Free-living and captive turtles and tortoises as carriers of new Chlamydia spp.

A variety of Chlamydia species belonging to the Chlamydiaceae family have been reported in reptilian hosts but scarce data about their occurrence in turtles and tortoises are available. In this study, research was conducted to acquire information on invasive alien species (IAS) of turtles and indigenous turtles and tortoises, living both free and in captivity, as possible reservoirs of Chlamydiaceae. Analysis of specimens (pharyngeal and cloacal swabs and tissues) from 204 turtles and tortoises revealed an overall Chlamydiaceae prevalence of 18.3% and 28.6% among free-living and captive animals respectively, with variable levels of shedding. Further testing conducted with a species-specific real-time PCR and microarray test was unsuccessful. Subsequently sequencing was applied to genotype the Chlamydiaceae-positive samples. Almost the full lengths of the 16S rRNA and ompA genes as well as the 16S-23S intergenic spacer (IGS) and 23S rRNA domain I were obtained for 14, 20 and 8 specimens respectively. Phylogenetic analysis of 16S rRNA amplicons revealed two distinct branches. Group 1 (10 specimens), specific to freshwater turtles and reported here for the first time, was most closely related to Chlamydia (C.) pneumoniae strains and the newly described Candidatus C. sanzinia. Group 2 (four specimens), detected in Testudo spp. samples, showed highest homology to C. pecorum strains but formed a separate sub-branch. Finally, molecular analysis conducted on positive samples together with their geographical distribution in places distant from each other strongly suggest that Group 1 specimens correspond to a new species in the Chlamydiaceae family. In-depth studies of Chlamydia spp. from turtles and tortoises are needed to further characterise these atypical strains and address arising questions about their pathogenicity and zoonotic potential.


Introduction
Chlamydiae are bacteria belonging to the Chlamydiaceae family. Their host specificity varies: e.g. C. caviae is found in guinea pigs, whereas C. pneumoniae, C. pecorum or C. abortus can infect different animals as well as humans. Besides the well-known Chlamydia species, new species and genotypes are constantly being discovered in different hosts and environments [1-a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 6]. The first report of a chlamydial agent in a reptile, the eastern fence lizard (Sceloporus undulatus), dates back to 1944 [7]. Since then, Chlamydiaceae have been observed among more reptiles including the Burmese python [8,9], emerald tree boa [10], western sand viper [11], chameleon [12], crocodiles (Nile, Indo-Pacific and Siamese) [5,13,14] and tortoises belonging to the Testudinidae family [15,16]. Species determination in Chlamydiaceae-positive samples was not always possible but retrospective study and the newest research have shown C. pneumoniae presence in snakes and chameleon, and other yet unclassified Chlamydia spp. in the remaining cases [17]. Recently, candidate species C. sanizinia was reported for the first time in a Madagascar tree boa (Sanzinia madagascariensis volontany) [3].
Threats arising from the introduction of invasive alien species (IAS) are well documented in different environments around the globe [18,19]. Pond sliders (Trachemys (T.) scripta) are turtles native to North America which have been imported in their thousands to European countries in recent years as companion animals. Numerous reports on their introduction to natural environments in different locations, e.g. France, Italy, Switzerland, and Japan have been published and cases of successful reproduction have been confirmed [20,21]. The pond sliders (including three subspecies, namely the red-eared slider (T. s. elegans), the Cumberland slider (T. s. troostii) and the yellow-bellied slider (T. s. scripta)) together with the common snapping turtle (Chelydra serpentina), the painted turtle (Chrysemys picta) and the false map turtle (Graptemys pseudogeographica) are now listed as reptilian IAS not only in Polish [22] but also in European Union legislation [23,24]. Negative impact of IAS on biodiversity may manifest in predation or competition with resident species and their often being a reservoir of new viral and/or bacterial pathogens such as the Chlamydiaceae or Enterobacteriaceae families [25]. The main threat in Poland is posed to the endangered and strictly protected European pond turtle (Emys (E.) orbicularis), the only native freshwater turtle. Significant populations of IAS, mainly T. scripta, have been observed in south-eastern regions for at least 10 years (B. Gorzkowski, personal communication). These species share an ecological niche with E. orbicularis and compete for food and basking places. Reptiles can often be found in recreational areas, such as zoos, parks and lakes, and their popularity as pets is ever-increasing nowadays. Taking into account the occurrence of Chlamydiaceae in these poikilothermic animals, zoonotic implications should also be considered [9,11]. Thus, a survey was conducted in Poland to acquire information on both free-living and captive turtle (semi-aquatic animal) and tortoise (land animal) species as possible reservoirs of Chlamydiaceae. Positive samples were subjected to further molecular and phylogenetic characterisation.

