Abstract
Epigenetic mechanisms are involved in regulating chromatin structure and gene expression through repression. In this study, we show that histone deacetylase inhibitors (DAIs) that alter the acetylation of histones in chromatin enhance the expression of several genes on tumor cells including: MHC class I, II, and the costimulatory molecule CD40. Enhanced transcription results in a significant increase in protein expression on the tumor cell surface, and expression can be elicited on some tumors that are unresponsive to IFN-γ. The magnitude of induction of these genes cannot be explained by the effect of DAIs on the cell cycle or enhanced apoptosis. Induction of class II genes by DAIs was accompanied by activation of a repressed class II transactivator gene in a plasma cell tumor but, in several other tumor cell lines, class II was induced in the apparent absence of class II transactivator transcripts. These findings also suggest that the abnormalities observed in some tumors in the expression of genes critical to tumor immunity may result from epigenetic alterations in chromatin and gene regulation in addition to well-established mutational mechanisms.
Although the majority of normal cells express surface MHC class I Ags, most normal nonhemopoietic cells express MHC class II only after activation by IFN-γ. However, a few cell types, including mature oligodendrocytes, sensory neurons, plasma cells, and trophoblasts, cannot be induced to express class II by IFN-γ. MHC class II expression is tightly regulated primarily at the level of transcription by multiple transcriptional activators binding to the W/S, X, and Y box promoter sequences. In addition, coactivators, such as class II transactivator (CIITA),3 Bob1, and CBP/p300, play important roles in the transcriptional regulation of class II (1, 2, 3). Recently, CIITA has been reported to facilitate IFN-γ induction of MHC class I in addition to class II Ags (4, 5), and this, together with the similarities in the structure of the class I and II promoters (6), suggests a commonality in the regulation of MHC genes.
Many tumors are MHC class II negative and, although some can be induced to express class II by IFN-γ, a subset of tumors is noninducible by cytokines. In a few noninducible tumors, defects in IFN-γ receptors or Janus kinase-STAT signaling pathway components have been described and have been associated with tumor progression (7). In plasmacytoma cell lines, and in selected melanomas and hepatocellular carcinomas, the lack of expression of all class II isotypes appears to be due to a defect in the transcription of the CIITA gene. In these cells, transfection of CIITA results in class II expression (8, 9) and, in certain tumors, this restores their ability to present intact Ag (9). That MHC class II may play a role in tumor immunity is suggested by studies demonstrating that transfection of MHC class II-negative tumor cells with class II genes inhibits tumorigenicity and elicits immunity to subsequent challenge with wild-type tumor cells (10). However, induced expression of MHC class II does not always enhance immunity and, in the absence of costimulatory factors, may even promote tumor progression by inducing anergy (11). Thus, differences in the functional role of class II may be related to the context in which the MHC genes are expressed in different tumors, i.e., the presence of associated “escape” defects (reviewed in Refs. 12, 13).
Considerable evidence has accumulated recently that gene expression is influenced by epigenetic mechanisms that alter chromatin structure (14). Acetylation of lysine residues on the N-terminal tails of the basic core histones in nucleosomes is generally associated with enhanced transcription, whereas deacetylation compacts chromatin and represses transcription. Thus, while targeting molecules with histone acetyltransferase activity to promoters results in gene activation, repression of certain genes is accomplished by transcriptional repressors and corepressors that bind and target histone deacetylases (HDACs) (15, 16). Several noncompetitive inhibitors of HDACs have been described, including sodium butyrate (NaBu) and the highly specific inhibitor trichostatin A (TSA), that arrest cells in both the G1 and G2-M phases of the cell cycle (17), influence growth, and induce apoptosis (18). These inhibitors have been shown to activate a number of important genes by altering histone acetylation and chromatin structure, although not all genes are activated by HDAC inhibitors (DAIs), many are unaffected and some are inhibited (19).
We show here that MHC class II mRNA and cell surface protein can be induced by DAIs in mouse and human tumor cell lines that cannot be activated to express class II by IFN-γ. Evidence is presented that, in some tumors, DAIs activate a repressed CIITA gene, whereas, in others, induction may occur via a route different from the classical CIITA-mediated pathway. Importantly, DAIs are also shown to enhance the constitutive expression of MHC class I, as well as the costimulatory molecule CD40, on some tumor cells that were not induced by IFN-γ. These data also suggest that failure of tumors to express certain molecules critical to immunity (tumor escape) may result from chromatin repression, which is reversed by HDAC inhibitors.
