Abstract Contagious cancers that pass between individuals as an infectious cell line are highly unusual pathogens. Devil facial tumor disease (DFTD) is one such contagious cancer that emerged 16 y ago and is driving the Tasmanian devil to extinction. As both a pathogen and an allograft, DFTD cells should be rejected by the host–immune response, yet DFTD causes 100% mortality among infected devils with no apparent rejection of tumor cells. Why DFTD cells are not rejected has been a question of considerable confusion. Here, we show that DFTD cells do not express cell surface MHC molecules in vitro or in vivo, due to down-regulation of genes essential to the antigen-processing pathway, such as β 2 -microglobulin and transporters associated with antigen processing. Loss of gene expression is not due to structural mutations, but to regulatory changes including epigenetic deacetylation of histones. Consequently, MHC class I molecules can be restored to the surface of DFTD cells in vitro by using recombinant devil IFN-γ, which is associated with up-regulation of the MHC class II transactivator, a key transcription factor with deacetylase activity. Further, expression of MHC class I molecules by DFTD cells can occur in vivo during lymphocyte infiltration. These results explain why T cells do not target DFTD cells. We propose that MHC-positive or epigenetically modified DFTD cells may provide a vaccine to DFTD. In addition, we suggest that down-regulation of MHC molecules using regulatory mechanisms allows evolvability of transmissible cancers and could affect the evolutionary trajectory of DFTD.

The adaptive immune system should prevent cancer cells passing from one individual to another. However, contagious cancers can emerge in nature and provide a unique opportunity to study mechanisms of immune escape and tumor evolution in cancers that are continually passaged between individuals. Devil facial tumor disease (DFTD) is one such contagious cancer that emerged 16 y ago in the Tasmanian devil, a marsupial carnivore found only on the island of Tasmania (1).

The emergence of DFTD triggered immediate comparisons to the only other naturally occurring contagious cancer, canine transmissible venereal tumor (CTVT) (1, 2). CTVT is a sexually transmitted tumor in dogs that has existed as a parasitic cell line for thousands of years (2, 3). Despite their shared ability to pass between individuals, DFTD and CTVT have contrasting relationships with their respective hosts. Although DFTD causes complete mortality among infected devils (4), CTVT does not usually kill host dogs (5). This contrasting impact on the host species must be linked to how DFTD and CTVT interact with the immune system of their respective hosts. Although there is some understanding as to how CTVT avoids the immune response and progresses in its host (6, 7), very little is understood about how DFTD evades the immune response of host devils.

Recognition of allograft cells and malignant cells by the adaptive immune system depends on the interaction of host T cells with highly polymorphic MHC class I and class II molecules (8). In the endoplasmic reticulum (ER), MHC class I heavy chains bind peptide and β 2 -microglobulin (β 2 m), a single domain protein essential for stabilizing the heavy chain, resulting in a stable trimeric complex trafficked to the cell surface where peptide is presented to CD8+ T cells (9). Other proteins are required for successful peptide binding, including the transporter associated with antigen processing (TAP; a heterodimer of TAP1 and TAP2) that pumps peptides from the cytoplasm into the ER, and tapasin, which facilitates peptide binding (9). MHC class II molecules present peptides to CD4+ T cells, and are heterodimers encoded by A and B genes, also requiring other proteins to facilitate peptide binding, including the chaperone DMB (10).

Regulation of MHC expression occurs through a combination of cytokines, transcription factors, and epigenetic modifications (11, 12). Both MHC class I and class II molecules can be up-regulated by cytokines such as IFN-gamma (IFN-γ), released by lymphocytes upon stimulation (12). The response to IFN-γ results in expression of the MHC class II transactivator (CIITA), a transcription factor that binds to the SXY site within the promoter elements of MHC class I, MHC class II, and β 2 m genes, inducing or up-regulating expression (13). In addition to transcription factors and cytokines, MHC gene expression is influenced by the physical state of chromatin within the promoter regions of relevant genes. Acetylated histones and demethylated DNA within promoter elements are generally associated with relaxation of chromatin structure, binding of transcription factors to DNA, and transcription (14). MHC class II expression is tightly regulated, and these molecules are only found on the surface of antigen presenting cells, whereas MHC class I molecules are expressed on the surface of nearly all cells.