Project presentation
The present survey is a part of a project titled "Invasive turtle species as a source and vector of animal and human pathogens" started to assess environmental and epidemiological threats posed by IAS turtles and their impact on biodiversity in Poland. This interdisciplinary project included trapping, clinical examination, quarantine, euthanasia and anatomopathological examination.

Samples
Samples were taken from free-living and captive turtles and tortoises from private households or zoo collections during the summers of 2015 and 2016. Cloacal and pharyngeal swabs as well as tissue samples were obtained from free-living, whereas only cloacal swabs were sampled from captive animals. Flocked swabs were used throughout the study. Table 1 summarizes data on results, animal species, free or captive status and type of samples collected. Altogether 204 animals were tested. These included 68 IAS turtles captured in Lubelskie voivodeship, eastern Poland (Fig 1) as well as three adult European pond turtles (E. orbicularis). These animals were submitted to a cloacal and pharyngeal swabbing (n = 68) during the medical examination at the beginning of quarantine. Seven pond sliders were resampled at the end of this period to follow chlamydiae shedding. Euthanasia was performed with the use of Morbital (Biowet Pulawy, Poland) administered intravenously after at least two weeks of clinical observation  and was followed by necropsy. Tissue samples were collected including specimens from the respiratory (n = 58) and reproductive (n = 57) tracts as well as sections from cloacae (n = 46) of 58 animals. The only adult European pond turtle available for necropsy showed signs of starvation and advanced autolysis and its reproductive tract was not sampled. Additionally, 19 European pond turtle hatchlings which had died in a breeding facility and were submitted for testing by Polesie National Park were included in the study. In this instance, all tissues were pooled together due to the scarcity and bad conservation of the material and were tested as a single sample. Tissue homogenates were prepared in saline solution or PBS at 1:2 dilution and 100 μL of homogenate was used for DNA isolation. Additionally, 44 captive turtles and 89 captive tortoises from private households and zoo collections were sampled. Cloacal swabs were collected from these animals during routine activities following standard procedures. All samples were stored at −20˚C prior to DNA extraction.

Ethics statement
According to Polish and European Union legislation, ethical approval is not necessary to capture and euthanize invasive alien species which threaten ecosystems, habitats or species [24,26]. Ethical approval for animal experiments was also not required according to the Local Ethical Committee on Animal Testing at the University of Life Sciences in Lublin (statement of 14 Oct 2014). The contribution of additional samples delivered by Polesie National Park (letter of 26 Jun 2016) and laboratory testing of samples originating from protected animal species were approved by Regional Directorate for Environmental Protection resolution no. WPN.6401.206.2016MPR.

DNA extraction
DNA isolation was performed on 100 μL of tissue homogenates, pharyngeal or cloacal swabs with a QIAamp DNA Mini Kit (Qiagen, Germany) according to the manufacturer's protocol, with a final elution volume of 200 μL. DNA was aliquoted and stored at −20˚C before use.