Materials and Methods
Cells and reagents
The human neuroblastoma tumor cell line SK-N-MC (MC) and mouse plasmacytoma J558 cell line were obtained from American Type Culture Collection (Manassas, VA). Mouse adenocarcinoma Colon 26, generously provided by Elizabeth A. Repasky (Roswell Park Cancer Institute, Buffalo, NY), was maintained in dMEM growth media supplemented with 10% FCS (Life Technologies, Grand Island, NY) as a cell line derived from the original transplantable tumor. RJ2.2.5 cells were a generous gift from Cheong-Hee Chang (University of Michigan, Ann Arbor, MI). Stock and working dilutions of TSA (Wako Biochemicals, Osaka, Japan) were prepared using ethanol as a diluent. Abs to murine surface markers were obtained from the following suppliers: 1) PharMingen (San Diego, CA): PE-conjugated anti-H-2Dd (clone 34-2-12), I-Ad (AMS-32.1), CD40 (3/23), CD44 (IM7), anti-CD45R (RA3-6B2), B7 integrin (M293), strepavidin; biotinylated anti-CD25 (7D4), CD29 (HA2/5), CD119 (GR20); purified anti-CD104 (346-11A), and FITC-anti-H-2Dd (34-2-12); 2) Caltag (South San Francisco, CA): PE-conjugated anti-H-2Kk (CTKk), I-Ak (14V.18), CD62L (MEL-14), CD80 (RMMP-1), CD86 (RMMP-2); FITC- and RPE-annexin-V; and 3) Southern Biotechnology Associates (Birmingham, AL): PE-conjugated anti-CD54 (KAT-1), CD154 (MRI). Flow cytometry reagents for human cells, PE-anti HLA-A, B, C (clone G46-2.6) were obtained from PharMingen, PE-anti HLA-DR (clone 243) from Becton Dickinson (Mountain View, CA), and PE-anti CD40 (clone mAb 89) from Immunotech/Coulter (Miami, FL).
Inhibition of HDAC activity
Toxicity at various concentrations and incubation times with TSA and NaBu were first determined using three techniques: 1) trypan-blue dye exclusion; 2) apoptosis as measured by DNA laddering on agarose gels; and 3) flow cytometric analysis with propidium iodide (PI) or annexin-V. Preliminary titration was performed on each cell line at 5, 10, 25, 50, 100, and 250 nM TSA for 6, 12, 24, 48, and 72 h and with the MC cell line at 1 mM NaBu for 12, 24, and 48 h, and concentrations and incubation times were selected to exclude toxicity and maximize expression. IFN-γ was titrated at 10, 50, 100, and 500 U/ml for each cell line. In most experiments, 100 U/ml was employed.
RT and quantitative real-time PCR
Standard RT-PCR was performed as previously described using primers for human HLA-DRα and CIITA (20) and mouse IAα (21). Other primers used included human and mouse GAPDH (Clontech Laboratories, Palo Alto, CA) and 5′-ACACCTGGACCTGGACTCAC-3′ (forward) and 5′-GCTCTTGGCTCCTTTGTCAC-3′ (reverse) for mouse CIITA, 5′-GAAACTGGTGAGTGACTGC-3′ (forward) and 5′-CACATTGGAGAAGAAGCC-3′ (reverse) for human CD40 (22), and 5′-GTTTAAAGTCCCGGATGCGA-3′ (forward) and 5′-CTCAAGGCTATGCTGTCTGT-3′ (reverse) for mouse CD40 (23).
The real-time fluorescent PCR technique was used to quantitate fold changes in GAPDH, CIITA, MHC class II, and CD40. The detection system and quantitative PCR have been described in detail elsewhere (24). The threshold cycle number (CT) corresponds to the cycle number at which the fluorescent emission reaches 10 SDs above the mean fluorescent emission measured during the early “baseline” cycles of the PCR. The CT values for triplicates were averaged and normalized to GAPDH levels (i.e., average CT for HLA-DR on sample cDNA minus average CT for GAPDH on same sample = normalized CT or ΔCT for that sample). Assuming reverse transcription efficiencies were approximately the same for each sample, this allowed us to determine relative amounts of mRNA using the comparative CT (ΔΔCT) method (Perkin-Elmer Taqman User Bulletin 2). Fold changes in mRNA levels were calculated as 2x, where x = the difference between the GAPDH normalized CT values of the control and experimental samples.
