Molecular and Immunophenotypical Characterization of a Feline Immunodeficiency Virus (FIV)-Associated Lymphoma: a Direct Role for FIV in B-Lymphocyte Transformation?

Abstract

We describe the histopathological, immunohistochemical, and molecular characterization of a lymphoma arising in a 7-year-old cat following experimental infection with feline immunodeficiency virus (FIV). The tumor was high grade and of B-cell lineage. The transformed cell had an immature phenotype (CD79a⁺, CD79b⁻, CD21⁻, immunoglobulin heavy and light chain negative), confirmed by antigen receptor gene analysis, which showed germ line configuration. Single-copy, clonally integrated FIV provirus was detected in tumor genomic DNA. FIV p24 antigen was not detected in tumor cells by immunostaining. This study provides the first evidence that the feline lentivirus may play a direct role in cell transformation under certain circumstances.

Lymphoma (lymphosarcoma) is the most common tumor in cats (19). Feline leukemia virus is the etiological agent of the majority of these tumors, especially in younger cats, but lymphoma can occur independently of feline leukemia virus infection (25). Recently, epidemiological studies have described lymphoproliferative malignancies in association with the feline lentivirus feline immunodeficiency virus (FIV) (1, 4, 5, 8, 23, 24, 40, 43, 44, 47), and an increased risk of lymphoma in FIV-infected cats has been demonstrated (15, 23, 45, 46).

Several common features have emerged from the examination of FIV-associated lymphomas. These lymphomas are predominantly extranodal, conforming to the least common anatomical classification of miscellaneous (7, 13, 23, 40). Where histological and immunophenotypical analyses have been carried out, the tumors have been found to be of B-cell lineage (7, 13, 40) and usually of the high-grade immunoblastic or centroblastic type (7). FIV-associated lymphomas thus resemble those seen in human immunodeficiency virus type 1 (HIV-1) infection of humans (21, 30) and in simian immunodeficiency virus infection of nonhuman primates (16, 41).

In contrast to the directly oncogenic retroviruses such as the murine and feline leukemia viruses, lentiviruses are generally considered to play an indirect role in tumor development. Most HIV-1-associated lymphomas are believed to develop secondary to one or a combination of factors, including immunodeficiency (38), polyclonal B-cell activation (34), and the involvement of other infectious agents, notably Epstein-Barr virus (14, 18). However, a direct causal role is implied for HIV-1 in a T-cell lymphoma in which an integrated provirus was demonstrated (22, 48).

The role of FIV in tumorigenesis is unknown, but the lack of integrated sequences in the small number of tumors examined by restriction fragment analysis to date has led to the provisional hypothesis that the virus facilitates tumor development by indirect mechanisms similar to those postulated for HIV-1 (50). In contrast, we have now detected FIV sequences in DNA extracted from another lymphoma, this one occurring in a cat experimentally infected with FIV. This tumor consisted of a clonal precursor B-cell population. Further analysis demonstrated a single integration site, indicating that viral infection occurred prior to clonal expansion. This work suggests that a lentivirus may be directly involved in B-cell transformation.

Histopathological and immunophenotypical characterization of the tumor.

A 10-month-old neutered male specific-pathogen-free cat was inoculated with FIV Glasgow-8 as part of a long-term pathogenesis study (9). Six years 3 months after infection, a preliminary diagnosis of neoplastic disease was made and euthanasia was carried out. For histopathological examination, tissues were preserved by fixation in 10% neutral buffered formalin, embedded in paraffin, cut into 5-μm-thick sections, and stained with hematoxylin and eosin. Selected sections were also stained with Giemsa stain. Microscopic examination confirmed the presence of tumor tissue in the liver, lymph nodes, and omentum. This tumor was composed of mononuclear cells with medium to large nuclei. The majority of cells had oval, cleaved, or indented nuclei with two or three small nucleoli. These cells resembled centroblasts. A smaller number of cells (<10%) had larger nuclei with a prominent, centrally located, large nucleolus. Some of these cells had prominent, intensely basophilic cytoplasm and resembled immunoblasts. The remainder had weakly staining or no cytoplasm and resembled paraimmunoblasts. By use of the updated Kiel classification system (29), this tumor was classified as high grade and of the monomorphic centroblastic subtype. Cells were arranged in sheets to form nodules within liver tissue and lymph nodes. Nodules in the liver were separated by compressed residual hepatic parenchyma which contained vacuolated hepatocytes. Nodules within lymph nodes were located in the cortex and coalesced or were separated by preexisting trabecular tissue. Medullary tissue was compressed. Neoplastic cells were present in a fine fibrous stroma, and numerous foci of necrosis, often in association with inflammatory cells, were observed. Aggregations of small lymphocytes were present both within nodules and at their edges, and irregular infiltrates of macrophages were also observed at the perimeter of many nodules. Mitotic figures, many of bizarre shapes, were frequently noted, and hyperchromatic nuclei were also present. Bone marrow was not infiltrated by the tumor.