Molecular analysis
Real-time PCR. A real-time PCR according to Ehricht et al. [27] was used as a screening method with primers specific for the Chlamydiaceae family. All primers used in the study are listed in S1 Table. The positive control for PCR, C. trachomatis (Genekam, Germany) and DNase-RNase free water (Qiagen, Germany) as the negative control were included in each run. To ensure that no inhibition was observed, an internal amplification control (TaqMan Exogenous Internal Positive Control, Applied Biosystems, USA) was added to all analysed samples according to the manufacturer's protocol. An analytical cut-off value of 38 was selected corresponding to the defined lower limit of detection of the test. Any Ct values above this defined limit were thereafter considered unreliable.
Species designation. All Chlamydiaceae-positive samples were re-tested with real-time PCRs specific for C. pneumoniae and C. pecorum. Primers and probe sets according to Kohlhepp et al. [28] and Pantchev et al. [29] were used.
Microarray. Chlamydiaceae-positive samples from 10 arbitrarily selected animals from among those with the lowest Ct values (Ct<34) were sent to Alere Technologies GmbH (Germany) and subjected to microarray testing.

Phylogenetic analysis
Sequencing data were subjected to a BLAST search against the NCBI database to identify the most similar entries and aligned with a set of selected sequences representative for known chlamydial species. To assess the phylogenetic relationship between Chlamydia spp. and the tested samples, phylogenetic trees were constructed for 16S rRNA (1180bp) and IGS-23S rRNA (920 bp) as well as for ompA (920 bp) using Geneious Pro 8.0 software (Biomatters, New Zealand). The MrBayes method with 1,000 replicates bootstrap was applied. The general time-reversible (GTR) model of DNA substitution was chosen for analysis of the 16S rRNA gene and Hasegawa-Kishono--Yano (HKY85) for 23S rRNA and ompA phylogeny, based on MEGA 6 calculations. An estimation of mean nucleotide distances was prepared with the MEGA 6 software package [34].

Chlamydiaceae detection and level of shedding
The results of Chlamydiaceae-specific real-time PCR in 11 different turtle and tortoise species included in the survey are presented in Table 1 and S2 Table. Examination of samples from 204 animals gave positive results in 51 cases (25.0%). European pond turtles from a breeding facility were also found positive. Cloacal swab specimens collected from turtles and tortoises kept in captivity gave prevalence of 28.6% with 38 positive results. Samples from 48 Horsfield's tortoises kept together were all negative whereas both marginated tortoises and one Hermann's tortoise acquired from different private owners were positive. The presence of Chlamydiaceae was detected in 65.7% (23 out of 35) of Testudo spp. specimens based on the qPCR test. Prevalence of 36.4% was noted among captive T. scripta, most of which came from one collection and had direct contact with each other. Among 71 free-living animals 13 (18.3%) were positive in at least one type of sample. Different ways of shedding were observed in IAS turtles. Chlamydial DNA was detected in both pharyngeal and cloacal swabs in four pond sliders, only in cloacal swab in seven turtles and exclusively in pharyngeal swab in a further two animals. In seven animals positive at the first sampling for the cloacal swab, cloacal tissue was confirmed positive for five specimens.
Variable levels of pharyngeal and cloacal shedding were noted in the course of the study ( Table 2). Seven pond sliders were sampled again at the end of quarantine exhibiting different loads of chlamydial DNA in swab specimens. An increase in pathogen shedding was observed in two animals (nos. 118 and 119) and a maintained level in one (no. 231) at both sampling sites for two time points. A decrease was shown in two cases (nos. 117 and 211), one of which was connected with the animal becoming negative in specimens tested at the second sampling point. However, examination of a further two T. scripta (nos. 116 and 237) at the end of quarantine revealed that chlamydia shedding had stopped in cloacae but had started in the upper respiratory tract. Attempts were also made to confirm if only one Chlamydia strain was present in the animal when multiple samples returned positive results. Thus, ompA sequences were acquired from samples 117A03 (cloacal swab) and 117X83 (respiratory tract homogenate) as well as 213X25 (cloacae tissue) and 213X34 (reproductive system homogenate). Resulting sequences were identical for each of the animals, supporting assumption of a single strain colonization.
Chlamydiae were detected in three out of 58 tissue samples from the respiratory tract (animals nos. 103, 117 and 213) as well as in two samples from the reproductive system (animals nos. 110 and 213). T. scripta no. 213 was the only one with all specimens positive in the Chlamydiaceae-23S assay. The pooled sample of E. orbicularis hatchlings (specimen no. 153) tested positive while three adult European pond turtles which were available for sampling were not carrying Chlamydia spp.