Validation experiments showed a linear relationship between input cDNA and CT values over a range of serial dilutions of cDNA extending from 1 to 10−6. The slopes of these lines generated for each primer/probe set indicated that PCR efficiencies for each of the primer sets were close to 100% and validated the use of the ΔΔCT (comparative CT) method for relative quantitation. These experiments also demonstrated the linear range of amplification and allowed determination of the maximum CT value at which linear amplification was reliable. For most of our primer/probe sets, this value was 40; if the samples did not amplify by 40 cycles, the gene for that sample was considered not detectable (ND), because CT values beyond this point would not meet the criteria for calculation of fold change by the comparative CT method. In such cases, to calculate a minimum fold change for other samples in the same group, the CT value for the not detectable sample was arbitrarily set at 40. Primer sets for real-time PCR are available at http://realtime.nwu.edu.
Flow cytometry
Flow cytometric analyses were conducted by published methods (25) with cells fixed with 1.0% paraformaldehyde and analyzed on a FACScan (Becton Dickinson). Cell cycle analyses were conducted by flow cytometry with PI staining of genomic DNA. Cells were fixed and permeabilized in methanol, treated with RNaseA and PI and analyzed for DNA content by determination of FL3-area on the FACScan. The FL3-area data were analyzed with the ModFit software for determination of the percent cells in each phase of the cell cycle. In some cases, FITC-conjugated Ab staining was conducted before PI treatment and surface staining was analyzed as a function of the cell cycle phase. Apoptosis was analyzed by staining with annexin V directly conjugated with either FITC or R-PE and by sub-G0 analysis of DNA staining. Data are representative of at least three independent experiments.
Results
Induction of MHC class II in human and mouse tumor cells
Our experiments were initially designed to explore whether the failure of certain tumor cells to express MHC class II might be a result of a repressive chromatin structure. Therefore, we tested the effect of treatment with the DAIs, NaBu and TSA, on the activation of class II genes in selected tumor cell lines. As shown in Fig. 1⇓, treatment of the human neuroblastoma cell line SK-N-MC (MC) with 100 nM TSA or 1 mM NaBu resulted in a substantial increase in the levels of mRNA for HLA-DR (DR) as measured by RT-PCR, whereas CIITA transcription remained not detectable. Levels of mRNA for the housekeeping gene, GAPDH, were not significantly changed after treatment. Low levels of mRNA for the DP and DQ isotypes of MHC class II were also induced by TSA (data not shown). Although TSA and NaBu activate transcription of MHC class II in MC tumor cells, these cells cannot be induced to express mRNA for DR by IFN-γ (Fig. 1⇓A). This is not due to the failure of MC cells to express IFN-γ receptors or to a defect in the IFN-γ signaling pathway, because mRNA for two prototype IFN-γ-inducible genes, IFN regulatory factor-1 and guanylate binding protein, is enhanced by IFN-γ treatment (data not shown). Essentially identical results to those of MC were obtained with JAR, a human trophoblast cell line, that was refractory to IFN-γ but inducible for class II by TSA and NaBu.
Activation of MHC class II gene expression by IFN-γ and DAIs in the SK-N-MC human neuroblastoma cell line. A, MC cells were treated with 100 U/ml of IFN-γ, 100 nM TSA, or 1 mM NaBu and mRNA for HLA-DR amplified by RT-PCR at the time of maximal expression (24 h for TSA and 48 h for NaBu). DR is induced in MC by TSA in the absence of detectable mRNA for CIITA. Raji is a constitutively positive B cell control. B, Real-time PCR experiments on the same RNA preparations demonstrate that TSA induced a 108-fold and NaBu an 88-fold increase in mRNA for DR, whereas IFN-γ failed to elicit significant increases in three separate experiments. CIITA mRNA was not detectable (ND) at 60 cycles in controls or after TSA or NaBu treatment. IFN-γ consistently induced low levels of CIITA mRNA that were observed (fold changes from 3.5 to 11) by real-time PCR in three experiments. GAPDH mRNA remained essentially unchanged in both the RT-PCR and real-time PCR experiments (data not shown).
