Evidence of an oncogenic Gammaherpesvirus in domestic Dogs

Abstract

In humans chronic infection with the gammaherpesvirus, Epstein-Barr Virus is usually asymptomatic however, some infected individuals develop hematological and epithelial malignancies. The exact role of EBV in lymphomagenesis is poorly understood partly because of the lack of clinically relevant animal models. Here we report the detection of serological responses against EBV capsid antigens in healthy dogs and dogs with spontaneous lymphoma and that dogs with the highest antibody titers have B-cell lymphoma. Moreover, we demonstrate the presence of EBV-like viral DNA and RNA sequences and Latent Membrane Protein-1 in malignant lymph nodes of dogs with lymphoma. Finally, electron microscopy of canine malignant B cells revealed the presence of classic herpesvirus particles. These findings suggest that dogs can be naturally infected with an EBV-like gammaherpesvirus that may contribute to lymphomagenesis and that dogs might represent a spontaneous model to investigate environmental and genetic factors that influence gammaherpesvirus-associated lymphomagenesis in humans.

Background

EBV is the prototypical, double stranded DNA gammaherpesvirus (GHV) that belongs to the lymphocryptovirus (LCV) subfamily (Bajaj, Murakami, and Robertson, 2007). As one of the most prevalent viral infections known, EBV infects up to 90% of the human population (Young and Rickinson, 2004). Following acute infection, viral latency is established within the B cell compartment and in the majority of immune-competent individuals chronic infection is asymptomatic (Callan et al., 1998; Munz et al., 2000; Rickinson and Moss, 1997). However, a subset of infected individuals develop EBV associated malignancies that reflect the restricted cellular tropism of EBV. These include hematological malignancies such as Hodgkin's Lymphoma (HL), Non-Hodgkin's Lymphoma (NHL) (including Burkitt's lymphoma (BL), diffuse large B cell lymphoma (DLBCL) and NK/T cell lymphoma), post transplant lymphomas and epithelial malignancies including nasopharyngeal and gastric carcinoma (Crawford, 2001; Kutok and Wang, 2006; Young and Murray, 2003). Despite intense study and the recognition that EBV proteins activate intracellular signlaing pathways that are associated with oncogenesis (Middeldorp, 2003; Rowe, 1994; Sivachandran, 2010) the exact mechanisms by which EBV contributes to the pathogenesis of lymphoid and epithelial neoplasms are not completely understood, hindering the development of effective strategies to treat or prevent EBV-associated neoplasia (Epstein and Barr, 1964; Thorley-Lawson and Allday, 2008). One of the primary obstacles to understanding the pathogenesis of EBV associated disease is the lack of a clinically relevant spontaneous animal model in which concurrent genetic and environmental factors thought to be necessary for virus-associated lymphomagenesis are recapitulated.

The domestic dog shares a close phylogenetic relationship with humans and has co-habited with them for 15,000 years (Lindblad-Toh et al., 2005; Sampson, 2006; Vila et al., 1997). As both genetic and environmental factors influence the initiation and progression of malignancy, it is not surprising that dogs develop spontaneous lymphoid malignancies including DLBCL and Burkitt-like lymphoma that share remarkable similarities in biological, behavioral, genetic and cytogenetic characteristics with their counterpart subtypes in humans (Breen and Modiano, 2008; Vail and MacEwen, 2000). For example, a chromosomal translocation observed in human BL that places MYC under the control of the IgH promoter, leading to constitutive MYC expression and further potentiation of EBV-associated lymphomagenesis is also observed in canine DLBCL and Burkitt-like lymphoma (Breen and Modiano, 2008; Dalla-Favera et al., 1982; Taub et al., 1982). In addition, the canonical NF-κB pathway is constitutively active in malignant lymphocytes of dogs with DLBCL (Gaurnier-Hausser et al., 2011) a finding that is shared with EBV-associated lymphoid malignancies in humans. Indeed, c-myc and NF-κB are the two main transcription factors activated in the latency III phase of EBV infection and they mediate the biological characteristics of EBV transformed cells (Faumont et al., 2009). Currently, there are no recognized gammaherpesviruses that infect dogs. However, previous studies have revealed that following transfection with human CD21, the receptor for EBV, canine epithelial cells (Mardin Darby Canine Kidney (MDCK)) can be infected with EBV and sustain this infection to establish stable EBV-infected canine clones (Chodosh et al., 2000; Yates, Warren, and Sugden, 1985). In these cells, the viral genome exists as an episome and a latency type III gene transcription profile is expressed (Yang, Maruo, and Takada, 2000). Furthermore, antibodies to an EBV-like virus and EBV-like DNA sequences have been identified in serum samples and peripheral blood lymphocytes respectively of healthy pet dogs (Chiou et al., 2005; Milman, Smith, and Erles, 2011). These findings suggest that an EBV-like gammaherpesvirus can naturally infect dogs and similar to the majority of humans infected with EBV, infection is asymptomatic.

Here we have investigated the hypotheses that pet dogs can be infected with an EBV-like herpesvirus and that this virus contributes to spontaneous B cell lymphoma in this species. We have detected serological responses against two EBV-viral capsid antigens (VCA) in privately owned dogs and have found that dogs with the highest VCA-IgG titers have spontaneous B cell lymphoma. Moreover, we show by in situ hybridization (ISH) and PCR that viral nucleotide sequences with high homology to EBV are present within the malignant lymph nodes of dogs with DLBCL and that Latent Membrane Protein-1 (LMP-1) a major oncogenic protein of EBV is expressed in the lymph nodes of some dogs with B cell lymphoma. Finally, we show the presence of classic herpesvirus particles in cultured malignant canine B cells using electron microscopy. Taken together, these findings raise the intriguing possibility that the domestic dog might serve as a long sought-after, spontaneous, large animal model in which to investigate EBV-associated lymphomagenesis and the environmental and genetic factors that influence tumor initiation and progression. Furthermore, our results begin to provide valuable insight into potential mechanisms of lymphomagenesis in the dog and pave the way toward development of a clinically relevant spontaneous model in which to determine the safety and efficacy of novel therapies for the treatment and prevention of EBV-associated lymphomas.