Genotyping and phylogenetic analysis
Microarray testing revealed that two out of 10 randomly selected Chlamydiaceae-positive samples (S2 Table) gave positive signals with both Protochlamydia amoebophila and Protochlamydia naegleriophila probes in microarrays but in the remaining eight cases results were negative. The remaining positive samples were not included in the microarray testing. No positive results were obtained with species-specific PCRs. Thus, sequencing of the ompA gene and ribosomal loci was performed to carry on further molecular characterisation of specimens (S2 Table). Partial ompA amplicons for 17 animals were obtained and aligned with sequences of Chlamydiaceae reference strains and others of reptilian origin. All sequences from the current study precisely overlie sequences within the Chlamydiaceae family and two distinct and clearly separated branches can be observed as shown on the constructed dendrogram (Fig 2). Group 1 includes sequences from freshwater turtles (pond sliders and European pond turtle hatchlings) and is divided into three sub-branches. One (7 amplicons from T. scripta samples KY928243, KY928245, KY928247, KY928249, KY928250, KY928253 and KY928255) shows minor heterogeneity (1 SNP) whereas the remaining two are formed by identical sequences from four (KY928242, KY928246, KY928251 and KY928252) and three animals (KY928239 to KY928241) respectively. In all three cases, the maximum posterior probability value supports branch division. All amplicons in the first sub-branch were of free-living pond sliders, the second also contained a captive animal specimen, and sequences incorporated in the last subgroup were all obtained from captive T. scripta from one collection. The second main branch, namely Group 2, includes three sequences (KY928236 to KY92838) acquired exclusively from Testudo spp. samples. Maximum support was calculated for this group and considerable variability can be noted. Short sequences (380 bp) obtained earlier from nasal lavage fluid of T. marginata (AY845422.1) and T. horsfieldii (AY845419. 1   Sequencing of the nearly full-length 16S rRNA gene was implemented to designate the chlamydiae species. Fourteen amplicons out of 52 positive animals were acquired, successfully sequenced and used for phylogenetic comparison. Sequences from this study were shown to have up to 38 SNPs in a 1180bp fragment of ribosomal loci. BLAST analysis carried out against the GenBank database revealed 96.8% to 97.8% nucleotide homology to C. pecorum strains E58 (NR102975.1) and P787 (CP004035.1) and to the Cand. C. sanzinia strain 2742-308 (CP014639.1). The phylogenetic tree constructed on the basis of the available 16S rRNA fragments is presented in Fig 3A and shows topology almost identical to the one obtained for the ompA gene. Group 1 is monophyletic and gathers 10 sequences (KY928260 to KY928270) from both free-living and captive T. scripta and E. orbicularis hatchlings and is supported by a posterior probability value of 1. The remaining four amplicons (Group 2, KY928256 to KY928259) obtained from captive Testudo spp. are more heterogeneous and show the highest homology to C. pecorum sequences, with up to 27 base pair differences to the Maeda strain (D85715). Posterior probability of 0.88 was calculated for this branch. The genetic distances (S3 Table) of each group described in this study to other Chlamydiales are noticeably larger (0.202-0.254) than those to members of the Chlamydiaceae family (0.028-0.70). Analysis of domain I of the 23S rRNA gene and IGS sequence (Fig 3B) displays a similar pattern to that of 16S rRNA with some differences in branch ordering, though only eight sequences were available.