The success of CTVT as a contagious cancer depends on a balance between tumor growth and the ability of the immune system to control tumor growth, which is intricately linked to MHC expression. After transmission of CTVT, tumors appear within 2 mo and cells undergo an initial growth stage where the tumor cells lack expression of class I and class II molecules, and lymphocytes fail to infiltrate the tumor (6, 15). This period of tumor growth does not continue indefinitely, and after 3 to 9 mo, tumor growth either stabilizes or begins to regress, which is associated with a significant increase in MHC class I and class II expression on the surface of the CTVT cells and infiltration of lymphocytes into the tumor mass (6, 15). Outside the laboratory setting, CTVT tumors often enter a stationary phase in which the tumor neither grows nor regresses (5).

In contrast to CTVT, the mechanisms that allow DFTD to transmit successfully have been a subject of confusion. It has been proposed that DFTD successfully passes as an allograft due to low genetic diversity at MHC genes (16). This view has been supported by evidence for expression of MHC genes in tumor cell lines and biopsies (16, 17), intact antigen presentation genes in the tumor genome (18), and a lack of tumor infiltrating lymphocytes (19). However, devils are not monomorphic at MHC (20, 21), so reduced MHC genetic diversity cannot explain the sustained lack of immune response to the tumor. Thus, how DFTD cells move between individuals without immune rejection has remained unknown.

It is predicted that the rapid spread of DFTD will cause extinction of Tasmanian devils in the wild (22). DFTD arose in a Tasmanian devil in northeastern Tasmania but has since spread to all devil populations in eastern and central Tasmania (1, 22). DFTD cells are thought to pass between animals when they bite each other during social interactions and, once transmitted, large tumors form around the face and neck that cause 100% mortality (1). The spread of DFTD has caused the population to decline rapidly (23, 24) and to be classified as endangered (www.iucnredlist.org). Without an accurate understanding of how DFTD escapes the host–immune response, any hope of conserving the devil in the wild appears slim.

Here, we report our investigation of MHC expression and regulation in DFTD cells. We find that DFTD cells do not express cell surface MHC molecules, due to a down-regulation of genes essential to the antigen-processing pathway. Loss of expression is not due to structural mutations, but regulation including epigenetic modification of histones. We show that MHC class I molecules can be restored to the surface of DFTD cells by using recombinant devil IFN-γ. These results demonstrate how DFTD passes as an allograft, revealing characteristics that may be important in the emergence and evolutionary success of contagious cancers more generally. Further, these results have implications for the development of a vaccine against DFTD.

