Although Burkitt’s lymphoma (BL) is a readily treated malignancy, recurrences, as well as disease arising in immunosuppressed patients, are notoriously resistant to conventional therapeutic approaches. The EBV is associated with a significant proportion of these lymphomas that evade immune surveillance through decreased expression of both viral and cellular antigens. Increasing the immunogenicity of BL cells may, therefore, represent a potentially beneficial therapeutic maneuver. Using in vitro models of EBV-transformed lymphoblastoid as well as BL cell lines, we demonstrate increased expression of genes coding for HLA class I and EBV latent proteins by the differentiation inducer phenylbutyrate (PB). The aromatic fatty acid also caused cytostasis associated with sustained declines in c-myc expression, a direct antitumor effect that was independent of the EBV status. We conclude, therefore, that differentiation therapy of BL with PB may lead to growth arrest with increased tumor immunogenicity in vivo. The findings may have clinical relevance because the in vitro activity has been observed with PB concentrations that are well tolerated and nonimmunosuppressive in humans, a desirable feature for the different patient populations afflicted with this disease.
BL,3 previously recognized mainly as an endemic childhood disease in Africa, but uncommon in developed countries, has become more prevalent with the spread of the AIDS. An aggressively growing, small noncleaved B-cell malignancy frequently associated with infection by the EBV, BL is often curable by cytotoxic chemotherapy in immunocompetent patients (1, 2). Despite high response rates, however, a significant proportion of patients experience relapse with a multidrug-resistant phenotype that carries a poor prognosis. Similarly, the high-grade lymphomas encountered in the setting of immunosuppression, induced by either the human immunodeficiency virus or therapeutic immunosuppressants, are notoriously resistant to conventional therapy (3). In the later context, less myelosuppressive regimens with new mechanisms of antitumor activity would be highly desirable.
Several molecular aspects that are characteristic of BL biology may lend themselves to therapeutic intervention. BL cells invariably harbor a deregulated c-myc, a nuclear phosphoprotein involved in the control of cell proliferation and differentiation. Chromosomal translocation that juxtapose the coding region of c-myc to one of the immunoglobulin genes, as well mutations in both its coding and regulatory sequences, has been observed (1, 4, 5). In addition to cellular oncogenes, latent infection by the EBV can also contribute to neoplastic transformation of B cells (1, 2, 3). In the significant proportion of EBV-infected BL, tumor cells have been shown to evade immune surveillance by cytotoxic T cells through decreased or absent expression of the EBNAs and LMPs, as well as declines in the cellular HLA class I antigens (6, 7, 8, 9, 10). Increasing the immunogenicity of BL cells may, therefore, represent a potentially beneficial therapeutic maneuver. Moreover, although typically latent, a lytic cycle can be induced in infected BL cells, providing an additional mechanism of cell kill. BL can, thus, be conceptualized a disorder of differentiation associated with aggressive tumor growth and evasion of the immune system. These could be reversed by targeting cellular and viral genes using gene therapy (11) or pharmacological agents (12, 13, 14, 15, 16). Aromatic fatty acids, the prototype of which is PB, make up one such class of therapeutic compounds.
PB and its metabolite, PA, are novel antitumor agents with a low toxicity profile in humans (17, 18, 19, 20, 21). In preclinical models, PB and PA cause cytostasis and differentiation of various hematological and solid neoplasms, including multidrug-resistant tumors (17, 18, 22, 23, 24, 25, 26, 27). Changes in tumor biology have been associated with modulation of gene expression occurring at both the transcriptional and posttranslational levels (28, 23). Of relevance to BL, increases in tumor antigens have been documented in treated melanoma and prostate cultures (18, 24). It was of interest, therefore, to characterize the effect of PB on cellular and viral activity in BL- and EBV-transformed LCLs.
MATERIALS AND METHODS
Cell Culture and Reagents.