To investigate the tumor cell phenotype, we carried out immunostaining of paraffin-embedded sections and cryostat sections as described in a previous study with the following panel of antibodies: anti-CD79a (mb-1), anti-CD79b (B29), anti-CD21, anti-immunoglobulin G (IgG), anti-IgM, anti-IgA, anti-κ light chain, anti-λ light chain, anti-CD3, anti-CD4, anti-CD5, anti-CD8, anti-major histocompatibility complex class II (MHCII), anti-bcl-2, and MAC387 (7). Normal feline lymphoid tissue stained with anti-CD4 or anti-CD8 antibodies was used as a positive control. The primary antibody was omitted to provide negative controls.

The neoplastic cells expressed the B-cell marker CD79a (Fig. 1) and were negative for all other markers tested (data not shown). The majority of small lymphocytes characterized as T-cell infiltrates expressed CD3, with smaller numbers of the population also expressing CD8 and CD5. The presence of macrophages was confirmed by use of MAC387 antibodies, and a proportion of infiltrating small lymphocytes were characterized as B cells by their expression of either IgG or IgA heavy chains and κ or λ light chains.

FIG. 1.

Anti-CD79a staining of neoplastic cells. Formalin-fixed, paraffin-embedded tissue sections were stained by the avidin-biotin complex technique, nickel enhancement, and a safranin (pink) counterstain. CD79a-positive tumor cells, which stain darkly, are visible invading the lymph node cortex. Bar, 20 μm. Objective, 40×.

The antibody used to stain CD79a labels B cells from several mammalian species (31), including domestic cats (7, 32). CD79a is expressed at a very early stage in B-cell differentiation, prior to immunoglobulin heavy chain [Ig(H)] constant region μ-chain gene rearrangement and CD79b expression, and may persist until the plasma cell stage (3, 26, 31, 32, 51). The finding that tumor cells were negative for other B-cell markers, i.e., CD79b, CD21, and surface immunoglobulin heavy and light chains, suggests that malignant transformation occurred at a differentiation stage earlier than the pre-B-cell stage, i.e., at the pro-B-cell or stem cell stage, in rodent models (20, 49). CD21-positive B cells form an important productive reservoir of FIV infection in vivo (12). Our results extend these observations by demonstrating that B cells can be infected in vivo at a very early stage of development. These combined investigations established that the tumor was a high-grade, extranodal lymphoma with B-cell characteristics, typical of those described previously in association with FIV infection (7, 40).

Genotypic characterization.

