The pathogenicity of frog virus 3 (FV3)–like ranavirus varies in adult chelonian species at different environmental temperatures, but differences in pathogenicity at different temperatures has yet to be determined in juveniles. Our objective was to determine the susceptibility to FV3-like ranavirus in four species of juvenile chelonians: red-eared sliders (RES; Trachemys scripta elegans), Mississippi map turtles (Graptemys pseudogeographica kohnii), false map turtles (FMT; Graptemys pseudogeographica), and eastern river cooters (Pseudemys concinna concinna) at two environmental temperatures. Two simultaneous trials (n=8 treatment and n=4 controls of each species) were conducted in separate temperature-controlled rooms with animals maintained at 22 C or 27 C. All of the inoculated animals of each species at each temperature died, but no mortality was observed in control animals. Median survival times varied between 8 d and 11 d, based on species and temperature, with RES in the 27 C trial surviving the shortest time and the FMT in the 22 C trial surviving the longest. Combining all species, turtles in the 27 C trial survived for fewer days than those housed at 22 C, despite all turtles in both trials having similar viral copies detected in postmortem tissues. Lesions in inoculated turtles resembled those noted in natural and experimental FV3-like ranavirus infections and included vasculitis, thrombosis, hemorrhage in multiple organs, renal tubular necrosis, and hepatic necrosis. Myositis was not present in any juvenile, infected turtles in this study. This study confirmed that juvenile chelonians have a high susceptibility to ranaviral disease.
Ranaviruses have caused mass mortality events in wild ectotherm populations worldwide, with mortality rates as high as 90–100% and is proposed as a significant threat to biodiversity (Allender et al. 2011; Miller et al. 2011). Ranaviral disease is a World Organisation for Animal Health reportable disease in amphibians and is known to infect fish and reptiles with the potential for interclass transmission by numerous routes (Gray et al. 2009; Lesbarréres et al. 2012; Brenes et al. 2014a, b). Infection with ranaviruses causes marked systemic disease; however, gross signs of infection may not be displayed depending on the host species and viral strain (Chinchar et al. 2001; Miller et al. 2015). Experimental infection of anurans and caudates demonstrated a range of outcomes from complete mortality to no detected infection (Hoverman et al. 2010, 2011; Haislip et al. 2011).
Environmental temperature has been shown to contribute to variable outcomes in fish (Whittington and Reddacliff 1995; Jun et al. 2009), amphibians (Rojas et al. 2005), and reptiles (Allender et al. 2013b) infected with ranaviruses. For adult tiger salamanders (Ambystoma tigrinum) experimentally inoculated with Ambystoma tigrinum virus, animals showed high mortality at 10 C and 18 C, and low mortality was observed at 26 C (Rojas et al. 2005). However, common frog (Rana temporaria) tadpoles experienced greater mortality at higher environmental temperatures, but tiger salamander larvae, wood frog (Rana sylvaticus) tadpoles, and northern leopard frog (Rana pipiens) tadpoles showed higher mortality at lower environmental temperatures (Bayley et al. 2013; Echaubard et al. 2014). Frog virus 3–like virus-infected, adult red-eared sliders had greater mortality at lower environmental temperatures (Allender et al. 2013b), but it is not known whether juvenile chelonians show similar susceptibility to FV3-like virus as adult chelonians or other juvenile ectotherms.
Our objective was to determine the pathogenicity of FV3-like ranavirus in four species of juvenile chelonians: red-eared sliders (RES; Trachemys scripta elegans), Mississippi map turtles (MMT; Graptemys pseudogeographica kohnii), false map turtles (FMT; Graptemys pseudogeographica), and eastern river cooters (RC; Pseudemys concinna concinna) at two environmental temperatures. The hypotheses were that the juvenile turtles would have similar susceptibility across species but show higher mortality and greater pathogenicity at the lower environmental temperature.
