Genotype data not consistent with clonal transmission of sea turtle fibropapillomatosis or goldfish schwannoma

Recent discoveries of transmissible cancers in multiple bivalve species suggest that direct transmission of cancer cells within species may be more common than previously thought, particularly in aquatic environments. Fibropapillomatosis occurs with high prevalence in green sea turtles ( Chelonia mydas) and the geographic range of disease has increased since fibropapillomatosis was first reported in this species. Widespread incidence of schwannomas, benign tumours of Schwann cell origin, reported in aquarium-bred goldfish (Carassius auratus), suggest an infectious aetiology. We investigated the hypothesis that cancers in these species arise by clonal transmission of cancer cells. Through analysis of polymorphic microsatellite alleles, we demonstrate concordance of host and tumour genotypes in diseased animals. These results imply that the tumours examined arose from independent oncogenic transformation of host tissue and were not clonally transmitted. Further, failure to experimentally transmit goldfish schwannoma via water exposure or inoculation suggest that this disease is unlikely to have an infectious aetiology.


Introduction
Cancer is an increasingly recognised cause of mortality in many domestic and wildlife animal species [1][2][3] . Clusters of neoplastic disease cases can be linked to species-specific genetic vulnerabilities 4 , environmental contaminant exposure 5 and infectious aetiologies 6,7 . However, in the latter case, the causative infectious agent often remains elusive 1,2 . One infectious modality that may be more frequent than previously assumed is the transmissible cancer cell 7,8 . Transmissible cancers are somatic cell lineages that are spread between hosts by the physical transfer of living cancer cells. These clones can 'metastasize' within populations, having adapted to transmit across external environments and evade host immune responses. Ten naturally occurring transmissible cancer lineages have been described: one in domestic dogs 9-11 , two lineages in Tasmanian devils 12,13 as well as multiple independent lineages in marine bivalves [14][15][16][17] . In this study, we assessed the hypothesis of clonal transmission in two animal cancers.
Fibropapillomatosis (FP) is a neoplastic disease reported in all seven sea turtle species [18][19][20][21][22] . FP results in fibroepithelial lesions that are often associated with the external soft tissues, with common sites including the flippers, inguinal and axillary regions, oral cavity and conjunctiva ( Figure 1, Table 1). Tumours affecting the visceral organs, such as lungs, kidneys, heart, and liver, are also reported. Although usually localised, secondary complications arising from tumour site and tumour burden can limit host lifespan by impairing vision, feeding, and internal organ function. The first report of FP was made in 1938, when disease was described in a captive green sea turtle (Chelonia mydas) from Key West, Florida 23 . The disease is now recognised in C. mydas populations worldwide and could threaten long-term population survival given higher disease prevalence in juvenile individuals 24,25 .
FP transmission studies in green turtles and spatial patterns of disease spread are consistent with an infectious aetiology 26,27 . Transmission to naive captive-reared green turtles via cell-free extracts has also been reported, supporting the possibility of a viral infectious agent 28 . The disease has been linked with herpesvirus infection, specifically chelonid alphaherpesvirus 5 (ChHV5), also known as fibropapilloma-associated turtle herpesvirus (FPTHV) 29 . However, a causal relationship between ChHV5 inoculation and disease has not yet been confirmed; ChHV5 infection is also reported in disease-free C. mydas populations 30 and a tumour-specific ChHV5 viral variant has not yet been identified. Higher disease incidence in turtles exposed to environmental pollutants has been reported 31,32 ; however, this observation might be explained by other features of nearshore and inshore environments, such as increased population densities 32-34 . Transmission of FP via marine leeches (Ozobranchus spp.) and reef cleaner fish has previously been suggested [35][36][37] . Such vector organisms could provide plausible physiological routes for cancer cell transmission 38 . Spontaneous regression of FP, an observation rarely made in human cancers, has also been described, similar to reports of spontaneous regression in transmissible cancers 39-41 .
Peripheral nerve sheath tumours (PNST), including neurofibromas and schwannomas, are the most commonly observed tumours in goldfish (Carassius auratus) 42,43 . Schwannoma presents as soft, frequently hemorrhagic, nodules on the skin and subcutaneous tissue (Figure 1). Tumours express S-100 protein and calretinin, and are proposed to be of Schwann cell 44 or fibrocyte 45 origin. Although the aetiology of goldfish schwannoma is unknown, a viral aetiology has been hypothesized for the damselfish schwannoma, in which experimentally transplanted tumour cells are capable of causing new growths in naive fish 46 . A relatively high incidence of schwannoma has been reported in isolated goldfish colonies, suggesting that this tumour may have an infectious origin, and previous reports have hypothesized that this cancer might be transmissible 42,43,47 . Interestingly, both transmissible cancers known in Tasmanian devils arose from the Schwann cell lineage 48,49 . It is possible that goldfish schwannoma may represent a transmissible clonal cancer wherein social behaviour during spawning could provide a mechanism of cancer cell transfer. Moreover, growing evidence suggests that aquatic environments may provide favourable conditions for transfer of genetic material, even transmissible cancer cells 50-53 .
Here, we analysed microsatellite repeat loci in host and tumour tissue from diseased C. mydas and C. auratus in order to determine whether either FP or goldfish schwannoma had clonal origins. We found that in both tumour types, genotypes of neoplastic cells matched those of their hosts, strongly arguing that cancer cells are host-derived and excluding clonal transmission of the analysed tumours.