Discussion Here we show that DFTD cells do not express functional MHC class I molecules in vitro and in vivo, explaining how DFTD escapes the T-cell response typical of allograft rejection. Further, we show that loss of MHC molecules from the cell surface of DFTD cells is due to coordinated down-regulation of genes essential to the antigen-processing pathway, and that this loss is by regulatory mechanisms including epigenetic modifications rather than structural mutations. Finally, we show that the class I-negative phenotype of DFTD cells can be reversed in vitro and in vivo. DFTD cells lack MHC molecules due to down-regulation of multiple components of the antigen processing pathway. Without β 2 m and peptide (pumped by the TAPs), MHC class I heavy chain produced by DFTD cells will be retained in the ER and degraded (9). Although more β 2 m RNA was detected in the DFTD biopsy samples compared with the cell lines, IHC using the same biopsy samples demonstrates the presence of β 2 m in connective tissue derived from the host devil. These results explain the higher level of β 2 m RNA we detected in biopsy samples compared with DFTD cell lines, as well as a previous analysis (17) of a DFTD biopsy that did not detect any differences in expression levels of β 2 m, TAP1, and TAP2 RNA between DFTD and host tissue. The lack of MHC class I molecules expressed on DFTD cells most likely represents a down-regulation of class I expression that occurred during the transformation of the ancestral DFTD cell to a malignant cell. DFTD cells are reported to be of Schwann cell origin (17), and MHC expression is known to be regulated in the nervous system to safeguard against hypersensitivity (26). Tasmanian devil Schwann cells have not been isolated. However, human and rodent Schwann cells constitutively express class I but not class II molecules, although both can be up-regulated with IFN-γ (27, 28). Thus, assuming that devil Schwann cells have a similar phenotype to those of placental mammals, modifications in the DFTD cell have resulted in loss of MHC class I expression, giving these cells the ability to move as an allograft. However, the absence of MHC class II RNA expression may represent the ancestral state of the DFTD cell. Loss of MHC class I expression in DFTD cells is not caused by structural mutations, but by regulatory modifications including epigenetic changes. The β 2 m, TAP1, and TAP2 genes are structurally sound, and their expression can be increased in vitro by treatment with either TSA or recombinant devil IFN-γ. However, the inhibition of histone deactylases with TSA only partially restores gene expression and does not up-regulate CIITA expression. In contrast, treatment with IFN-γ results in expression of CIITA, a more pronounced expression of β 2 m and TAP genes, and cell surface expression of MHC class I molecules. As well as facilitating the binding of other transcription factors to MHC class I, class II, and β 2 m promoters (13, 29), CIITA also has intrinsic acetyltransferase activity and, upon binding, it relaxes chromatin structure, allowing transcription factors to access DNA (30). Thus, although inhibition of histone deacetylation results in a small increase in gene expression, CIITA may be required for full acetylation and gene expression. Alternatively, IFN-γ may control other transcription or epigenetic factors. Regardless, the down-regulation of MHC molecules via regulatory rather than structural mutations has implications for the interaction of the tumor with the devil immune system and may be exploited to design an effective vaccine against DFTD. We propose that priming the devil immune system with MHC-positive and TSA-treated DFTD cells could provide an effective vaccine against DFTD. A whole-cell vaccine would expose devil T cells to antigenic peptides derived from the DFTD cells and presented by foreign MHC molecules. Upon subsequent challenge with wild-type DFTD cells, host cells should be activated against those antigens found even at low levels on the surface of DFTD cells and/or intracellular antigens released by DFTD cells during tumor growth. Once an immune response is initiated, the release of cytokines such as IFN-γ should stimulate wild-type DFTD cells to express MHC molecules, as we have shown, potentially leading to a more significant and protective immune response. The regulation of MHC gene expression we describe for DFTD also has implications for the evolution of DFTD, perhaps representing an advantage for long-lived transmissible tumors. The only other naturally occurring contagious cancer, CTVT, has existed for as much as 2,500 y (2) and is rarely fatal to dogs. CTVT down-regulates MHC class I expression during tumor transmission and growth before up-regulating expression as it enters a stationary phase associated with lymphocyte infiltration and IFN-γ release (7). Like DFTD, CTVT has not switched off MHC expression permanently by structural mutations, but regulates MHC expression, although the mechanisms of regulation are not fully understood (2). Evolutionary pressure may have favored CTVT subclones that can subsist in the population through a balance between tumor growth and the host–immune response (2). It remains to be seen whether DFTD will evolve into a less aggressive cancer and ensure its own survival, but in any case, control of gene expression by regulatory, rather than structural mutations, gives DFTD cells the ability to adjust MHC expression in response to changing cellular environments. The results presented here explain why the adaptive immune system fails to reject DFTD cells and provide the basis for answering other important questions. For instance, down-regulation of class I molecules should make DFTD cells good targets for NK cells (31). Do DFTD cells use regulatory mechanisms to alter the balance of activating and inhibitory NK ligands as an additional mechanism of immune escape? In a more general sense, dogs control CTVT whereas devils do not control DFTD, although both tumors down-regulate MHC molecules and up-regulate them upon treatment with IFN-γ. What additional mechanisms of immune evasion (speed of replication, release of immunosuppressive cytokines, manipulation of the tumor environment) make DFTD tumor cells so difficult for the devil immune system to control? And finally, DFTD and CTVT are contagious cancers, but they share some features with trophoblasts in the fetus of placental mammals (32). To what extent are the regulatory mechanisms shared between transmissible tumors and normal cells in specialized situations? The mechanisms of immune escape used by DFTD have the potential to provide a greater understanding of the complex interaction of a tumor with its host, in addition to more general mechanisms of immune surveillance and regulation.

