A second transmissible cancer in Tasmanian devils

DFTD was first observed in northeast Tasmania in 1996, and rapidly spread south and west across the island (Fig. 1A). Tasmanian devils with signs of DFTD were first recorded in the D'Entrecasteaux Channel area, a ∼550 km² peninsula in Tasmania’s southeast, in December 2012. Since then, there have been 12 confirmed DFTD reports in the area (Fig. 1A and Table S1). Tumors in these animals ranged in appearance from small foci involving the oral mucosa and/or facial skin (e.g., JV, NR, LV; Fig. 1B), to locally extensive or disseminated masses deforming facial structures (e.g., RV, SN, GV; Fig. 1B). Metastases are common in DFTD (5), and were found in submandibular lymph node (SMLN) in NS, and in SMLN, lung and kidney in RV. Gross features of tumors in these animals were suggestive and typical of DFTD associated with the first described DFTD clone, DFT1.

Initial histological assessment of two of the Channel DFTD cases from 2014, RV and SN, revealed that tumors from these animals presented atypical features. DFTD tumors associated with the DFT1 clone are generally composed of pleomorphic round cells arranged in distinct bundles, cords or packets (5). In contrast, the tumors from RV and SN were characterized by sheets of pleomorphic (amorphic to stellate and fusiform) cells arranged in a solid pattern (Fig. 2). Furthermore, tumors from RV and SN were negative for periaxin (PRX), an immunohistochemical marker that is diagnostic for DFT1 (12, 17) (Fig. 2). Although it was initially considered plausible that tumors from RV and SN were DFT1 variants or spontaneous tumors arising from the tissues of their hosts, further cytogenetic and genetic analyses were performed to confirm the nature of these two tumors and other tumors derived from devils in the Channel Peninsula.

The clonal nature of DFTD was first suggested based on cytogenetic evidence indicating that DFTD tumors carry a distinctive aneuploid karyotype (9). This DFT1 cytogenetic profile differs markedly from the normal devil karyotype, and is characterized by the absence of identifiable chromosome 2 homologs, the presence of four marker chromosomes and missing sex chromosomes (Fig. 3). However, cytogenetic analysis of samples from the Channel revealed that tumors from five of the devils (those derived from RV, SN, JV, NR, and NS) shared an identical aneuploid karyotype that was clearly distinct from that of DFT1 (Fig. 3 and Fig. S1). Chromosomes from these tumors all exhibited identical complex structural abnormalities, including the presence of additional material on chromosomes 1, 2, and 4, a deletion involving chromosome 5 and monosomy for chromosome 6. Both X and Y sex chromosomes were present. This cytogenetic evidence presented the possibility that these five tumors were derived from a clone, which we have named DFT2, and that is distinct from DFT1.

Despite the absence of cytogenetically identifiable sex chromosomes in DFT1, studies using fluorescence in situ hybridization and DNA sequencing have found evidence for two X chromosome copies in DFT1 (13, 14). Furthermore, the number of single point substitution variants mapping to the X chromosome in DFT1 suggested that the two X chromosomes were germ-line homologs rather than recent somatic duplicates (13). We further investigated the origins of the X chromosomes in DFT1 by genotyping a panel of 10 X chromosome variants that were heterozygous in DFT1 in a group of male and female devils. At each locus, we found that both alleles could be detected in the devil population, confirming that these variants are germ-line single nucleotide polymorphisms (SNPs) rather than somatic mutations (Fig. 4A). Furthermore, our analysis confirmed that these SNPs indeed map to the X chromosome as only females, and not males, were found to be heterozygous at these loci (Fig. 4A). These data confirm that DFT1 carries DNA from two homologous X chromosomes, indicating that this lineage probably arose in a female devil. The discovery that DFT2 carries a Y chromosome is thus incompatible with a single clonal origin for DFT1 and DFT2.

To investigate the genetic relationship between DFT1 and DFT2, we analyzed the genotypes of these two lineages at nine polymorphic microsatellite loci (10, 12, 18, 19). Our analysis confirmed that two DFT1 tumors, 87T and 88T, shared an identical genotype with each other and with a previously analyzed panel of 27 DFTD tumors collected from geographically dispersed areas of Tasmania (12) (Fig. 4B and Table S2). Analysis of DFT2, however, revealed that the tumors derived from RV and SN had different genotypes from DFT1 (Fig. 4B). Furthermore, these two tumors shared identical genotypes with each other and were genetically distinct from their hosts (Fig. 4B). Indeed, the microsatellite analysis indicated that DFT2 tumors were no more similar to DFT1 (identical genotypes at two of nine loci) than they were to their hosts (identical genotypes at four and three of nine loci for RV and SN, respectively), or to other devils in the population (e.g., 53H, a devil from northern Tasmania) (Fig. 4B).

We further characterized DFT1 and DFT2 by analyzing panels of tumors for the presence of 12 putative somatic structural variants that had previously been identified in DFT1 (13), as well as for the presence of two polymorphic germ-line structural variants (Fig. 4C). None of the 12 putative somatic structural variants were found in DFT2 tumors, although they were all found in DFT1 tumors 87T and 88T (Fig. 4C). One of the polymorphic germ-line structural variants was present in DFT2, but was absent in DFT1 and host tissue derived from RV (Fig. 4C).

