COCCIDIAL INFECTION OF THE ADRENAL GLANDS OF LEATHERBACK SEA TURTLES (DERMOCHELYS CORIACEA)

The globally distributed leatherback sea turtle (Dermochelys coriacea) is the largest living chelonian (Eckert et al. 2012). The species is classified as critically endangered, largely because of severe declines in the Pacific (Spotila et al. 1996; Eckert et al. 2012). Harvest of eggs and turtles for human consumption and incidental capture by fisheries are the principal threats to populations. Leatherbacks occupy a unique ecologic niche among sea turtles, feeding almost entirely on gelatinous zooplankton and their associated symbionts and parasites (Eckert et al. 2012). Relatively little is known or understood about diseases or parasites of leatherbacks because of their predominantly oceanic life history, wide range, and the poor postmortem condition of most specimens available for necropsy (Stacy et al. 2015).

There are very few protozoal parasites reported in free-ranging sea turtles with only two species of coccidia described (Laukner 1985; Duszynski and Morrow 2014). The most significant coccidian in terms of disease is Caryospora cheloniae, which is found in green turtles (Chelonia mydas) and has been associated with epizootics and mortality in captive and wild turtles (Leibovitz et al. 1978; Gordon et al. 1993). The other reported sea turtle coccidian, Eimeria caretta, was found in a loggerhead sea turtle (Caretta caretta) and is of unknown pathologic significance (Upton et al. 1990).

We observed histologic lesions suspected to be caused by protozoa within the adrenal glands of dead and moribund leatherback turtles that were recovered from beaches and near-shore waters of North America. A multi-institutional review of cases found these lesions to be extremely common among subadults and adults that were presented for necropsy. We describe these lesions and present molecular evidence that they are associated with infection by previously undescribed coccidia within the Eimeriidae.

We reviewed records for leatherback sea turtles necropsied by collaborating institutions (see Acknowledgments) from August 2001 through April 2014 to identify cases in which adrenal glands were examined histologically (n=21). We compiled summary information for each case, including the date and location of discovery, curved carapace length (CCL: measured from nuchal notch to the tip of the pygal), gender, and the cause and circumstances of death. Tissues for histology were preserved in formalin and were processed into paraffin, sectioned, and stained with H&E by routine methods. Formalin-fixed adrenal gland from a representative case was also examined by transmission electron microscopy. Briefly, tissues for transmission electron microscopy were prepared by overnight immersion in McDowell-Trump fixative (McDowell and Trump 1976) and microwave processing adapted from the protocol described by Giberson et al. 1997, 2003. Ultrathin sections were stained with the use of lead citrate and uranyl acetate, and examined with the use of a Phillips EM 300 electron microscope (Pye Unicam, Cambridge, UK) operated at 80 kV. Adrenal samples from a subset of turtles (n=10) were frozen at −20 C or −80 C for genetic studies. In addition, centrifugal fecal flotation with microscopic examination was performed opportunistically on a fecal sample from a live leatherback turtle captured and released during a field study off the coast of Massachusetts, US.

We extracted DNA from frozen adrenal gland and feces with the use of a commercial kit (DNeasy^® Blood and Tissue Kit, Qiagen, Valencia, California, USA). Based on the unusual morphology of the adrenal lesions and some morphologic resemblance to Besnoitia-like organisms (Gardiner et al. 1998), we used previously described consensus PCR methods to amplify the 18S rRNA gene of Eucoccidia (Garner et al. 2006). Specific primers were then designed to obtain additional sequence. All primers used in this study were designed manually and examined with the use of Primer3 (Rozen and Skaletsky 2000). Specific primer 1147R (5′-GCAGGAGAAGCCAAGGTAGG-3′) was used with a previously described pan-coccidial primer 18F (5′-CTGGTTGATCCTGCCAGTAGTC-3′; Innis et al. 2007). Another specific primer 1424F (5′-AACGAACGAGACCTTAGCCT-3′) was used with a previously described primer 1898R (5′-GATCCTTCYGCAGGTTCAC-3′; Innis et al. 2007). This larger sequence, which is relatively conserved within Eucoccidia, was amplified from an index case with adrenal lesions and the fecal sample with oocysts.

