Author Correction: Novel Chlamydia species isolated from snakes are temperature-sensitive and exhibit decreased susceptibility to azithromycin

Chlamydia species have recently been recognized as emerging pathogens in snakes. However, isolation of novel snake chlamydiae is critical and their growth characteristics are largely unknown. In this study, two novel chlamydial species are described: Chlamydia serpentis and Chlamydia poikilothermis, isolated after attempts on 23 cloacal and choanal swabs from 18 PCR-positive captive snakes originating from different Swiss snake collections. Isolation success, growth curve and infectivity rates over a 48-hour time period were dependent on temperature (37 °C for C. serpentis, 28 °C for C. poikilothermis). C. serpentis and C. poikilothermis were sensitive to tetracycline and moxifloxacin during evaluation by in vitro antibiotic susceptibility assay but intermediate to resistant (2–4 μg/ml) to azithromycin. Whole genome sequencing of the isolates provided proof of the novel species status, and gives insights into the evolution of these branches of genus Chlamydia.

Phenotypic characterization of field isolates from captive snakes. Growth experiments were performed at 28 °C and 37 °C with and without cycloheximide for the four snake isolates, using the human C. pneumoniae strain K6 as a control. Initial growth curve experiments were performed using timepoints 24, 32, 40 and 48 hpi (Supplementary Figure S1), but revealed very small inclusions at 24 hpi rendering further evaluation difficult. Further experiments thus used only 32 and 48 hpi. Infected monolayers were evaluated for titre and by IF to determine inclusion morphology and size, at 32 and 48 hpi (from early reticular body (RB) phase to mature inclusions). Regardless of timepoint and temperature, the C. pneumoniae strain K6, isolates H15-1957-3C and H15-1957-10C formed larger inclusions at earlier timepoints when cycloheximide was added (Fig. 1) as compared to S15-834C and S15-834K (Fig. 2). At 37 °C, the C. pneumoniae strain K6 (Fig. 1, panel a-d), H15-1957-3C (Fig. 1, panel i-l) and H15-1957-10C (Fig. 1, panel q-t) formed round, regular inclusions over the course of 48 hours. At 28 °C, inclusions remained smaller at both investigated timepoints and contained mostly RB forms (Fig. 1, panels e-h, m-p, u-x).  Table 1. Details of snakes positive by real-time PCR for Chlamydiaceae. Samples were collected from asymptomatic alive snakes or taken at necropsy from dead snakes. 1 C = swab samples from choana; K = swab sample from cloaca. 2 Samples diluted 1:10 for analysis. 3 UT = undetermined. 4 Suspected genus/species based on Arraymate probes. 5 Possible false-positive real-time PCR result.
Titration by sub-passage confirmed IF observations, showing that C. pneumoniae K6, H15-1957-3C and H15-1957-10C are significantly more infectious when grown at 37 °C compared to 28 °C. The addition of cycloheximide increased the infectivity in these strains (Fig. 4a). In contrast, the infectivity of S15-834C and S15-834K EBs was decreased after growth at 37 °C compared to 28 °C, and cycloheximide inhibited growth of these strains (Fig. 4b).
TEM of all isolates was performed at 48 hpi at 28 °C and 37 °C. The snake isolates displayed inclusions at various stages of the developmental cycle, similar in ultrastructure to C. pneumoniae strain K6 inclusions. In all four snake isolates, putative EBs are round, electron dense and, 0.25-0.5 nm in diameter. RBs are larger (0.5-1 mm), round to oval and more electron lucent. Binary fission of RBs and, occasionally, intermediate bodies (IBs, electron dense centre and more lucent periphery) were observed. Fully developed inclusions containing EBs, RBs and IBs were observed for the isolates at their respective optimum temperature, namely 37 °C for C. pneumoniae K6, H15-1957-3C and H15-1957-10C (Fig. 5, left panel) and 28 °C for S15-834C and S15-834K (Fig. 5, right panel). Interestingly, while inclusions at 28 °C resulted in small, RB-dominant inclusions for C. pneumoniae K6, H15-1957-3C and H15-1957-10C (Fig. 5, right panel), S15-834C and S15-834K also formed 2 µm and larger sized RB-like structures, which were interpreted as aberrant bodies (ABs; Fig. 5, left panel).

Figure 2.