DFTD tumors are allogeneic grafts within their hosts. Previous studies have confirmed that DFT1 is able to colonize hosts carrying different genotypes at MHC loci (10, 20). However, MHC molecules are not expressed on the cell surface of most DFT1 cells, and this has been proposed as a mechanism whereby this lineage escapes destruction by host T cells (11). To investigate whether DFT2 is able to grow in hosts with disparate MHC genotype, we cloned and sequenced part of the polymorphic second exon of MHC class I loci from two DFT2 tumors and their corresponding hosts. We identified five MHC class I exon 2 haplotypes in DFT2, and confirmed that DFT2 has a different MHC class I genotype from DFT1 (Fig. 4D). It is important to note that the MHC class I genotype found in DFT2 was distinct from the genotypes found in DFT2 hosts, confirming that DFT2 is not restricted to hosts with an identical MHC class I genotype (Fig. 4D).

Wild Tasmanian devils in the Channel area with signs of DFTD were either trapped or found dead from road trauma or other causes. Live devils with visible signs of distress were euthanized for welfare reasons. Tissue biopsies and tumor fine needle aspirates were either collected post mortem, or from live devils, which were subsequently released. All animal procedures were performed under a Standard Operating Procedure approved by the General Manager, Natural and Cultural Heritage Division, Tasmanian Government Department of Primary Industries, Parks, Water and the Environment (DPIPWE), in agreement with the DPIPWE Animal Ethics Committee.

Standard hematoxylin and eosin (H&E) and immunohistochemical staining with a periaxin antibody (HPA001868; Sigma Aldrich) were performed on 3-μm paraffin sections from tumor tissues fixed in 10% neutral buffered formalin as previously described (17).

Karyotyping was performed as previously described (31). Chromosomes were banded with Leishman’s stain (Sigma-Aldrich) according to standard methods. Analysis and imaging were performed using Metasystems karyotyping software.

Candidate SNPs mapping to the X chromosome that were heterozygous in DFT1 tumors as well as in the germ line of a female devil (91H) were identified from whole genome sequence reads (13). Ten of these SNPs were genotyped in 74 devils (37 males and 37 females), as well as in 10 DFT1 tumors and 2 DFT2 tumors, by PCR and capillary sequencing; DNA was only available for 2 DFT2 tumors (those from RV and SN) during preparation of the manuscript. Tables S3 and S4 have details of PCR primers, samples, and genotypes. All PCRs were performed under standard conditions with annealing temperature of 60 °C, extension time of 45 s and 31 cycles.

Microsatellite genotyping was performed as previously described (12). In brief, PCR was performed with primers listed in Table S2; each forward primer had a 19-bp M13F sequence (5′-AGG-AAA-CAG-CTA-TGA-CCA-T-3′) attached to its 5′ end. PCR was performed with 57 °C annealing temperature, reducing the annealing temperature by 1 °C per cycle for 6 cycles, followed by 31 cycles with annealing temperature 50 °C; at this step, 8 pmol of 5′-6FAM or 5′-HEX labeled M13F primer was added to each reaction. The reaction was then continued for eight additional cycles with annealing temperature 48 °C. Products were separated on an ABI 3730XL instrument and analyzed using GeneMarker software. Concordance of alleles shown in Fig. 4B with previously published microsatellite alleles (12) is provided in Table S2.

Whole genome sequence reads derived from two DFT1 tumors and a male and female devil (31H and 91H) were analyzed using an algorithm that uses discordantly mapped read pairs to identify putative structural variants, as previously described (13). A set of 14 putative structural variants were analyzed by PCR with DNA from DFT1 and DFT2 tumors and as well as with germ-line DNA from devils. PCRs were performed under standard conditions with annealing temperature of 60 °C and 35 cycles. Primer sequences can be found in Table S5.

A 274-bp DNA segment was amplified from genomic DNA using primers recognizing the polymorphic exon 2 of devil MHC class I loci. Primer sequences, Sahaα1F (5′-TCT CAC TCC TTG AGG TAC TTC G-3′) and Sahaα1R2 (5′-CTC GCT CTG GTT GTA GTA GCC G-3′) were modified from ref. 32; modifications to Sahaα1R (32) were made based on sequence alignments available in ref. 25. The number of loci amplified by these primers is not known, and may vary between individuals, but is predicted to be between four and five (33, 34); furthermore, we cannot exclude the possibility that these primer sequences select for only a subset of haplotypes due to polymorphisms present within primer binding sites. Libraries were prepared from amplicons from each individual and sequenced on the Illumina MiSeq platform with 150-bp paired end reads. Between 100,000 and 900,000 read pairs were sequenced per individual. Forward and reverse sequences were merged for each read pair, discarding any read pairs with mismatches in the overlapping region as well as any read pair that did not exactly match both primer sequences, yielding between 10,000 and 23,000 unique haplotypes per individual. Of these, only 17 haplotypes that were present with at least 1% frequency in at least one individual were considered further. We further validated these haplotypes by searching for reads exactly matching the haplotypes within the available whole genome sequences from 87T and 91H (13). One haplotype could not be validated using this approach and was discarded. The sequence composition of two further haplotypes meant that they could only be validated in conjunction with another haplotype, and the pattern of haplotype read counts made it uncertain if these two haplotypes were present in some individuals; thus these two haplotypes were excluded from Fig. 4D. Tables S6 and S7 contain nucleotide and predicted amino acid sequences for the MHC class I haplotypes presented in Fig. 4D. Each of the 14 haplotypes in Fig. 4D are predicted to encode a unique peptide sequence. Two haplotypes that had not previously been described were named SahaI*97 and SahaI*98 and their sequences submitted to GenBank with accession numbers KT188437 and KT188438, respectively.