We evaluated cause of injury and quantified levels of three potential mycoplasmal pathogens (Mycoplasma agassizii, Mycoplasma testudineum, and an emydid mycoplasma) in three-toed box turtles (Terrapene carolina triunguis) from the greater St. Louis, Missouri, US area, brought to and housed at the Wildlife Rescue Center (Ballwin, Missouri, US) in 2015 and 2016. We created a probebased quantitative PCR (qPCR) assay for the emydid mycoplasma, with a similar specificity and sensitivity as the existing qPCR assays for M. agassizii and M. testudineum. All three microbes have been implicated in the development of upper respiratory tract disease in turtles and tortoises. We assessed whether signs of respiratory disease, sex, type of trauma, or treatment (administration of antibiotics) affected the presence of pathogens. We found that the most common types of injury experienced by turtles (n=85) were due to motor vehicles and other types of machinery, and that injuries due to motor vehicles were the most severe. We found a 61% prevalence of emydid mycoplasma (n=28) but M. agassizii or M. testudineum were not detected. Prevalence of disease and antibiotic treatment was too low to statistically relate to levels of mycoplasma. Sex and type of trauma were not associated with levels of emydid mycoplasma. The box turtle population we sampled did not experience signs of respiratory disease due to the fairly widespread prevalence of emydid mycoplasma. However, mycoplasmal diseases can be pathogen load-dependent. The qPCR we designed can be used to assess levels of emydid mycoplasma in other emydid species, populations, and individuals, in which there might be a positive association between the microbe and expression of respiratory disease.
Upper respiratory tract disease (URTD) is thought to be a cause of decline in wild and captive populations of many turtle and tortoise species (Jacobson 2007). There is strong evidence that URTD is caused primarily by Mycoplasma agassizii in the Mojave desert tortoise (Gopherus agassizii) and the gopher tortoise (Gopherus polyphemus; Brown et al. 1994, 1999). Many healthy Mojave desert tortoises carry low loads of M. agassizii, near the limit of detection, and only higher loads are associated with disease (Weitzman et al. 2017b). In the past, seroprevalence to M. agassizii was interpreted as evidence of the pathogen occurring in eastern box turtles (Terrapene carolina carolina; Calle et al. 1998; Adamovicz et al. 2015). However, genetic techniques suggest that box turtles and other turtles within the Emydidae family carry another, currently unnamed, species of mycoplasma (emydid mycoplasma; Feldman et. al. 2006; Farkas and Gal 2009; Ossiboff et al. 2015). A number of species of Testudines (turtles and tortoises) from the eastern US carry this unique emydid mycoplasma without signs of disease (Ossiboff et al. 2015). However, other studies suggest that this emydid mycoplasma might be associated with disease in Eastern and three-toed box turtles (T. c. carolina and Terrapene carolina triunguis; Feldman et al. 2006; Palmer et al. 2016). It is not known how widespread this emydid mycoplasma is and whether it can have a loaddependent effect on disease. Both sequencing and other genetic techniques to detect emydid mycoplasma have provided valuable information about the presence of the mycoplasma turtles (Ossiboff et al. 2015; Archer et al. 2017). However, probe-based quantitative PCR (qPCR) assays are unique in being able to quantify very small amounts of specific DNA and are extremely useful in detecting subclinical as well as clinical disease (Boyle et al. 2004; Bustin et al. 2009).
We studied three-toed box turtles that had been brought to a wildlife rehabilitation clinic in the summers of 2015 and 2016, or that had been admitted in a previous year and were still in recovery. In addition, we assessed records from all box turtles submitted during this time (including animals which required euthanasia) to first determine the primary causes of death and injury to box turtles which were brought to the Center from the greater St. Louis, Missouri, US, area. We then used nasal lavage to test all turtles housed at the center for M. agassizii, Mycoplasma testudineum, and emydid mycoplasma by a qPCR assay (Braun et al. (2014) and one we created to quantify emydid mycoplasma. Because respiratory diseases in Testudines are often chronic with intermittent signs of disease, assessing URTD in wild animals can lead to an inaccurate assessment of clinical disease (Aiello et al. 2016; Sandmeier et al. 2017a). We took advantage of the fact that turtles at the center were usually in recovery for longer than a week and up to several years, with carefully maintained health records. Furthermore, many turtles had experienced relatively severe injuries, suggesting that if they had a latent disease, signs would likely emerge due to the animals' weakened state. Hence, we had a relatively clear picture of the physiological health of the animals. At the same time, the nature of this study also warranted some caution in interpretation, because turtles were collected opportunistically and our analyses do not address health in turtles that had not come in contact with humans.