Real-time fluorescent PCR, capable of detecting low levels of mRNA, was employed, and fold changes in transcript levels were calculated using CT values standardized to GAPDH (see Materials and Methods). As shown in Fig. 1⇑B, treatment of MC cells with 100 nM TSA for 24 h resulted in a 108-fold increase in expression of DR, but no increase in CIITA was detectable after 60 cycles of PCR. NaBu treatment (1 mM, 48 h) resulted in an 88-fold enhancement in DR and no activation of CIITA (Fig. 1⇑B). Thus, within the limits of these quantitative measurements, the activation of DR gene expression by DAIs appears to occur with no change in the level of CIITA transcripts. Although IFN-γ did not induce HLA-DR transcripts on MC cells at any concentration or time point studied, this cytokine did elicit low levels of CIITA mRNA detectable by real-time PCR. Preliminary studies showed that the sequence of transcripts stimulated by IFN-γ in MC cells are very similar to a minor alternatively spliced Raji transcript, which has a deletion in the N-terminal activation domain, and is functionally inactive (26).
To determine whether the effect of DAIs on MHC class II was manifested in other tumors and in different species, we tested the effect of TSA on two commonly used mouse tumor models: the Colon 26 adenocarcinoma and the J558 plasmacytoma. Analysis by standard RT-PCR at 35 cycles showed that IFN-γ induced expression of CIITA and class II in Colon 26 and, similar to the human MC cell line, CIITA was not detected after TSA treatment (Fig. 2⇓A). Real-time PCR analysis of Colon 26 cells was consistent with the RT-PCR data and demonstrated that TSA induced a 74-fold increase in Ia in the absence of CIITA, whereas IFN-γ elicited both CIITA (20-fold) and Ia (58-fold) expression. The J558 cell line behaved quite differently from Colon 26 in that IFN-γ did not activate either CIITA or class II expression (Fig. 2⇓B). Because the classical IFN-γ-inducible genes IFN regulatory factor-1 and guanylate binding protein, as well as MHC class I mRNA, were not induced by IFN-γ in J558 (data not shown), this cell line may have a signaling defect in the IFN-γ pathway. However, unlike either the MC or Colon 26 cells, TSA induced both CIITA (102-fold) and Ia (537-fold) in J558 as shown in Fig. 2⇓. In different experiments, maximum expression at 12 h occurred at TSA concentrations between 50 and 100 nM and at 24 h between 25 and 50 nM. At higher concentrations or longer incubation times, decreasing expression was associated with increasing apoptosis. The contrasting patterns in regard to CIITA induction suggest the possibility that TSA may function by both CIITA-dependent (in J558) and -independent (in MC and Colon 26) mechanisms in different cell types.
Contrasting patterns of CIITA and class II expression in response to IFN-γ and TSA in two murine cell lines. A, Colon 26 cells treated for 48 h were examined by RT-PCR (upper left) and real-time PCR (upper right). IFN-γ induced expression of both CIITA (20-fold) and IAα (58-fold). TSA elicited mRNA for IAα (74-fold), whereas CIITA remained at control levels and was not detectable (ND) at 60 cycles. B, J558 cells treated for 24 h were not responsive to IFN-γ, but CIITA was induced (102-fold) and IAα (537-fold) by TSA. A20 is a constitutively positive B cell control.
To assess the ability of TSA to induce MHC class II expression in the absence of functional CIITA, we analyzed the human CIITA mutant cell line RJ2.2.5 (27) that has been thoroughly characterized as functionally deficient in CIITA. The RJ2.2.5 CIITA mutation results in deletion of the activation domain (28) of CIITA, and this cell line does not express HLA-DR in response to IFN-γ. TSA treatment of this cell line resulted in a 51-fold increase in mRNA for HLA-DR (Fig. 3⇓) and in the appearance of substantial levels of cell surface DR (data not shown). These data support the existence of a CIITA-independent TSA induction of MHC class II expression.