Immunofluorescence images of S15-834C and S15-834K at 32 and 48 hours post infection (hpi). Shown are immunofluorescence images illustrating the morphology of S15-834C (a-h) and S15-834K (i-p) at Antibiotic susceptibility to tetracycline, moxifloxacin and azithromycin was evaluated for the four snake isolates. The control C. pneumoniae strain K6 was sensitive to all tested antimicrobial substances with MIC/MBC values ranging from 0.06-0.5 μg/mL ( Table 2). All snake isolates were equally sensitive to tetracycline and moxifloxacin but showed an intermediate to resistant phenotype to azithromycin ( Table 2).
Identification of the novel snake isolates. Initial identification using 16S rRNA genotype indicated that strains H15-1957-3C, H15-1957-10C are most closely related to C. pneumoniae (99% identity), whereas S15-834C and S15-834K are closer to C. psittaci, C. caviae and C. abortus (99% identity to each). Due to limitations with the resolution of 16S rRNA gene identification, and to gain further insights into the phylogenetic position . The average inclusion size of C. pneumoniae K6, H15-1957-3C and H15-1957-10C is significantly decreased following incubation at 28 °C compared to 37 °C. Shown is a boxplot comparing the inclusion size (µm 2 ) distribution for strains a) C. pneumoniae K6, H15-1957-3C and H15-1957-10C, and b) S15-834C and S15-834K following 48 h of incubation at 28 °C (dark grey) or 37 °C (light grey). Filled circles represent outliers (>1.5x interquartile range), while asterisks represent extreme values (>3x interquartile range). The inclusion size was determined using a Leica DMLB fluorescence microscope (Leica Microsystems, Wetzlar, Germany) and a UI-2250SEC-HQ camera (uEye, IDS Imaging Development Systems GmbH, Obersulm, Germany) and analysed with the BonTec measuring and archiving software (BonTec, Bonn, Germany). Boxplots were created by the SPSS Statistics software. An unpaired t test was used for statistical analysis, and a p-value of less than 0.05 was considered significant.   and lifestyle of these bacteria, WGS was then performed on the strains used in the study, followed by finishing to generate complete chromosomes and manually improved automated annotation.
Using an established scheme based on nine conserved taxonomically informative gene products 16 , it was determined that the strains belong to two new species of the genus Chlamydia, with strains H15-1957-3C and H15-1957-10C most closely related to C. pneumoniae, and S15-834C and S15-834K most closely related to C. caviae (Fig. 6). Digital DNA-DNA hybridization (dDDH) analysis comparing their genomes to their respective closest relative species also indicated that these isolates represent novel species (22.6% and 30.9% respectively; Table 3). A scheme using pairwise comparisons of the nine gene products used above among all Chlamydia species, also confirmed these as new species 16 (Supplementary Figure S2).
Whole genome characteristics of novel snake species. Features of the genomes of these novel snake species are given in Table 3, compared against the most closely related species, and the other sequenced species from a snake host, Ca. C. sanzinia 14 . Both novel genomes were compared to representative genomes of fifteen species of the genus Chlamydia ( Fig. 7a and b). Some regions of the chromosome exhibited lower sequence conservation as compared to other Chlamydia genomes, including the PZ, pmp operons and tandem arrays of repetitive genes (Fig. 7). Both the new species contain plasmids.
The genome of C. serpentis isolate H15-1957-10C is syntenic with that of the comparator C. pneumoniae CWL029, but has many whole gene differences, the overwhelming majority of which encode hypothetical proteins, including some families with Chlamydia-specific domains of unknown functions (DUF) (Supplementary Table S1). It also possesses fewer predicted pseudogenes than the genome of CWL02 (Table 3).
The genomes of the novel species C. serpentis and C. poikilothermis each contain pmp genes in four clusters in the same genomic locations (Fig. 8). The genome of C. pneumoniae contains the highest number of pmp gene copies, but also the greatest number predicted to be inactivated.
Much of the variation between strains occurs within the plasticity zone (PZ), also known as the replication termination region (RTR). While both the novel snake species have accBC and guaAB-add gene clusters, often used to describe the ends of the PZ, neither carry genes encoding cytotoxins, phospholipase D, or the tryptophan operon. The C. serpentis genome of H15-1957-10C possesses a 2.5 kb putative MAC/perforin gene at Csp10C2_00153, and appears to have a PZ more intact than that found in the genomes of C. pneumoniae strains. In comparison, the C. poikilothermis genome of S15-834K has a reduced PZ relative to that within the genome of C. caviae GPIC.
To rationalize the azithromycin results for the snake strains, alignments of the 23S rRNA gene were performed, using sequences extracted from the genomes of S15-834K, H15-1957-10C, K6 and C. caviae GPIC. Of these, the former two possess intermediate resistance, and the latter two are sensitive. A single base substitution showing a distribution between these two groups was found, at base 1230 (AAGGAGTA/GCTGGAGC), which does not correlate with previous findings in which mutations at 2058 and 2059 have been implicated in conferring azithromycin resistance 17,18 .

