Sage Journals: Discover world-class research

Abstract

German

Spanish

French

Examining zebrafish populations for the presence of disease is an integral component of managing fish health in research facilities. Currently, many different strategies are used for zebrafish fish health inspections, which is a scenario that may result in subjective and biased diagnostic evaluations. The goal of this study was to compare the success of pathogen detection between a sample size of randomly selected fish (n = 60) that provides 95% confidence in pathogen detection based on a presumed pathogen prevalence level ≥5%, and other subpopulations and sample numbers commonly submitted for diagnostic testing within a 1000 tank, 30,000 fish, recirculating research system. This included fish collected from a sump tank (n = 53), sentinel fish (n = 11), and fish that were found moribund or freshly dead (n = 18). Additionally, five fish from each subpopulation were collected for histopathologic examination. A second study used retrospective data to examine pathogen distribution between systems (n = 2−5) in multi-system facilities (n = 5) using a sample size of 60 fish per system. For the pathogens detected, results supported the use of representative sample numbers rather than smaller numbers of populations considered more at risk. The exception to this is for the moribund/mortality group, which may be a resource for targeted surveillance of select pathogens. Each system within multi-system facilities should be considered separate units in terms of fish health inspections and biosecurity. Development of these evidence-based standards for fish health inspections in zebrafish systems enhances fish welfare, provides identification of potentially zoonotic pathogens, and ensures scientific integrity and reproducibility of research results.

Keywords

Disease surveillance fish health inspection zebrafish sampling

Introduction

The relatively rapid growth of fish as model organisms in the laboratory exemplifies their value in medical and biological research. Most notable are zebrafish (Danio rerio), although medaka (Oryzias latipes),¹ killifish and mummichogs (Fundulus spp.),² goldfish (Carassius auratus),³ and sticklebacks (Gasterosteus spp.),⁴ among others, are commonly utilized. This emergence has led to major scientific discoveries but has also challenged our ability to keep pace and develop proper standards for husbandry of fish in research facilities. Fish health management is important not only from an animal welfare standpoint, but also for ensuring scientific integrity and reproducibility of research results. Inconsistent or improper husbandry of these research models can affect physiologic and immunologic responses that may in turn impart undefined variables into experiments. Thus, adequate time and resources should be put into standardizing and maintaining health within fish populations.

One essential component of managing fish health is monitoring populations for the presence of infections and non-infectious disease. Fish within research systems are at risk for disease, as aquatic conditions favor the growth of disease-causing organisms due to the natural biofilms that harbor pathogens, relatively high fish densities that promote transmission of disease, and elevated fish stress secondary to handling and manipulation.⁵ In research fish, there is a growing list of infectious and non-infectious diseases described in laboratory settings, particularly for zebrafish. This includes various Gram-negative bacteria likely representing both primary and opportunistic infections, including Edwardsiella ictaluri, which has been associated with septicemia in previously healthy fish,⁶ and Aeromonas spp., which have been linked to skin lesions and subsequent septicemia, likely secondary to water quality issues and increased fish stress.⁷ At least six species of mycobacteria have been isolated from zebrafish with varying degrees of chronic morbidity and mortality.⁸ Parasitic diseases diagnosed in research facilities include enteritis associated with the nematode Pseudocapillaria tomentosa,⁹ skin and gill infections by the dinoflagellate Oodinium pillularis,¹⁰ and internal infections caused by the mxyozoan Myxidium streisingeri¹¹ and the microsporidians Pseudoloma neurophilia and Pleistophora hyphessobryconis.¹² Non-infectious diseases observed in zebrafish include neoplastic lesions¹³ and egg-associated inflammation often in older zebrafish.¹⁴

To date, there are very few guidelines for monitoring pathogens and diseases in zebrafish facilities. Current methods for sampling and testing fish often vary between facilities and lack evidence-based strategies. There is wide variation in the number of fish sampled and the population of fish submitted for testing. Fish number is often arbitrarily determined or centered on disease testing costs. Although practical considerations will always be a part of these decisions, the effect that these biases may have on testing results must also be noted. Based on the Office International des Epizooties (OIE) Aquatic Animal Health Surveillance¹⁵ guidelines and American Fisheries Society¹⁶ guidelines, fish health inspections should use a sample size that provides 95% confidence in pathogen detection using approved diagnostic tests. For practicality, an assumed pathogen prevalence (APPL) of ≤5% is often used. By those standards, to test a population of 2000 fish or more, 60 fish would need to be submitted for testing to provide 95% confidence in pathogen detection at a 5% APPL or 120 fish for a 2% APPL. This does not negate a value in sampling fewer fish. However, negative results need to be interpreted cautiously because of the lower confidence level in pathogen detection. These standards were largely developed for food fish and wild fisheries, and to date, application of these sample sizes have not been examined in zebrafish facilities, which represents a different dynamic in terms of fish husbandry, genetics, and diseases.

Zebrafish sampled for disease testing often come from different subpopulations within the system. This includes randomly sampled fish from the general population, which can provide a wide sample distribution, depending on the number of fish selected. Additionally, facilities may choose to utilize “sentinel” fish housed within isolated tanks or the sump tank that are exposed to effluent water.¹⁷ These programs are designed to evaluate the efficacy of ultraviolet sterilization or as a strategy to monitor fish health within the system. The assumption is that increased exposure to organisms shed from fish in the general population makes sentinel fish more likely to harbor pathogens. More rarely, facilities may utilize moribund or freshly dead fish for risk-based surveillance. There is evidence to suggest that this distribution increases the odds of identifying select pathogens within commercial aquaculture systems.¹⁸ However, this methodology and the effect of subpopulation on success of pathogen prevalence have not been comprehensively explored in zebrafish facilities.

