1. Introduction
Several
Tenacibaculum bacterial species are putative pathogens associated with te- nacibaculosis in fishes [
1,
2,
3,
4,
5,
6]. In British Columbia (Canada; BC),
Tenacibaculum species are commonly isolated from Atlantic salmon (
Salmo salar L.) with mouthrot; a regional, clinical presentation of tenacibaculosis, often characterized by epidermal ulcerations and development of yellow plaques around the mouth [
1,
5,
6,
7,
8].
Previous research exposing Atlantic salmon smolts to BC isolates of
T. maritimum described epidermal ulceration and bacterial plaques on the mouth and less frequently on other tissues [
6]; wherein exposure to
T. maritimum (isolate TmarCan15-1, 16-1, 16-5) at 10
5–10
7 cells mL
−1 induced >84% and >27% mortality of the exposed and cohabitant population [
6]. Atlantic salmon infected with
T. maritimum TmarCan16-1 predominately had ulcerations of the perioral epidermis and dissemination of the bacteria into the tooth pulp, and the authors proposed this as a potential route for the development of systemic infections [
5,
6].
T. dicentrarchi TdChD05 and 35/09
T, and
T. finnmarkense HFJ
T and Tsp.2 have likewise been used in exposure models [
4,
9]. Bath immersion with 10
5–10
6 colony-forming-units/cells mL
−1 resulted with 65% (TdChD05 [
4]), 20% (35/09
T [
4]), 80% (HFJ
T [
9]) and 10% (Tsp.2 [
9]) cumulative mortality, respectively, in exposed Atlantic salmon [
4,
9]. Unlike a previous report [
6], where mainly perioral lesions were noted, other studies recorded the presence of ulcerations at multiple locations on the fish [
4,
9]. These results indicate that
Tenacibaculum spp. may be able to establish and induce disease in multiple tissues. Variation in tissue tropisms could be related to
Tenacibaculum species being opportunistic pathogens, where dysbiosis [
10,
11], variation in isolate virulence, and host genetics may facilitate tenacibaculosis. More work is needed to support how tenacibaculosis is established and how the bacteria become systemic, as well as understanding the mechanisms for mortality in Atlantic salmon smolts.
Tenacibaculosis at BC aquaculture sites is treated through antibiotic application, primarily florfenicol; however, sulfadimethoxine and ormetoprim are also used. As of 2021, only one commercial vaccine for
T. maritimum is available in Spain for use in turbot (
Psetta maxima L.) (ICTHIOVAC
®-TM, Hipra Laboratories) [
12]. Vaccines for other aquaculture fishes (e.g., Atlantic salmon) are warranted as
Tenacibaculum outbreaks lead to direct profit loss associated with fish mortality in addition to the associated cost of serial applications of antibiotics [
13]. Before new vaccines can be developed, repeatable experimental infection models are required in target species such as Atlantic salmon.
This work aims to create a repeatable experimental infection model in Atlantic salmon by exposure to isolates most similar to T. maritimum NLF-15 (Tmar), T. dicentrarchi TdChD04 (Tdic), or T. finnmarkense Tsp.2 (Tfinn), and tracking fish pathology and mortality over time. The development of a repeatable infection model will be beneficial for vaccine testing as well as to help understand the pathogenesis of mouthrot in BC.
3. Discussion
Exposure to
Tmar and
Tdic resulted in mortality and similar clinical signs of mouth-rot. Unlike the
Tmar treatments, there were large variations between survivorship for the
Tdic treatment replicates for exposed fish. Inherit variations for survivorship have been recorded before for
Tenacibaculum-related exposures; in a previous study, variation between replicates was at or below 20% [
6]. Variations for survivorship, outside of the pathogen’s influence, may be related to host responses and the environment. The environment may be an important consideration, as variation could be related to the design of the RAS system, where tank-dependent effects may have occurred (e.g., differences in flow and aeration). However, with no major differences between survivorship for cohabitants between replicated, more research should occur to ensure that the variation between replicates for the survival of exposed and cohabitant fish is consistent. Interestingly, the
Tfinn treatments did not result in noticeable mortality or clinical signs. The contrasting results are likely not species-dependent but isolate and dose-dependent, as
Tfinn has been shown previously to induce tenacibaculosis in other experimental infection models [
9]. The lack of clinical signs produced by
Tfinn in this study could be influenced by genetic differences between isolates; the isolate of
Tfinn used here may not have been virulent. This is likely because a selection of isolates were largely random and isolates collected from fish with lesions can contain both virulent and non-virulent isolates; for example, fish with coldwater disease [
14]. The lack of demonstrated virulence may also be attributed to the methodology used, where bacteria may have not been in the correct conditions to produce lesions. Variable isolate pathogenicity has been described in other pathogens [
15,
16], including
Tenacibaculum [
6]. This could explain why
Tfinn and even
Tmar type-strain NCIMB 2154
T [
6,
9,
17] have been related to mortality in fishes on-site, but not in-vivo, and could also explain why some isolates that are genetically identical using 16S rDNA have variable results in inducing disease in lab studies [
6]. A third reason why
Tfinn did not induce mouthrot was that it was outcompeted by a similar
Tenacibaculum species for the same niche, as deceased fish for the
Tfinn treatment were positive using qPCR for
Tdic, but not
Tfinn. Often when a new niche is available for colonization, bacteria that have similar resource requirements will compete, where three long-term consequences are often described (exclusion [a winner], assignment of metabolic niches, or spatial separation of bacteria) [
18]. Contamination between tanks could have occurred, potentially through contaminated water, with
Tdic as the winner and can be supported by the fact that
Tdic is a common isolate from fish with mouthrot in BC [
8]. This potential phenomenon requires further investigation, perhaps using dual infections and monitoring isolates over time with qPCR.