MATERIALS AND METHODS
Ninety-six RES (n=24), MMT (n=24), FMT (n=24), and RC (n=24) were acquired from a commercial farm within 10 d of hatching and were randomized by computer into control (n=4) and inoculation (n=8) groups at each environmental temperature for each species. The two temperature groups were run simultaneously in July–August 2014 in two separate, 4.8×3.7-m environmental chamber rooms set at 22 C and 27 C. Uninfected controls and inoculated animals were kept on opposite sides of the same room for each temperature, separated by a double plastic barrier that extended from floor to ceiling. All animals were subjected to an 8-d acclimation period. They were housed in hard-plastic enclosures (48×27×20 cm) with approximately 2 L of water and a dry brick for basking. They were subjected to a 12-h light cycle with a timer and were fed a commercial aquatic turtle hatchling diet (Fluker Farms, Port Allen, Louisiana, USA) ad libitum. Complete water changes with dechlorinated water were performed twice a week. All activities were approved and monitored by the University of Illinois Institutional Animal Care and Use Committee (protocol 14071).
A FV3-like ranavirus isolate originally from a wild eastern box turtle (Terrapene carolina carolina), frozen at that time and thawed for this project, was inoculated onto a confluent layer of a Terrapene heart cell line (TH-1) as previously described (Allender et al. 2013b). When cells exhibited 100% cytopathic effects, the flasks were frozen and thawed three times, being thoroughly vortexed before and after each freeze cycle. The contents were then transferred to 50-mL conical tubes and centrifuged at 4,000 × G for 20 min. The cell pellet was discarded and quantitative PCR was then performed on the supernatant to confirm presence and quantity of viral DNA. Viral titers (50% tissue culture infective doses [TCID50]) were determined from serial dilutions of virus in TH-1 cell culture in four replicates (Reed and Muench 1938). The virus was frozen at −80 C in 5×105 TCID50 aliquots until the day of inoculation, when they were thawed to room temperature.
At 8 and 5 d before inoculation, all turtles were examined, weighed, and confirmed to be negative for FV3-like virus via quantitative PCR in swabs from the oral cavity or the cloacal. The swabs (plastic-handled, cotton-tipped applicators, Thermo Fisher Scientific, Hanover Park, Illinois, USA) were stored in separate 2.0-mL polypropylene Eppendorf tubes and stored dry at −20 C until analyzed. Quantitative PCR of the swabs confirmed all subjects to be negative using a previously established protocol (Allender et al. 2013a).
On day 0 of the experiment, each treatment animal was inoculated intramuscularly in the right forelimb with 5×105 TCID50 of FV3-like virus (0.1 mL). Each control animal was administered an equal volume of uninfected TH-1 cell lysate in the same manner. Turtles were examined and weighed on days 0, 2, 5, 9, 12, and 16. Clinical signs were monitored and documented daily, including lethargy, nasal and ocular discharge, oral plaques, and swelling at the injection site; each sign was classified as absent or present. When two inoculated animals of the same species and temperature died or were humanely euthanized, one matched control animal was humanely euthanized for comparison.
Postmortem examination occurred after euthanasia or natural death. Sterile procedures were used to collect samples of spleen, liver, kidney, and intestine. Control animals were euthanized, and collection of tissue was performed before euthanizing and sampling of inoculated animals. The samples were stored dry at −20 C in 2.0 mL polypropylene Eppendorf tubes until analysis of all samples could be performed. We extracted DNA from postmortem samples of kidney, liver, and spleen (DNeasy, Qiagen, Valencia, California, USA) using the manufacturer's instructions. The DNA extracts were evaluated for concentration and purity (ratio of absorbances at wavelengths of 260 and 280 nm) using a spectrophotometer (Nanodrop, Wilmington, Delaware, USA). Quantitative PCR was performed as described for the swabs (Allender et al. 2013a). The final viral copy number was standardized to total DNA concentration per microliter by dividing the copy number by 2.5 (volume of extract used in each reaction), and the DNA concentration was determined by spectrophotometry.
After sterile collection of tissues and removal of the plastron, turtles were fully immersed in 10% neutral-buffered formalin. Tissues selected for histopathology included liver, kidney, spleen, skeletal muscle from the right hind limb, heart, lung, and intestine, as available. For some individuals, residual tissue after sampling for PCR was not present. Unfortunately, the size of these animals prevented collection of spleen for both molecular and pathologic investigation; thus, we were unable to perform histopathology of the spleen on any animal. In many cases, adjacent tissues were also collected and included thymus, thyroid, esophagus, trachea, stomach, pancreas, remnant yolk sac, peripheral nerve, and skin. Tissues were routinely processed and embedded for histopathology, cut at 3 μm and stained with H&E. Twenty-four individuals were selected for complete histopathologic investigation, which comprised a control and two inoculated animals from each species-temperature cohort. Selection of inoculated animals for evaluation consisted of the animal with the highest and lowest viral copy number per cohort. All animals were evaluated by one board-certified veterinary pathologist (K.A.T.), who was blinded to the treatment and temperature data.