Fish health inspections outlined in this study utilized a combination of targeted and broad-based diagnostic techniques. Targeted diagnostic techniques provided sensitive and pathogen-specific inspection for previously described zebrafish pathogens, such as Mycobacteria spp. and P. neurophilia, using polymerase chain reaction (PCR) assays. Broad-based methods used strategies to ensure breadth in diagnostic scope with the awareness that the list of known etiologies of zebrafish is still growing. This included bacterial culture, virus isolation, histopathology, and wet-mount cytology.

The objective of this study was to compare diagnostic results between commonly used sampling strategies. A sample size of 60 fish was used as the standard that would allow 95% confidence in pathogen detection based on a presumed pathogen prevalence of 5%. Gross signs of disease and pathogen identification were compared to a relatively small number of sentinel fish, to a larger group of fish collected from the sump tank, and from fish observed as moribund or freshly dead. Additionally, diagnostic results were retrospectively analyzed within facilities that contained multiple independent recirculating systems to examine the boundaries of biosecurity and to guide testing protocols within these potentially more complex establishments. Our goal was to help define the advantages and limitations of targeted and broad-based diagnostic methods currently employed in zebrafish facilities. Examining these facets will help define best practices that provide confidence in testing results without unnecessary economic expenses and fish sacrifice.

Methods

Ethical statement

The use of these animals in these trials was approved by the Institutional Animal Care and Use Committee at Boston Children’s Hospital (IACUC Protocol# 17-08-3511R).

Effect of subpopulation and fish number on success of pathogen detection

Fish health inspection was performed on a recirculating system consisting of 1000 tanks and approximately 30,000 adult zebrafish of mixed genetic lines. A complete description of the husbandry and environmental conditions in housing for the fish used in this study is available as a collection at dx.doi.org/10.17504/protocols.io.mrjc54n. All equipment that comes into contact with research animals or water are cleaned and disinfected prior to and after each use. Disinfection is accomplished by rinsing at a temperature of ≥82.2℃ for a minimum of three minutes, and when objects have been exposed to animals or water continuously for periods in excess of 24 hours, they are disinfected by using a surfactant-free disinfecting solution followed by ≥82.2℃ deionized water or fresh water rinse to remove chemical residues.

Tanks are cleaned/disinfected on average of once every 16 weeks or as needed, and floors are cleaned and disinfected on a weekly basis using Virkon S following manufacturers’ protocols.

Rotifer and Artemia culturing vessels are disinfected with 50% Wescodyne solution with a contact time of five minutes or more between each use.

Fish within the system were divided into four experimental subpopulations. (a) The randomly sampled group consisted of 65 fish sampled by technicians by hand-netting one fish out of 65 different tanks. Based on OIE and AFS Standards, 60 fish sampled from a system of 30,000 fish represents a sample number that provides 95% confidence in pathogen detection, presuming ≥5% pathogen prevalence.^15,16 The additional five fish sampled from this group were submitted for histopathology. (b) The sentinel fish group consisted of 16 fish sampled from a single sentinel tank continuously exposed to effluent water for approximately six months. (c) The sump tank group consisted of 58 fish collected from the system sump tank. These fish represented various ages and effluent water exposure time. (d) The morbidity/mortalities group consisted of seven fish found freshly dead with minimal autolysis and 16 fish interpreted as moribund due to either reddened skin, emaciation, anorexia, lethargy, or an inability to maintain constant buoyancy in the water column.

Targeted and broad-based diagnostic methods

Following collection, fish were euthanized with an anesthetic overdose of tricaine methanesulfonate (MS-222; Sigma–Aldrich, St. Louis, MO) at 200 mg/L.¹⁹ Five fish from each group were sampled for histopathology. Fish were fixed whole in 10% neutral-buffered formalin for 72 hours and serially cross-sectioned. Tissues were processed routinely, and 5 µm sections were stained with hematoxylin and eosin and acid-fast stain. Digital slides were produced with a Leica SC2 scanner (FishVet Group, Portland, ME), and tissues were examined using ImageScope software (Leica Biosystems, Richmond, IL) by a U.S. residency and graduate program trained aquatic veterinary pathologist.

Necropsies and diagnostic testing were performed immediately on remaining fish with a combination of targeted and broad-based diagnostic methodologies (Table 1). Skin scrape, gill clip, and intestinal wet mounts were taken from each fish to look for parasites and filamentous bacteria. The surface of each fish and necropsy tools were sprayed with 70% ethanol and passed through a flame prior to sampling of internal tissues. Bacterial swabs were then performed on the cranial kidney and plated on tryptic soy agar (TSA) and blood agar and incubated at 28℃ for 72 hours. When positive growth was observed, individual colonies based on morphologic appearance were re-streaked onto TSA or blood agar and incubated at 28℃ for 48 hours and a Gram stain was performed. Single colonies from pure cultures were then incubated at 28℃ for 24 hours in tryptic soy broth and DNA isolation (DNEasy Kits, Qiagen, Valencia, CA), and 16 S rRNA PCR sequencing²⁰ was performed. Resultant sequences were edited using Chromas software (http://technelysium.com.au) and searched against the BLASTn database for identification.