Activation of MHC class II gene expression by TSA in the RJ2.2.5 human B cell line deficient in CIITA activity. A, RJ2.2.5 cells were treated with 100 U/ml of IFN-γ, or 50 nM TSA and mRNA for HLA-DR amplified by RT-PCR. IFN-γ did not activate CIITA for HLA-DR. However, DR is induced in RJ2.2.5 by TSA in the absence of CIITA. Raji is a constitutively positive B cell control. B, Real-time PCR experiments on the same RNA preparations demonstrate that TSA induced a 51-fold increase in mRNA for DR, whereas IFN-γ failed to induce expression. Neither IFN-γ nor TSA induced CIITA.
Activation of the CD40 gene by TSA
Because of the importance of costimulatory molecules in the cellular activation mechanism involving MHC complexes, we explored whether DAIs altered the expression of B7-1, B7-2, or CD40 genes. Although no consistent changes in B7-1 and B7-2 were detected, CD40 expression was enhanced by TSA in all three cell lines (Fig. 4⇓). MC cells constitutively expressed low levels of CD40 and IFN-γ enhanced CD40 mRNA levels 5-fold, whereas TSA elicited a 12-fold increase by real-time PCR (Fig. 4⇓A). Colon 26 cells did not constitutively express CD40 and mRNA levels remained undetectable at 60 cycles after IFN-γ treatment, although TSA induced a 147-fold increase in CD40. J558 cells showed low constitutive levels of CD40 mRNA that did not change after treatment with IFN-γ, whereas TSA induced a 161-fold increase in CD40. The lack of induction of CD40 by IFN-γ in J558 cells is consistent with the failure of IFN-γ to activate other IFN-γ inducible genes. The failure of IFN-γ to induce CD40 mRNA in Colon 26, which has no apparent defect in the IFN-γ signaling pathway, is unexplained, but may be related to the ability of TSA to activate factors in addition to those elicited by IFN-γ that are required for CD40 gene expression.
Expression of the costimulatory receptor CD40 gene is enhanced by TSA. A, Neuroblastoma MC cells express low levels of CD40 mRNA in untreated control cells as detected by real-time PCR, but not by RT-PCR. CD40 expression was enhanced 5-fold by IFN-γ and 12-fold by TSA at 24 h. B, Colon 26 CD40 mRNA was not detectable by RT-PCR or real-time PCR in control cells. Treatment of Colon 26 with 100 nM TSA for 24 h, elicited maximum CD40 mRNA levels (147-fold induction). Following 100 U/ml of IFN-γ for 24 h, CD40 mRNA remained undetectable at 60 cycles. Experiments using Colon 26 cells at different time points (12 and 48 h) also showed enhanced CD40 with TSA but not with IFN-γ (data not shown). C, CD40 mRNA was detected in control samples of J558 cells and the low levels did not change significantly after incubation with IFN-γ. After treatment with TSA (25 nM for 24 h), CD40 was enhanced 161-fold. Two different experiments with each cell type yielded similar results.
Flow cytometry analysis of cell surface protein expression of MHC class I, II, and CD40 on tumor cell lines
To determine the effect of TSA on cell surface protein expression, the J558 cell line was subjected to FACScan analyses. Of 13 markers analyzed, only class I, II, and CD40 were enhanced at low concentrations of TSA. Levels of several nonexpressed Ags were not elicited (including CD25, β7 integrin, CD40L, H-2Kk, and I-Ak). Other markers were expressed, but were not altered (including CD44, CD54, CD80, CD86, and CD119) by TSA treatment. In each cell line tested, TSA significantly induced the expression of MHC class II protein on the cell surface with induction ranging from 3- to 21-fold in different cells, as indicated by the percent positive cells shown in Fig. 5⇓. The highest expression (21-fold induction) was achieved on J558 cells after 24 h of treatment with 25 nM TSA. IFN-γ did not induce class II on any of the cell lines described here. MHC class I expression was also found to be enhanced by TSA in each cell type, but, in contrast to MHC class II, TSA was less effective than IFN-γ in MC and Colon 26, whereas J558 was unresponsive to IFN-γ. However, similar to the class II data, TSA induced expression of CD40 on all three cell types and did so more effectively than IFN-γ.
