy broth (TSB) (Bacto TSB, BD) prepared according to manufacturer's directions was used as the culture medium for planktonic growth.
Bio lm media was prepared by adding 0.2% agar powder (AlfaAesar) to TSB media prior to sterilization. Approximately 70 ml of molten bio lm agar was poured into deep well petri dishes (Fisher) and allowed to solidify. Bacterial strain vAh ML09-119 was removed from cryogenic storage and inoculated into 25 ml TSB media and grown overnight at 30°C with shaking. A 1 ml aliquot of overnight culture was transferred to 70 ml of TSB and grown at 30°C on an orbital shaker to mid-log phase, approximately 16 hours. Bio lm agar plates were inoculated from overnight culture by stab inoculation. Plates were sealed with para lm and incubated at 30°C for 72 hours, until an adherent bacterial lm covered the agar surface.
Planktonic and bio lm cultures were performed in triplicate.

Secretome Preparation
Planktonic Secretome: vAh ML09-119 was cultured as described above. Cells were pelleted by centrifugation at 20,000 x g for 15 minutes at 4°C and supernatant was decanted and retained. Cells were washed twice with cold, sterile PBS, pelleted as above, and the wash was added to the supernatant. Remaining cells were removed by passage through a low-binding 0.22 µm vacuum lter (VWR). Cell-free supernatants were used as the starting point for puri cation of extracellular proteins (ECPs).
Bio lm Secretome: vAh ML09-119 cells were gently removed from the bio lm media surface with a sterile cell scraper, transferred to 50 ml conical tube, and washed twice with cold, sterile PBS as described above. The cell wash was decanted and retained. To collect secreted proteins within bio lm media, the plates were disrupted using a sterile disposable probe until the soft agar had formed a slurry. The agar slurry was transferred to a sterile 50 ml conical tube and centrifuged at 20,000 x g for 15 min at 4°C to pellet the agar. Following centrifugation, the liquid media was decanted from the agar plug and retained. The agar plug was then resuspended in 20 ml cold sterile PBS, centrifuged as above, and the wash solution decanted and retained. All wash solutions and liquid media were combined and ltered, rst through a low-binding 0.45 µm vacuum lter (VWR), then through a low-binding 0.22 µm vacuum lter to remove any residual agar and bacterial cells. This cell-free supernatant was used at the starting point for bio lm ECP puri cation.
Ammonium Sulfate Precipitation: ECPs were precipitated from cell-free supernatants by the addition of ammonium sulfate crystals (Fisher Scienti c) to achieve 65% saturation, followed by incubation at 4°C on a rotary platform shaker with gently mixing for 24 hours. Precipitated proteins were collected by centrifugation at 30,000 x g for 45 min at 4°C, then dissolved in 10 ml cold Tris buffer (20mM Tris-Hcl, pH 7.6) + protease inhibitor (Complete tablets, mini, EDTA-free (Roche)). Resuspended proteins were dialyzed twice, for 18 hours and 12 hours, respectively, against the same buffer in 10 Kda dialysis cassettes (Slide-A-Lyzer (Thermo Fisher)). After dialysis, the total volume was adjusted to 20 ml by the addition of cold Tris buffer. The protein concentration of each sample was determined by the Bradford assay (Pierce Coomassie Plus Protein Assay, Thermo Fisher). These concentrated proteins were used for all assays.