Table 1.

Diagnostic assays used to perform pathogen surveillance and screening on sampled fish.

	Assay/tissue
Bacteriology
General aerobic bacteria (ID by 16S rRNA sequencing)	Culture on TSA and blood agar/kidney
Mycobacteria spp.	PCR/pooled tissue^21–24
Edwardsiella ictaluri	PCR/pooled tissue²⁵
Flavobacterium columnare	PCR/pooled tissue²⁵
Aeromonas hydrophila	PCR/pooled tissue²⁵
Virology
Cell culture	Pooled tissue on FHM and BF-2 cell lines
Parasitology
External parasites (Costia, Trichodina, Piscinoodinium, etc.)	Microscopy/gill and skin wet mounts
Ichthyophthirius multifilis	PCR/pooled tissue²⁶
Internal parasites (gastrointestinal parasites)	Microscopy/caudal intestine
Pseudocapillaria tomentosa	Microscopy and PCR/caudal intestine²⁷
Pseudoloma neurophilia	PCR/pooled tissue²⁸
Histopathology	Formalin-fixed whole fish

Tissues used for PCR and viral cell culture consisted of brain, gill, heart, liver, spleen, and kidney pooled from five fish per sample.

TSA: tryptic soy agar; PCR: polymerase chain reaction; FHM: fathead minnow.

Virology and pathogen-specific PCR was performed on homogenates of the brain, gill, heart, liver, spleen, and kidney pooled from five fish per sample. Homogenate (100 mg) was used for virus isolation on epithelioma papulosum cyprini and fathead minnow cell lines incubated at 30℃ for 28 days with a blind pass at 14 days. Tissue homogenate (25 mg) was used for PCR for targeted disease surveillance of select pathogens of zebrafish outlined in Table 1 following published protocols. Each assay was verified for adequate sensitivity and run with appropriate positive and negative controls. All positive PCR results for P. tomentosa were confirmed by microscopy.

Effect of system on diagnostic results in multi-system facilities

A retrospective analysis was performed on data collected from fish health inspections in five facilities containing two or more systems with 1000–25,000 fish per system (Table 5). Sixty-five fish were randomly sampled from each system, and diagnostic testing was performed as described above.

Statistical analysis

A univariate odds ratio was used to compare the likelihood of detecting each observed disease sign and pathogen within the experimental subpopulations compared to the randomly sampled group. The effect of sample size on success of pathogen detection was calculated using epitools (http://epitools.ausvet.com.au/). The pathogen prevalence needed to provide a 95% confidence interval based on sample size of each group was determined, as previously described by Murray et al.,²⁹ with an assumed assay sensitivity of 90–95%.

For the retrospective analysis exploring pathogen detection between systems in multi-system facilities, a proportions test (prop.test function in R; The R Foundation, Vienna, Austria) was used to explore pathogen distribution and determine whether the proportions were equal. Additionally, a simple linear regression analysis (lm() function in R) was used to determine the relationship between number of fish per system and total number of pathogens detected in a system to asses if there is an increased risk of pathogen presence in systems with more fish. All statistics were run with a significance level of p < 0.05.

Results

Association of subpopulation and disease signs

Gross signs of disease observed during necropsy included emaciation, spinal curvature, skin reddening, splenomegaly, ascites/dropsy, and gill epithelial hyperplasia (Table 2 and Figure 1(a) and (b)). The incidence of disease signs was significantly different among the various subpopulations. Emaciation and skin reddening were 4.2 times and 28.6 times, respectively, more likely to be seen in the morbidity/mortality group compared to the randomly sampled group. Splenomegaly was 23.8 times more likely to occur in the morbidity/mortality group and 3.9 times more likely to occur in the sump fish group compared to the randomly sampled group. Gill hyperplasia was 14.9 times more likely to occur in the sump group and 10.9 times more likely to occur in the sentinel group than the randomly sampled group. Spinal curvature, ascites, and dropsy were not significantly associated with any one experimental group.

Figure 1.

Gross and microscopic findings in zebrafish. (a) Zebrafish demonstrating generalized dropsy and reddening of the skin. (b) Splenomegally (arrow). (c) Multifocal, variably sized granulomas (arrowheads) throughout the kidney. 10× magnification, bar = 200 µm. (d) Intraspinal pseudocyst filled with microsporidia consistent with Pseudoloma neurophilia. 40× magnification, bar = 50 µm.

Table 2.

Incidence of disease signs observed on gross necropsy for each zebrafish subpopulation.

Group	Emaciation	Spinal curvature	Skin reddening	Enlarged spleen	Ascites/ dropsy	Gill hyperplasia
Randomly sampled (n = 60)	5/60 (8%)	2/60 (3%)	5/60 (8%)	3/60 (5%)	2/60 (3%)	2/60 (3%)
Sentinel fish (n = 11)	1/11 (9%)	0/11 (0%)	0/11 (0%)	1/11 (9%)	0/11 (0%)	2/11 (18%)
Sump fish (n = 53)	3/53 (6%)	2/53 (4%)	2/53 (4%)	9/53 (17%)	2/53 (4%)	17/53 (32%)
Moribund/freshly dead (n = 18)	5/18 (27%)	0/11 (0%)	13/18 (72%)	10/18 (55%)	2/18 (18%)	4/18 (22%)

Numbers in bold represent a statistically significant increase compared to the randomly sampled group (p ≤ 0.05).