The BL cell lines and LCLs used in this study are described in Table 1 (see also Refs. 29, 30, 31, 32). Stock cultures were maintained in RPMI 1640 supplemented with 15% heat-inactivated FCS (BioWhittaker Inc., Walkersville, MD), penicillin, and streptomycin and incubated at 37°C in a humidified, 10% CO2 atmosphere. For the duration of the experiments, cells were plated at a density of 1–1.5 × 105 cell/ml in 10% FCS, unless otherwise indicated. Growth inhibition curves were obtained by cell enumeration using a hemocytometer, and cell viability was determined by trypan blue exclusion. The sodium salts of phenylacetic acid and phenylbutyric acid (provided by Elan Pharmaceutical Research Corporation, Gainesville, GA) were dissolved in distilled water, and aliquots were stored at −20°C. For studies involving cell attachment to fibronectin, human fibronectin (Biomedical Technologies Inc., Stoughton, MA) was dissolved in 2 M urea to 0.1 mg/ml, aliquoted, and stored at −20°C, unless used immediately. Tissue culture plates (100 mm) were coated with fibronectin (3.5 ml, 20 mg/ml in PBS) by incubation at room temperature for 1–3 h and washed with PBS before cell plating.
RNA Isolation and Northern Blot Analysis.
Cytoplasmic RNA was isolated from cells lysed in buffer containing NP40 (0.5%) and analyzed as described previously (33). Briefly, RNA samples (10 μg) were fractionated by electrophoresis through 1% agarose-formaldehyde gels, transferred onto nitrocellulose, and hybridized with 32P-labeled probes. The membranes were stripped and rehybridized following standard methods. Quantitation was by PhosphoImager (Molecular Dynamics). Probes were: (a) the ClaI-EcoRI fragment of c-myc (34); (b) the HindIII/EcoRI fragment of HLA-A3, which is homologous to HLA A11, for HLA class I (35); and (c) the 1.5-kb EcoRI fragment of Cμ, for μ heavy chain. LMP (7) and Zebra (36) probes were prepared by the PCR from AKR and Raji DNA, respectively. All probes were labeled with 32P-dCTP using the Random Prime kit (Bethesda Research Laboratories, Bethesda, MD).
Detection of EBV DNA Replicating (Linear) Forms.
Cells were incubated in proteinase K containing lysis buffer and subjected to Southern blot analysis, according to published procedures (33). Briefly, nucleic acids were recovered by phenol and chloroform extraction, precipitated in ethanol, and dissolved in distilled water. BamHI restriction digests of DNA samples (10 μg) were electrophoresed through 1% agarose gel and transferred onto nylon membranes (Nytran). The probe used to detect EBV linear forms was a 5.2-kb BamHI-EcoRI fragment of the NJ-het region of the circular EBV genome, which contains unique sequences from both ends of the linear genome and tandem repeats with the LTR (37).
Pellets from PB-treated and untreated cells were fixed in 2% glutaraldehyde in PBS and then postfixed for 12 h in 1% osmium tetroxide at 4°C. Specimens were counterstained with uranyl acetate and lead citrate before examination by transmission electron microscopy.
Growth Inhibition of BL Cells and LCLs by PB.
A wide range in sensitivity to PB was noted (Fig. 1). Typically, 50–80% growth arrest was evident after 3 days of continuous exposure to 1–2 mm PB, drug concentrations that are within the human pharmacological range. A representative dose-response curve (Daudi, IC50 1 + 0.2 mm) is shown in Fig. 1. Compared with PB, significantly higher concentrations of the metabolite PA were required for tumor cytostasis (Daudi, IC50 5 + 0.4 mm; see Fig. 1). In additional studies aimed to reproduce the rapid growth rate of BL in vivo, cultures were split daily to maintain logarithmic growth. Under these conditions, the Daudi cells failed to grow after a few days of exposure to 2 mm PB (Fig. 2), whereas the less sensitive CA46 cells sustained their ability to grow, albeit at a slower rate. In the sensitive lines JLP, AG876, ST486, and MC116, treatment with PB also reduced cell viability by 15–50% compared with untreated controls, as determined by trypan blue exclusion. No correlation was found between the sensitivity to PB and the presence of the EBV in cells. In all cases, cytostasis was dependent on the continuous presence of the drug, and cells resumed proliferation within 48 h after treatment was discontinued (data not shown).
Morphological Changes of PB-treated Cells.