Results

Serological evidence of infection with an EBV-related gammaherpesvirus in domestic dogs

To test the hypothesis that dogs are naturally infected with an EBV-related herpesvirus, we evaluated plasma samples from 41 dogs without lymphoma and 48 dogs with spontaneous lymphoma for the presence of IgG antibodies against the highly immunogenic, lytic phase, small VCA p18 encoded by reading frame BFRF3, by ELISA (van Grunsven et al., 1993). Thirty six different dog breeds were represented in the analysis with mixbreeds, Golden Retriever dogs, Labradors and Minature Schnauzers over-represented in the sampled population. Golden Retrievers, Labradors and Minature Schnauzers were also over-represented in the lymphoma group. Mixbreeds were over-represented in the non-lymphoma group. ELISA revealed a scattered dataset with the relative O.D. of samples from 8/48 dogs with lymphoma and 3/41 dogs without lymphoma higher than the human positive control (Fig. 1A). There was no apparent breed pre-disposition to high anti-VCA-p18 titers. Dogs with high anti-VCA-p18 titers in the lymphoma group included 2 mixbreeds, 1 Labrador, 1 Golden Retriever, 1 Minature Schnauzer, 1 Malamute, 1 English Setter and 1 Bichon Frise. Dogs with high anti-VCA-p18 titers in the non-lymphoma group included 1 mixbreed, 1 Golden Retriever and 1 Minature Schanuzer. These findings indicate that a subset of non-lymphoma dogs and dogs with lymphoma have generated antibody responses against EBV VCA p18. End-point dilution titers of dogs with lymphoma in this dataset ranged from <1:64 to 1:32,768 and in non-lymphoma dogs ranged from <1:64 to 1:512. Plasma samples taken from 34 adult SPF beagles were also evaluated and 32/34 dogs had end point dilution titers <1:64. Although end-point dilution antibody titers of both cohorts of privately owned dogs varied, those dogs with the highest VCA-p18 specific end-point dilution titers were dogs with spontaneous lymphoma (Fig. 1A). Furthermore, when the relative O.D. of samples from this group of dogs with known lymphoma immunophenotype (B cell: 21/48, T cell 5/48) were analyzed the highest relative O.D. values were found in dogs with B cell lymphoma (Fig. 1B).

Figure 1. Detection of VCA-p18 specific IgG responses in dogs.

A. Plasma samples from 41 non-lymphoma dogs and 48 dogs with lymphoma were evaluated for IgG responses against VCA-p18 antigen by ELISA. Relative optical density (O.D.) results were determined by dividing the average sample O.D. by the average O.D. of a single, constant seronegative human sample. B. Plasma samples from 21 dogs with B cell lymphoma and 5 dogs with T cell lymphoma were evaluated for IgG responses against VCA-p18 antigen by ELISA. Relative O.D. results are shown. C. Plasma samples from 2 dogs with lymphoma and a human positive and negative control sample were blocked with VCA-p18 antigen prior to evaluation by ELISA (*, p<0.05, student's t-test).

To confirm the specificity of VCA-p18 reactivity in this assay, plasma samples from 2 dogs with B cell lymphoma and high anti-VCA-p18 IgG titers were pre-incubated with VCA-p18 antigen. This was done to occupy p18 specific antigen binding sites of antibodies in plasma samples and therefore demonstrate the specificity of ELISA for p18 protein. In both samples, pre-incubation with VCA-p18 significantly reduced the O.D. reading of the samples indicating that detected antibody responses were specific for VCA-p18 in these dogs (Fig. 1C). Similar blocking led to a significant reduction in antibody binding in seropositive human plasma but had no effect on a negative human control. Conversely, boiling p18 antigen prior to its use as the target antigen in ELISA had no effect on the O.D. readings suggesting that, as in humans, canine IgG antibodies generated against VCA-p18 are most likely specific for linear epitopes within VCA-p18 (data not shown) (van Grunsven, Spaan, and Middeldorp, 1994).

In addition to VCA-p18 seropositivity, serological responses directed against p23, another small VCA encoded by the BLRF2 reading frame have been reported as highly specific indicators of EBV infection in humans (Reischl et al., 1996). To determine whether dogs have detectable IgG responses against EBV VCA-p23 and whether these responses correlate with antibody responses against VCA-p18, plasma from 25 dogs with spontaneous lymphoma (evaluated in Fig. 1A) were evaluated for p23 immunoreactivity by ELISA. ELISA results revealed a scattered distribution of data points that correlated significantly with anti-VCA-p18 IgG responses in this group of dogs (R²=0.817) (Fig. 2A).

Figure 2. Detection of VCA-p23 specific IgG responses in dogs.

A. Plasma samples from 25 dogs with lymphoma were evaluated for IgG responses against VCA-p23 antigen by ELISA . A correlative analysis of O.D. of the VCA-p18 and VCA-p23 specific IgG responses as determined by ELISA is shown. B. Western blot analysis of antibodies to recombinant VCA-p23 in dog plasma. Each strip contains recombinant VCA-p23 antigen. Strips were incubated with plasma from 4 dogs with lymphoma and high anti-p18/p23 IgG titers and 2 non-lymphoma dogs with low anti-p18/p23 IgG titers as indicated. Human EBV seropositive and seronegative plasma were used as controls.