Association between success of pathogen detection, subpopulation, and sample size

Targeted and broad-based diagnostic methods identified the presence of multiple pathogens within the system (Table 3). This includes isolation of Aeromonas veronii and an Aeromonas spp. that could not be identified to species from kidney cultures. PCR assays positively identified Mycobacteria haemophilum, Mycobacteria marinum, and Mycobacterium fortuitum, and P. neurophilia. There was no evidence to support the presence of other bacteria, viruses, or external or internal parasites via culture, microscopy, or molecular testing.

Table 3.

Organisms detected in each subpopulation of zebrafish within the same system.

Group	Calculated pathogen prevalence needed for detection based on sample size	Bacterial culture	Mycobacteria PCR	P. neurophilia PCR
Randomly sampled (n = 60)	≥5.5%	Aeromonas veronii 3/60 (5%) Aeromonas spp. 2/60 (2%)	Mycobacteria haemophilum 11/12 (92%) Mycobacteria marinum 4/12 (33%) Mycobacterium fortuitum 1/12 (8%)	8/12 (67%)
Sentinel fish (n = 11)	≥26.5%	Aeromonas spp. 1/11 (9%)	0	2/2 (100%)
Sump fish (n = 53)	≥6.1%	A. veronii 5/53 (9%) Aeromonas spp. 4/53 (8%)	M. haemophilum 11/11 (100%) M. marinum 1/11 (9%) M. fortuitum 2/11 (18%)	11/11 (100%)
Moribund/freshly dead (n = 18)	≥18.5%	A. veronii 7/18 (44%) Aeromonas spp. 1/18(6%)	M. haemophilum 4/4 (100%) M. marinum 2/4 (50%)	3/4 (75%)

PCR assays were performed on the gill, brain, liver, spleen, and kidney pooled from five fish per sample. Text in bold represents a statistically significant increase in pathogen detection compared to the randomly sampled group (p ≤ 0.05).

The greater quantity of pathogens was detected in the groups with the larger sample sizes—specifically, the 60 randomly sampled fish and the 53 fish from the sump tank (Table 3). Based on a sample size of 60 fish from the randomly sampled group and an assumed assay sensitivity of 95% confidence, this sample size was calculated to be sufficient for detection of pathogens with ≥5.54% prevalence. Within the 53 zebrafish from the sump tank, pathogens with ≥6.1% prevalence should be detected. There were no significant differences between the disease results of these groups. Representative results of disease testing within the randomly sampled group are broken down in groups of 10 fish in Table 4 to illustrate the association between sample number and pathogen detection better. Specifically, pathogens with relatively low prevalence such as Aeromonad bacteria were not detected within the first 20 fish cultured, M. fortuitum was not detected until fish numbers 31–40, and P. neurophilia was not detected until fish numbers 11–20. This pattern was similar to that found in the sump fish group.

Table 4.

Organisms detected within the randomly sampled subpopulation (n = 60) broken into sequential increments of 10 fish to examine the effect of sample number on detection success.

Randomly sampled fish	Bacterial culture	Mycobacteria spp. PCR	P. neurophilia PCR
Fish # 1–10	0	M. haemophilum 1/2 (50%) M. marinum 1/2 (50%)	0/2 (0%)
Fish # 11–20	0	M. haemophilum 2/2 (100%) M. marinum 1/2 (50%)	2/2 (100%)
Fish # 21–30	A. veronii 1/10 (10%) Aeromonas spp. 1/10 (10%)	M. haemophilum 2/2 (100%)	1/2 (50%)
Fish # 31–40	A. veronii 1/10 (10%)	M. haemophilum 2/2 (100%) M. fortuitum 1/2 (50%)	1/2 (50%)
Fish # 41–50	A. veronii 1/10 (10%) Aeromonas spp. 1/10 (10%)	M. haemophilum 2/2 (100%) M. marinum 1/2 (50%)	2/2 (100%)
Fish # 51–60	0	M. haemophilum 2/2 (100%) M. marinum 1/2 (50%)	1/2 (50%)

Within the subpopulations with a relatively small sample size, sampling 11 zebrafish within the sentinel group should provide detection of pathogens that had ≥26.5% prevalence. In this group, only Aeromomas spp. and P. neurophilia were detected. There was lack of detection of the three Mycobacterium spp. or for Aeromomas veronii. Within the moribund/mortality group, a sample size of 18 fish should provide detection for pathogens with ≥18.5% pathogen prevalence. Both Aeromonad isolates, M. haemophilum, M. marinum, and P. neurophilia were identified, while M. fortuitum was not found. Relative prevalence of A. veronii was higher in the morbidity/mortality group (5/13 moribund fish and 2/5 dead fish; total of 7/18 fish, 44%), and there was a 12.1 times greater risk of infection for A. veronii in the morbidity/mortality group compared to the randomly sampled group.

Histopathology

Histopathologic findings outlined in Table 5 included multi-organ granulomas found in 7/20 (28%) fish (Figure 1(c)). These were found in at least one fish from each group except for the sentinel group. Granulomas were associated with intracellular acid-fast organisms consistent with mycobacteria in 3/7 (43%) fish. Also present on histopathology was a granulomatous (n = 1) and eosinophilic granular cell peritonitis (n = 1) not overtly associated with any observed pathogens and branchial hyperplasia (n = 1) associated with water mold. Intraspinal pseudocysts containing microsporidia consistent with P. neurophilia were present in one fish in each of the randomly sampled group and the sentinel group (Figure 1(d)). There were no observations of microspores in the sump fish group or the moribund/mortality group. Mild, multifocal, renal protein deposits were present in 6/20 (30%) fish and were represented in all groups.