Expression of MHC class II, class I, and CD40 proteins on the cell surface analyzed by flow cytometry. Expression levels without treatment (control cells) are shown as shaded peaks, IFN-γ-treated cell expression is plotted by a dotted line, whereas the level of expression found on TSA-treated cells is represented by a heavy line. Isotype controls were analyzed and, in each case, fell on or slightly below the levels shown for control cells (data not shown). Each histogram presents FL2 fluorescence intensity on the logarithmic x-axis and events on the linear y-axis. SK-N-MC and Colon 26 treatments shown were 50 nM TSA for 48 h, whereas J558 treatment was 10 nM TSA for 24 h. All IFN-γ treatments were 100 U/ml for the same length of time as the corresponding TSA treatment. The percent of cells positive for MHC class II or CD40 expression is indicated below each histogram. Because all cells were positive for MHC class I expression, the number under the class I histogram represents fold-change of expression calculated from the mean channel shift after treatment relative to untreated cells.
Effect of TSA on the cell cycle and apoptosis
TSA has been described as a cell cycle inhibitor that induces both G1 and, especially, G2-M blocks (17). MHC class II, but not class I, has been reported (29, 30) to be regulated differentially during the cell cycle. We therefore analyzed the cell cycle of our tumor lines (by DNA staining and flow cytometry) with each of the treatment conditions under study. From these studies, we learned which TSA concentrations significantly altered the cell cycle, and therefore designed our experiments to include conditions that did not affect cell cycle progression in the cell lines examined. In Fig. 5⇑, the conditions employed did not significantly alter the cell cycle (see cell cycle analysis Fig. 6⇓A) and yet 29% of cells were positive for class II expression in MC, 53% in Colon 26, and 75% in J558. Two additional methods were used to analyze cell cycle effects on MHC expression. Because TSA arrests the cell cycle mainly in G2/M, and class II has been reported to accumulate in G2-M (30), we studied the effects of the cell cycle inhibitor nocodazole, which arrests cells in G2-M (Fig. 6⇓B). Nocodazole arrested nearly 100% of J558 cells in G2-M and resulted in only 23% MHC class II-positive cells (Fig. 6⇓C). However, 10 nM TSA resulted in only 19% of cells in G2-M and yet this treatment produced 75% MHC class II-positive cells (Fig. 6⇓D). This indicated that a significant proportion of the MHC class II expression in the TSA-treated population was independent of G2-M. Also, using double staining with PI and Ab to class I on J558 cells, we found TSA induced expression throughout the cell cycle and enhancement was not restricted to a specific phase. From the above experiments, we conclude that TSA has specific effects on MHC gene expression that are not solely a result of its action on the cell cycle. Although our data demonstrated CD40 expression at TSA concentrations that do not alter the cell cycle, more detailed studies are required to clarify the cell cycle regulation of CD40.
Cell cycle regulation of MHC class II by TSA. J558 cells were treated for 24 h with 10 nM TSA or 10 μM nocodazole and stained for cell surface expression of MHC class II. The cell cycle was analyzed by DNA staining with PI. A, The DNA staining after 10 nM TSA treatment (heavy line) superimposed on the DNA staining of untreated control cells (dotted line). The TSA and control curves are essentially identical. B, The DNA staining of 10 μM nocodazole-treated J558 cells (heavy line) superimposed on the DNA staining of control cells (dotted line) showing that nearly 100% of the cells have accumulated in G2-M after nocodazole treatment. C, The MHC class II staining of the nocodazole-treated cells. D, The MHC class II staining after TSA treatment. C and D, The shaded peak represents isotype control staining, the dotted line untreated cells, and the heavy line treated cells (data not shown).
It became clear during these studies, as reported by others (18, 31), that DAIs induce apoptosis. To evaluate a potential role for apoptosis in the altered gene expression by TSA, apoptosis was monitored by flow cytometry through DNA staining with PI (the “sub-G0 ” population) and cellular staining with annexin-V. At the low concentrations of TSA employed in Fig. 5⇑, selected for a normal cell cycle, apoptosis was not significantly different from untreated controls. To demonstrate that the MHC class II expression induced by TSA was, in large part, independent of apoptosis, we stained J558 cells for I-Ad and annexin V. Subsequent analyses were gated on the annexin V-negative population and assayed for class II expression (Fig. 7⇓). We conclude that TSA has specific effects on gene expression over and above any effects that may result from apoptosis. However, at higher concentrations (differing for each cell type) apoptosis may result in changes in gene expression and complicate the interpretation of experiments using DAIs.