We designed a qPCR assay that was specific to emydid mycoplasma, determined the prevalence of emydid mycoplasma in T. c. triunguis in the greater St. Louis area, and tested for patterns of emydid mycoplasma prevalence based on signs of URTD, sex, type of trauma experienced, and administration of antibiotics. We hypothesized that the emydid mycoplasma would be widespread, signs of URTD would be relatively rare, and box turtles would not carry M. agassizii or M. testudineum. If emydid mycoplasma is associated with disease, we also hypothesized that animals that experienced more severe injury would have higher loads of emydid mycoplasma due to their greater susceptibility to disease.
MATERIALS AND METHODS
Records of all box turtles sampled as well as turtles submitted to the Animal Rescue Center (Ballwin, Missouri, US) in 2015 and 2016 (including animals released without further treatment or euthanized due to inoperable injury) were categorized by primary cause of injury: motor vehicle, other human machinery, predation by domestic animals, disease, injury not directly associated to human activity, and uninjured but kept as pets. Veterinarians and trained staff had recorded treatments and administration of antibiotics. Sex was determined by tail length and eye color (Conant and Collins 1998). If sex was indeterminate, it was recorded as unknown and the animal was not included in sex-specific analyses. When possible, the exact recovery location was determined from records that included coordinates, crossroads, or addresses. All turtles were checked once a day by staff, who noted any physical abnormalities, including signs of URTD, and who also noted any signs of disease at time of sampling. Locations of turtles were mapped in ArcGIS 10.3 (ESRI, Redlands, California, USA). We displayed cause of injury on a base map of roads as well as the presence or absence of emydid mycoplasma.
We spent a total of 4 d collecting nasal lavage samples from all box turtles kept at the center during that time. We sampled all the turtles at the center in 2015, including animals that had been admitted in previous years. Because animals were released as soon as they were deemed healthy, the turtles we sampled were a subset of all turtles from which we obtained records. In 2016, samples were collected only from newly admitted turtles. Nasal lavage consisted of a gentle flush of the nares with 1–1.5 mL of sterile saline solution using a sterile syringe, and fluids were collected in a sterile container (Sandmeier et al. 2017b). Recovered fluid was added to RNAlater (Qiagen, Valencia, California, USA) at a ratio of 200 µl RNAlater to 500 µL of recovered lavage. The samples were kept on ice and frozen within 6 h. The DNA was extracted from 500 µL of each nasal lavage sample using the gram-negative protocol of a DNEasy kit (Qiagen) and frozen at _40 C. Samples were tested separately for M. agassizii and M. testudineum (multiplexed reaction; Braun et al. 2014) and for emydid mycoplasma. Work was conducted under the following permits: Missouri Department of Conservation (16427, 16723) and Colorado State University–Pueblo Institutional Animal Care and Use Committee (2015–17).
Quantitative PCR for emydid mycoplasma
To create a specific qPCR assay for emydid mycoplasma, we used the published sequence for the 16S-23S ribosomal intergenic spacer region (GenBank ascension no. KJ623617; Ossiboff et al. 2015) to identify the best primers and hydrolysis probes using Primer Express 3.0.1 (Thermo Fisher Scientific, Waltham, Massachusetts, USA). Each possible sequence was then BLAST searched (National Center for Biotechnology Information 2018) to identify a sequence found only in all described strains of emydid mycoplasma (Ossiboff et al. 2015), but not found in any other published sequence (Table 1). Thus, we optimized assay specificity and secondarily identified the most stable set of primers and probe to amplify a target sequence unique to emydid mycoplasma.
We used a 25 µL reaction mixture, containing 12.5 µL TaqMan® Environmental Master Mix, 0.2 µL forward primer, 0.2 µL reverse primer, 0.025 µL TaqMan® probe, 9.575 µL RNase-free water, and 2.5 µL DNA (Thermo Fisher Scientific). The thermal profile consisted of 2 min at 50 C and 10 min at 95 C, followed by a 45-cycle repetition of 15 sec at 95 C and 1 min at 58.8 C. All reactions were run on a QuantStudio 3 Real-Time PCR System and analyzed with the accompanying software (Thermo Fisher Scientific). We used a plasmid containing the target sequence (Integrated DNA Technologies, San Diego, California, USA) as a positive control, to optimize assayperformance and convert cycle threshold value to copy numbers of DNA. In particular, we tested a variety of annealing temperatures ranging from 58 to 60 C to increase sensitivity and reduce intraassay variation. Dilution curves of the plasmid ranged from 5×108 to 5×101 copies (Fig. 1). For all assays we used a baseline fluorescence of 0.02. To test for in vitro specificity, we attempted to amplify extracted DNA from other species of mycoplasma and other microbes that might occur in turtle respiratory tracts, obtained from the American Type Culture Collection (ATCC, Rockville, Maryland, USA), including M. agassizii (ATCC 700616), M. testudineum (ATCC 700618), Pasteurella testudinis (ATCC 33688), Pasteurella multicida (ATCC 43137), and Escherichia coli (ATCC 87446). The DNA was directly extracted from pellets obtained from ATCC with a DNEasy blood and tissue kit (Qiagen) and quantified with a Nanodrop (Thermo Fisher Scientific), and diluted to either 5 ng/µL or 0.5 ng/µL, depending on yield.