Effect of system and fish number on pathogen prevalence in multi-system facilities

Diagnostic results for the five multi-system facilities are outlined in Table 6. In summary, 13 different bacteria were identified. This included six distinct Gram-negative bacteria purportedly identified as Vibrio spp., Aeromonas spp., Aeromonas hydrophila, A. veronii, Pleisomonas shigelloides, and a Pseudomonas spp. PCR testing identified Mycobacterium abscessus, Mycobacterium chelonae, M. fortuitum, M. haemophilum, P. tomentosa, and P. neurophilia within at least one system. Of the 25 instances where a pathogen was identified, the pathogen was absent from at least one system in the facility on 10/25 (60%) occasions. For the remaining 15/25 (40%) occurrences, there was one instance when pathogen proportions were significantly different between systems within the same facility. This occurred with P. tomentosa in facility 5 for systems 2 and 5 (0.22, 0.05; p = 0.03). There was poor correlation between overall pathogen presence and fish number (p = 0.6798, R²= 0.01352), suggesting that systems with larger numbers of fish do not appear to be at increased risk of pathogen presence compared to systems with fewer fish numbers.

Table 5.

Histopathologic findings in each zebrafish subpopulation.

Group	Histopathologic findings
Random sampled (n = 5)	Multifocal hepatic granulomas with intralesional acid-fast bacteria (n = 1) Mild renal protein deposits (n = 1) Intraspinal pseudocyst containing microsporidia (n = 1)
Sentinel fish (n = 5)	Mild renal protein deposits (n = 1) Intraspinal pseudocyst containing microsporidia (n = 1)
Sump fish (n = 5)	Multifocal renal and hepatic granulomas (n = 1) Moderate branchial hyperplasia with intralesional water mold (n = 1) Mild eosinophilic granular cell peritonitis (n = 1) Mild renal protein deposits (n = 2)
Moribund/freshly dead (n = 5)	Severe systemic granulomas with intralesional acid-fast bacteria (n = 5) Granulomatous peritonitis (n = 4) Mild renal protein deposits (n = 2)

Table 6.

Organisms isolated from zebrafish sampled from multiple independent recirculating systems within the same facility.

	Gram-negative bacteria						Mycobacteria					Parasites
	Vibrio spp.	Aeromonas spp.	A. hydrophila	A. veronii	Plesiomonas shigelloides	Pseudomonas spp.	Mycobacterium spp. generic	Mycobacterium abscessus	Mycobacterium chelonae	M. fortuitum	M. haemophilum	P. tomentosa	P. neurophilia
Facility 1
System A, n = 13,500	+	–	–	–	–	–	–	–	–	–	–	–	–
System B, n = 13,500	+	+	–	–	–	–	–	–	–	–	–	–	+
Facility 2
System A, n = 1000	–	–	–	–	–	–	+	–	+	–	–	–	+
System B, n = 21,000	–	–	+	–	+	–	+	–	+	–	–	–	+
Facility 3
System A, n = 14,000	–	–	–	–	+	–	+	–	–	–	–	+	–
System B, n = 14,000	–	–	–	–	+	+	+	–	–	–	+	+	–
System C, n = 14,000	–	–	–	+	+	–	+	–	–	–	–	+	–
Facility 4
System A, n = 2000	–	–	–	–	–	+	–	–	+	+	–	+	+
System B, n = 25,000	–	–	–	–	–	+	–	+	–	+	–	–	+
System C, n = 25,000	–	–	–	–	–	–	–	–	–	+	–	–	+
Facility 5
System A, n = 25,000	–	+	–	–	+	+	–	–	–	–	–	–	–
System B, n = 20,000	–	+	–	–	–	–	–	+	–	–	–	–	+
System C, n = 10,000	–	+	–	–	–	–	–	–	–	–	–	–	+
System D, n = 10,000	–	+	–	–	+	–	–	+	–	–	–	–	+
System E, n = 10,000	–	+	–	–	–	–	–	–	–	–	–	–	+

Sample size was 60 fish/system.

Discussion

The goal of this study was to evaluate how the success of pathogen detection varies between diagnostic sampling strategies within zebrafish research systems. We used a combination of targeted and broad-based diagnostic methodologies on different subpopulations of fish using sample numbers that are commonly submitted for diagnostic testing. Results demonstrated that within this system and for the specific pathogens isolated, there are significant differences in disease signs and success of pathogen detection, depending on the sampling strategy. Results suggest that for most pathogens, the number of fish sampled may be more important to diagnostic success than the subpopulation. The exception to this was for an Aeromonas spp., where increased risk of infection in the moribund/mortality group suggests more targeted surveillance for the Aeromonas spp. can be achieved by sampling fish that are moribund or freshly dead. This may be a result of the role of Aeromonads in morbidity and mortality as either an opportunistic or primary pathogen.³⁰

The relatively higher detection rate of M. haemophilum and, to a lesser extent, M. marinum was consistent with previous studies that describe high prevalence levels for these species within zebrafish systems.³¹ M. haemophilum and M. marinum have relatively high pathogenicity in zebrafish compared to other mycobacterium species,^8,32 which may also result in increased transmission and persistence within the zebrafish population. In theory, this would result in a higher detection rate, as demonstrated in this study. It is unknown why positive identification of mycobacteria was lacking in fish in the sentinel group. Based on the high prevalence rates of M. haemophilum and M. marinum in other groups, it would have been assumed that prevalence was higher than the 26.5% threshold that statistically limited the sample size of 11 fish.