Discussion
Evidence for the existence of two additional members of the genus Chlamydia. Analysis    FtsK < 98%, PepF < 96%, Adk < 95%, HemL < 95%) of the the Chlamydia genus (DnA ≥ 70%, SucA ≥ 64%, Hyp325 ≥ 57% and Fabl ≥ 78%) 16 . Whole genome average nucleotide identity (ANI) analysis of the two new species with their closest relatives are also lower than the 95% generally considered to delineate bacterial species 19 ( Table 3). Previously, 16S and 23S rRNA sequences 20 or multilocus sequence approaches 21 were used to assign new chlamydial species. The scheme used in this study to classify novel snake isolates is based on a set of highly reliable protein sequences that were shown to be good markers of whole genome relatedness 16 . Comparative genomic analysis shows that C. serpentis, most closely related to C. pneumoniae, appears to have a less disrupted genome than the latter, having a more extensive PZ and fewer pseudogenes, perhaps therefore being more representative of the common ancestor of the two species. C. poikilothermis, in contrast, has a more reduced PZ than the closest comparator C. caviae. Comparing with the genomes of other known Chlamydia found in snakes 14,15 , none carry the cytotoxin or trp operon in the PZ, possibly implying that this improves their colonisation of snakes. Presence of the purine biosynthesis genes guaAB-add, however, is not required for growth within snakes, as this operon is absent from Ca. C. sanzinia.
Both novel species behaved intermediate to resistant to azithromycin. According to Ellington et al. 22 and Tagini et al. 23 , the genotypic to phenotypic congruence for antibiotic susceptibility may vary quite extensively from species to species and from antibiotics to antibiotics. Thus, the fact that these new chlamydia-related species exhibits a low susceptibility to azithromycin in absence of the typical mutations described in other species is somehow expected and highlights the limitation and difficulties faced when trying to infer AB susceptibility from genomic data. H15-1957-10C). C. serpentis (serpentis L. fem. Gen. pl. serpentum, of the snake, because snakes are the currently known host).

Description of C. serpentis sp. nov. (
C. serpentis strains occur in snakes belonging to the families Colubridae and Viperidae and were isolated from captive asymptomatic Patherophis guttatus and Atheris squamigera. The presence of the agent in other snake species as well as in free-ranging snakes and even other reptiles seems possible, but has yet to be investigated. C. serpentis can be recovered from choanal and cloacal swabs and might be also detected in inner organs of infected reptiles. The natural route of transmission and potential reservoirs are unknown to date. The carrier snakes were clinically asymptomatic, but a facultative pathogenic role has to be considered in concert with other bacterial or viral infections, or induced by stress due to capture and transportation, high-density farming and hibernation 12 . The potential for zoonotic infection of humans, in particular snake owners, is unknown. C. serpentis can be grown in LLC-MK2 cells, a rhesus monkey epithelial kidney cell line, which has been successfully used to isolate C. suis strains from fecal swab samples 24 and is able to survive and replicate at lower temperature such as 28 °C and 12 °C. The replication of C. serpentis is enhanced by adding cycloheximide after the infection to block de novo host protein synthesis similar as shown for C. pneumoniae K6. Shape, size and distribution of inclusions including their production of infectious EBs measured as IFU per mL at 32 and 48 hpi resemble those seen in C. pneumoniae-infected LLC-MK2 cells. Size of inclusions and productivity of C. serpentis at 28 °C is diminished compared to 37 °C in line with C. pneumoniae K6 in this study. By TEM, the typical bi-phasic developmental cycle can be observed for C. serpentis including EBs and RBs comparable in size and morphology to C. pneumoniae. C. serpentis is susceptible to tetracycline and moxifloxacin but has an intermediate sensitivity of azithromycin (ranging from 2 to more than 4 μg/mL). The type strain is H15-1957-10C T . Two strains H15-1957-10C (DSM 106151) and H15-1957-3C (DSM 106152) have been deposited at the DSMZ (Deutsche Sammlung Description of C. poikilothermis sp. nov. (S15-834K). C. poikilothermis (poikilothermis adj., of a poikilotherm species, because this species was isolated from a poikilotherm reptile which is an organism whose internal temperature varies considerably).