With the use of the sequence from the index case, a specific semi-nested PCR assay was designed to detect and characterize the eucoccidian 18S sequence in the remaining frozen adrenal samples, and to compare genetic diversity among cases. This assay amplified the V4 region, which contains areas that are less conserved in the Eucoccidia (Hadziavdic et al. 2014). The primers utilized for the first amplification were 18F (5′-CTGGTTGATCCTGCCAGTAGTC-3′; Innis et al. 2007) and LB929R (5′-CCCCCTACTGTCGTTCTTGA-3′). The second amplification was performed with the use of LB462F (5′-ATTGGAATGATGGGAATCCA-3′) with the same LB929R primer.

For all PCR reactions, we used Takara SpeedStar HS DNA polymerase (Takara Bio Inc., Otsu, Japan) to perform amplifications as follows: denaturation at 94 C for 5 min, then 40 cycles of denaturation at 94 C (45 s), annealing at 58 C (45 s), and extension at 72 C (45 s). The final extension step was at 72 C for 10 min. Direct sequencing of amplicons from the second amplification was performed at the University of Florida Interdisciplinary Center for Biotechnology Research Sequencing Facilities using the Big-Dye Terminator Kit (Applied Biosystems, Foster City, California, USA) and then analyzed by an ABI 3130 automated DNA sequencer.

Predicted homologous 18S nucleotide sequences of related coccidian species were aligned using MAFFT, as described (Katoh and Toh 2008). Bayesian analyses of each alignment were performed with the use of Mr. Bayes (Ronquist and Huelsenbeck 2003) on the CIPRES server (Miller et al. 2015) with gamma distributed rate variation and a proportion of invariant sites, and a general time-reversible model. Adelina bambarooniae (Apicomplexa: Eucoccidiorida), which infects insects, was used as an outgroup. Four chains were run and statistical convergence was assessed by looking at the standard deviation of split frequencies and potential scale reduction factors of parameters. The first 10% of 1,000,000 iterations were discarded as a burn in, based on examination of trends of the log probability versus generation. Two independent analyses were performed to avoid entrapment on local optima.

Maximum-likelihood analysis of the alignment was performed with the use of RAxML on the CIPRES server as described by Stamatakis et al. (2008), with the use of a gamma distributed rate variation, a proportion of invariant sites, and a general time-reversible model. Again, A. bambarooniae was used as an outgroup. We used bootstrap analysis with 1,000 resamplings to test the strength of the tree topology (Felsenstein 1985).

We used a combination of quantitative PCR (qPCR) and comparative tissue density of the adrenal lesions to investigate an association between the detected coccidian sequence and adrenal histology more directly. The qPCR assay targeted a specific conserved region of the V4 sequence. Specificity was evaluated with the use of the following well-characterized coccidia: Toxoplasma gondii, Sarcocystis calchasi, Eimeria southwelli, Eimeria gruis, Cryptosporidia varanii, Cryptosporidia serpentis, and Cryptosporidia muris, and tortoise intranuclear coccidia (Alvarez et al. 2013). The following primers were designed with the use of BLASTN (Altschul et al. 1997): forward primer LBcoccF (5′-GTTCTCCGGTACCGCCTT-3′), the reverse primer LBcoccR (5′-CAATGAAGAGCGGAAGGACAG-3′), and the probe LBcoccProbe (5′-[6-FAM] GGTGTGCACTTGGTTGTACG [BHQ1a-Q]-3′). A 10-fold serial dilution of extracted DNA from a representative case was used to measure the standard curve. The 7500 Fast Real-Time PCR System (Applied Biosystems) amplified the reactions with the following protocol: denaturation at 95 C for 20 s, 50 cycles at 95 C for 3 s, finishing with 60 C for 30 s. A control for DNA quality for each sample was added into a separate well with the use of a Eukaryotic 18S rRNA Endogenous Control primer/probe set (Applied Biosystems). The ABI 7500 fast equipment was used to assess target DNA presence/absence and relative quantity.