Compared to M. haemophilum and M. marinum, M. fortuitum is believed to be relatively less virulent and may act as more of an opportunistic pathogen.³¹ The lower pathogenicity of M. fortuitum may result in lower prevalence within the system, thus making detection more challenging compared to more pathogenic mycobacterial species. This was demonstrated by a lack of positive identification of M. fortuitum within the sentinel and moribund/mortality groups or within the first 30 fish tested in the randomly sampled group. This highlights the importance of sample size in adequately characterizing less prevalent pathogens in zebrafish systems.

Granulomatous inflammation and acid-fast bacteria consistent with mycobacterial infection were seen on histopathology in each subpopulation of fish except fish in the sentinel group, which is consistent with PCR results. Histopathology provided evidence of manifested mycobacteriosis and lesion development, but alone, it did not provide a diagnosis of which mycobacterial species were present in the fish. Molecular testing for mycobacteria provided speciation of pathogens present but only inference of disease risks and impact on the health of the fish. However, together, molecular testing and histopathology were valuable in fully characterizing the scope and impact of infection within the system. We suggest that histopathology be used as a tool to confirm and characterize disease due to infectious agents and as a means for screening for non-infectious disease, but not as a method for targeted pathogen surveillance. The relative low sensitivity of histology and relatively higher costs compared to being able to pool five fish per sample for PCR renders it a less optimal strategy.

The chronic pathogenicity³³ and widespread prevalence²⁹ of P. neurophilia provided a largely uniform diagnosis across all subpopulations using PCR and five fish/sample pools. Positive identification of the microsporidium was lower using histopathology, with negative results for the sump group and morbidity/mortality group. This was not unexpected, as cross-sectioning the fish limits the coverage of the spinal cord, although in the authors’ opinion, it is the preferred method for ensuring adequate visualization of other major organs.

There were six distinct Gram-negative bacteria isolated from kidney swabs within the subpopulation study and the retrospective analysis of data from multi-system facilities. Bacterial culture cannot differentiate between colonization and true tissue invasion, and postmortem interval may have affected bacteriology results for dead fish within the moribund/mortality group. With the exception of highly virulent E. ictaluri,⁶ the prevalence and significance of Gram-negative bacteria in zebrafish systems are largely unknown. Based on limited reports in zebrafish and what has been described in fish in more traditional aquaculture settings, A. hydrophila, A. veronii, P. shigelloides, and Pseudmonas spp. are believed to reflect the microflora of the aquatic habitat but with roles as primary and opportunistic pathogens.^7,34–38 A statistically increased detection rate of Aeromonas spp. in the moribund/mortality group, including isolation from 38% of tested moribund fish, suggests its role as a pathogen. Adequate interpretation of these findings requires further research to provide genomic comparisons to previously described fish pathogens and to fulfill Koch’s postulates through challenge studies to describe pathogenicity. Isolates deemed to pose substantial risks to zebrafish health could then be prioritized for development of more sensitive methodologies for targeted surveillance.

Fish health inspections may also be useful for identifying potentially zoonotic pathogens present in the system. Mycobacteria and Gram-negative bacteria isolated in this study, specifically, M. marinum, M. fortuitum and M. chelonae, as well as Vibrio, Aeromonad, Pseudomonad, and Plesiomonas species, have been linked to infections in humans through exposure to fish or freshwater.^39–45 Identification of potentially zoonotic pathogens provides risk assessment data for technical and research personnel who work within the systems, and it also promotes the use of personal protection equipment.

Within multi-system facilities, the pathogen landscape was different between systems more often than it was the same. When pathogens were present within the same facility, there were relatively few differences in disease detection rates, with the exception of P. tomentosa in one system. This trend provides evidence that each system within a facility should be considered unique with regards to disease testing and biosecurity. There is limited information identifying biosecurity risks within zebrafish facilities, but based on fundamental principles of aquaculture, fish movement, equipment disinfection, handling of sick and dead animals, and reducing fish stress⁴⁶ all likely play major roles in the spread and impact of infectious disease on zebrafish recirculating systems. Results from this study did not find that there was an association between the number of fish in a system and the subsequent number of pathogens, although larger systems may pose different biosecurity challenges compared to smaller systems. However, a more comprehensive and multi-faceted study is needed to characterize risk factors such as tank number, fish density, movement of fish, and so on that may affect infectious disease dynamics in zebrafish systems.

Through studies that determine best practices, we can begin to utilize resources to support fish health more efficiently. In the absence of established guidelines, the evaluation of each health program is subjective, which makes it difficult to share data and support movement of fish between facilities. The first step to developing these guidelines is to understand the disease landscape to help prioritize pathogens for elimination and exclusion. Based on results from this study, we suggest that this be performed for each system within multi-system facilities and that OIE standards for fish health inspections be used that allow >95% confidence in pathogen detection with a presumed pathogen prevalence of 5%. The use of small numbers of fish, even if sentinel, moribund, or freshly dead, were not sufficient to characterize system-wide fish health, especially when pathogen prevalence was relatively low. We also suggest that histopathology be utilized to confirm infectious disease and characterize its effect on fish health, in addition to screening fish for non-infectious causes of disease.