Two highly similar strains of this species occurred in a captive Pantheropis guttatus belonging to the family Colubridae. The agent can be recovered from choanal and cloacal swab and is possibly also present in other snake families, captive or free-ranging. A pathogenic potential cannot be differentiated from the cause of death (salmonellosis) in the actual case. The mode of transmission and zoonotic potential are unknown. Like other Chlamydiaceae species, C. poikilothermis can be isolated and grown in cell culture but requires lower temperatures such as 28 °C. Isolation at 37 °C is less successful, growth curves (Supplementary Figure S1) over time show the ability of C. poikilothermis to replicate at 37 °C but inclusions are significantly smaller and morphologically similar to ABs. The replication of C. poikilothermis is not enhanced by adding cycloheximide. It grows better in the absence of cycloheximide regardless of the temperature (28 °C, 37 °C). By IF, the inclusion morphology at 28 °C is heterogenous and inclusions tend to grow around host cell nuclei. The ultrastructural features of C. poikilothermis display EBs and RBs replicating by binary fission.

Material and Methods
Snake collections, sample collection and DNA extraction. Samples (n = 23) investigated in this study comprised choanal and cloacal swabs (FLOQSwabs ® , Copan Italia, Brescia, Italy) taken from 18 captive snakes belonging to the families Boidae, Colubridae, Pythonidae and Viperidae. The majority of the snakes was sampled in a previous study 11 and included six private snake collections in Switzerland. These snakes were clinically inconspicuous at the time of sampling. Samples were collected as dry swabs for DNA extraction and subsequent chlamydial screening. Chlamydia-positive snakes were sampled again, whereas swabs were stored in sucrose phosphate (SP) transport medium at −80 °C for isolation as described 24 . Additionally, two individual captive snakes, which had been submitted to the Institute of Veterinary Pathology (Vetsuisse Faculty, University of Zurich) for diagnostic purposes, were sampled during necropsy. Two swabs were collected per snake and anatomical location, of which one swab was stored in SP medium.
Chlamydial screening. DNA of dry swab samples was extracted using the QIAamp DNA mini kit (Qiagen, Hilde, Germany), following the supplier's recommendations. Extracted DNA of all samples (n = 23) was examined using real-time PCR based on Chlamydiaceae family-specific 23S rRNA gene primers performed on an ABI 7500 instrument, as previously described 25 including internal amplification controls 26 . All samples were tested in duplicate and the cycle threshold was set at 0.1 for each run. A mean cycle threshold (Ct value) < 38 was considered positive, and was used to calculate the corresponding chlamydial load as Chlamydiaceae 23S rRNA gene copy number per µl. If the amplification of internal control DNA was inhibited, the run was repeated following 1:10 dilution of the sample. A positive control containing a sevenfold dilution series of C. abortus DNA and a negative control of water instead of the template DNA were included in each run 27 .
All samples (n = 23) were further investigated using a species-specific 23S rRNA Arraymate microarray assay (Alere, Jena, Germany), as established by Borel et al. 28 . The current version carries 34 probes for eleven Chlamydiaceae species, three genus-specific probes, four family markers and 15 probes for Chlamydia-like organisms. Additionally, there are four internal control DNA probes and an internal staining control (biotin marker) 27 . Each sample, including internal control DNA (Intype IC-DNA, Qiagen Labor, Leipzig, Germany), was amplified and biotin-labeled using a biotinylation PCR, as described by Borel et al. 28 , with 10 minutes (min) of initialization (96 °C) and 40 cycles of 94 °C (denaturation), 50 °C (annealing), and 72 °C (elongation) for 30 seconds each. 2-4 µl of amplification product was loaded on the chip, which was processed according to manufacturer's instructions.
To determine the mean inclusion size, images were taken using a Leica DMLB fluorescence microscope (Leica Microsystems, Wetzlar, Germany) and a UI-2250SEC-HQ camera (uEye, IDS Imaging Development Systems GmbH, Obersulm, Germany) and analysed with the BonTec measuring and archiving software (BonTec, Bonn, Germany). Depending on the inclusion size, a magnification of 1000x (oil immersion) or 40x was chosen and the area was measured with either the "Fläche(Kreis) [Area(circle)]" or "Fläche(Polygon) [Area(polygon)]" function depending on the shape of the inclusion. The area of at least 50 inclusions was measured from one or two coverslips in one or two independent experiments. Boxplots were created with the SPSS Statistics software. The QuickCalcs unpaired t test of the GraphPad software (https://www.graphpad.com/quickcalcs/ttest1.cfm) was used for statistical analysis.