To confirm the specificity of canine antibody responses for VCA-p23, Western blot analysis was performed using purified VCA-p23 antigen to detect the presence of VCA-p23 specific IgG responses in dogs. Plasma samples from 4 dogs with B cell lymphoma and high anti-VCA-p23 relative O.D.s and two healthy, non-lymphoma dogs with low anti-VCA-p23 relative O.D.s were evaluated. Seropositive human plasma was used as a positive control (Fig. 2B). Bands of ~8KDa and 18 KDa were identified in human seropositive plasma and plasma from dogs with B cell lymphoma. In contrast, no activity was identified in either of the healthy, non-lymphoma dogs analyzed. These results indicate that IgG responses specific for VCA-p23 antigen can be detected in domestic dogs and that these responses correlate with antibody responses specific for VCA-p18 antigen. This data suggests that a subpopulation of domestic dogs have developed immunological responses against two well-characterized and highly specific GHV associated VCAs and supports the hypothesis that pet dogs can be naturally infected with an EBV-like gammaherpesvirus.

Indirect Immunofluorescence detects seropositivity in dogs with high VCA-p18 antibody titers

The gold standard for serological diagnosis of EBV infection remains the indirect IgG immunofluorescence assay (IFA) that detects antibodies directed against clusters of Early Antigen (EA), EBNA1 and VCA antigens expressed by EBV-transformed lymphoblasts (Henle, Henle, and Horwitz, 1974). As these tests are labor intensive and subjective they have largely been replaced by ELISA assays. However, they are still useful in validating ELISA results (Klutts et al., 2009). Therefore to confirm the presence of serological responses directed against EBV-like viral proteins in dogs, IFA was performed using EBV transformed lymphoblastoid cell lines as targets and plasma from 3 dogs with lymphoma and high VCA-p18 titers and 1 healthy, non-lymphoma dog with a low VCA-p18 titer. Results were compared to human positive and negative serological controls. In 2 dogs with high VCA-p18 antibody titers bright, granular, primarily cytoplasmic fluoresence was present and comparable with the human positive serological control. In contrast, plasma from the healthy non-lymphoma dog with a low VCA-p18 titer showed no increase in immunofluorescence above the human negative control (Fig. 3).

Figure 3. Indirect Immunofluorescence confirms seropositivity in dogs with high anti-VCA-p18 IgG titers.

Indirect immunofluorescence was performed on canine plasma samples using human lymphoblastoid cells as targets. Bound IgG was detected using a FITC conjugated rabbit anti-canine IgG. A. Positive control using human EBV-positive serum. B. Negative control using non-reactive human serum. C. Plasma from a dog with a 1:32,768 anti-VCA-p18 IgG titer and B cell lymphoma. Results are representative of 2/3 dogs with high anti-VCA-p18 IgG titers D. Plasma from a healthy, non-lymphoma dog with a low anti-VCA-p18 IgG titer.

Detection of herpes-like viral DNA sequences in malignant lymph nodes of dogs with spontaneous B cell lymphoma

The results of our serological studies suggest that domestic dogs may be infected with an EBV-related GHV and that those dogs with the highest anti-VCA IgG titers have spontaneous lymphoma. Based on these results we hypothesized that infection with an EBV-like GHV may contribute to lymphomagenesis in the domestic dog. In humans, the detection of virus-specific sequences within lymphoid malignancies is used to indicate an association between viral infection and oncogenesis. Therefore, we sought to determine whether herpesvirus DNA sequences were present in malignant lymphocytes of dogs with DLBCL. We performed PCR using previously described degenerate primers that amplify a highly conserved region in the DNA polymerase gene of the herpesviridae family (Rose et al., 1997). RNA was extracted from the cryopreserved malignant lymph nodes of three dogs with DLBCL and lymph nodes from three healthy dogs. cDNA generated using random hexamers was used as template DNA in the reaction. An initial PCR product of approximately 536bp was obtained from control LCL cDNA (Fig. 4A). No clear amplicons were identified in the lymph nodes of the 3 dogs with DLBCL or the 3 healthy control dogs or in a water only control. The presence of cDNA within all samples was confirmed by PCR amplification of cyclophilin (Fig. 4B). PCR using previously described degenerate nested primers and the amplified products as a template produced a ~250bp product in dog 083-001 only (data not shown). The expected size of the nested product of the target herpesvirus DNA polymerase is 236bp (Rose et al., 1997). The DNA sequence of the amplified 250bp PCR product showed high sequence homology with the BALF5 gene of EBV that encodes the viral DNA polymerase in the 5’ region with disparity in the 3’ region. Additional nested primers were designed against the canine sequence in the conserved 5’ sequence and in the disparate 3’ sequence that would yield a product of 150bp. These nested primers were used following the first round PCR amplification to determine whether this viral DNA polymerase sequence could be detected in additional lymphoma samples. Nested PCR using primers designed on the canine sequence was performed on the first round PCR product (shown in A). Amplicons of ~150bp were identified in two of the dogs with DLBCL (including 083-001) but not in the healthy dogs (Fig. 4C). These primers also produced amplicons in the LCL control. Amplicon nucleotide sequences were almost identical between the two dogs and showed disparity compared to the EBV sequence in the 3’ region as identified using the original nested PCR primers (Fig. 4D). ORF translation of the sequenced products showed high homology to the EBV BALF5 protein (Fig. 4E). A BLAST search of the translated amino acid sequence revealed that it is most closely related to human herpesvirus 4 (EBV). Furthermore, minimal sequence homology was found with the cloned sequence of canine herpesvirus-1 (CHV-1) DNA polymerase (Fig. 4F; GenBank Accession No. EU531507) indicating that the amplified viral sequence does not reflect infection with CHV-1.

Figure 4. Detection of EBV-like DNA polymerase sequences in pet dogs with spontaneous B cell lymphoma.