The density of adrenal lesions that could be associated with protozoal infection was quantified in 10 randomly selected 1,144×858–μm fields, the maximum area available to avoid overlap. For a single juvenile turtle, only two fields could be examined because of the small size of the adrenal gland. A variety of morphologic variants of protozoal-suspect lesions (described in the upcoming text) were counted and compared to consider possible parasite-stage–associated differences in the quantity of parasite DNA present. Statistical analysis of data collected from the qPCR assay and histologic scores was performed with the use of the Kolmogorov-Smirnov test and Spearman correlation in InStat (GraphPad Software, San Diego, California, USA). We considered correlation to be significant for P values <0.05.

Postmortem examination of 21 leatherback turtles found on the Atlantic (n=17), Pacific (n=3), and Gulf of Mexico (n=1) coasts of North America included adrenal gland histology. Seventeen turtles were subadults or adults with CCL>120 cm (Eckert et al. 2012) and four were juveniles (CCLs 14.5, 20.0, 21.5, and 25.0 cm). We identified a traumatic cause of death for 12 cases, including vessel strike (n=5) and entanglement in fishing gear or other material (n=4). Three cases did not have any apparent abnormalities and drowning was suspected based on exclusion of other findings. Six turtles had inflammation involving one or more organs, most often in combination with emaciation. The adrenal glands were grossly unremarkable in all cases.

Adrenal lesions suspected to be the result of protozoal infection were observed microscopically in all 17 subadults and adults, including 10 females and seven males, but were not found in juveniles. In affected adrenal glands, 250–300-μm-diameter round bodies were diffusely distributed throughout the parenchyma and often bulged into thin-walled vascular spaces (Fig. 1A). A prominent outer capsule surrounded large, multinucleated cells with a central core of granular basophilic material and variably sized small eosinophilic globules. The multinucleated cells had lacy, finely vacuolated cytoplasm, indistinct cell borders, and large, oval, pale-staining nuclei. In addition, many cases exhibited regression of these cellular forms into shrunken residual lesions with dissipated central granular material, degeneration and loss of multinucleated cells, and complete replacement by sparsely cellular, homogeneous, eosinophilic stroma (Fig. 1B). Small numbers of lymphocytes surrounded these fibrotic lesions. Lesion density and relative proportions of cellular and regressed forms varied among cases. Notably, the residual forms were the only phase observed in the three turtles from the Pacific Coast. In addition, in three turtles, rare, individual suspected protozoal lesions identical to those found in the adrenal glands were observed by histology in other tissues, including the submucosa of the small intestine and connective tissues of the dermis, phallus, and tongue.

One leatherback from the Atlantic Coast also had a focal adrenal lesion consisting of numerous groups of eight or more 1–2-μm tightly clustered zoites, resembling meronts with merozoites, interspersed with necrotic material, macrophages, and heterophils (Fig. 1C). These zoites were not observed in any other cases or other tissues, and could not be found in additional sections of adrenal gland from the same turtle.

Electron microscopy revealed that the centers of the round bodies shown in Figure 1A, B were composed of variably sized vesicular profiles and granular material with no discernible cellular structure (Fig. 2). The droplets of brightly eosinophilic material observed under light microscopy were electron-dense aggregates on electron microscopy. The central vacuole was surrounded by a multinucleated cell composed of a variably wide collar of cytoplasm that was elaborated into multiple, radially oriented protuberances. At the base of these protuberances were nuclei with dispersed clumps of heterochromatin, often located along the inner aspect of the nuclear membrane, and occasionally a nucleolus. These cytoplasmic protuberances had undulating plasma membranes that formed numerous, slender, microvillous-like projections. The cytoplasm contained many vesicles and occasionally whorls of membranous material, resembling myelin figures. The surrounding capsule was composed of admixed fibrillar and granular material.

Small numbers of unsporulated coccidian oocysts were detected by centrifugal flotation in a fecal sample that was opportunistically collected from a live leatherback turtle (Fig. 1D). Three were measured and were 26×25, 25×20, and 35×22 μm. Oocysts were recovered from flotation and suspended in 15 mL of either artificial seawater or 2% potassium dichromate. Air was circulated through the suspension for 5 d, but sporulation did not occur. These oocysts were included in genetic analyses for comparison with parasite sequence amplified from the adrenal glands. Fecal samples were not available for flotation from deceased leatherbacks.