To investigate the clonality and lineage of this lymphoma, antigen receptor gene arrangements were investigated by Southern blot analysis. Genomic DNA was obtained from snap-frozen tumor tissue, autologous salivary gland, and FIV-infected cell line FL4 (52) by the use of standard phenol-chloroform extraction and ethanol precipitation. Southern blot hybridization analysis of HincII- and HindIII-digested high-molecular-weight DNA was performed as described previously (35, 50). The status of the genes encoding the T-cell-receptor β-chain and the Ig(H) constant region was determined by use of a Cβ probe (FeTCRCβ) with HincII-digested DNA and a Cμ probe (FeCμ) with HindIII-digested DNA, respectively (50). Probes were radiolabelled to a high specific activity with ³²P by random priming. Autoradiographs were developed for 1 to 4 days at −70°C. With this technique, rearranged antigen receptor genes in clonal T- or B-cell populations gave different germ line patterns of hybridizing fragments. We found that, in each digest, DNA from the tumor, salivary gland, and FL4 cells generated identical restriction fragments (Fig. 2). The sizes of the observed fragments, i.e., 9, 3, and 8 kb, representing FeTCRCβ1, FeTCRCβ2, and FeCμ, respectively, are consistent with previously published data for the feline germ line (50). This result corroborates the findings of the phenotypical analysis, which showed that the neoplastic cell had characteristics of an early B cell, by indicating that antigen receptor rearrangement had not occurred prior to transformation.

FIG. 2.

Analysis of antigen receptor gene configuration. Tumor (T), autologous salivary gland control (C), and FL4 (P) DNAs were digested with HincII and probed with FeTCRCβ (A) or digested with HindIII and probed with FeCμ (B) to determine the status of the genes encoding the T-cell-receptor β-chain and Ig(H) constant regions, respectively. The three DNA samples gave identical germ line patterns. Lane M contains molecular size standards (sizes, in kilobases, are on the left).

Clonal integration of FIV proviral sequences.

To determine whether FIV might be involved directly in lymphomagenesis, we carried out Southern blot analysis to look for FIV sequences in tumor DNA, as described previously (35, 50). DNAs prepared from autologous salivary gland and from FL4 cells were used as controls. PvuII-digested DNA was probed with a FIV-specific probe consisting of a 1.2-kb FIV Glasgow-14 (42) gag-pol fragment (1,242 to 2,431 bp from the 5′ end of the genome) and a 1.3-kb FIV Petaluma 34TF10 (39) env fragment (6,978 to 8,090 bp from the 5′ end of the genome). Filters were washed under high-stringency conditions (0.1% SSC [1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate]–0.5% sodium dodecyl sulfate). Two hybridizing bands were detected in the tumor tissue but were absent from the autologous control tissue (Fig. 3), demonstrating the presence of significant levels of tumor-specific FIV DNA.

FIG. 3.

Detection of FIV sequence in an FIV-associated lymphoma. Tumor (T), autologous salivary gland control (C), and FIV-infected FL4 cells (P) were analyzed by Southern blotting. DNA from these tissues was digested with PvuII and probed with a mixed FIV gag-pol/env probe. FIV DNA was detected as two hybridizing bands (arrows) in tumor DNA but not in salivary gland DNA. Lane M contains molecular size standards (sizes, in kilobases, are on the left).

To determine the integration status of the FIV DNA in tumor tissue, the technique was modified to analyze undigested genomic DNA as follows. High-molecular-weight DNA was run on a 0.6% agarose gel and transferred to a charged nylon filter (Hybond-N+; Amersham International, Amersham, United Kingdom). The filter was neutralized in 0.5 M Tris (pH 7.0) for 5 min and then probed and washed as described above. The hybridizing sequence comigrated with the bulk of the undigested DNA (Fig. 4), indicating that the FIV DNA was present as a high-molecular-weight complex or as integrated proviral sequences.

FIG. 4.

Determination of the integration status of FIV DNA sequence in tumor DNA. Undigested tumor (T), autologous salivary gland control (C), and FL4 (P) DNAs were probed with a mixed FIV gag-pol/env probe on a Southern blot. FIV DNA comigrates with tumor genomic DNA (arrow). Lane M contains molecular size standards (sizes, in kilobases, are on the left).

The status of these sequences was investigated further by digestion of tumor cell and control DNA with EcoRI, which does not cut within the provirus of the inoculum strain, FIV Glasgow-8, and with SstI, which recognizes a single site in the provirus, and analyzed with the FIV probe as described above. The SstI site in FIV Glasgow-8 is located approximately 500 bp downstream of the 5′ long terminal repeat, proximal to gag, and therefore does not interfere with hybridization of the probe. Digestion with these enzymes should therefore generate one fragment per integration site if the FIV sequences in the tumor cell DNA retain the structure of the infecting virus. A single, intense hybridizing band was seen at 11.5 and 21 kb in the EcoRI and SstI digests, respectively (Fig. 5), indicating monoclonal integration of FIV provirus in tumor cell DNA.