Enzymatic Activity
The in vitro activity of secreted proteins was measured using speci c substrates to determine the degradative and toxigenic potential of planktonic and bio lm secretomes, as described below: Hemolysis: Hemolytic potential was measured using the method of Peatman et al. (2018) with some modi cations. In brief, heparinized blood from three channel cat sh was pooled and diluted 1:10 in sterile phosphate buffered saline (PBS). A suitable dilution of protein in 150 µl PBS buffer was added to 25 µl diluted blood in sterile microcentrifuge tubes. Tubes were incubated at 30°C in an orbital shaker for 2 h. Positive control tubes representing 100% hemolysis contained 150 µl sterile distilled water in place of protein samples. Negative control tubes contained 150 µl sterile PBS in place of protein samples. Controls were incubated with 25 µl diluted blood as above. Following incubation, tubes were centrifuged at 1,000 x g to pellet un-lysed cells and 150 µl of supernatant was transferred to 96-well at bottom plates. Erythrocyte lysis was quanti ed by measuring absorbance of released hemoglobin at 415 nm in multi-mode plate reader (Synergy HTX, Bio-Tek) and hemolysis was reported as percent of positive control.
Universal Protease Activity: Non-speci c proteolytic activity was measured using HiLyteFluor 488-labeled casein as the substrate, following manufacturer's protocol with minor modi cations (Sensolyte Green Fluorimetric Protease Assay Kit, AnaSpec, Inc.). Brie y, a suitable concentration of protein in 50 µl deionized water was added to triplicate wells of black, at-bottom 96-well plates with nonbinding surface (Greiner Bio-One). Trypsin, diluted 50-fold in deionized water, acted as a positive control and sterile deionized water served as a substrate control. Following the addition of 50 µl labeled casein substrate, plates were mixed brie y and uorescent intensity was measured at Ex/Em = 490 nm/520 nm every ve minutes for one hour in a multi-mode plate reader (Synergy HTX, Bio-Tek) with 30°C incubation temperature. Data were plotted as relative uorescence units versus time for each sample.
Elastase Activity: Elastase-speci c activity was measured using 5-FAM/QXL TM 520 labelled elastin as the substrate, following the manufacturer's protocol with minor modi cations (Sensolyte Green Fluorimetric Elastase Assay Kit, AnaSpec, Inc.). Brie y, a suitable concentration of protein in 50 µl deionized water was added to triplicate wells of black, at-bottom 96-well plates with non-binding surface. Elastase, diluted 50-fold in assay buffer, acted as a positive control and sterile, deionized water was a substrate control. Following the addition of 50 µl labeled elastase substrate, plates were mixed brie y and uorescent intensity was measured continuously at Ex/Em = 490 nm/520 nm, and data recorded every ve minutes for one hour in a multi-mode plate reader (Synergy HTX, Bio-Tek) with 30°C incubation temperature. Data were plotted as relative uorescence units versus time for each sample.
In vivo Proteolysis Extracellular protein activity was measured in vivo using channel cat sh ngerlings to determine potential proteolytic and cytotoxic tissue effects.
Cat sh: Speci c-pathogen free channel cat sh ngerlings maintained under Auburn University IACUC-approved protocol 2018-3251 (Cat sh Production and Maintenance) were used for challenges. All challenges were performed adhering to the guidelines of AU-IACUCapproved protocol 2016-2900 (Identi cation of toxigenic proteins of virulent Aeromonas hydrophila and evaluation of host response).
Protein Preparation: Ten microgram aliquots of secreted planktonic and bio lm-associated proteins, prepared as above, diluted in 100 µl sterile PBS were used for injection challenges.
Challenge Model: Channel cat sh ngerlings were transferred to 57-liter glass aquaria containing dechlorinated municipal and acclimated at 30°C for two days prior to challenge. Triplicate tanks containing ve sh each represented planktonic ECP, bio lmassociated ECP, and injection control groups. Prior to injection, ngerlings were transferred to sedation aquaria containing 70 mg/ L tricaine methanesulfonate (MS-222) buffered to neutrality with sodium bicarbonate. Following sedation, characterized by decreased opercular movement and loss of equilibrium, 100 µl of sterile PBS containing 10 µg of total protein was injected intramuscularly just below the dorsal n using tuberculin syringes tted with 26 gauge needles. Control sh were injected with 100 µl sterile PBS. Fish were then returned to the appropriate aquarium and monitored until fully recovered. Fish were maintained in aquaria at 30°C for 7 days under ow-through conditions at 1 gallon per hour water replacement. Moribund sh or sh developing severe external lesions were euthanized by prolonged exposure to buffered MS-222, the tissues were collected and xed in 10% neutral-buffered formalin. After 7 days, remaining sh were humanely euthanized and samples were collected and prepared as above.
Histology: Formalin-xed tissues were para n-embedded and 4 micron sections were prepared and stained with hematoxylin and eosin according to standard methods (61). Slides were evaluated and photographed using an Olympus BX53 microscope tted with an Olympus DP26 digital camera. Secretome Analysis.
To determine how vAh niche occupancy might in uence protein production, secreted protein pro les of vAh cultured within a bio lm and in broth were compared by liquid chromatography with tandem mass spectrometry (LC MS/MS) analysis at the UAB Mass Spectrometry/Proteomics shared facility to identify and quantify proteins present in each sample, as previously described in detail (62). In brief, raw MS Spectra were collected and data were converted to searchable les. Data were searched using SEQUEST (Thermo Fisher) with a species-speci c subset of the UniRef 100 database. Peptide IDs generated based on SEQUEST search results were ltered and normalized using Scaffold (Protein Sciences, Portland, Oregon), which retained only high con dence IDs and allowed for relative quanti cation across all samples. Relative quanti cation across samples were performed via spectral counting and, when relevant, spectral count abundances were normalized between samples. Proteins present in at least two experimental replicates were included in analyses. To determine statistical signi cance, two non-parametric statistical analyses were performed between each pair-wise comparison, including reproducibility-optimized test statistic (ROTS) (bootstrapping value = 1000) combined with single-tail t-test (p < 0.05) (63,64). These were then sorted according to the highest statistical relevance in each comparison. For protein abundance ratios determined by normalized spectral counts, a fold change threshold ≥1.5 was set for signi cance. Protein abundance of proteins present in only one experimental group was set as the average of the normalized quantitative value.
Protein Function: To de ne the potential function of secreted proteins, major biological processes of statistically signi cant proteins were determined from gene ontology annotation in UniProt (Consortium, T.U. 2018) and QuickGO (65). Predicted protein function was assessed by determining major biological processes through gene ontology. Using these data, eight functional groups were established, and proteins were sorted into these groups based on their primary biological function. A further comparison was made by compiling all proteins in each functional group from both bio lm and planktonic secretomes and expressing as parts of a whole, with side-by-side comparisons for each secretome type.