In vitro, many LCLs and BL cells (e.g., AG876 and JLP) form aggregates of increasing size as a function of cell density over time (38). This phenomenon has been associated with expression of the transforming proteins LMP and EBNA-2 phenotype. Formation of tumor aggregates was enhanced by growth-inhibiting concentrations of PB (2 mm), despite the lower cell density of treated cultures. Cells excluded from the aggregates were dead, as determined by trypan blue exclusion. PB induced aggregate formation, although smaller is size, even in cultures that usually grow as single cell suspension. No changes were noted in the resistant line CA46.
The most prominent changes in both morphology and cell behavior were observed in Namalwa cultures. While untreated cells grew in suspension (Fig. 3,A), a subpopulation of treated cells adhered to the plastic and projected cytoplasmic processes several-fold longer than the cell body diameter (Fig. 3,B). Precoating the dish with fibronectin further enhanced this drug-induced phenomenon, resulting in the formation of cellular monolayer. In contrast, in untreated cultures only few cells adhered lightly to the fibronectin coat and did not form cytoplasmic processes. These morphological alterations were studied by electron microscopy (Fig. 3, C and D), which demonstrated a more abundant endoplasmic reticulum and a decreased nucleus:cytoplasm ratio, characteristic of a more mature, plasma cell-like phenotype.
Changes in Cellular Gene Expression.
Pharmacologically induced changes in the expression of c-myc, HLA-A3 (class I MHC gene), and the μ heavy chain are of specific interest in BL. In tissue culture, expression of these genes may be affected by cell density itself. Thus, steady-state levels of the respective mRNA species were determined in both low- and high-density cell cultures after 2 and 4 days of treatment with PB, respectively. As shown in Fig. 4, the levels of c-myc transcripts were reduced in 4-day cultures compared with 2-day cultures, regardless of whether treatment was applied. However, because cell density had remained low in PB-treated cultures, the decline in c-myc could not be explained solely on the basis of changes in cell density. The inhibitory effect of PB on cell proliferation and myc expression were maintained for the duration of treatment, which is in marked contrast to the transient effect seen with other differentiation inducers (39).
In untreated cultures, HLA class I gene expression increased up to 3-fold as a function of cell density (Fig. 4). In contrast, PB induced a 3-fold increase in the amounts of HLA class I surface antigens, as determined by fluorescence-activated cell sorter analysis of JLP and AKR cells,4 whereas cell density remained low. In Namalwa cells, PB also increased the expression of the μ heavy chain by 3–5 fold (data not shown). These changes in c-myc, HLA, and μ heavy chain are consistent with the more extensive endoplasmic reticulum imaged by electron microscopy and suggest that cellular differentiation has been induced by PB.
Activation of EBV DNA Replication in Response to PB.
Cultured BL cells that are latently infected with EBV are permissive for virus replication. This is associated with a change in EBV DNA conformation from circular to linear, the expression of replication antigens, and the induction of a lytic cycle. In a small number of BL lines (e.g., P3HR1), EBV particles are being produced in 2–10% of the cells (40). Spontaneous abortive cycle can also occur in nonproducer lines (41). Because these phenomena can be enhanced by n-butyrate and phorbol esters (16), it was of interest to characterize the effect of PB on the activation of EBV replication. EBV DNA replication can be identified by the presence of LTR containing “linear” DNA rather than circular plasmid. Digestion of the linear DNA with BamHI yields fragments that contain varying numbers of the tandem repeat unit with the LTR and can, thus, be separated by size from the larger fragment that originates from the fused termini of the EBV plasmid (42). On treatment with PB, a quantitative increase in the linear form of EBV DNA was detected in two cell lines (AKR and P3HR1) of the three in which viral DNA synthesis could be detected before treatment (Fig. 5). In addition, mRNA levels coding for Zebra, a lytic cycle inducer protein (36), were significantly augmented in the PB-treated P3HR1 cultures (Fig. 6 B). Activation of a lytic cycle by PB may be limited to EBV-producing cells because the drug failed to induce de novo EBV DNA synthesis in nonproducing lines.
Up-Regulation of Viral LMP Expression.