Expansion of this research to include additional pathogens and replicate systems will further elucidate how best to utilize resources for diagnosing infectious disease in zebrafish systems and set standards for exploring these dynamics in other laboratory models of fish. Our hope is that this will lead to more standardized and efficient guidelines for disease testing.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

References

Kirchmaier

Naruse

Wittbrodt

, et al. The genomic and genetic toolbox of the teleost medaka (Oryzias latipes). Genetics 2015; 199: 905–918.

Burnett

Bain

Baldwin

, et al. Fundulus as the premier teleost model in environmental biology: opportunities for new insights using genomics. Comp Biochem Physiol D Genom Proteom 2007; 2: 257–286.

Ota

Abe

. Goldfish morphology as a model for evolutionary developmental biology. Wiley Interdiscip Rev Dev Biol 2016; 5: 272–295.

Peichel

Sullivan

Liachko

, et al. Improvement of the threespine stickleback genome using a Hi-C-based proximity-guided assembly. J Hered 2017; 108: 693–700.

Hall-Stoodley

Costerton

Stoodley

. Bacterial biofilms: from the natural environment to infectious diseases. Nat Rev Microbiol 2004; 2: 95–108.

Hawke

Kent

Rogge

, et al. Edwardsiellosis caused by Edwardsiella ictaluri in laboratory populations of zebrafish Danio rerio. J Aquat Anim Health 2013; 25: 171–183.

Pullium

Dillehay

Webb

. High mortality in zebrafish (Danio rerio). Contemp Top Lab Anim Sci 1999; 38: 80–83.

Kent

Matthews

Bishop-Stewart

, et al. Mycobacteriosis in zebrafish (Danio rerio) research facilities. Comp Biochem Physiol C Toxicol Pharmacol 2004; 138: 383–390.

Kent

Bishop-Stewart

Matthews

, et al. Pseudocapillaria tomentosa, a nematode pathogen, and associated neoplasms of zebrafish (Danio rerio) kept in research colonies. Comp Med 2002; 52: 354–358.

10.

Westerfield

. The zebrafish book: a guide for the laboratory use of zebrafish (Danio rerio), 4th ed. Eugene: The University of Oregon Press, 2000.

11.

Whipps

Kent

Murray

. Occurrence of a myxozoan parasite Myxidium streisingeri n. sp. in laboratory zebrafish Danio rerio. J Parasitol 2015; 101: 86–90.

12.

Sanders

Watral

Kent

. Microsporidiosis in zebrafish research facilities. ILAR J 2012; 53: 106–113.

13.

Spitsbergen

Buhler

Peterson

. Neoplasia and neoplasm associated lesions in laboratory colonies of zebrafish emphasizing key influences of diet and aquaculture system design. ILAR J 2014; 53: 114–125.

14.

Kent ML, Spitsbergen JM, Matthews JM, et al. Non-infectious and idiopathic diseases. ZIRC Health Services Zebrafish Disease Manual, https://zebrafish.org/wiki/health/disease_manual/start (2012, accessed 5 June 2018).

15.

World Organisation for Animal Health. Aquatic animal health surveillance. In: Aquatic animal health code. 13th ed. Paris: World Organisation for Animal Health, 2010, pp.9–43.

16.

USFWS and AFS-FHS (U.S. Fish and Wildlife Service and American Fisheries Society-Fish Health Section). Section 2.2, Procedures for the detection and identification of certain finfish and shellfish pathogens, http://afs-fhs.org/bluebook/bluebook-index.php (2016, accessed 5 June 2018).

17.

Lawrence

. Advances in zebrafish husbandry and management. Methods Cell Biol 2011; 104: 429–451.

18.

Oidtmann

Peeler

Lyngstad

, et al. Risk-based methods for fish and terrestrial animal disease surveillance. Prev Vet Med 2013; 112: 13–26.

19.

AVMA Panel on Euthanasia. Guidelines for the euthanasia of animals, Schaumburg, IL: American Veterinary Medical Association, 2013, pp. 32–32.

20.

Loch

Fujimoto

Woodiga

, et al. Diversity of fish associated Flavobacteria of Michigan. J Aquat Anim Health 2013; 25: 149–164.

21.

Pourahmad

Nemati

Richards

. Comparison of three methods for detection of Mycobacterium marinum in goldfish (Carassius auratus). Aquaculture 2014; 422–423: 42–46.

22.

Kee

Kim

, et al. Multiplex PCR assay for identification of mycobacterial species located from liquid cultures. Chonnam Med J 2009; 45: 19–26.

23.

Whipps

Dougan

Ken

. Mycobacterium haemophilum infections of zebrafish (Danio rerio) in research facilities. FEMS Microbiol Lett 2007; 270: 21–26.

24.

Kim

Cha

. Multiplex-real time PCR assay and melting curve analysis for identifying Mycobacterium tuberculosis complex and nontuberculous mycobacteria. J Clin Microbiol 2012; 50: 483–487.

25.