TSA induces MHC class II expression in nonapoptotic cells. J558 cells were treated with 10 nM TSA and stained with annexin-V-FITC, anti-I-Ad-PE, and 7-AAD. Flow cytometric analyses were conducted and the annexin-V-negative cells were analyzed for MHC class II expression. The shaded peak in the histogram is an isotype control, the dotted line represents I-Ad expression on the untreated sample, whereas the heavy line represents I-Ad expression by nonapoptotic (annexin-V-negative) TSA-treated J558 cells. In this experiment, 68% of nonapoptotic cells expressed MHC class II after TSA treatment, as compared with 6% of annexin-negative, untreated cells. CD40 expression was found on 41% of annexin-V-negative TSA-treated cells compared with 8% of annexin-V-negative untreated J558 (data not shown). The mean channel fluorescence from H-2Dd staining showed a 1.2-fold induction of expression in nonapoptotic TSA-treated cells relative to nonapoptotic control cells.
Discussion
Using cells treated with low concentrations of DAIs that produced little or no apoptosis, and maintained an essentially normal cell cycle, we demonstrated that a human neuroblastoma cell line and two mouse tumor cell lines could be induced to express MHC class II mRNA by the deacetylase inhibitor TSA. In all three tumors, IFN-γ failed to induce MHC class II Ags on the cell surface, but, following TSA treatment, each cell line elicited substantial expression detected by flow cytometry. Enhanced expression of class I and CD40 were elicited by low concentrations of TSA, whereas the expression of 10 other surface Ags remained essentially unchanged. Higher concentrations of TSA induced Fas and FasL expression and apoptosis. Thus, as expected, various genes appear to differ in their sensitivity to DAIs that likely reflects inherent differences in chromatin structure and nucleosomal positioning in the promoters regulating these genes.
The current paradigm for MHC gene expression suggests that CIITA is required for constitutive and induced class II transcription, IFN-γ enhanced MHC class I expression, and perhaps plays a nonessential role in maintaining constitutive levels of transcription of class I. This study, together with other recently reported evidence (21, 32, 33, 34), suggests that, under specific experimental circumstances and in certain cell types, class II may be expressed independently of CIITA. One explanation of our data is that CIITA expression is defective or repressed in MC and Colon 26, and an alternative pathway exists that is activated by DAIs that completely bypasses CIITA. This is especially evident in the CIITA-mutant cell line RJ2.2.5. In normal IFN-γ responsive cells, the CIITA protein interacting with other factors may function to facilitate an open chromatin structure at the class II promoter. Consistent with this thesis are the reports that, in certain cells lacking endogenous CIITA, the DR, Ii, and DM promoters are not occupied (35), and that transfection of CIITA results in the appearance of DNase hypersensitivity sites in the class II promoter and expression of the DR gene (36). These observations, together with the finding that CIITA does not alter the stability of the higher order multiprotein complexes that bind to the class II promoter (37), supports the thesis that the CIITA complex enhances access of transcription factors to DNA binding sites by altering chromatin. Several recent studies have reported (2, 38, 39) that the histone acetyltransferase cofactor CBP/p300 binds to CIITA and significantly enhances its activity. TSA, at least in part, could replace the function of a CIITA-CBP/p300 coactivator complex by acetylating histones and relieving nucleosome repression allowing promoter occupancy and some level of transcription in the absence of CIITA.
An additional model that is consistent with current dogma and with the data presented here is that, in certain cell types, DAIs release chromatin repression of the CIITA gene. The activation of CIITA and MHC class II after treatment of the J558 plasmacytoma with TSA is consistent with this hypothesis. Active suppression of CIITA and class II has been postulated during the differentiation from B cells to plasma cells (8) and in trophoblast cells (20, 40). Inhibition of CIITA transcription could be mediated by a repressor that recruits HDACs to the CIITA promoter, analogous to the repression mediated by HDAC recruitment by nuclear receptors, the Mad/Max and Rb-E2F complexes, and by the methyl-CpG-binding proteins (15, 16). A likely candidate is the Blimp-1 transcriptional repressor, a regulator of terminal B cell differentiation that has been shown to associate with HDACs (41). The surface expression of class II on TSA-treated J558 cells is high, approaching that on B cells and IFN-γ inducible cells; compared with the relatively low levels found in the CIITA-negative MC and Colon 26 cells. This suggests that, although class II expression may occur without CIITA expression, CIITA may be required for high levels of expression. The difference in inducibility of CIITA by TSA observed in the cell lines in this study could be related to the distinct promoters (PI→PIV) and their differential usage in regulating CIITA in various cell types (42). J558 is of B cell lineage and should preferentially utilize PIII, the B cell promoter. One possibility is that, in the transition from a B cell to a plasma cell, the PIII promoter is repressed by chromatin and that this is reversed by TSA, leading to the activation of CIITA and class II in plasma cells.
