Recognition of tumor cells by cytolytic T lymphocytes depends on cell surface MHC class I expression. As a mechanism to evade T cell recognition, many malignant cancer cells, including those of prostate cancer, down-regulate MHC class I. For the majority of human cancers, the molecular mechanism of MHC class I down regulation is unclear, although it is well established that MHC class I down-regulation is often associated with the down-regulation of multiple genes devoted to antigen presentation. Since the promyelocytic leukemia (PML) proto-oncogene controls multiple antigen-presentation genes in some murine cancer cells, we analyzed the expression of proto-oncogene PML and MHC class I in high-grade prostate cancer. We found that 30 of 37 (81%) prostate adenocarcinoma cases with a Gleason grade of 7-8 had more than 50% down-regulation of HLA class I expression. Among these, 22 cases (73.3%) had no detectable PML protein, while 4 cases (13.3%) showed partial PML down-regulation. In contrast, all 7 cases of prostate cancer with high expression of cell surface HLA class I had high levels of PML expression. Concordant down-regulation of HLA and PML was observed in different histological patterns of prostate adenocarcinoma. These results suggest that in high-grade prostate cancer, malfunction of proto-oncogene PML is a major factor in the down-regulation of cell surface HLA class I molecules, the target molecules essential for the direct recognition of cancer cells by cytolytic T lymphocytes.
This article was published in Cancer Immunity, a Cancer Research Institute journal that ceased publication in 2013 and is now provided online in association with Cancer Immunology Research.
Cytotoxic T lymphocytes (CTLs) are the major effector cells against malignant cancer cells (1, 2). CTLs recognize antigenic peptides presented by the human leukocyte antigen (HLA) class I (3) and cause cognate destruction of cancer cells in vivo (4). For this reason, a major focus of cancer immunotherapy is to induce high numbers of CTLs specific for cancer cells (5). This effort is greatly facilitated by the identification of a large number of cancer-associated antigens and by the understanding of the molecular basis of T cell activation (6, 7, 8, 9). However, it has been established that many human cancers down-regulate cell surface expression of HLA class I (10). In several types of cancers, such as small cell lung carcinoma (11), laryngeal carcinoma and breast carcinoma (12, 13), loss of HLA is associated with poor prognosis. Down-regulation of HLA class I expression was observed in 34% of primary prostate adenocarcinomas and in 80% of lymph node metastases (14, 15).
Loss of cell surface HLA class I may allow tumor evasion of immune surveillance (16) and in theory may render tumor cells unresponsive to immunotherapy aimed at augmenting cancer-specific CTLs. It is therefore of great interest to identify the molecular basis of HLA down-regulation. Cell surface expression of HLA requires the coordinated expression of several antigen-presentation genes (17, 18, 19, 20), such as the transporter proteins associated with antigen-processing TAP-1 and -2, the low molecular proteins LMP-2 and -7, and a TAP-associated protein Tapasin. These gene products assist in the production, transport and loading of antigenic peptide onto the HLA heavy chain and beta-2-microglobulin (beta2M). In selected cases, the molecular defects in antigen-presentation, including mutations in beta2M, HLA heavy chain, TAP-1 and TAP-2, have been reported (21, 22, 23, 24). However, the mechanisms involved in the down-regulation of MHC class I are largely unknown for most cancer cases.
Interestingly, down-regulation of cell surface HLA is associated with the malfunction of multiple antigen-processing genes (25, 26, 27). This raised the intriguing possibility that molecule(s) that coordinate the expression of antigen-processing genes are defective in malignant cancer cells. We have previously reported that proto-oncogene PML can control the expression of multiple antigen-presentation genes in a murine tumor model and that malfunction of PML is responsible for the coordinated down regulation of multiple antigen presentation genes and for tumor evasion from T cell immunity (16). The PML protein is located in PML nuclear bodies that consist of multi-protein domains. It has been reported that there is a highly non-random association of the gene-rich MHC on chromosome 6 with PML bodies in interphase nuclei (28). PML enhances antigen presentation by MHC class I molecules in some human lung cancer cell lines (29). However, an in vitro study using human cell lines suggested that the PML gene is not required for the regulation of MHC class I expression (30). The relevance of these controversial observations in human cancers is unclear. An important question is whether HLA loss in human cancer is associated with loss of PML expression. To address this question, we analyzed 37 cases of high-grade prostate cancer with different histological patterns. Our data indicate a strong association between loss of PML expression and down regulation of cell surface HLA in prostate cancer.