For titre analysis, infected monolayers were scraped into 1 ml fresh infection medium at 32 and 48 hpi. Supernatant and scraped cells were pooled before storage at −80 °C. Chlamydial titration by sub-passage was evaluated to determine the infectivity of the isolated strains at their respective optimum temperature (C. pneumoniae K6, H15-1957-3C and H15-1957-10C at 37 °C; S15-834C and S15-834K at 28 °C) after 48 hpi. As previously described 24 , samples were vortexed for 1 min before serial dilution in incubation medium and subsequent infection of the prepared LLC-MK2 cells was carried out as described for infection of host cells. Fixation and immunostaining was performed as described for IF. The number of inclusions in 30 random microscopic fields for duplicate coverslips per condition was determined using a Leica fluorescence micro-scope at 200x magnification with a 20x objective (PL FLUOTAR 20x/0.50 PH 2, '/0.17/B) and a 10x ocular objective (Leica L-Plan 10x/25 M, Leica Microsystems). Inclusion forming units (IFU) per ml of undiluted inoculum was then calculated.
For TEM analysis, cells were fixed with 2.5% glutaraldehyde (Electron Microscopy Sciences, Ft. Washington, USA) for 1 h and embedded in epoxy resin (Fluka; Sigma-Aldrich).
Briefly, we collected chlamydiae-positive cultures in SP media and determined the number of inclusion forming units per ml (IFU/ml) via 10-fold dilution as described 24,30 Each antibiotic susceptibility determination was performed with approximately 5 × 10 3 IFU/mL per investigated isolate 30,31 . Following inoculation of sixteen confluent monolayers and centrifugation (1 h, 2385 g, 33 °C), inocula were replaced with incubation medium containing a serial two-fold dilution of the antimicrobial agent in question. After a 48-hour incubation period at their optimum temperature (28 °C or 37 °C), monolayers were fixed in methanol for 10 min, immunolabelled and processed with a fluorescein-conjugated monoclonal antibody specific for the chlamydial LPS genus-specific antigen (IMAGEN Chlamydia K610111-2, Thermo Fisher Scientific) as described 24 .
The minimum inhibitory concentration (MIC) of the respective antibiotic substance was defined as "the lowest concentration preventing the detection of more than 90% of the chlamydial inclusions compared with the drug-free control" 30,31 . In parallel, media of the remaining Trac bottles was replaced with antibiotic-free chlamydiae cultivation medium after washing the coverslips with PBS. Monolayers were fixed after 48 h of incubation and immunolabelled for the evaluation of the minimum bactericidal concentration (MBC), which was identical to MIC determination. Chlamydial isolates with an MIC/MBC of 4 μg/mL were defined as resistant, whereas cultures with 2 μg/mL ≤ MIC/MBC < 4 μg/mL were considered intermediate, and isolates with a MIC/MBC of <2 μg/mL sensitive. Identification using 16S rRNA gene analysis. Initial genotyping was performed by sequencing the 16S rRNA gene using the Chlamydiales-specific primer pair 16SIGF (5′-CGGCGTGGATGAGGCAT-3′) and 16SB1 (5′-TACGGYTACCTTGTTACGACTT-3′) 32 targeting almost the entire gene (approximately 1400 bp) as described 11 . PCR products were purified with the QIAquick PCR Purification Kit (Qiagen) according to manufacturer's instructions. Purified DNA was Sanger sequenced by Microsynth (Balgach, Switzerland).
ScieNtific REPORTS | (2018) 8:5660 | DOI:10.1038/s41598-018-23897-z WGS, mapping, assembly and annotation. Sequencing of the four snake isolates (H15-1957-3C, H15-1957-10C, S15-834C and S15-834K) and the human C. pneumoniae isolate K6 was performed on the Illumina Miseq platform with 250 bp paired end reads at the Functional Genomics Center Zurich (FGCZ), following NEBNext library creation. Coverage data is shown in Supplementary Table S2. Assembly was performed using SPAdes in multi-cell mode 33 , followed by ordering of the chlamydial contigs against C. pneumoniae CWL029 (accession number AE001363) or C. psittaci 01DC12 (HF545614) within ACT 34 . The contig order was confirmed, and genomes finished, by amplifying and capillary sequencing across gaps using the primers listed in Supplementary Table S3. The resulting single contig genome assemblies were checked and compared using BWA 35 . The genome of K6 remained in two contigs. Automated annotation was performed using AnnotateBacteria [https://github.com/sanger-pathogens/Bio-AutomatedAnnotation/] with further manual curation in Artemis 34,36 . All read data, with associated assembly and annotation where relevant, has been submitted to ENA under project PRJEB19768.