A. First round PCR using degenerate primers. B. Cyclophilin PCR performed on cDNA synthesized from lymph nodes and used in A. C. Nested PCR reaction performed on first round PCR products (in A) using primers designed to amplify the canine viral sequence. D. Nucleotide sequences of nested PCR amplicons and alignment to EBV DNA polymerase gene (GenBank Accession NO: V01555). E. ORF translation of nested PCR products and alignment with EBV DNA polymerase protein.F. ORF translation of nested PCR products and alignment with Canine Herpesvirus 1 DNA polymerase (GenBank Accession No: EU31507); 083-001, 083-004 and 083-009 samples from dogs with B cell lymphoma; NLN Normal Lymph Node

To determine whether additional EBV-like sequences were present in malignant canine lymph node tissue, we performed PCR for EBNA3C on genomic DNA isolated from the malignant lymphocytes of 9 dogs with confirmed B cell lymphoma. EBNA3C is one of six nuclear proteins that is expressed during latent EBV infection and unlike DNA polymerase it lacks homology with genes from any other human herpesvirus (Sample, 1990). Samples from three dogs (083-001, 083-004 and 083-009) yielded amplicons of approximately 150bp, the expected product size of EBV-1 (Moon et al., 2004) (Fig. 5A). Nucleotide sequences of these amplicons revealed high sequence homology with EBNA3C (Gen Bank Accession No. V01555.2) (Fig 5B). Open reading frame translation of the canine products showed almost complete identity with EBNA3C with only a single amino acid change of isoleucine (B95-8) to valine in all three dogs (Fig. 5C). No amplicons were identified in the other 6 dogs evaluated or in the negative BJAB control.

Figure 5. Detection of EBNA3C nucleotide sequence in canine malignant lymph node tissue.

A. Amplification of EBNA3C sequence (upper panel) and control cyclophilin (lower panel) from genomic DNA isolated from lymph node tissue of dogs with spontaneous B cell lymphoma. Genomic DNA from LCLs and BJAB cells was used as a positive and negative control respectively. B. Nucleotide sequences of canine EBNA3C PCR amplicons and alignment to the EBNA3C coding region (partial) of EBV-1 (B95-8) (GenBank Accession No: V01555). C. ORF translation of canine amplicons and alignment with EBNA3C protein sequence.

ISH detects EBERs in lymph nodes of dogs with spontaneous B cell lymphoma

The use of in situ hybridization (ISH) to detect virus-specific sequences within lymphoid malignancies is considered the gold standard to confirm that lymphoproliferative disease is associated with EBV (Gulley and Tang, 2008). In particular, the detection of EBV-encoded small non-polyadenylated RNAs (EBERs) by ISH is used routinely to confirm EBV-associated disease. To determine whether canine B cell lymphoma is associated with an EBV-related GHV, lymph node biopsies from 9 dogs with spontaneous DLBCL were evaluated by ISH for the presence of EBERs (Fig. 6). Three out of 9 dogs with DLBCL showed evidence of EBERs in malignant nodes although the amount of staining/hybridization varied from diffuse staining throughout the malignant node (Fig. 6, 083-004) to staining of single cells within sheets of neoplastic cells (Fig. 6, dog 083-16 and data not shown). However, while the staining for EBERs in the human EBV-associated lymphoma positive control was intense and restricted to the nucleus (Fig. 6, positive control), staining in the canine lymph node tissues was comparatively weak and appeared to be in the nucleus and the cytoplasm. The disparity in staining may be due to different sequences between the human EBV-associated EBER and canine EBER-like viral sequences leading to less intense binding and weaker staining. Nevertheless, taken together these data suggest that domestic dogs harbor an EBV-like GHV that may play a role in B cell lymphomagenesis.

Figure 6. Detection of EBERs in canine malignant lymph node tissue using in situ hybridization.

Diffuse, positive staining (DLBCL-4) and single cell staining (DLBCL-16) (arrows) within sheets of neoplastic lymphocytes in the lymph nodes of two dogs with spontaneous DLBCL. The positive control shows strong nuclear staining in an EBV-positive human lymphoma. The negative control was performed with a scrambled probe on an EBV-negative human lymph node.

Detection of LMP-1 protein in affected lymph nodes of dogs with spontaneous DLBCL

Latent Membrane Protein–1 (LMP1) is consistently expressed in EBV-associated lymphoproliferative malignancies where it plays an important role in malignant transformation of lymphocytes through its ability to up-regulate Bcl-2 and inhibit apoptosis (Henderson et al., 1991). To determine whether this key oncogenic viral protein is expressed in the malignant lymph nodes of dogs with DLBCL, formalin-fixed, paraffin-embedded lymph node sections taken from 6 dogs with DLBCL at the time of diagnosis were analyzed for LMP-1 expression by immunohistochemistry using the murine S12 antibody. Clear cytoplasmic/membranous staining was identified in approximately 1% of the large, malignant B cells in 2 out of the 6 dogs evaluated (Fig. 6A). A similar LMP-1 staining pattern was observed in Reed-Sternberg cells within a human EBV+ HL lymph node sample used as a positive control (Fig. 6B). Occasional plasma cells also stained positive in canine lymph node sections although this has previously been attributed to non-specific staining. Staining of the plasma cells which were charaterized morphologically by a condensed, hyperchromatic nucleus, was predominantly perinuclear whereas canine lymphoblasts characterized morphologically by large vesicular nuclei, had distinctly more granular diffuse cytoplasmic staining. No staining was observed in either human or canine samples using a murine IgG isotype control. To determine whether other EBV associated proteins are expressed in malignant lymphocytes of dogs with DLBCL, lymph node sections taken from 9 dogs were analyzed for EBNA-1 and EBNA-2 proteins by immunohistochemistry. No staining was identified in any of the tissue sections examined (data not shown).

Taken together these results suggest that an LMP-1 like protein is expressed in the lymph nodes of some dogs with DLBCL and provide a possible mechanistic link between viral infection and lymphomagenesis in this species.