Materials and Methods Cells and Cell Culture Conditions. A devil fibroblast cell line (18) was used as a control. Three cell lines derived from DFTD primary tumors (1426, 4906 and C5065) are described (33, 34). Complete medium and cell culture conditions are described in SI Appendix. An additional DFTD cell line (DFTD_NV) was derived from a DFTD biopsy taken from a wild Tasmanian devil as follows. A fine needle aspiration from the DFTD tumor mass was placed into complete medium (but with kanamycin at 200 μg/mL). The cells were centrifuged (350 g for 10 min) and the resulting pellet was resuspended in complete medium before plating at ∼2 × 106/mL followed by incubation at 35 °C with 5% (vol/vol) CO 2 . Cell lines 1426, 4906, and C5065 have been kept in culture since 2005, whereas cell line NV was in culture for only 3 wk before analysis. All field procedures were carried out with approval from the University of Tasmania's Animal Ethics Committee (AEC Ref # A0011696). Cell Treatments. DFTD cells were treated with the histone deacetylase inhibitor trichostatin A (Sigma; T1952) at 10 ng/mL in culture for 72 h. A range of concentrations (5–20 ng/mL) and culture times (24, 48, and 72 h) were trialed to ensure minimal cell death during treatment. Three DFTD cell lines (1426, 4906, and C5065) were treated with recombinant devil IFN-γ. Briefly, devil IFN-γ was identified in the Tasmanian devil genome sequence (ref. 16; www.ensembl.org/Sarcophilus_harrisii; Location: GL861606.1:1664620–1670021:1) and amplified by using the following primers (F – 5′ AGCGGATCCGCCATGAATTATTCAAGCTACCTCTTAGC 3′ and R - 5′ TATCTCTAGATTACTGTGTGATTTTTCCTTGGCTTTT 3′). The amplicon was cloned into the pcDNA3.1 expression vector (Invitrogen) by using standard molecular biology procedures. The resulting construct was sequenced in both directions to ensure no errors were introduced during amplification or cloning. The construct (pcDNA3-IFN-γ) was transiently transfected into Chinese hamster ovary (CHO) cells (cultured as described in SI Appendix) by using FuGENE transfection reagent (Promega). A construct-only control, transfection reagent-only control, and untreated cells were also included in the experiment. Cells were cultured for 30 h, before the supernatants from transfected and control cells were harvested and filtered by using a 0.45-μm filter (Millipore). DFTD cells were cultured for 48 h in 50% (vol/vol) culture supernatant from transfected or control Chinese hamster ovary cells, before cells were harvested for RT-PCR and flow cytometry analysis, as described below. RT-PCR and DNA PCR. RNA was extracted from cells by using the Nucleospin RNA II kit (Macherey-Nagel). One microgram of RNA was reverse transcribed to cDNA by using Verso cDNA synthesis kit (Thermo Scientific). DNA was extracted from cultured cells by using DNeasy blood kit (Qiagen). Primers were designed for RPL13A, TAP1, TAP2, β 2 m, MHC class I heavy chain, tapasin, CIITA, MHC class IIB, class II A, and DMB, and the promoters of β 2 m, TAP1, and TAP2 (500 base pairs upstream of the translation start sites), by aligning sequences from a range of species or where possible from the reference genome for the Tasmanian devil. All primers with the reactions and cycling conditions are in SI Appendix, Tables S1–S4). All gel products were purified (QiaQuick gel purification kit; Qiagen) and cloned into pJET plasmid (CloneJet; Fermentas). Six clones for each PCR were selected and sequenced in both directions by using T7 and bovine growth hormone