For statistical analysis, animals were classified as control (inoculated with uninfected TH-1 cell lysate) or inoculated (with FV3-like virus), then further by temperature (22 C or 27 C). Normality was assessed by the Shapiro-Wilks test for all continuous variables (mass and viral copy number). Descriptive statistics were calculated for standardized viral copy number, including mean, SD, and minimum and maximum. Kaplan-Meier estimates were used to determine any difference in median survival time (MST) based on temperature. Because of uneven sample sizes (because of death occurring at unequal rates), body mass was grouped into four categories: preinoculation (pre; average of days −8 and −5), day 0, day 2, and terminal body mass (last body mass before death, which occurred from day 2 to day 16). Differences in mass over time were evaluated using repeated-measures analysis of variance with post hoc Tukey's test for differences between treatment groups. Further post hoc tests of mass between day 0 and the terminal sample were evaluated with a paired-sample t-test. Those analyses were completed with SPSS software (version 22, IBM, Chicago, Illinois, USA). A Cox proportional-hazards test was performed individually for three variables: species, temperature, and standardized viral quantity. An additional model combined the variables to determine their effect on death after inoculation (days). The models were tested for agreement with proportional assumptions and overall fit. Those analyses were performed with the “survival” package (Therneau 2015) in the statistical program R (R Core Team 2016). To compare each combination, the “multcomp” package was used to create general linear-hypothesis models (Hothorn et al. 2008). Analysis of histopathologic findings was performed by coding the presence (1) or absence (0) of the following lesions: vasculitis or thrombus at any site, intracytoplasmic inclusions, hepatic necrosis, renal tubular necrosis, pneumonia, and myositis. The χ2 or Fisher's exact test was used to evaluate the association of lesions with species, treatment, temperature group, and high/low viral copy number. Statistical significance was set at α=0.05. Statistical analysis was performed with commercial software (SPSS).
Lethargy was the only clinical sign observed and was only present in a single, inoculated RES in the 22 C chamber. That individual was obtund after appearing clinically normal approximately 24 h prior. No control animal was observed with clinical signs. Mass increased significantly over all four times (pre, day 0, day 2, and terminal) when combining all species, temperatures, and treatment groups (P<0.001). Weight did not differ significantly between control and inoculated RES (P=0.985), RC (P=0.997), FMT (P=1.000), and MMT (P=1.000) at the preinoculation times. Similarly, weight did not differ significantly between control and inoculated RES (P=0.983), RC (P=0.997), FMT (P=0.976), and MMT (P=0.981) at the terminal time point. Mean±SD body mass increased significantly in RES from 9.4±0.88 g to 10.1±1.0 g (P<0.0001), in FMT from 8.4±0.61 g to 8.8±0.85 g (P<0.0001), and in MMT from 8.6±0.75 g to 9.3±0.91 g (P<0.0001), but nonsignificantly in RC from 11.1±0.48 g to 11.3±0.82 g (P=0.181).
Most inoculated turtles died before the development of clinical signs or were euthanized because of the severity of clinical signs (particularly lethargy). Deaths of all infected turtles occurred within 6–16 d of inoculation, resulting in a 100% morality rate. We detected FV3-like ranavirus DNA in all inoculated animals (Table 1). None of the control turtles had molecular evidence of FV3-like ranavirus; thus, there was a significant positive association between inoculation and ranavirus detection (P<0.001).
The Cox proportional hazards test for individual factors identified temperature and species as significant relative to mortality (P=0.002), whereas viral quantity standardized by DNA quantity at time of death was not significant (I=0.540; Table 2). The same pattern held true for the combined model, with MMT and 22 C representing the baseline hazard (Table 3). Turtles of all species housed at 27 C had significantly shorter MSTs (8 d) than did turtles housed at 22 C (11 d; P<0.0001; Fig. 1A, B). The combined test did not meet the assumption of proportional hazards, specifically for the temperature variable and likely because of the rapid death over a relatively short time to event occurrence. However, the concordance value of 0.779 and R2=0.405 indicate an acceptable model fit for the observed data, which was supported further by the individual factor models.