Due to nonnormal distributions, we used a Wilcoxon and a Kruskal-Wallis sign-rank test to evaluate the effects of sex and type of trauma on quantities of emydid mycoplasma.
We obtained records from or handled 96 box turtles, 85 of which had complete records and 28 of which were being housed at the center over our 4 d of sampling in the summers of 2015 and 2016. Eleven animals were euthanized on arrival due to the severity of injuries; eight of those were hit by vehicles. Of the animals with complete records, causes of trauma primarily included vehicle hits (44%; 37/85) and other injuries due to machinery (24%; 2/85). Another 12% (20/85) had injuries not directly associated to human activities, 11% (9/85) sustained injuries from domesticated animals (primarily dogs), 6% (5/ 85) were found uninjured but trapped inside a dwelling or picked up and kept as pets, and 5% (4/85) had signs of ocular disease (swollen eyes). Animals were not submitted for any other diseases. Point locations were known for 71 of the 85 animals with complete records (Fig. 2A).
The qPCR did not amplify any of the DNA obtained from pure cultures of the other microbes it was tested against, showing in vitro specificity. Using a 10-fold dilution curve (ranging from 5 to 5 × 108 copies of DNA), the detection limit of the assay was on the order of 50 copies, but intra-assay variability increased greatly at 50 copies. The linear range for which the assay was optimized was 5×102 to 5 × 108 copies of DNA, including an optimization of annealing temperature (Fig. 1 and Table 1). Figure 1 shows mean values calibration curves, used to calculate assay specifications presented in Table 1.
In general, the animals we sampled by nasal lavage included those being treated for longterm care, versus those rereleased into the wild after minor treatment. For animals from which we obtained a nasal lavage, one had ocular signs of disease (swollen eyes) at the time of admission to the center, one had nasal signs of disease (mucous drips from the nares) during sampling, and two received general antibiotics effective against gram-negative bacteria. All animals tested negative for M. agassizii and M. testudineum. A total of 61% (17/28) tested positive for emydid mycoplasma (Fig. 2B). Within samples from positive animals, we detected between 41 and >30,000 copies of DNA per 2.5 µL of DNA, with a mean of 5,325 (SD=14,098). Of the positive samples, 93% (26/28) were below 5,000 copies of DNA. The turtle with exudate, but not the one with a history of ocular signs of disease, tested positive for emydid mycoplasma, and values (DNA copy number) fell within the range of those of nondiseased turtles. The two turtles that received broad-spectrum antibiotics tested negative for emydid mycoplasma. The distribution of turtles with emydid mycoplasma did not show any distinct pattern and emydid mycoplasma was detected in animals from throughout the greater St. Louis area (Fig. 2B). Neither sex nor type of trauma had a significant effect on emydid mycoplasma copy number (Z=0.623, P=0.609; χ2=5.314, P=0.256, respectively).
Given low rates of disease and low rates of antibiotic treatment, we were not able to statistically assess how these two measures influenced quantities of emydid mycoplasma. However, qPCR data did indicate that this microbe occurs at a high prevalence in otherwise healthy three-toed box turtles, supporting conclusions reached by Ossiboff et al. (2015) that this microbe might be common and commensal in many emydid turtles. Sex, type of trauma, and location of the turtle on the landscape also did not appear to influence levels of emydid mycoplasma. Therefore, our study suggested that emydid mycoplasma was not pathogenic in these box turtles. Whether or not higher loads of this microbe, or higher loads in conjunction with other microbial infections can cause disease remains to be determined through experimentation or broader field studies, and our qPCR assay provides the tool for assessing load-dependence in diseased animals. In Mojave desert tortoises, the ability to quantify infection intensity of M. agassizii within tortoise respiratory tracts greatly increased the understanding of M. agassizii prevalence, pathogenicity, and transmission (Aiello et al. 2016; Weitzman et al. 2017b). Low cost, ease of use, and specificity to microbial species make qPCR assays valuable to the study of wildlife diseases.