Identification of herpesvirus particles in malignant canine B cell cultures

Given the identification of viral DNA, RNA and protein in primary malignant canine B cells we next sought to visualize assembled viral particles in malignant canine B cell cultures. Cryopreserved, single cell suspensions of malignant lymph nodes from 2 dogs with positive ISH-EBER results and 1 dog with a low p18 antibody titer were thawed and cultured with lethally irradiated KtCD40L cells as previously described for the expansion of healthy canine B cells from PBMCs (Mason et al., 2008). Efforts to expand cryopreserved, malignant canine B cells in the absence of CD40L were unsuccessful. Short-term cultured malignant B cells were evaluated by light microscopy (Fig. 8A). Electron microscopy of these cultures revealed the presence of healthy malignant B cells that could be clearly distinguished from the remaining, highly degenerate KtCD40L cells by size (Fig. 8B). Particles with an electron dense central core and icosahedral nucleocapsid were visualized in an extracellular location in cultures from 1 of the EBER positive dogs (Fig. 8C) and in both EBER positive dogs similar particles were identified in endocytic vacuoles (Fig. 8D). In contrast, electron microscopy of KtCD40L stimulated malignant B cells from the dog with a low VCA-p18 antibody titer did not reveal the presence of viral particles. Taken together these results suggest that replication competent herpesvirus particles are present in canine malignant B cell cultures and can infect canine B cells via endocytosis, similar to EBV infection of human lymphocytes (Nemerow and Cooper, 1984).

Figure 8. Detection of herpesvirus particles in malignant canine B cell cultures.

A. Light microscopy of cultured malignant canine B cells (1000x magnification). B cells are seen clustered around KtCD40L feeder cells. Arrowheads indicate mitotic figures within a proliferating aggregate of canine B cells. Arrows indicate large (30-35 μm diameter) irradiated KtCD40L cells undergoing degeneration characterized by cytoplasmic vacuolization and blebbing, and dispersed chromatin. B. Electron microscopy of canine malignant B cells C. Electron microscopy of cultured canine B cells showing herpesvirus-like particles. D. Viral particles present in endocytic vacuoles within malignant canine B cells.

Discussion

Advances in the understanding of the pathogenesis of EBV and its role in lymphomagenesis together with the development of effective therapeutics to treat EBV-related diseases have been hampered by the lack of spontaneous models of EBV-associated lymphoproliferative diseases. EBV naturally infects only humans and previous efforts to understand its role in lymphomagenesis have relied upon the use of species-specific LCV of non-human primates to approximate EBV infection and pathogenesis. The domestic dog develops spontaneous lymphoma that shares biological, behavioral, genetic and cytogenetic similarities to DLBCL and Burkitt-like lymphoma subtypes in humans and is widely recognized as a model to evaluate the safety and efficacy of novel therapies for these disease entities in humans (Bergman et al., 2003; Breen and Modiano, 2008; Fournel-Fleury et al., 1997; Lin et al., 2008; MacEwen, 1990; Mason et al., 2008; Paoloni and Khanna, 2008; Paoloni et al., 2009; Thomas et al., 2003). Here we use ELISA, ELISA blocking experiments, Western blot and IFA to show that dogs develop specific serological responses against viral capsid antigens of an EBV-like gammaherpesvirus and thus extend previous findings of EBV-like seropositivity in dogs (Chiou et al., 2005; Milman, Smith, and Erles, 2011). Furthermore, for the first time we demonstrate the presence of viral DNA sequences and protein in the malignant lymphocytes of dogs with spontaneous B cell lymphoma. Previous efforts using PCR to identify viral sequences (BamHiW fragment) within 33 canine lymphoma samples failed to identify the presence of an EBV-like virus. In that study, a wide variety of lymphoma samples were analyzed, including samples from skin (12), gut (5) and liver (2). 8 malignant lymph nodes were evaluated although the phenotype of these malignancies was not determined. The identification of viral DNA sequences and the viral oncogenic protein LMP-1 in spontaneous canine malignant B cell lymphomas outlined in this report suggest for the first time that this virus may contribute to B cell lymphomagenesis in a subset of infected dogs. Therefore we propose that the dog may provide a unique model in which to study EBV-related lymphomagenesis, including the genetic factors that influence its development and novel therapies that may treat or prevent EBV-related disease.

In humans, infection with EBV is marked by an early and persistent IgG antibody response directed against the highly immunogenic, lytic phase small VCAs p18 and p23 (van Grunsven, Nabbe, and Middeldorp, 1993). IgG titers against these lytic antigens are extremely robust, minimally affected by immune suppression and persist for the lifetime of the individual (Klutts et al., 2009). As such, their identification is used routinely to confirm infection with EBV in humans (reviewed in (Pattle and Farrell, 2006). Likewise, the identification of antibodies that cross react with EBV-VCA is used to confirm infection with closely related LCVs that express homologs of EBV-VCA p18 and p23 in the rhesus monkey and common marmoset (Jiang, Cho, and Wang, 2000; Moghaddam et al., 1998). Homologs of EBV-VCA p18 and p23 have also been identified in the rhadinovirus Bovine Herpes Virus 4 (BHV4) (NP 076557.1). In this report, we show for the first time that domestic dogs have serological IgG responses that cross-react with both EBV-VCA-p18 and EBV-VCA-p23. Since dogs without lymphoma have detectable serological responses, our findings suggest that as with EBV infection in humans, infection with an EBV-like virus in dogs is likely asymptomatic in many cases. Importantly as there are no closely related homologs of VCA p18 or p23 in alphaherpesviruses (including the only known canine alphaherpesvirus CHV-1) or betaherpesviruses, these serological findings suggest that privately owned dogs have been naturally exposed to an EBV-like GHV.