In summary, DAIs could function in activating MHC class II in different cell types by one of several mechanisms that are not mutually exclusive: 1) Acetylating histones at the MHC promoter, thus providing a function normally performed by the CIITA-CBP/p300 coactivation complex. 2) Activating expression of the CIITA gene that is repressed by chromatin. This would include CIITA promoters that are methylated on CpG islands and bind methyl CpG binding proteins that recruit HDACs (16). In the trophoblast cell line, Jar, the PIV CIITA promoter is methylated and, although 5-azacytidine alone does not activate CIITA and class II, such treatment does support low level expression after IFN-γ treatment (43). In view of the potential synergy of demethylation and DAIs in activating certain epigenetically silenced genes (44), future studies of the effect of combinations of DAIs and demethylating agents will be of interest. 3) Altering the levels and/or activation state of essential cofactors such as CBP/p300. 4) Direct acetylation of the CIITA protein, or other factors required in class II transactivation, by DAIs could also affect their activity or nuclear transport.
Because the class I and II promoters show significant sequence homologies (6), and both utilize CIITA as a coactivator (4, 5), DAIs may be functioning by similar mechanisms in both class I and II. Tissue specific expression of class I is determined, in significant part, by transcriptional mechanisms that depend on upstream negative regulatory elements in the class I promoter (45) and factors binding to these elements may recruit HDACs and repress transcription. The mechanism of TSA activation of CD40 is unclear and we are not aware of data on the promoter structure of CD40 that would help clarify this issue. The observation that TSA can induce surface expression of CD40 has potential biological implications. CD40 ligand binds to the CD40 receptor and promotes tumor killing by CTLs, directly induces apoptosis of certain tumor cells (46), as well as enhances the ability of normal DC to present Ag and elicit CTLs (47). Thus, up-regulating CD40 expression could potentially be useful in future attempts to treat cancers by the administration of soluble CD40L or cells genetically modified to express CD40L genes.
Epigenetic phenomena including methylation and histone acetylation are critical in regulating gene expression through repression (14), and complex patterns of chromatin activation and repression are undoubtedly involved in many fundamental immune processes (48, 49, 50). These alterations in chromatin structure would likely involve changes in the pattern of histone acetylation. Thus, agents that hyperacetylate histones may be able to substitute for natural signals that release repression during cellular differentiation and, as shown in this study, these same agents may elicit expression of repressed genes critical in the immune response against tumors. The work reported here also suggests that some of the variation in immunogenicity and tumorigenicity of different tumors may be mediated by regulatory events involving reversible acetylation patterns, and these could potentially be altered by treatment or targeting with agents that alter acetylation.
Acknowledgments
We thank Todd Vogt, Michelle Detwiler, and Thomas Isac for expert technical assistance. We thank Garron Solomon for his participation in our apoptosis analyses. We thank Soldano Ferrone, Shawn Murphy, and Sara Schneider for critical review of the manuscript.
Footnotes
-
↵1 This work was supported in part by Grant HD 17013 from the National Institutes of Health. The Biopolymer and Flow Cytometry Facilities are supported in part by a Cancer Center Support Grant CA 16056.
-
↵2 Address correspondence and reprint requests to Dr. Thomas B. Tomasi, Department of Immunology, Elm and Carlton Streets, Roswell Park Cancer Institute, Buffalo NY 14263. E-mail address: thomas.tomasi{at}roswellpark.org
-
3 Abbreviations used in this paper: CIITA, class II transactivator; HDAC, histone deacetylase; DAIs, HDAC inhibitors; TSA, trichostatin A; PI, propidium iodide.
- Received June 2, 2000.
- Accepted September 21, 2000.
- Copyright © 2000 by The American Association of Immunologists
References
In this issue