primers, to determine whether the RT-PCR primers amplify all sequences from the devil MHC loci recently identified (25). RT-qPCR. RT-qPCR was carried out for RPL13A, MHC class I, and β 2 m genes on the Biorad iCycler (Biorad) with cDNA generated as described above, using the Absolute Blue Sybr Green Fluorescein qPCR mix (Thermo Scientific). Details of reaction conditions, housekeeping genes, and analysis can be found in SI Appendix. RACE PCRs. 5′ and 3′ RACE cDNAs were constructed from total RNA of DFTD cells and fibroblast cells (isolated as described above) by using the GeneRacer kit according to the manufacturer’s instructions (GeneRacer; Invitrogen), with primers and PCR conditions described in SI Appendix, SI Methods. All gel products were cloned, and 12 clones were sequenced as described above. Bisulphite Sequencing of β 2 m, TAP1, and TAP2 Promoters. DNA from three DFTD cell lines (1426, 4906, and C5065) and a fibroblast cell line was treated with bisulphite to convert cytosines to thymines by using the Epitect Kit according to the manufacturer’s instructions (Qiagen). Primers were designed to amplify the CpG islands across the promoter sequences of β 2 m and TAP1, using amplification conditions in SI Appendix, Tables S3 and S4. Amplified sequences were cloned, and 12 clones for each sample were sequenced as described above. The promoter of TAP2 was not amplified as no CpG dinucleotides were detected. Development of Antibodies. Mice were immunized s.c. three times with 25 μg of a peptide representing the cytoplasmic region of devil MHC class I [GGKGGDYVPAAGN: based on Saha*01, a previously published full-length class I transcript (National Center for Biotechnology Information accession no. EF591089)] coupled to diphtheria toxoid by using glutaraldehyde. The antigen was adsorbed to Al(OH) 3 and mixed in 1:1 ratio with incomplete Freund’s adjuvant. Four days before the fusion, the mice received an i.v. injection with 25 μg of antigen administered with adrenalin. Spleen cells and SP2/0-AG14 myeloma cells were used for fusion. Positive clones were selected by screening against the peptide coupled to ovalbumin in ELISA, with specificity of the MHCI-mAb clone TD50 illustrated in SI Appendix, Fig. S12. Full-length devil β 2 m was amplified from cDNA derived from fibroblast cells with primer B2mF (5′ TTGCCATATGGTCACAAGTCCTCCCAGAGTTC 3′) and B2mR (5′ GCACCAAGTTCTGTTCTGGATCCCATTTAATTAC 3′). The subsequent amplicon was cloned into the pET22b+ vector (Novagen) and transformed into Rosetta pLysS cells (Novagen) according to the manufacturer’s instructions. Details of the expression induction and protein purification can be found in SI Appendix. Rats of the Sprague–Dawley strain were immunized s.c. at 2- to 3-wk intervals by using ∼30 μg recombinant devil β 2 m in 100 μL of PBS mixed with 100 μL of the GERBU 10 adjuvant (Gerbu Biotechnik). Two weeks after each immunization, the rats were bled from the tail vein and antibodies were recovered as EDTA plasma. Specificity of the β 2 m-Ab is shown in Fig. 1 by preincubation of the antibodies with purified recombinant devil β 2 m. Flow Cytometry. Cells were incubated on ice with protein-G purified β 2 m-Ab (3 μg/mL) or protein-G purified preimmune rat serum for 20 min, followed by secondary antibody (goat anti-rat IgG conjugated to FITC; Sigma, F6258) for 20 min. In addition to secondary antibody-only and no-antibody controls, the specificity of the antibody was determined by adding 1 mg of devil β 2 m protein to the β 2 m-Ab and incubating