Although serological responses against EBV-specific thymidine kinase, EBV encoded DNA binding protein, EBV-specific DNA polymerase and LCLs have been reported in healthy pet dogs suggesting the presence of an asymptomatic GHV carrier state (Chiou et al., 2005; Milman, Smith, and Erles, 2011) an association between viral infection and lymphomagenesis in the dog has not been identified. Here we show that dogs with the highest anti-VCA antibody titers have spontaneous B cell lymphoma. In humans, serological antibody titers against VCA remain relatively constant over the lifetime of the individual (Khan et al., 1996). Increases in anti-VCA titer reflect active viral replication and high anti-VCA titers have been identified in patients with EBV-associated lymphoproliferative disorders including Hodgkin's lymphoma, BL and PTLD (Fan et al., 2005; Gallagher et al., 1999; Khan et al., 2005; Riddler, Breinig, and McKnight, 1994). Previous retrospective studies have shown that high anti-VCA titers and viral loads may be predictive of lymphomagenesis in human patients (de-The et al., 1978; Mueller et al., 1989). Given that certain breeds of dog, such as the Golden Retriever, have a very high lifetime risk of developing lymphoma (1 in 8 dogs) (Glickman L, 2000), identification of an EBV-like virus that contributes to canine lymphomagenesis provides a unique opportunity to prospectively evaluate whether increasing anti-VCA antibody titers predict onset of lymphomagenesis in a genetically predisposed group of dogs.

Confirmation of the association between EBV infection and tumorigenesis is based in large part on demonstration of viral proteins and/or viral nucleotide sequences within malignant cells (Weiss et al., 1991; Wu et al., 1990). As EBERs are expressed in all stages of viral latency, their identification in biopsy specimens, using ISH represents the gold standard for identifying latent EBV infection and demonstrating an association between infection and tumorigenesis (Gulley and Tang, 2008). Our EBER-ISH findings are consistant with those reported in human patients with B-NHL. Several reports indicate that between 4-7% of B-NHL cases are EBV positive as demonstrated by EBER-ISH (d'Amore et al., 1996; Ohshima et al., 1990). Furthermore significant heterogeneity in EBER staining patterns occur amongst individuals with B-NHL (d'Amore et al., 1996). The vast majority of EBER positive B-NHL cases contained only a few scattered EBER-positive cells (19/25) whereas 4/25 and 2/25 had intermediate and extensive positivity respectively (d'Amore et al., 1996). In contrast, in human tumors such as EBV associated Burkitt lymphoma (positive EBER-ISH control) and post transplant lymphoproliferative disease, virtually all the tumor cells are EBER positive. IHC to identify LMP-1 is also used to confirm the presence of latent infected B cells in lymphoid malignancies (Pallesen et al., 1991; Young and Murray, 2003). Here we demonstrate the presence of nucleotide sequences that are highly similar to EBERs, EBV DNA polymerase and EBNA3C together with proteins that cross react with an anti-LMP-1 antibody (suggesting high similarity to LMP-1) within the malignant lymph nodes of a subset of dogs with DLBCL. The high level of nucleotide sequence homology between canine viral sequences and EBV type I suggests that these dogs are infected with EBV or a highly homologous EBV-like virus. Future deep sequencing of the canine virus will allow for this distinction to be made. Importantly, the difference in nucleotide sequence of the DNA polymerase gene make it unlikely that our findings are due to contamination with laboratory strains of EBV.

Although the function of EBERs is unknown, LMP-1 plays an important role in lymphocyte transformation by driving the lymphocyte cell cycle and inhibiting apoptotic pathways through over-expression of Bcl-2 (Henderson et al., 1991). Our findings of an association of LMP-1 with canine malignant lymphocytes is consistent with our hypothesis that a naturally occurring EBV-like herpesvirus may contribute to lymphomagenesis in this species. Interestingly, the frequency of LMP-1 positive malignant cells in the canine tissue samples was very low and more consistant with findings in NHL of immune competent humans (Jiwa et al., 1995). This is in contrast to Hodgkin's Lymphoma where the frequency of LMP-1 positive Reed-Sternberg cells varies between 10-80%. The low frequency of LMP-1 stained cells in canine NHL tissue samples may be associated with targeted immunological destruction of LMP-1 positive cells, suppression of LMP-1 expression or decreased detection of a canine homolog by the S12 antibody. Likewise, failure to identify EBNA-1 or EBNA-2 staining in canine DLBCL samples may be associated with failure of the antibody to cross react with a putative canine GHV, or a stage of latency where EBNA-1/2 is not expressed Future studies correlating EBER ISH with LMP-1 expression in canine malignant tissue samples may provide important insights into the clinical relevance of EBV-like infection and LMP-1 expression in this spontaneous dog model.

Our electron microscopy findings suggest that infectious, replication competent viral particles are present in malignant canine B cell cultures. While these cells are grown in the presence of human KtCD40L cells, these feeder cells do not express CD21, are lethally irradiated prior to use and repeatedly test negative for EBV DNA by PCR. Furthermore, we did not detect the presence of viral particles in a third KtCD40L- stimulated B cell culture using malignant lymphocytes from a dog with a low VCA-p18 titer. As such, it is extremely unlikely that visualized viral particles derive from the KtCD40L cells. The observation of intact viral particles within endocytic vacuoles of cultured canine B cells is highly reminiscent of the initial stages of EBV infection of human B cells and suggests that the virus present in these cultures can infect canine B cells in vitro. However, in the absence of immunoelectron microscopy, the identification of these viral particles as EBV or EBV-like cannot be confirmed. The generation of spontaneous canine LCLs from primary tumor specimens in the absence of feeder cells has generally been unsuccessful and only a few canine B cell lines are available. This may reflect a low incidence of GHV-associated lymphomagenesis in the dog or sub-optimal culture methods for LCL generation. Although LMP-1 expression mimics CD40 signaling and should intrinsically enable LCL production, recombinant canine IL-4 and CD40-CD40L signaling appear to be required for canine malignant B cell growth and culture. It should be noted that cyclosporine is used in these cultures to suppress T cell activation and proliferation, this is important since CD40L-treatment of EBV infected cells increases their sensitivity to CD8 T cell mediated lysis (Gulley and Tang, 2008). Nevertheless, the identification of KtCD40L feeder cells as an efficient system for canine malignant B cell culture should now enable this herpesvirus to be cultured, isolated and fully characterized.