Although several viral genes coding for latent proteins such as LMP and EBNA-2 can be activated in BL and EBV-transformed LCLs on propagation in tissue culture, their expression is highly restricted in vivo (7, 8). In EBV-positive lines expressing LMP transcripts, the specific mRNA levels increased in untreated cultures with increased cell density. However, LMP expression was further increased by PB treatment, whereas cell density remained low (Fig. 6 A). In treated P3HR1, a cell line that harbors a heterogeneous population of EBV strains with a partially deleted, transformation-defective genome, we noted the induction of two LMP mRNA species smaller in size compared with the gene transcript of the other cell lines. The increases in viral LMP and cellular HLA could augment tumor immunogenicity in vivo.
Using in vitro models, we herein show that PB can induce cytostasis and maturation of various BL cells. The antitumor activity seems to be independent of the degree of cell differentiation before treatment, oncogene activation, or the presence of the EBV. In addition to modulating the expression of cellular genes implicated in deregulated growth and evasion of the immune system (c-myc and HLA class I), PB also up-regulated the expression of EBV genes, such as LMP and Zebra, which promote BL tumor immunogenicity and cell lysis in vivo. Of note is the fact that this activity was observed with drug concentrations that are pharmacologically achievable in humans.
Considering the genetic heterogeneity of BL tumors in patients, we have characterized the antitumor activity of PB in 11 cell lines that differ in their growth characteristic, morphology, and homotypic interactions, as well as in the mode of c-myc deregulation and the expression of sIg. Of these cell lines, seven harbored the EBV genome (Table 1 and Ref. 34). Significant inhibition of tumor cell proliferation by pharmacological concentrations of PB (1–2 mm) was documented in 10 BL cell lines and LCLs. Consistent with previous findings with other tumor types (17, 18, 22, 23, 24, 25, 26, 27), cytostasis was not limited to virally infected cells, nor did the presence of the virus increase tumor vulnerability in vitro. In contrast to other tumor models, in which the ability of the aromatic fatty acids to induce cytostasis and terminal differentiation correlated with the degree of cell maturation before treatment (18), no such correlation was found in the BL lines. The immature, sIg-negative Raji line had an intermediate cytostatic response to PB, whereas the more sensitive ST486 and the relatively resistant CA46 both express sIgM. In all cases, the cytostatic effect was dependent on the continuous presence of the drug. Although no terminal differentiation was induced in the BL cells, there were marked changes observed within 4 days of treatment. These included plasma membrane ruffling, excessive cell aggregation indicative of modifications of homotypic interactions via cell surface molecules, increases in HLA and μ, and significant declines in c-myc. Changes in HLA and myc were also seen in untreated cultures after they had reached a stationary stage due to high cell densities. In the PB cultures, however, such molecular changes occurred in low-density cultures, suggesting that these are drug-induced. Moreover, cell aggregation and marked phenotypic changes consistent with maturation toward a plasma cell, as documented in the Namalwa cultures, were induced only with treatment.
In addition to cytostasis and differentiation, high concentrations of PB can promote apoptosis of leukemic cells (22).5 At pharmacologically achievable concentrations, there was no significant cell death induced in the majority of tested BL lines. Tumor cell kill may, however, be achieved in vivo, resulting from activation of tumor and viral antigens. Concomitant to the declines in c-myc, treated cells had increased HLA class I mRNA levels in response to PB, which can augment responses to the immune system or immunotoxin therapy. Our findings are in agreement with previous studies, indicating that HLA class I expression is inversely related to that of various members of the myc gene family (43). In addition to HLA, the drug increased the expression of LMP, an EBV transmembrane protein implicated in immune recognition, in all of the cell lines that harbored EBV. We also detected an increase in EBV DNA replication of both P3HR1 and B95-8 strains in response to PB. Viral DNA replication implies preceding synthesis of the immunogenic early EBV lytic cycle proteins, as confirmed by the up-regulation of Zebra transcripts observed PB-treated P3HR1 cells. Although the induction of a lytic cycle by differentiation agents is often aborted (12), recent evidence suggests that early lytic cycle proteins may contribute to the recognition of BL cells by CTLs through HLA class I molecules presenting self peptides (9). High circulating concentrations of antibodies