Preliminary mathematical modeling of survival and bacterial enumeration in de-ceased fish may provide important insights to understanding mouthrot and tenacibaculosis. Prior research into aquatic pathogens has shown that cycle quotients from qPCR assays can differentiate disease states of fish and determine thresholds of bacteria required to cause infection, where similar applications have occurred for salmon gill poxvirus [
19] and bacterial coldwater disease [
20]. However, the qPCR assays developed by a previous study [
7], which were used in the present study, are not suitable for determining such a threshold in uncontrolled conditions as there are limitations relating to the specificity and copy number for the 16S rDNA gene in
Tenacibaculum. For
Tdic, mortality of exposed fish within 48 h is similar to [
4] and may indicate that the results of the
Tdic exposure are not solely based on the bacteria. In this study, and work conducted previously [
6], stress from the transition between freshwater and saltwater and exposure to bacteria could both play a significant role in the progression of disease in Atlantic salmon. This experimental model mimics the environmental challenges that salmon experience during the transition to netpen sites; in Canada, fish likely undergo considerable stress when directly transferred from freshwater hatcheries to marine sites. Shortly after the transition, mouthrot manifests and induces mortality in salmon until the fish reach a size threshold. Detection of
Tdic in the present study was at low numbers after exposure and after mortality had ceased indicating that
Tenacibaculum bacteria may become a commensal part of the microbiome as demonstrated for Atlantic salmon [
10] and Asian sea bass (
Lates calcarifer) [
21]. However, without a greater understanding of the microbiome of Atlantic salmon, it is difficult to define what a ‘healthy microbiome’ is and understanding if that includes
Tenacibaculum. In contrast,
Tmar was detected at low numbers, which increased, fluctuated marginally, and then decreased. Despite these fluctuations, mortality and clinical signs of disease persisted until the end of the trial, with a low probability of survival for any group.
Tmar may also induce disease in the host through exploiting dysbiotic changes to the microbiome; however, the ongoing or subacute disease profile, in contrast to the peracute one of
Tdic, strongly suggests marked differences in pathogenesis between
Tenacibaculum isolates. Additional complete genome sequencing, annotations, and comparisons would help identify why there could be differences in pathogenicity.
The pathogenesis of mouthrot could not be easily compared in this study, as fish were removed when moribund or dead, rather than sampling at a routine frequency throughout the trial similar to a previous experimental setup [
22]. The current sampling protocol can show what the fish looked like at or near death but provides limited insights into the development of disease. Subsequent studies will use predetermined routine sampling to investigate the pathogenesis of tenacibaculosis. Even though the sampling routine was sub-optimal, important key ideas about the pathogenesis of tenacibaculosis were identified, such that from a bath exposure trial,
Tdic and
Tmar managed to spread internally. Further, this occurred for both the exposed fish and cohabitants indicating that tenacibaculosis is horizontally transmissible and can become systemic. Important key ideas are supportive of previous research [
6], with a need for a greater understanding of the mechanisms for clinical sign presentation and mortality. Previous bioinformatics analysis of
T. maritimum identified numerous genes and potential virulence factors related to secretion and uptake systems, mobility, adherence, and biosynthesis of compounds (i.e., hemolysins, proteases, glycoside hydrolases) [
23]. These genes and resultant proteins require investigation to understand the pathogenesis of tenacibaculosis. In addition, complete genome comparisons between
Tenacibaculum species would reveal if these same genes and protein targets can be used to develop criteria to identify which isolates may induce disease.