There are a number of reports of interspecies herpesvirus transmission including transmission of the pseudorabies virus, suis herpesvirus 1 from pigs to dogs. However, currently LCVs are thought to be restricted to humans and non-human primates. Given the close contact between humans and dogs during the last 15,000 years of domestication, and recent evidence that traces the origins of EBV back to over 5 million years ago it is possible that interspecies transmission has occurred and a novel, closely related EBV-like virus has emerged in the canine population (Ehlers et al., 2010). Future deep sequencing of this canine virus will provide important insight into the phylogeny and epidemiology of this virus.

Conclusions

The findings reported here of high antibody titers to EBV-VCA p18 and p23 in dogs with spontaneous B cell lymphoma and EBV-related viral components within the malignant lymph nodes have important implications for both the human and veterinary medical fields and raise the intriguing possibility that the long sought after spontaneous model for EBV-associated lymphomagenesis might reside in man's best friend, his dog.

Methods

Cells, cell lines, animals and reagents

Canine peripheral blood mononuclear cells (PBMCs) were isolated by Ficoll density gradient centrifugation. Primary malignant canine B cells were obtained from the lymph nodes of dogs with untreated NHL after owner's informed consent and with approval from the University of Pennsylvania's Institutional Animal Care and Use Committee. The human EBV positive lymphoblastoid cell lines (LCLs) and the EBV-negative cell lines B-JAB were grown in standard RPMI media containing 10% FCS and 100mg/ml of Penicillin and Streptomycin as previously described (Menezes et al., 1975; Rutgen et al.). Genomic DNA was extracted using the DNeasy kit (Qiagen). Archived canine plasma samples and tissue samples were obtained from the University of Pennsylvania's Tumor Tissue Bank and from specific pathogen free (SPF) Beagles at Liberty Research, Inc. (Waverly, NY). Recombinant EBV proteins p18-VCA and p23-VCA were obtained from GenWay Biotech Inc. (San Diego, CA).

ELISA and End Point Dilution Titers

ELISA plates (MaxiSorb Nunc) were coated with 0.5μg of either recombinant p18-VCA protein or p23-VCA protein (GenWay Biotech Inc. San Diego, CA) in PBS at 4°C overnight. Plates were blocked with 3% BSA in PBS for 1 hour at 37°C. Canine plasma samples were diluted 1:21 in 3% BSA in PBS and incubated on the plates for 2 hours at 37°C. Plates were washed seven times with 1x PBST. Bound antibody was detected using Horseradish peroxidase-conjugated AffiniPure Rabbit anti-human or rabbit anti-dog IgG (H+L) (1:50,000 dilution in 5% milk/PBS, Jackson ImmunoResearch Laboratories Inc., West Grove, PA). Plates were washed seven times and developed using 3,3,5,5-tetramethylbenzidine (TMB) peroxidase substrate (Kinkergaard and Perry Laboratories, Gaithersburg, MD, USA). Optical density (OD) was read at 650 nm after 30 minutes using a Spectrafluor plus^® spectrophotometer (Tecan US, Inc. Mannedorf, Switzerland). Commercially available standard human positive and negative controls were included on each plate (Calbiotech Inc., Spring Valley, CA.). For antigen blocking experiments, 1:21 dilutions of plasma were incubated with 8μg of either p18-VCA or p23-VCA for 1 hour at 37°C, prior to performing ELISA. To account for inter-plate variability, all samples O.D.s were expressed as a ratio relative to the O.D. of the human negative control. All samples were run in duplicate. For end point dilution titers, serial 1:2 dilutions of all samples were made in 3% BSA in PBS up to 1:32,768 and ELISA was performed as above. The end point dilution titer was defined as the reciprocal of the last sample dilution that gave a positive reading above three times the background O.D.

Western blot

1 μg of recombinant p23-VCA protein was subject to electrophoresis on a 10% polyacrylamide gel and transferred to a PVDF membrane (Bio-Rad, Hercules, CA, USA). The membrane was blocked with 5% milk in TBST with 1% rabbit serum for 1 hour at room temperature. Canine plasma samples were diluted 1:100 in blocking buffer and incubated with the PVDF membrane overnight at 4°C. Membranes were then washed 3 times with 1x TBST and incubated for 1 hour with a 1:50,000 dilution of HRP conjugated rabbit anti-dog IgG or rabbit anti-human IgG antibody (Jackson ImmunoResearch) in blocking buffer at room temperature. Membranes were then washed 3 times with 1x TBST and blots were developed using ECL Plus (Amersham Biosciences, Piscataway, NJ, USA). Human EBV-positive plasma and a commercially available human standard negative control (Calbiotech Inc.) were used as positive and negative controls respectively.

Polymerase Chain Reactions

For degenerate DNA Polymerase PCR reactions, total RNA was extracted from the lymph nodes of healthy dogs and dogs with DLBCL using the RNeasy Mini Kit (Qiagen). Reverse transcription was performed using random hexamers and Superscript II reverse transcriptase (Invitrogen Corp, Carlsbad. CA, USA) according to the manufacturer's instructions. PCR was performed using degenerate primers and cycling conditions previously described to amplify the highly conserved region of herpesvirus DNA polymerase (Rose et al., 1997). Nested primers were designed based on the canine DNA polymerase sequence as follows: sense: 5'-GGGGTGGCCAACGGCCTCTTT-3'; anti-sense: 5'-TMMTYCGTAGCTGACTCGGGTGA-3.' PCR was performed using 2ul of the first round PCR reaction; 0.5U Vent DNA polymerase, 0.4mM dNTP, 0.4μM primers and 1X Thermo buffer in a 50ul reaction volume. Cycling conditions used to amplify the canine specific sequence were as follows: 94°C for 3 min; 94°C for 20 sec, 60°C for 20 sec, 72°C for 45 sec (42 cycles); 72°C for 10 min.