FIG. 5.

Demonstration of monoclonal integration of FIV provirus in tumor DNA Tumor (T), autologous salivary gland control (C), and FL4 (P) DNAs were digested with EcoRI (A), which does not recognize a site within FIV provirus, or with SstI (B), which cuts once in FIV provirus, and probed with a mixed FIV gag-pol/env probe. A single sharp band in both digests (11.5 [A] and 21 [B] kb) indicates monoclonal integration of FIV provirus in tumor tissue. The presence of fragments smaller than the virus genome (9 kb) in the FL4 EcoRI digest has been observed consistently (50) and may represent the integration of incomplete proviral genomes and the presence of partial unintegrated sequences. Lane M contains molecular size standards (sizes, in kilobases, are on the left).

Investigation of FIV core antigen expression in tumor cells.

We used immunocytochemical staining to determine whether the FIV capsid protein, p24, was expressed in tumor cells. Cells were dried onto flat-bottom, 96-well plates (5 × 10⁴ cells/well) by incubation at 37°C overnight and then fixed and stained with an anti-p24 monoclonal antibody (Serotec Ltd., Oxford, United Kingdom) as described previously (37). FL4 cells and the uninfected feline lymphoid cell line MYA-1 (33) served as positive and negative controls, respectively. While FL4 cells were strongly positive and MYA-1 cells were negative for FIV p24 antigen, we could not detect FIV p24 antigen expression in tumor cells (data not shown). This result suggests that the integrated provirus is not expressed in the tumor cells but does not exclude the possibility of a transcriptionally active but defective provirus.

Concluding remarks.

The most significant finding of this study is the demonstration of a monoclonally integrated FIV provirus in tumor DNA. This observation indicates that a single infection event preceded clonal expansion of the transformed cell and raises the possibility of a direct oncogenic role for FIV.

Although a direct role has been postulated for HIV-1 in the initiation of B-cell lymphoma and the results of some studies (2, 17), including the demonstration that HIV-1 has transforming properties in vitro (28), are consistent with this theory, Southern blot analysis of lymphoma tissue has to date failed to lend support. In contrast, a single-copy clonally integrated HIV-1 provirus was identified by Herndier and colleagues in a rare HIV-1-associated T-cell lymphoma (22).

At this stage, we can only speculate as to the mechanism(s) by which FIV may have contributed to cell transformation in this case. Important in this regard is the observation that viral gene expression could not be detected in tumor cells. This result suggests that, while FIV may be involved in an early lymphomagenic event, viral gene expression is not necessary to maintain the transformed phenotype. Based on our findings, FIV has features in common with the human T-cell leukemia virus–bovine leukemia virus group, which typically induces tumors with clonally integrated but transcriptionally inactive provirus (6, 10). The favored explanation for the oncogenic action of the human T-cell leukemia virus–bovine leukemia virus group is an initiation-progression model in which early expression of a viral oncoprotein, Tax, induces cellular proliferation but expression is lost in malignant cells where secondary mutations fix the transformed phenotype (11, 36).

Alternatively, a transcriptionally active, defective FIV provirus might be acting as an insertional mutagen analogous to the type C retroviruses (27) and as postulated for HIV-1 integration within fur upstream of the c-fes/fps proto-oncogene (48). Transduction of a cellular oncogene has never been shown for a lentivirus but remains a theoretical possibility in the light of the rapid clinical course and aggressive nature of this lymphoma. Further analysis of virus transcription and the provirus integration site should help to distinguish between these possible mechanisms.

Although we have demonstrated proviral integration in a single case of FIV-associated lymphoma, only a relatively small number of cases have been analyzed at the molecular level to date (50). More extensive analysis will be necessary to determine the frequency with which this occurs and its wider implications for lentiviral oncogenesis.