Discussion
Threats arising from IAS are well documented in different environments. The knowledge of invaders' presence in new habitats, population range, behaviour and carried microflora is important in management and eradication strategies [35]. Thus, this study on Chlamydiaceae dissemination among free-living reptiles captured in lakes and rivers of eastern Poland was performed. The investigation yielded a high prevalence rate of 18.3%. There are no sufficient data available on Chlamydia occurrence in IAS turtles and most published research works concerning reptiles include animals living in captivity or on farms, zoos or in private collections [5,11,14,15,36]. In our study 28.6% of captive turtles and tortoises were found positive in Chlamydiaceae-specific real-time PCR. What is worth noting is that 48 Horsfield's tortoises from a zoo collection did not carry Chlamydiae while prevalence in the T. scripta group kept altogether reached 36.4% and for Testudo spp. was even 65.7%. Similar results were obtained by Frutos et al. reporting a prevalence of 57.9% among reptiles in a recreational park in a central area of Argentina [9]. Contrastingly, in a German study on companion animals of the family Testudinidae only 10.3% of cases were found to be Chlamydiaceae-positive [15]. The high prevalence noted both in our and the Argentinean survey could be due to a shared environment and potential pathogen transfer through personnel tending the animals in tested collections, which was not the case for animals kept separately in private households.
Chlamydiaceae in T. scripta were shed intermittently as shown in Table 2, like other species of this family, e.g. C. psittaci [37]. Also positive results mainly obtained for swab specimens without simultaneous detection in tissue samples can prove them to be commensals or conditionally pathogenic rather than pathogenic microflora, which was suggested by Hotzel et al. [15]. A decrease in animal immune response caused by temperature changes, stress, malnutrition or other pathogens [38,39] combined with Chlamydia colonization may trigger systemic infection. In our study, chlamydial DNA was detected in all collected samples only for pond slider no. 213. This animal was caught in a small water basin without a basking place or way out and died on the third day of quarantine. Poor overall condition with signs of exhaustion was noted during its clinical examination (N. Chlebicka, personal communication). Klebsiella spp. was found in swabs and respiratory tract specimens (A. Kędrak-Jabłońska, personal communication). Moreover, Mycobacterium sp. was identified in internal organ samples but no histopathological changes were observed in the liver (M. Lipiec, personal communication). Severe conjunctivitis, which is a typical manifestation of most Chlamydia spp. infections, was observed in T. scripta no. 213 accompanied by accumulation of fibrin (N. Chlebicka, personal communication). Similar signs of chlamydiae presence were also reported by other authors in crocodiles [14,40] and tortoises [41]. The symptoms described imply that a chlamydial agent is at least one of the components of the disease cause.
High mortality of juvenile European pond turtles was observed at the beginning of 2016 in a breeding facility at Polesie National Park. As an outcome of the investigation into the cause, Chlamydiaceae DNA was detected in animal no. 211 (T. scripta), which was a resident of this facility and was handled by the same personnel as handled freshly hatched European pond turtles. No clinical signs of infection were observed in the pond slider, whereas numerous hatchling deaths were recorded (specimen no. 153). Laboratory testing for Salmonella, Yersinia and Mycobacterium spp. gave negative results (W. Iwaniak and M. Lipiec, personal communication) while the Ct value in Chlamydiaceae-23S real-time PCR was 28.7, indicating high bacterial load in tissues. Sequences of the ompA gene (950 bp) from animal 211's (T. scripta) and 153's (E. orbicularis hatchlings) specimens were aligned and confirmed to be identical. These data could point to transfer of chlamydiae from T. scripta to E. orbicularis hatchlings. The immaturity of the immune system in juveniles or a possible concurrent infection (e.g. viral) could result in the systemic spread of Chlamydiae and a fatal outcome. A similar case was reported by Huchzermeyer et al. in juvenile crocodiles [13]. Uncorroboratively, none of the 13 adult European pond turtles, neither free-living nor captive originating from different locations, tested positive. Detailed research is needed to acquire information about whether E. orbicularis local populations are harbouring Chlamydia spp. or if these bacteria carried by IAS turtles can further endanger protected species.