For EBNA3C PCR, genomic DNA was isolated from the malignant lymphocytes of dogs with DLBCL and from LCL-1 cells (positive control) and BJAB cells (negative control) using the QIAamp DNA Mini kit (Qiagen). PCR was performed using 200ng of genomic DNA and primers previously described (Moon et al., 2004). Cycling conditions were as follows: 94°C for 5 min; 94°C for 10 sec, 58°C for 5 sec, 72°C for 10 sec (35 cycles); 72°C for 5 min. Amplification of canine cyclophilin was used as an internal control and was performed using primers and cycling conditions previously described (Campbell et al., 2001).

Immunohistochemistry

Immunohistochemical analysis for LMP-1, EBNA-1 and EBNA-2 was performed on formalin fixed, paraffin-embedded tissue sections of lymph nodes from healthy dogs and dogs with DLBCL. A human EBV+ HL section was used as a positive control. Sections were deparaffinized in xylene and rehydrated through graded alcohols to distilled water. Sections were subjected to heat-induced epitope retrieval by boiling the slides in 0.1M citrate buffer (pH 6.0) for 10 mins. Endogenous peroxidase was blocked with 0.3% hydrogen peroxide in methanol for 10 minutes. Slides were incubated with a murine anti-LMP-1 monoclonal antibody (S12) (Mann, Staunton, and Thorley-Lawson, 1985; Rowe, 2001) at room temperature for one hour, followed by a 1:100 dilution of biotinylated horse anti-mouse IgG (Vector Labs, Burlingame, CA) at room temperature for 30 minutes. For EBNA-1 and EBNA-2 staining, slides were incubated with either a rabbit polyclonal anti-EBNA-1 antibody (a kind gift from Dr. Paul Lieberman) (1:250 dilution) or a mouse polyclonal anti-EBNA-2 antibody (1:300 dilution) at room temperature for one hour, followed by a 1:100 dilution of biotinylated horse anti-mouse/anti-rabbit IgG at room temperature for 30 mins. In all cases, bound antibody was detected using the streptavidin-biotin-peroxidase ABC method (Vector Labs) as per the manufacturer's instructions using 3’,3’ diaminobinzidine (DAB) as chromagen (Invitrogen). Tissue sections were counterstained with hematoxylin. Sections were dehydrated through graded alcohols to xylene prior to mounting with Histomount (Invitrogen).

Indirect Immunofluorescence

The commercially available EBV-VCA IgG immunofluorescence assay (IFA) (Fuller Laboratories, Fullerton, CA) based on chemically induced, acetone fixed P3HR-1 human lymphoblastoid cells that express the full range of EBV early and late antigens was modified to detect canine IgG. Briefly, canine plasma was diluted 1:10 in PBS and 10ul of sample was added to lymphoblastoid cells on IHC slides for 30 mins at 37°C. Slides were rinsed three times with PBS and incubated with a FITC conjugated rabbit anti-canine IgG antibody (1:1600 dilution; Jackson Immunoresearch) for 30 mins at 37°C. Slides were rinsed again and visualized using a Nikon E600 Infinity corrected upright microscope. Images were captured with a Nikon Digital Sight DS-Fi1 Color camera using NIS-Element BR3.0 software.

EBER in situ hybridization

ISH was performed on a Ventana XT (http://www.ventanamed.com/product/page?view=benchmarkxt) sytem using the iVIEW Blue detection system (http://www.ventanamed.com/product/list/12) using the Ventana EBER DNP RNA Probe according to the manufacturer's instructions.

Electron microscopy

Cryopreserved malignant lymphocytes were thawed and cultured with lethally irradiated K562 cells transfected with human CD40L (KtCD40L) at a ratio of 5:1 in the presence of recombinant canine IL-4 (R&D Systems, Minneapolis, MN) and cyclosporine as previously described (Mason et al., 2008). Cells were re-stimulated every 5-7 days with fresh, lethally irradiated KtCD40L. Four days following the third round of stimulation, cells were harvested and evaluated by light microscopy. For electron microscopy cells were fixed in 2.5% glutaraldehyde and 2.0% paraformaldehyde. Samples were post fixed in osmium tetroxide, washed and stained with 2% aqueous uranyl acetate. Cells were rinsed in distilled water, dehydrated and embedded in Embed 812 (Electron Microscopy Science, Fort Washington, PA). Sections were examined using a JEOL 100CX electron microscope and digital images were recorded using a Hamamatsu camera.

Highlights.

• Animal models of spontaneous GHV-associated lymphomagenesis are needed

• Dogs with high antibody titers against EBV antigens have spontaneous lymphoma

• EBV-nucleotide sequences and LMP-1 are present in malignant canine lymph nodes

• EM shows herpesvirus particles in malignant canine lymph nodes

• Pet dogs may serve as a good spontaneous model for EBV-associated lymphomagenesis

Figure 7. Detection of LMP-1 protein in affected lymph nodes of dogs with spontaneous DLBCL.

Formalin-fixed, paraffin-embedded lymph node sections from A. a dog with DLBCL and B. a human with EBV+ HL were analyzed for LMP-1 expression by immunohistochemistry using the murine S12 monoclonal antibody. Results shown in A are representative of findings from 2 individual dogs.

List of Abbreviations

BL

Burkitt's Lymphoma

DLBCL

Diffuse Large B Cell Lymphoma

EBV

Epstein-Barr Virus

EBNA-1

Epstein Barr Nuclear Antigen-1

EBNA-2

Epstein Barr Nuclear Antigen-2

HL

Hodgkin's Lyphoma

LCV

Lymphocryptovirus

LMP-1

Latent Membrane Protein-1

NHL

Non-Hodgkin's Lymphoma

PTLD

Post-Transplant Lymphoproliferative Disease

VCA

Viral Capsid Antigen