Methods
The animal studies described below adhered to the Animal Research: Reporting of In Vivo Experiments (ARRIVE) guidelines 54-56 . All efforts were made to minimize the animals' suffering and to reduce the number of animals used for experiments.

Green Sea Turtle Fibropapillomatosis
Ethics. Tissue sampling was carried out under permit number MTP-18-236 from the Florida Fish and Wildlife Conservation Commission (FWC) and with ethical approval from the University of Florida Institutional Animal Care and Use Committee (IACUC), under protocol number 201909289. No additional stress or suffering was experienced by the sea turtle patients in relation to this sampling, and the research sampling in no way interfered with the veterinary care and rehabilitation of these wild animals. All sampling was at the discretion of the attending veterinarian, and samples were obtained during necropsy or during rehabilitation-related tumour removal surgeries during which patients were anesthetized.
Sample collection and diagnosis. Tissues from six juvenile green sea turtles were collected in 2017 and 2018 at the Whitney Laboratory for Marine Bioscience and Sea Turtle Hospital, St Augustine, Florida, as previously described 57 . The sex of juvenile individuals is not readily determinable. Tail fin tumour indicated with an arrow. Right, representative tumour histology from a haematoxylin and eosin stained section. Nuclear palisading, which is diagnostic for goldfish schwannoma can be observed (arrows) (right). Turtles were held in 240 cm diameter circular tanks, holding 2,270 litres of continuously filtered sea water. The researchers had no role in the husbandry or housing of the turtles, these are not experimental animals, rather endangered animals undergoing rehabilitative care with the ultimate goal of their release back to the ocean. Sample collection was opportunistic, without explicit sampling design. The only inclusion criteria applied was that only stranded sea turtles afflicted by external FP tumour growth were eligible for the study. As a patient-matched (tumour and non-tumour samples) analysis approach was employed six turtles were deemed to be an appropriate number to confirm whether tumour genotypes matched that of the host animal.
Internal tumour and host tissue samples were obtained during routine necropsy of animals euthanised due to inoperable internal tumour burdens (as per governing FWC rehabilitation guidance). Researchers involved in the study played no role in euthanasia decisions. Euthanasia was performed on a case-by-case basis at the discretion of the attending veterinarian with express permission of FWC and in line with disease severity, quality of life and likely rehabilitation outcome considerations, as per the governing sea turtle rehabilitation-related FWC guidance. External fibropapilloma tumours were surgically removed by laser resection as part of routine rehabilitative care.
Tumour and host tissue biopsies (Table 1) were stored at -80°C until processing. Gross and histopathological examinations were performed by veterinary pathologists to confirm FP diagnosis. Non-tumour biopsy sites were identified by gross examination by the attending veterinarian; such regions were confirmed visually to be tumour-free and not bordered by any tumour regions by attending veterinary technicians and researchers, as previously described 57 .
For each individual, FP severity was assessed according to the Southwest Atlantic score system 58 ( Table 1). The turtles included in this study had a tumour score range of mild to severe (>6.6 to 185.5). However, individual GT3 was not scored upon admittance to the hospital (Table 1).

DNA extraction.
Representative tissue sampled from tumour and host biopsies was used for DNA extraction using the Qiagen DNeasy Blood and Tissue Kit (Qiagen, Hilden, Germany) according to manufacturer's instructions. DNA was quantified using a Thermo Scientific NanoDrop 2000 UV-Vis Spectrophotometer (Thermo Fisher Scientific, USA). Experimental transmission via scarification. Five female PNST-affected goldfish and five female healthy goldfish were anaesthetized with 120 mg ml -1 MS 222 at pH 8. Using scalpels, 50 mm 2 of skin was scarified in each of the healthy group fish and about 50mm 3 incisional biopsies were taken from the tumours in the PNST-affected group. Biopsies were rubbed on the scarified areas. Fish were then placed in fresh water and observed at weekly intervals for the presence of externally visible lesions. After one year, all fish were euthanized and necropsies performed. Euthanasia was performed as described above and was necessary for complete pathological examination.