C. pneumoniae and C. psittaci are the chlamydial species most often encountered in reptiles [8,9,11,42]. Research by Kabeya et al. on zoo animals in Japan showed a significantly higher prevalence of Chlamydia in reptiles than in birds and mammals as well as the dominance of C. psittaci over C. pneumoniae. However, Sariya et al. [5] confirmed the presence of Chlamydia sp. most closely related to C. caviae in Siamese crocodiles (Crocodylus siamensis) on farms in Thailand. The results of the current study confirm that the variety of Chlamydia spp. encountered in reptilian hosts is even broader. Phylogenetic analysis of ribosomal gene sequences from these specimens allowed the distinction of two genotypes. Group 1 seems to be specific to semi-aquatic species, namely T. scripta and E. orbicularis. To date, no related sequence from other hosts has been found in GenBank. The 16S rRNA homologies of these sequences with C. pecorum E58 and Cand. C. sanzinia are 96.8% and 97.4% respectively. Moreover Group 1 clearly branches away from both mentioned species. In support of these data, 23S rRNA sequences as well as the distribution of this group of specimens in geographically separated places also suggest their classification as new species in the Chlamydiaceae family.
Group 2 on the other hand is most closely related to C. pecorum strains but still forms a separate sub-branch. All specimens included in this group were detected in Testudo spp. These data are in accordance with previous results published by Hotzel and colleagues [15]. Amplicons from our survey belong to a branch displaying the same phylogenetic position on the 16S rRNA and ompA dendrograms and reinforce assignment of the agent, as suggested in previous publications, to a new species with tortoises as hosts. Two sequences, namely 16-10076 and 16-10083, share 100% similarity across 214 bp of the 16S rRNA signature region with sequence AY845424 from a previous report, although corresponding ompA fragments display notable variation [15].
Although available phylogenetic data are sufficient for preliminary species recognition, in both cases further work is needed for an extensive characterisation of turtle and tortoise-specific Chlamydiaceae. Attempts were made at strain isolation but no successful growth was observed. Difficulties in culture may be due to the poikilothermic character of reptiles influencing the microbiome of these animals and the requirements of strongly adapted bacteria. Thus, adjustments of standard isolation procedures used for chlamydial strains infecting higher vertebrates have to be made.
Many earlier reports draw attention to zoonotic risks connected with reptiles kept as companion animals. Salmonella spp. [43,44] and Mycobacterium spp. [45] were found in captive animals sometimes representing a serious hazard for people's health [46][47][48]. It can also be the case with Chlamydia spp., as has already been suggested, especially when C. pneumoniae and C. psittaci were noted. Nonetheless our knowledge of newly described species is insufficient to establish their zoonotic implications, which still cannot be excluded. The numbers of reptiles imported to Poland in recent years are overwhelming and a considerable part of them might at some point be freed into the wild, joining the IAS population and consequently disturbing the local biodiversity. At the same time they can pose a direct threat to the public, as lakes and rivers often used as recreational zones are usually chosen for the release of the turtles.
Our research revealed the presence of two different chlamydial agents in turtles and tortoises with a considerable level of shedding for some of them. Phylogenetic analysis showed that even though a variety of Chlamydia spp. was already reported in reptiles, Group 1 encountered in semi-aquatic turtles is reported here for the first time. It is not known if the Chlamydiaceae described in our research was already present in the population of the endangered native species E. orbicularis or otherwise if it was transferred from an alien species of T. scripta. Moreover, no detailed data on the microbiome of IAS turtles is available, thus the zoonotic character of more pathogens than only chlamydial species has to be kept in mind. Further studies are needed to expand our knowledge on all presented issues.
Supporting information S1