Panangala

Shoemaker

Van Santen

, et al. Multiplex-PCR for simultaneous detection of 3 bacterial fish pathogens, Flavobacterium columnare, Edwardsiella ictaluri, and Aeromonas hydrophila. Dis Aquat Organ 2007; 74: 199–208.

26.

Jousson

Pretti

Di Bello

, et al. Non-invasive detection and quantification of the parasitic ciliate Ichthyophthirius multifiliis by real-time PCR. Dis Aquat Organ 2005; 65: 251–255.

27.

Floyd

Rogers

Lambshead

, et al. Nematode-specific PCR primers for the 18S small subunit rRNA gene. Mol Ecol Notes 2005; 5: 611–612.

28.

Sanders

Kent

. Development of a sensitive assay for the detection of Pseudoloma neurophilia in laboratory populations of the zebrafish Danio rerio. Dis Aquat Organ 2011; 96: 145–156.

29.

Murray

Dreska

Nasiadka

, et al. Transmission, diagnosis, and recommendations for control of Pseudoloma neurophilia infections in laboratory zebrafish (Danio rerio) facilities. Comp Med 2011; 61: 322–329.

30.

Chandrarathna

HPSU

Nikapitiya

Dananjaya

SHS

, et al. Outcome of co-infection with opportunistic and multidrug resistant Aeromonas hydrophila and A. veronii in zebrafish: identification, characterization, pathogenicity and immune responses. Fish Shellfish Immunol 2018; 80: 573–581.

31.

Whipps

Lieggi

Wagner

. Mycobacteriosis in zebrafish colonies. ILAR J 2012; 53: 95–105.

32.

Watral

Kent

. Pathogenesis of Mycobacterium spp. in zebrafish (Danio rerio) from research facilities. Comp Biochem Physiol C Toxicol Pharmacol 2007; 145: 55–60.

33.

Spagnoli

Xue

Murray

, et al. Pseudoloma neurophilia: a retrospective and descriptive study of nervous system and muscle infections, with new implications for pathogenesis and behavioral phenotypes. Zebrafish 2015; 12: 189–201.

34.

Meyer

Collar

. Description and treatment of a Pseudomonas infection in white catfish. Appl Microbiol 1964; 12: 201–203.

35.

Sazakli

Leotsinidis

Vantarakis

, et al. Comparative typing of Pseudomonas species isolated from the aquatic environment in Greece by SDS-PAGE and RAPD analysis. J Appl Microbiol 2005; 99: 1191–1203.

36.

López

Diéguez

Doce

, et al. Pseudomonas baetica sp. nov., a fish pathogen isolated from wedge sole, Dicologlossa cuneata (Moreau). Int J Syst Evol Microbiol 2012; 62: 874–882.

37.

Nisha

Rajathi

Manikandan

, et al. Isolation of Plesiomonas shigelloides from infected cichlid fishes using 16S rRNA characterization and its control with probiotic Pseudomonas sp. Acta Sci Vet 2014; 42: 1195–1202.

38.

Ran

Qin

Xie

, et al. Aeromonas veronii and aerolysin are important for the pathogenesis of motile aeromonad septicemia in cyprinid fish. Environ Microbiol 2018; 20: 3442–3456.

39.

Brenden

Miller

Janda

. Clinical disease spectrum and pathogenic factors associated with Plesiomonas shigelloides infection in humans. Rev Infect Dis 1988; 10: 303–316.

40.

Dong

Wang

, et al. Characteristics of Vibrio parahaemolyticus isolates obtained from crayfish (Procambarus clarkii) in freshwater. Int J Food Microbiol 2016; 238: 132–138.

41.

Gracia-Cazaña

Milagro

Queipo

, et al. Mycobacterium fortuitum infection after acupuncture treatment. Dermatol Online J 2017; 23: 21–21.

42.

Bhowmick

Bhattacharjee

. Bacteriological, clinical and virulence aspects of Aeromonas-associated diseases in humans. Pol J Microbiol 2018; 67: 137–149.

43.

Nascimento

Viana-Niero

Nogueira

, et al. Identification of the infection source of an outbreak of Mycobacterium chelonae keratitis after laser in situ keratomileusis. Cornea 2018; 37: 116–122.

44.

Silva

Marcoleta

Silva

, et al. Prevalence and susceptibility pattern of bacteria isolated from infected chronic wounds in adult patients. Rev Chilena Infectol 2018; 35: 155–162.

45.

Sunil

Harris

Sine

, et al. Investigation of a community cluster of cutaneous Mycobacterium marinum infection, an emerging zoonotic pathogen in aquaculture industry, Haliburton, Kawartha, Pine Ridge District Health Unit, Ontario, Canada, July–August 2015. Zoonoses Public Health 2019; 66: 164–168.

46.

Scarfe

Lee

O’Bryen

. Aquaculture biosecurity: prevention, control and eradication of aquatic animal diseases, Ames, IA: Blackwell, 2006.

Exploring the advantages and limitations of sampling methods commonly used in research facilities for zebrafish health inspections

Abstract

Keywords

Introduction

Methods

Ethical statement

Effect of subpopulation and fish number on success of pathogen detection

Targeted and broad-based diagnostic methods

Effect of system on diagnostic results in multi-system facilities

Statistical analysis

Results

Association of subpopulation and disease signs

Association between success of pathogen detection, subpopulation, and sample size

Histopathology

Effect of system and fish number on pathogen prevalence in multi-system facilities

Discussion

Footnotes

Declaration of Conflicting Interests

Funding

References