Samples used for genetic analyses included adrenal gland from 10 turtles, including eight subadults/adults and two juveniles, and frozen feces from the live leatherback. The PCR amplification and sequencing of a partial segment of the eucoccidian 18S rRNA gene from the adrenal of index case: RAB2006120201 yielded a 1,752–base pair (bp) product, excluding primer sequences (GenBank accession KT956976). An overlapping 830-bp fragment amplified from the fecal oocysts exhibited 98.67% sequence homology with the index adrenal sequence (GenBank accession KT956977).

No protozoal DNA was detected in adrenal samples from two small juvenile turtles that lacked histologic lesions. With the use of the seminested assay targeting the V4 region, 447-bp 18S sequence identical to the index case was amplified from cryopreserved adrenal gland from six of eight subadult/adult turtles with suspect protozoal lesions. A seventh case with adrenal lesions yielded a mixed sequence. Upon close review of the chromatograph, ambiguities corresponded to specific differences between sequences derived from the adrenal glands and fecal sample (at eight of 447 positions), suggesting concurrent infection of the adrenal gland by two distinct, but closely related, protozoa. The case with microscopic adrenal lesions in which no sequence was amplified was a Pacific leatherback with small numbers of acellular lesions consistent with the scarred forms observed in other animals. Because this was the only frozen adrenal sample available from a Pacific turtle, genetic sequence was only obtained from Atlantic leatherbacks. In addition, cryopreserved tissue was not available for PCR testing from the turtle with zoites in the adrenal gland. Extraction of DNA from formalin-fixed, paraffin-embedded tissues was attempted in this case, but the DNA was too degraded, as indicated by lack of amplification of eukaryotic 18S rRNA endogenous control in the qPCR.

Results of Bayesian analysis showed that the sequences obtained from the leatherback adrenal glands and fecal sample were the most closely related and formed a separate clade within the Eimeriidae (Fig. 3).

The qPCR standard curve was linear with an R² value of 0.998 and a slope of 3.45, indicating an efficiency of 95.03%. None of the other coccidia used for evaluation of assay specificity, including the oocysts from the leatherback fecal sample, were amplified. Despite the high homology of the leatherback adrenal and fecal coccidian sequences, differences at the 3′ ends of both primers resulted in specific binding of the targeted adrenal sequence and did not amplify the fecal-derived sequence.

Sufficient adrenal samples were available for seven cases (six subadult/adult turtles with lesions and one juvenile without lesions) to be included here. When the qPCR data was correlated with results from microscopic examination, the K-S test showed that the data were not normally distributed. When compared with the use of Spearman's correlation, histologic density and copies detected by qPCR were significantly correlated. This correlation was significant when all morphologies of the suspected protozoal round bodies were considered, and when the fibrotic forms were excluded.

Consistent amplification of coccidian-specific genetic sequence from adrenal glands with protozoal-suspect lesions, inability to detect this sequence in unaffected adrenal glands, and correlation between suspect parasite lesion density on histology and the quantity of targeted coccidian DNA detected supports our hypothesis that the adrenal lesions are the result of coccidial infection. The detection of one of the sequences from an affected adrenal and from unsporulated oocysts in feces also supports this conclusion. However, absence of clear morphologic features of protozoa development in the most commonly encountered forms leaves unanswered questions with regard to the identity of the taxa beyond the level of family.

We assume that the capsule and cellular component are pathologically altered host cells. The granular material contained within the lesions lacked any intact parasitic stages and may be residual degenerate material from previous parasitic infection. Our interpretation is that these lesions may represent end-stage meronts based on the absence of intact protozoa and apparent regression into fibrotic or scarred lesions. The productive phase of infection simply may have been completed by the time of the death of the turtle in most cases. Unfortunately, the leatherback with zoite formation could not be definitively linked with the other cases because of degradation of DNA and lack of material for additional study. In situ hybridization would be required to localize eucoccidian DNA within the adrenal lesions more clearly, and would be a worthwhile effort, but likely it would require a different genetic target than 18S sequence to achieve specific detection.