Even though other experimental trials have exposed fish to
Tenacibaculum isolates [
4,
6,
9], the described model is important because it uses isolates collected from BC waters and can provide important contrasts to other isolates inside [
6] and outside of Canada [
4,
9]. The current research also demonstrated that several
Tenacibaculum species can induce disease with the tested methodology.
4. Materials and Methods
4.1. Fish Husbandry
Atlantic salmon smolts (N = 400, 30 g) were raised in partial saltwater (11‰) by Grieg Seafood BC LTD. Prior to transport to the Centre for Innovation in Fish Health (CIFH), at Vancouver Island University in Nanaimo (BC), the salmon population was screened for and tested negative for infectious salmon anemia virus, infectious pancreatic necrosis virus, and piscine orthoreovirus. Water from the recirculating aquaculture systems (RAS) systems at the CIFH, were tested using qPCR for
Tmar,
Tdic, and
Tfinn prior to the introduction of fish following previously developed protocols [
7].
Following arrival at CIFH, fish were transferred from 11‰ saltwater to 30‰ saltwater in a RAS (~1200 L) containing five circular tanks (Tank [T] 1-T5; 100 L each, n = 40 per tank) in parallel, in two identical, but separate temperature- and humidity-controlled rooms (Room 1, 2) for 24 h. Water quality (temperature [12 °C], dissolved oxygen [~105% sat], pH [~8.0], oxidative-reductive potential [~350 mV], salinity [~30‰], flow rate [gal per hour]) was measured every 10 min using a real-time water quality monitoring system (Neptune Apex Controller System, Morgan Hill, CA, USA). Water from each tank was collected in the sump and treated by ultraviolet exposure (7–12 mW*cm−2) to reduce bacterial contamination before going back to the tanks. Fish were fed pellets (Nutra 1.5 mm, Skretting Vancouver, Vancouver, BC, Canada) twice daily (0.5% average body-mass).
Humane endpoints were used and fish exhibiting clinical signs of illness (excessive ulcerative lesions, increased respiratory rates, loss of equilibrium, exophthalmia) at any point were euthanized using an overdose of tricaine methanesulfonate (Aqualife, 250 mg L−1, DIN 02168510, Syndel Canada, Nanaimo, BC, Canada) and counted as mortalities. To facilitate this, a high frequency of observation was used: pre-exposure, fish were checked three times a d; post-exposure, fish were checked six times a d for seven d, and four checks each d thereafter.
4.2. Bacterial Propagation
Tmar 2.1C, Tdic 20-4116-9, and Tfinn 20-4106-2 isolates used in this study were collected during 2019–2020 from Atlantic salmon during mouthrot outbreaks in BC waters. The three isolates were sequenced at the University of Alberta for the 16S rDNA sequence using the universal 27F and 1492R primers. Sequencing indicated that the three bacteria were most similar to T. maritimum NLF-15, T. dicentrarchi TdChD04, and T. finnmarkense Tsp.2; all three bacteria were compared using approximately 1400 bp and had percent identities above 98.6%. The three different Tenacibaculum species were used in separate exposures to compare mortality rates, clinical signs, and pathology.
Isolates (Tmar, Tdic, and Tfinn [preserved in 1.4 ml aliquots (25% glycerol) at −80 °C]) were cultured on Flexibacter maritimus medium (FMM) supplemented with kanamycin (FMM+K) (50 µg mL−1) at 12 °C; downstream FMM+K broth cultures were mixed at 200 rpm. Standard curves were developed for each isolate using absorbances at 600 nm and colony-forming units (CFU) per mL of FMM+K broth (data not shown). For Tdic and Tfinn, 1 L of the medium, at an absorbance of 600 nm of 0.5 (2.0 × 109 CFU*mL−1), was mixed with seawater (80 L) to a final concentration of 2.5 × 107 CFU*mL−1. For Tmar, 100 mL of medium at an absorbance at 600 nm of 0.05 and 1.5 (2.0 × 109 CFU*mL−1; 2.0 × 1011 CFU*mL−1) was cultured and mixed with seawater (80 L) to a final concentration of 2.5 × 106 and 2.5 × 108 CFU*mL−1.