Pairwise comparisons of significant factors using the general linear-hypothesis models indicate statistical differences in MST between FMT-MMT and FMT-RES at 27 C, and differences within species between temperatures for MMT and RES (Fig. 1C–F). The FMT survived the longest (12 d), followed by MMT (9.5 d), RC (9.4 d), and RES (8.1 d). The MMT turtles housed at 27 C survived a significantly shorter time (7.7 d) than they did at 22 C (11.4 d; P<0.0001; Fig. 1D). Additionally, RES at 27 C survived fewer days (7.2 d) than at 22 C (9.7 d; Fig. 1F). There were no differences in survival time between turtles housed at 27 C and 22 C in FM (P=0.111) or RC (P=0.112; Fig. 1C, E).
Significant differences in the presence of most histopathologic lesions were noted between inoculated and control animals for vasculitis/thrombi presence (P<0.0001), hepatic necrosis (P<0.0001), renal tubular necrosis (P=0.0001), and pneumonia (P<0.0001) but not for the presence of intracytoplasmic inclusions (P=0.119) or myositis (P=1.000). No significant differences in histopathologic findings were noted among species or temperature groups, although that may have been due to the low power (<0.8) of the comparisons. Both control and inoculated turtles had vacuolated hepatocytes (glycogen and lipid). No significant lesions were noted in any of the control turtles. All inoculated turtles of both temperatures were observed with vasculitis and/or fibrin thrombi. Associated with vasculitis and thrombi were multifocal areas of hepatic necrosis, renal tubular necrosis, gastrointestinal submucosal hemorrhage, and heterophilic pneumonia (Fig. 2). None of the turtles in either temperature group were observed with myositis, although one turtle had intramuscular hemorrhage, and no lesions were noted within the thymus. More turtles housed at 27 C (n=3; 100%) were observed with renal tubular necrosis than were seen in the turtles housed at 22 C (n=2; 67%), but the difference was not significant (P=0.343). Conversely, more turtles housed at 22 C (n=3; 43%) were observed with intracytoplasmic inclusions than were seen in the turtles housed at 27 C (n=1; 17%), but the difference was not significant (P=0.273). Inclusions were noted most often in MMT (n=3; 75%) than RES (n=1; 25%), RC (n=0; 0%), or FMT (n=0; 0%; P>0.05). Renal tubular necrosis was least observed in RC (n=1; 33%) compared with FMT (n=4; 100%) or MMT (n=4; 100%; P>0.05).
Ranaviral disease has caused mass mortality in wild ectotherm populations worldwide (Gray and Chinchar 2015). However, the susceptibility, clinical signs, and mortality have shown variability with species, temperature, and age of the infected individual (Duffus et al. 2015; Miller et al. 2015). Despite several cases in free-ranging chelonians under various conditions (Duffus et al. 2015), few studies have experimentally challenged chelonians with ranaviruses, thus, making it difficult to understand how individuals of this reptilian order fit into the disease ecology. We observed 100% mortality after experimental FV3-like ranaviral infection in four juvenile aquatic turtle species at two environmental temperatures. Mortality rates in previous experimental challenge studies in turtles have ranged from 0% (Brenes et al. 2014a) to 100% (Allender et al. 2013b). However, those studies differed in route of transmission (bath; Brenes et al. 2014a) or age class (adults; Allender et al. 2013b). Many other factors are known to influence ranavirus disease occurrence, including isolate, dose, and route. We chose the exact same isolate, dose, and route from our previous study in adult RES to allow direct comparison of mortality. We aimed to determine what specific effect age had on disease susceptibility if those three factors (isolate, dose, and route) were controlled.
Age is a significant predictor of mortality in amphibians, with higher mortality in larval or metamorphic age classes in North American species and adults in the UK. Specifically, salamanders (Ambystoma spp.), wood frogs, green frogs (Rana clamitans), and American bullfrog (Rana catesbeiana) juveniles have higher mortality rates than adults have (Harp and Petranka 2006; Duffus 2008; Miller et al. 2011). We found that juvenile RES in our study had a higher mortality rate than adults had at 22 C (100% vs. 50%, respectively) and the same mortality rate at 27 C (100%; Allender et al. 2013b). However, Brenes et al. (2014b) observed no mortality using a bath inoculation in juvenile soft-shelled turtles (Apalone ferox). The MMT, RC, and juvenile RES in that study (Brenes et al. 2014b) exposed to FV3-like ranavirus via a shared environment with infected fish or amphibians similarly did not die. Because comparisons to adult age classes were not performed in those studies, it is difficult to elucidate whether differences in those studies and our findings were due to age class or route of infection.