Experimental transmission via inoculation.
Five female PNST-affected goldfish and five female healthy goldfish were anaesthetized as described above and submitted to another experimental transmission of tumoural cells. Using a 20 G needle connected to a 2.5 ml syringe, tumours were gently aspirated in order to collect a narrow cylinder of live tumoural tissue, which was quickly inoculated under the dorsal fin skin of healthy fish. Fish were then placed in fresh water and observed at weekly intervals for the presence of externally visible lesions. After one year, all fish were euthanized and necropsies performed. Euthanasia was performed as described above and was necessary for complete pathological examination.

Results
We analysed matched tumour and host samples from six FP-affected green sea turtles at five polymorphic microsatellite loci. In total, 24 alleles were identified in the population ( Table 2) 56,62 . At all loci, FP tumour genotypes were identical to matched host genotypes ( Table 2).
In C. auratus, we analysed the genotypes of four matched tumour and host samples across four polymorphic microsatellite loci. Overall, 16 alleles were identified (Table 3). At all loci, tumour and matched host tissues shared identical genotypes ( Table 3). Transmission of PNST from affected to naive goldfish through water exposure was not observed during laboratory proximity experiments. Moreover, inoculation of healthy goldfish with schwannoma cells by rubbing scarified skin with tumour biopsies, or by subcutaneous implantation of tumour biopsies, did not result in engraftment. Furthermore, no post-challenge complications were recorded.

Discussion
Microsatellite genotyping confirmed that the examined fibropapillomatosis tumours in green sea turtles are of host origin, and indicated that these tumours were not clonally transmitted between animals. Investigating the interaction of viral and environmental cofactors, such as water temperature, ultraviolet (UV) radiation and marine toxin exposure, that may lead to FP pathogenesis will be an interesting area for future study and may provide valuable information about how host genetics, host immunity, and ecological environments influence cancer growth and how disease spreads through marine environments 57,63 . ChHV5 has been detected in the tank water of FP-afflicted green sea turtles, and FP-afflicted turtles exhibiting weaker immune activation had worse clinical outcomes 37,63 . Recently, green sea turtle papillomavirus 1 (CmPV1) was reported in 47% of FP tumours analysed, suggesting the potential of multiple viruses as cofactors in FP disease 64 . It is also interesting to note that FP lesions, along with concurrent ChHV5 infection, have been reported in the eastern box turtle (Terrapene carolina), a terrestrial turtle species 65 .
Analysis of polymorphic microsatellite loci showed that goldfish schwannoma tumour genotypes consistently match  213 192, 192 192, 192 192, 202 192, 202 202,247 202,247 corresponding host genotypes, and implies that these tumours did not derive from a single clonal origin. Results of the cohabitation and inoculation experiments indicated that these tumours were not readily transmitted by contact with water from affected goldfish, or by implantation of tumour cells, and suggest that this disease may not have an infectious aetiology. Instead, genetic susceptibility, perhaps influenced by reduced genetic diversity may contribute to disease in domestic goldfish 66 .
While we did not find evidence of transmissible cancer, the genotyping and experimental transmission studies described here are limited to a small set of samples and cannot exclude the existence of different tumour subtypes in these species, some of which may be transmissible. Indeed, the co-occurrence of transmissible and non-transmissible forms of bivalve haemic neoplasia in mussels confirms that larger-scale sampling and genetic identification may be required in order to definitively rule out direct cancer cell transmission 17,67 . Furthermore, although care was taken to biopsy neoplastic sites in both cancers, it is worth noting that we were unable to confirm the proportion of neoplastic cells in the samples analysed in this study.
Although transmissible cancer clones are thought to emerge rarely, their numbers and distributions in wildlife populations are difficult to assess 68,69 . The current work, together with a previous study of urogenital carcinoma in California sea lions, argues against transmissible cancer aetiologies for three well-recognised animal cancers 8 .
Testing the hypothesis of transmissible cancer is an important step in understanding pathological processes involved in animal cancers and provides new research opportunities for animal disease biomonitoring and control 2 . Like pathogens and parasites, cancer, especially transmissible cancer, can have a negative impact on host fitness in wildlife populations and may be an important, but often overlooked, feature of animal ecosystems 70 . Future analysis of goldfish schwannoma and green turtle fibropapillomatosis will further reveal the mechanisms of these diseases and improve our understanding of how cancer occurs in animals in aquatic environments.