4.3. Direct and Cohabitation Bath Exposures
Three separate bath exposures on Atlantic salmon smolts were conducted simultaneously to determine clinical and pathogenic differences between
Tenacibaculum species. Room 1 was used for
Tdic and
Tfinn exposures while Room 2 was used
Tmar exposures (
Table 2). For bath exposures, each room contained three, separate, static, aerated, 120 L exposure-tanks (ET) filled with 80 L of saltwater. Two ETs (ET1 & ET2) were used for bacterial exposures while ET3 was used as control exposure and contained an equivalent amount of sterile medium to the other ETs. Forty fish were placed in each 120 L ET at a stocking density of 12 kg m
−3 and FMM broth (with and without bacteria) was added to obtain the final concentrations as described in
Table 2. All ETs were covered, and fish were left for 5 h with aeration. Dissolved oxygen was measured every 30 min during the exposure (data not shown). After 5 h, 40 fish from each bacterial ET tank were split evenly into two 100 L RAS tanks (tank 1 [T1] & tank 2 [T2] or tank 3 [T3] & tank 4 [T4]) and were grouped with 20 naïve non-exposed cohabitants identified by adipose fin clipping (
Table 2). The 40 fish from ET3 were put directly into RAS tank 5 [T5] with no cohabitant fish (
Table 2). The experiment continued until there were no mortalities for 5 consecutive d and lasted for 21 d total
4.4. Pre- and Post-Exposure Sampling
Three pre-exposure fish from each room were sampled after the 24 h acclimation, euthanized as described, grossly examined, and swabbed from the mouths, gills, and head-kidneys for bacterial culture on FMM+K. Bacterial plates were incubated at 12 °C for a minimum of 7 d and colony morphologies were recorded. From each fish, external (jaw and gills), and internal (spleen and head-kidney) tissues were collected into RNAlater (Ambion Inc., SIGMA: R0901-500ML, Austin, TX, USA), cooled overnight at 4 °C before being stored at −80 °C. Sagittal sections of the fish, such as head, body, and internal organs were placed into 10% neutral buffered formalin (NBF) for histopathology.
Post-exposure, deceased and or moribund fish were removed and processed similar to that of pre-exposure fish. At each time check, a maximum of three deceased or moribund fish from each tank was processed for downstream testing. For each fish, gross lesions were identified and recorded prior to necropsy examination and retrieval of tissues for downstream qPCR and histopathology. At the end of the trial, three fish from each tank (if remaining) were euthanized as described and processed similar to pre-exposure fish.
4.5. Quantitative PCR
DNA from collected tissues and bacterial isolates were extracted using the OMEGA E.Z.N.A Tissue DNA extraction kit (Omega Bio-tek, Inc., Norcross, GA, USA) according to the manufacturer’s recommendations, with the following modification: fish tissues were incubated at 56 °C in lysis buffer for 24 h before extraction.
Three qPCR assays specific to
Tmar [
24,
25],
Tdic, and
Tfinn [
7] were used on tissue samples and bacterial isolates. The procedure and thermal profiles designed previously [
7,
24] were used with minor modifications: the
Tmar probe (0.05 μM) the
Tmar primers (0.5 μM); and the
Tdic probe (0.125 μM), per reaction. Primers were obtained from SIGMA-ALDRICH (Canada) and probes were obtained from Eurofins Genomics (USA). For each reaction, each well received 100 ng of sample DNA. Similar to a previous study [
7], a cycle quotient of 35 was used as a cut-off to represent no bacteria.
Standard curves, using spiked muscle and kidney tissues (100 ng of salmon tissue DNA and 0.001–100 ng of bacterial DNA) were used for each assay to ensure proper amplification efficiency (data not shown). With no differences between the standard curves based on tissue type (data not shown), an average of both standard curves was applied. Three equations based on generated standard curves (one for the
Tmar specific assay, a second for the
Tdic specific assay, and a third for the
Tfinn specific assay) were used to change the qPCR cycle-quotient output to a theoretical number of bacteria per g of fish tissue (
Table 2). The equations stem from a previous study [
7]. For each equation, ‘x’ is the Log-number of bacteria (LNOB) per gram of fish tissue, ‘c’ is the concentration of the extracted sample and ‘v’ is the volume of the extracted sample (
Table 3).
4.6. Histopathology
Tissue processing, sectioning (7 µm) of paraffin-embedded tissue, and hematoxylin and eosin staining were performed by a commercial laboratory (Animal Health Laboratory, Abbotsford, BC, Canada). Each cassette contained a sagittal section of the head and any additional lesions to the body such as the tail.
4.7. Statistics
Statistical analysis was performed on R-Studio version 3.6.1 [
26]. Normal distributions and homogenous variances were tested using Q-Q plots, the Shapiro-Wilcoxon test, and the Levene test. Kaplan–Meier analyses were conducted using the R package ‘
survival’ for each tank population (exposed or cohabitant) using the daily mortality data. Kruskal–Wallis tests for the qPCR results, were conducted between: assays, cohabitants, and exposed fish, and internal and external tissues to subset the data for regression modeling and subsequent comparison by ANOVA.