to these antigens, as well as the presence of linear replicative forms of EBV DNA in tumor biopsies from BL patients, provide evidence that this phenomenon elicits an immune response in vivo (6). In addition, patients who fail to develop, or who lose immunity against these antigens, often go on to develop BL clinically (1, 3). Increases in specific EBV and cell recognition antigens induced by PB may, thus, improve the potential of eliciting a HLA class I-restricted, anti-EBV-specific cytotoxic immune response in vivo (8, 9). This immunogenicity of PB-treated tumor cells is currently under investigation. In humans, administration of large doses of PB could lead to tumor cell death also by depletion of circulating glutamine, an amino acid that is essential for the survival of malignant B cells (44). However, the in vitro activity cannot be explained by glutamine starvation because conjugation of the amino acid is limited to the liver and kidney of high primates. Other mechanisms may account for the molecular and biological changes observed. Both PB and PA can inhibit DNA methylation and up-regulate the expression of methylation-dependent genes (21, 45). Expression of the viral LMP and cellular HLA can, indeed, be augmented by hypomethylating agents such as 5-azacytidine and the differentiation inducer n-butyrate (46, 16). Transcriptional control by the aromatic fatty acids has been linked to activation of the PPAR subtypes α and γ (28, 47), members of the steroid nuclear receptor superfamily that bind to specific PPAR response elements. Differences in the baseline expression of these PPARs in cancer cells have recently been correlated with tumor sensitivity to growth arrest induced by PB and related aromatic fatty acids.6 The heterogeneity in tumor response noted among the BL cell lines tested may, thus, be related to differences in the expression of PPARs and/or other transcription factors. Sharing some other mechanisms of action with the aliphatic fatty acid PB can also affect histone acetylation (48) and the expression of genes driven by butyrate-response element (45). As would be expected, the biological end points brought by these two compounds in BL cell lines are similar: like PB, butyrate has been shown to inhibit cellular DNA synthesis, down-regulate c-myc expression, induce the expression of various differentiation markers, and up-regulate latent, as well as lytic, EBV gene transcription (12, 13, 14, 15, 16). PB offers some advantages for clinical use, including the ability to sustain plasma concentrations of PB in humans (20, 48) with an extended half-life (PB,1–2 h; butyrate, 5–6 min) and unique pleiotropic mechanisms of action that include PPAR activation, inhibition of protein prenylation (23), and depletion of circulating glutamine.
Taken together, our findings document the ability of pharmacological concentrations of PB to induce cytostasis and maturation of BL cells, an effect associated with changes in gene expression and dependent on continuous drug administration. Increases in the expression of EBV antigens and cellular recognition molecules may enhance BL tumor cell immunogenicity, leading to tumor eradication in vivo. In humans, tumor cell death could result also from glutamine starvation. PB is currently in Phase I clinical trials involving adults and children with cancer (20, 48, 49, 50, 51, 52). Of particular interest are recent findings that show complete clinical and cytogenetic remission, associated with histone hyperacetylation, in a child with acute promyelocytic leukemia who was treated with PB (48). In view of the antitumor activity and lack of immunosuppressive effects, we propose that PB be tested, alone or in combination with other therapeutic approaches, for the treatment of lymphoid malignancies such as BL and AIDS-associated lymphoproliferative disorders.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Supported in part by funds from Elan Pharmaceutical Research Corporation.
The abbreviations used are: BL, Burkitt’s lymphoma; LCL, lymphoblastoid cell line; PB, phenylbutyrate; PA, phenylacetate; LMP, latent membrane protein; EBNA, EBV nuclear antigen; LTR, long terminal repeat; PPAR, peroxisome proliferator-activated receptor; sIg, surface immunoglobulin.
D. Samid, unpublished observations.
J. A. DiGuisepe, L-J. Weng, K. H. Yu, S. Fu, M. B. Kastan, D. Samid, and S. D. Gore. Phenylbutyrate-induced G1 arrest and apoptosis in myeloid leukemia cells: structure-function analysis, submitted for publication.
D. Samid, M. Wells, M. Greene, W. Shen, C. N. A. Palmer, and A. Thibault. PPARγ as a novel target in cancer therapy: binding and activation by an aromatic fatty acid with clinical antitumor activity, submitted for publication.
|Cell line .||Type .||EBV .||sIg .||Reference .|
|Cell line .||Type .||EBV .||sIg .||Reference .|
We thank Kishor Bahtia and Michael Kuel for helpful discussions and Michelle Zavitz for technical assistance.