We recommend reducing interassay variability by using dilution curves of plasmids on each plate to calculate DNA copy numbers, as we did here (Bustin et al. 2009). Values at or below 50 copy numbers should be interpreted with caution, because the assay can detect these low levels in samples run in triplicate but is not optimized for such low amounts of DNA, and replication efficiency is reduced at these concentrations (Table 1). Given the limited genetic sequence available, we created the assay to be specific to emydid mycoplasma while sacrificing some optimal chemistry (e.g., hairpins and dimers formed by primers and probe; Table 1). If larger portions of both the emydid mycoplasma and related species are sequenced, it might be possible to create a slightly more efficient qPCR, or one that is better optimized to accurately quantify lower quantities of DNA.
However, quantitative PCR still is more sensitive in determining the presence or absence at low microbe loads than is conventional PCR plus sequencing in most experimental systems (e.g., Braun et al. 2014; Weitzman et al. 2017b), although this has not yet been explicitly tested for our assay. If this emydid mycoplasma-specific qPCR is more sensitive than conventional PCR, this possibly explains some discrepancies between our study and others that either found an association between emydid mycoplasma and URTD (Feldman et al. 2006; Palmer et al. 2016) or documented lower prevalence of emydid mycoplasma in box turtles than we found in this study (Archer et al. 2017). If emydid mycoplasma only causes disease at high infection intensities, conventional DNA sequencing might only have resulted in positive test results in symptomatic eastern box turtles, with high mycoplasma loads (Feldman et al. 2006). The study by Palmer et al. (2016) found that two diseased box turtles in Forest Park in St. Louis tested positive for emydid mycoplasma by DNA sequencing, but healthy animals were not similarly tested. Finally, Archer et al. (2017) found lower rates of emydid mycoplasma using a multiplexed qPCR, aimed at detecting 13 different microbes. However, their assay was developed to detect presence or absence, and sensitivities for individual microbe-specific amplification were not determined by their technique (Archer et al. 2017).
It is possible that different species or populations of box turtles experience different relationships with the same microbes and that coinfection dynamics among microbes might also differ across host species. For example, Weitzman et al. (2017a) found that quantities of M. agassizii and M. testudineum interacted differently with each other across the four US species of Gopherus tortoises. Possibly emydid mycoplasma occurs at a higher prevalence, without associated signs of disease, in T. c. triunguis, which is genetically distinct from the eastern T. carolina clade (Martin et al. 2013) surveyed by Feldman et al. (2006) and Archer et al. (2017). The host response to pathogens also might vary by species or even by population, because disease is combination of damage by the pathogen and damage by the host immune response (Wobeser 2006).
Interestingly, no turtle tested positive to M. agassizii and M. testudineum, which are relatively commonly found in US species of Gopherus (Weitzman et al. 2017b). The fact that none of the turtles here tested positive for M. agassizii or M. testudineum, yet box turtles can be seropositive to M. agassizii (Calle et al. 1998; Adamovicz et al. 2015), might be interpreted as evidence of cross-reactive antibodies in previously tested animals. Induced antibodies, especially immunoglobulin M in the early stages of infection, can cross react with similar pathogens (Murphy 2012). In addition, many ectothermic vertebrates, including species of turtles, have been shown to possess high levels of constitutive, natural antibodies (Sandmeier et al. 2012; Zimmerman et al. 2013). Although their function is less well understood than that of induced antibodies across vertebrate species, they have many functions in controlling disease and maintaining homeostasis in vertebrate animals (Baumgarth et al. 2005). These natural antibodies often cross react with a number of common pathogens and could explain why box turtles could test seropositive for M. agassizii (Baumgarth et al. 2005). High levels of natural antibodies that bind to M. agassizii could be important to disease dynamics in Mojave desert tortoises, and similarly might afford some protection against pathogens in box turtles as well (Sandmeier et al. 2013).
Motor vehicles and machinery were the primary cause of injury in this population of box turtles and far exceeded the quantifiable threat by respiratory disease in this population. The entire T. carolina group have recently been listed as vulnerable by the International Union for Conservation of Nature due to continuing threats and declining populations (van Dijk 2011). In St. Louis, it is clear that development and road traffic within developed areas is a primary threat. Public education to advise caution when operating machinery, discourage collection of turtles to keep as pets, and support of programs such as the Wildlife Rescue Center hopefully will allow for the persistence of box turtles within the greater St. Louis area.
We thank the Animal Rescue Center for access to data and animals, with special thanks to Lauren Caruso and Kim Rutledge. We thank undergraduate students (Amanda Reno, Chris Cannon, and Danielle Wildermuth) from Lindenwood University–Belleville with help in sampling animals. Funding was provided through the Colorado State University–Pueblo's C-BASE Communities to Build Active STEM Engagement (Department of Education Title III, Award P031C160025), and supported both undergraduate salaries and costs of materials.