on ice for 30 min before incubation with cells. Cells were analyzed on the FACScan cytometer (BD Biosciences), with data analyzed by using FlowJo software. Western Blots. Cells were detached by using PBS with 2 mM EDTA, cell pellets were lysed on ice for 30 min in a lysis buffer [100 mM TrisCl, 150 mM NaCl, 1 mM MgCl 2 , 0.5 mM 4-(2-Aminoethyl) benzenesulfonyl fluoride hydrochloride, and 1% digitonin], and the lysates were clarified by centrifugation to give 108 cells per mL, as described (35). The total protein in lysates was measured by using Bradford Reagent (Sigma) following the manufacturer’s instructions. Electrophoresis and blotting was performed as described (35) with the primary antibody, a tissue culture supernatant containing mAb TD50, and incubated overnight at 4 °C. To ensure equal loading of all samples gels were stained with Commassie Brilliant Blue posttransfer and membranes were stripped (Restore Stripping Buffer; Thermo Scientific) and blotted with a primary actin antibody (clone AC-15; Sigma) according to the manufacturer’s instructions. Immunohistochemistry. DFTD primary tumors and metastases were fixed in 10% (mass/vol) PBS-buffered formalin solution for 2 to 4 d. Tissues were processed and embedded in paraffin blocks, which were cut onto 3-aminotriethoxysilane–coated slides at 3-µm thickness. Sections were deparaffinized in xylene and rehydrated through graded alcohol solutions to water and antigen epitopes were retrieved by using heat treatment with citrate buffer solution (pH 6) for 15 min. Endogenous peroxidase and nonspecific protein binding were blocked by incubation of the slides with 3% (mass/mass) hydrogen peroxide (Analar) and serum-free block solution (Dako). Sections were then incubated with protein G-purified anti-devil β 2 m-Ab (1.5 mg/mL), protein G-purified preimmune rat serum, anti-periaxin (Sigma; diluted 1:300), or anti-CD3ε (Sigma; A0452) (list of antibodies in SI Appendix, Table S5), all diluted in antibody diluent (Dako) and incubated overnight at 4 °C. Primary antibody binding was detected with peroxidase-coupled secondary antibody (Envision kit; Dako). Sections were counterstained with hematoxylin for 40 s, dehydrated through graded alcohol solutions to xylene and cover-slipped. Sections were visually analyzed by using a Leica DM 2500 microscope, and selected micrographs were obtained with a Leica FireCam DFC320 camera.

Acknowledgments We thank Narelle Phillips for preparing tissue sections; Stephan Beck, Clive Tregaskes, and Ashley Moffett for valuable discussion; and Anne Cooke, Gillian Griffiths, and Clive Tregaskes for critically reading the manuscript. H.V.S. was supported by a National Health and Medical Research Council Overseas Postdoctoral Fellowship and is currently supported by a European Molecular Biology Organisation long-term fellowship. A.K. is supported by an Australian Research Council-linkage grant. The experiments in this paper were funded by a University of Tasmania Dr. Eric Guiler Tasmanian Devil Research Grant (to H.V.S.) and by Wellcome Trust programme Grant 089305 (to J.K.).

Footnotes Author contributions: H.V.S. and J.K. designed research; H.V.S. and C.K.Y. performed research; H.V.S., A.K., C.T., Y.C., K.B., K. Swift, A.-M.P., R.H., M.E.J., K. Skjødt, and G.M.W. contributed new reagents/analytic tools; H.V.S., C.K.Y., and J.K. analyzed data; and H.V.S. and J.K. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1219920110/-/DCSupplemental.

Freely available online through the PNAS open access option.