Statistical Analyses
Reproducibility-optimized test statistic (ROTS) analysis of differentially secreted proteins was performed in R (66). All other statistical analyses were performed in Prism 8.2.0 (Graphpad). One-way ANOVA followed by Tukey's multiple comparisons post-test were performed on triplicate data with signi cance set at p < 0.05. Graphical representations of data were produced in Prism 8.2.0.

Declarations
Ethics approval and consent to participate All animal challenges were approved by the Auburn University Institutional Animal Care and Use Committee (AU-IACUC) and were performed adhering to the guidelines of AU-IACUC-approved protocol 2016-2900 (Identi cation of toxigenic proteins of virulent Aeromonas hydrophila and evaluation of host response).

Consent for publication Not Applicable
Availability of data and materials The datasets generated and analyzed during the current study are included as supplementary les (Additional File 1) or from the corresponding author on reasonable request. General proteolytic potential of vAh extracellular proteins (ECPs) secreted under bio lm and planktonic growth. The general proteolytic potential of bio lm and planktonic secretomes was measure using HiLyteFluor 488-labeled casein as a substrate. Secreted protein from each condition was incubated at 30°C with labeled casein and uorescent intensity was measured at Ex/Em = 490nm/520nm every ve minutes for one hour. Data were plotted as relative uorescence units versus time for each sample. Three individual experiments were performed, and all samples were performed in triplicate. Statistical analysis consisted of one-way ANOVA followed by Tukey's multiple comparisons post-test with signi cance set at p < 0.05.

Figure 2
Elastase-speci c degradative potential of vAh extracellular proteins (ECPs) secreted under bio lm and planktonic growth. The elastase activity of bio lm and planktonic secretomes was measure using 5-FAM/QXLTM 520-labeled elastin as a substrate. Secreted protein from each condition was incubated at 30°C with labeled elastin and uorescent intensity was measured at Ex/Em = 490nm/520nm every ve minutes for one hour. Data were plotted as relative uorescence units versus time for each sample. Three individual experiments were performed, and all samples were performed in triplicate. Statistical analysis consisted of one-way ANOVA followed by Tukey's multiple comparisons post-test with signi cance set at p < 0.05.

Figure 3
Hemolytic potential of vAh extracellular proteins secreted under bio lm and planktonic growth. The hemolytic ability of vAh secreted proteins was measured using channel cat sh erythrocytes as the substrate. 2.5µg secreted proteins from each culture condition was incubated with 25µl cat sh blood diluted 1:10 in sterile PBS. at 30°C with shaking. Sterile, deionized water served as positive control and sterile PBS served as a negative control. Lysis was calculated by measuring sample absorbance at 415nm, and reported as percent positive control. All samples were assayed in triplicate. Statistical analysis consisted of one-way ANOVA followed by Tukey's multiple comparisons post-test with signi cance set at p < 0.05.