Ariberto Fassati
Division of Infection & Immunity, Institute of Immunity and Transplantation, University College London, London, UK In this paper, the authors examined the possible clonal origin of two tumours of unknown etiology and suspected transmissibility in the sea turtle Chelonia mydas and the gold fish. The authors obtained DNA from several affected individuals (tumour biopsies and control normal tissue) and carried out DNA microsatellite polymorphisms analysis. The results showed that, in every case examined, the tumour matched the host and therefore the clonal origin of these cancers was excluded.
The paper is well written, the rationale clearly explained and even if the results are negative, it will be useful to other colleagues to know about these findings. Actually, I would welcome a systematic analysis of other potential candidates.
The manuscript may be improved by providing some sense of how diverse microsatellite polymorphisms are in these species, if known, and more info in the Results section on why these particular regions were selected. Lastly, based on the microsatellite polymorphisms diversity, can a p value be calculated for the likelihood that these tumours are not clonal?

Are sufficient details of methods and analysis provided to allow replication by others? Yes
If applicable, is the statistical analysis and its interpretation appropriate? I cannot comment. A qualified statistician is required.

Michael Metzger
Pacific Northwest Research Institute, Seattle, WA, USA The Ní Leathlobhair paper investigates the hypothesis that two animal cancers (in green sea turtles and goldfish) could be transmissible cancers, like those found in dogs, Tasmanian devils, and bivalves. They clearly articulate the reason that this hypothesis was suggested in each case (primarily that there is evidence that both diseases occur in outbreaks and likely have an infectious etiology). The authors used microsatellites and convincingly showed that it is highly unlikely that these cancers are transmissible.
There are two main technical reasons that these studies might be a false negative result (ie. that the cancer could still possibly be transmissible). The first is that the cancer sample could have an insufficient amount of cancer cells. If the cancer sample was only partially cancer cells (<50%) then it can be difficult to detect the alleles corresponding to the cancer. The second is that the primers might not amplify the corresponding alleles in the cancer genome, due to mutations in the primer sequence. It is less likely that this would occur at multiple loci, so the use of multiple loci makes this less of a concern. This could also occur in the case of a cross-species transmissible cancer, but that is currently believed to be rare. The authors address the limitations of the study appropriately.
We know from the finding of transmissible cancers in many cases of disseminated neoplasia that there remain many transmissible cancer lineages that have yet to be found within marine bivalves, but it is unclear if there are other unknown transmissible cancers in other species, particularly those in the marine environment. Despite some speculation, there is no real way to determine whether there are unknown transmissible cancers in different species without experimental work like this study. It is an important addition to the field.  8,9 . It this therefore important to improve our ability to detect them to establish baseline data.

○
The following paper is cited in the introduction "Land Use, Macroalgae, and a Tumor-Forming Disease in Marine Turtles". I recommend to read the following paper, Work et al. (2014) 10 before deciding to cite this study since it is considered to be a controversial study.

Material and methods:
The choice of fibropapillomatosis (FP) as a potential candidate for being a transmissible cancer seems a bit odd, since so far, a significant body of literature is pointing toward a viral cause. The main issue with ChHV5, is it is difficult to satisfy the Koch postulate because the virus is hard to cultivate in the lab. It is very likely the virus is initiating the tumours; but we are still unclear what are the environmental condition acting as promotor (reviewed in Jones et al. 2016 11 ). As such I was not expecting the authors to find that FP was a transmissible cancer, which they confirmed with their analyses.
I am not fully convinced microsatellites alone are sufficient to be able to differentiate transmissible cancers from the host (and that the risk of false negatives is not greatly increased). For example, Hammel et al. (2021) 12 relied on a large number of SNPs to distinguish between transmissible cancer and the host in mussels. I think the authors must justify of their choice of using microsatellites and why they think the method is powerful enough to detect transmissible cancers.

Discussion:
A limitation of the study (mentioned by the authors and which they could develop a bit more) is that for sea turtles a small number of tumours samples were obtained from only two study sites while FP has now reached a panzoonotic status. Similarly, a small number of samples were obtained from goldfishes. In the following paper, Bramwell et al. (2021) 8 sensitivity analyses were performed to investigate, given a transmissible cancer prevalence, how many individuals you would need to sample to be able to detect it with a probability of 0.95 or 0.99. For a low prevalence 100's of individuals are often required (assuming the method to detect the transmissible cancer works perfectly).
The transmission experiments were also performed on a small number of individuals, but I can see the worth of those experiments as preliminary results.
For the fourth paragraph of the discussion see the following paper, Dujon et al. (2021) 9 I mentioned before since it provides a quantitative estimation of the number of transmissible cancers in bivalves and mammals.
In conclusion I think we need more studies systematically investigating if cancers observed in wildlife are transmissible or not (alike Ní Leathlobhair et al. study, but with an adequate sample size). Even if they return "negative" results, such studies allow to establish a baseline which we are currently lacking.

References:
Species name in references 24, 27, 36, 60 are not in italics. Worth double checking the references.