Coccidian development within the adrenal glands has been described in Virginia opossums (Didelphis virginiana) infected with Besnoitia darlingi and in golden hamsters (Mesocricetus auratus) experimentally infected by Besnoitia jellisoni (Frenkel 1977; Smith and Frenkel 1977). The reason for apparent adrenal tropism in leatherbacks is unknown. If adrenal infection is productive, stages likely would erupt into venous circulation given the perivascular location of the lesions. Possible co-infection of the adrenal glands in one turtle with the genotype derived from fecal oocysts, and relatedness between the organisms suggest that there could be an alimentary component to the life cycle that we did not detect on histology, possibly because of late stage of infection, limited tissue distribution, or low intensity of gastrointestinal parasitism. This explanation is speculative given our inability to demonstrate shedding oocysts and adrenal parasitism in the same individual host. It is also possible that the fecal sample was contaminated by urine and that oocysts originated from the urinary tract; however, we observed no protozoa by histology in the kidneys or urinary bladder in any deceased turtles.

The two coccidian parasites we found in leatherbacks form a distinct clade (Fig. 3). Although these organisms are closely related, the differences were greater than those among Hammondia, Neospora, and Toxoplasma, which suggests interspecific variation, and not simply different stages of the same organism. The leatherback coccidian clade is phylogenetically near the genera Eimeria and Caryospora, which includes C. cheloniae of green turtles as currently classified based on morphology (Leibovitz et al. 1978; Gordon et al. 1993). Without sporulated oocysts, we cannot propose taxonomic identity.

The occurrence of adrenal lesions in all examined subadult and adult turtles suggests that infection is relatively common and occurs in both the Atlantic and Pacific leatherbacks. We cannot draw conclusions on potential health effects from adrenal parasitism, but there was no evident relationship between parasitism, cause of death, and other indicators of general health, such as nutritional condition. A large proportion of turtles were nutritionally robust animals killed by sudden traumatic events. The turtle with the most intensely affected adrenal glands was in good nutritional condition when it drowned in a trawler net and the turtle had no other significant abnormalities.

There are numerous avenues for potential future study that hopefully will be facilitated by this work. Necropsy and parasitologic examination of additional animals are necessary to identify additional stages of infection, which is critical to characterize tissue distribution, parasite propagation, pathophysiology, and shedding, and potential pathologic significance for the turtle host. The qPCR assay we describe provides a cost-effective, sensitive, specific, and efficient means of surveying tissues, feces, and other samples from live and dead turtles. In addition, genetic characterization of coccidia from leatherbacks collected from the North Atlantic provides a valuable reference for other regions, and a basis for defining global distribution and diversity.

We thank participants in the Sea Turtle Stranding and Salvage Network (US) and the Canadian Sea Turtle Network (Kathleen Martin), Department of Fisheries and Oceans Canada, Population Ecology Division (Mike James and officers of the Conservation and Protection Division), Florida Fish and Wildlife Conservation Commission, staff and volunteers of New England Aquarium, the Marine Mammal Center, Moss Landing Marine Laboratories Marine Operations, Jennifer Keene, Robert Prescott, Kara Dodge, Erin Burke, Michael Dodge, Mark Leach, George Purmont, Molly Lutcavage, Sea Rogers Williams, Bridget Dunnigan, Don Lewis, Michael Moore, Darlene Ketten, Betty Lintell, Kate Sampson, Scott Benson, Craig Harms, Emily Christiansen, Sue Barco, Matthew Godfrey, International Fund for Animal Welfare, Massachusetts Department of Marine Fisheries, Massachusetts Audubon Society, Provincetown Center for Coastal Studies, Paul Doshkov and staff of the Cape Hatteras National Seashore, and the California Department of Fish and Wildlife. We thank Chris Gardiner for review of earlier manuscript drafts and Stephen Daniels for the electron microscopy services. S.D.F. was supported by the Morris Animal Foundation Summer Student Scholar Program. Partial funding was provided by US Department of Commerce, National Oceanic and Atmospheric Administration, National Marine Fisheries Service (NA10NMF4720028). Live turtle capture was conducted under the authority of the National Marine Fisheries Service Endangered Species Act Section 10 Permit 15672, and approved by the Institutional Animal Care and Use Committees of the New England Aquarium (06-03) and the University of Massachusetts (2010-0019).