Originally, the injectable route was supported because of the potential for this pathogen to be transmitted through mosquito vectors (Kimble et al. 2015). However, exposing turtles to a bath solution is more likely to be the natural source of infection. In this study, we were able to confirm infection occurred in all individuals, thus improving our ability to draw conclusions about age. Future research should focus on establishing the relevant routes of transmission and then follow up with age-specific challenge studies using those routes.
Despite having a similar mortality rate to adult RES, there was a much shorter time to death at both temperatures. The MST of adults at 22 C (17 d; Allender et al. 2013b) was longer than juveniles (11 d) at the same temperature. Similarly, the MST of adults at 27 C (30 d; Allender et al. 2013b) was longer than for juveniles (8 d) at the same temperature. Unlike previous studies in adult RES that observed shorter MST at colder environmental temperatures (Allender et al. 2013b), we observed warmer temperatures led to shorter MST in both RES and MMT and made no difference in FMT or RC. There is significant variation among ectotherms in disease susceptibility and severity based on temperature. Higher mortality in larval salamanders housed at colder temperatures followed direct bath exposure to Ambystoma tigrinum virus (Rojas et al. 2005), whereby in European perch (Perca fluviatilis) and rainbow trout (Oncorhynchus mykiss) exposed to epizootic hematopoietic necrosis virus, mortality was greatest at higher temperatures (19–21 C; Whittington and Reddacliff 1995). It is possible that the host immune response is responsible for the observed temperature differences. However, the effect of temperature on adult reptiles is poorly studied, and in juveniles, it is mostly unknown (Zimmerman et al. 2010). The larval immune system of amphibians is poorly developed compared with adults (Flajnik 1996) and further suppressed in response to ranaviruses (Salter-Cid et al. 1998; Gantress et al. 2003; Robert 2005). It is plausible to consider that temperature, in addition to age, could modulate disease progression in chelonians. Further investigation into the immunology of ectotherms, specifically that of juvenile chelonians and the pathogenicity of ranavirus in those juvenile chelonians at different temperatures is warranted.
Adult chelonians are commonly observed with lethargy, nasal and ocular discharge, and oral plaques in both natural and experimental infection with FV3 (Johnson et al. 2008; Allender et al. 2011; Allender et al. 2013b; Currylow et al. 2014). Only a single individual demonstrated lethargy in our study, and none showed upper respiratory signs. Wood frog tadpoles and larvae have specifically been found to have positive FV3 results by PCR in the wild without clinical evidence of infection (Duffus et al. 2008). Similarly, African clawed frog (Xenopus laevis) larvae had increased susceptibility, but fewer signs of morbidity, compared with adults (Gantress et al. 2003). In the case of Xenopus, it is hypothesized that nonclinically affected animals are carriers of ranavirus (Soto-Azat et al. 2016). Chelonians may also be reservoirs of the disease (Brenes et al. 2014b). With the survival times of inoculated turtles in our study being between 6 d and 16 d, it is unlikely that juvenile chelonians have a role as a nonclinical reservoir. As juvenile turtles die in aquatic systems, they may release millions of copies of virus, thereby contributing to an outbreak in free-ranging chelonians, but the contamination would be a single point and not a continuing source of virus. Future studies can be directed at measuring ranavirus in water to establish whether death and viral release into water is a viable source of infection.
Histologic lesions in inoculated juvenile terrapins were similar among species and temperatures to the spectrum of lesions reported in experimental and natural infections of adults (Johnson et al. 2007, 2008; Allender et al. 2013b). In comparison with inoculated adults, none of the juveniles had myositis. It is not certain whether that was due to inoculation site selection (hindlimb vs. forelimb), generally more acute time course, or age-related differences. The necrosis of hematopoietic tissue in the kidney, liver, and bone marrow that we observed has been described (Johnson et al. 2008), but interestingly, we observed no lesions of the thymus. Because of sampling for molecular assays, the presence of splenitis could not be confirmed histologically. Inclusions, an inconsistent finding in ranavirus infections, were most common in MMT but were unrelated to disease time course or viral copy number; therefore, it is not certain whether that is a true species difference.
We thank Merial for summer student support of this project.