Down regulation of HLA in prostate cancer tissues
We used monoclonal antibody HC10, which binds the cytoplasmic domain of HLA-B and C, to evaluate HLA-expression in paraffin sections (31). Of the 37 cases examined, 14 cases had no detectable HC10 reactivity, 16 cases lacked HC10 reactivity in more than 50% of cancerous cells, while 7 cases had normal HC10 reactivity in essentially all cancerous tissues (Table 1). Our results are consistent with previous reports (14, 15).
The four cases of prostate cancer shown in Figure 1 reveal different patterns of HLA down regulation. Figure 1a depicts prostate adenocarcinoma cells that express a normal level of HLA. In the case depicted in Figure 1b, some adenocarcinoma cells show normal HLA expression, while other tumor glands show a complete loss of HLA expression. The HLA+ and HLA- cancerous cells are found in distinct clusters that are adjacent to each other. Figures 1c and 1d are examples of loss of HLA expression in entire cancerous regions. In Figure 1c a Gleason grade 3 carcinoma comprising a well-circumscribed cribriform tumor mass shows no detectable HLA expression, while the adjacent benign prostate glands, blood vessel endothelial cells, and scattered fibroblast cells still express high levels of HLA. Figure 1d shows an example of individual tumor glands that completely lack HLA expression. In the right upper corner there is a benign gland with normal HLA expression.
The existence of clusters of cancerous cells with different degrees of HLA loss indicates that the absence of HLA on prostate cancer cells is not likely to be due to transformation of HLA- precursor cells. Since loss of MHC class I confers an advantage to tumor cells in the presence of a T cell response (32), it is likely that HLA loss in cancer cells is due to selection by the immune system.
Down-regulation of PML expression in cancerous tissues: association with HLA down-regulation
Since HC10 reacts with HLA-B and C at both alleles, the HLA loss observed in this study is likely due to either an abnormality in critical antigen presentation genes, such as TAP-1 and TAP-2, or to the malfunction of genes that co-ordinate multiple antigen-presentation genes, such as PML (16). To evaluate this possibility, we evaluated PML expression using immunohistochemistry. PML resides primarily in the nucleus and forms a characteristic promyelocytic oncogenic domain (POD) (33). As shown in Table 1, among the 30 cases with complete or partial loss of HLA expression, 22 cases had no detectable PML, while 4 cases had partial loss of PML expression. PML expression is high in normal prostate tissue and other non-cancerous cells, such as vascular endothelial cells.
A strong correlation between PML down regulation and HLA loss was observed. As summarized in Table 2, of the 30 cases showing complete or partial HLA down-regulation, 26 cases showed complete or partial loss of PML expression. Conversely, all seven cases with normal HLA expression showed normal PML expression. The correlation between PML expression and expression of HLA on the cell surface became more striking when the distribution of HLA+ and PML+ cells is considered. Figures 2a and 2b show a section of case 31 consisting of HLA+ vascular endothelial cells and prostatic intra-epithelial neoplasia (PIN). The cells that express HLA show the typical nuclear staining of PML, while the HLA- cancer cells are devoid of PML. Prostate epithelial cells in PIN expressed a high level of HLA. Correspondingly, strong nuclear staining of PML was found in the cells residing in the PIN (Figure 3). The correlation between PML and HLA expression also holds in malignant cancer cells that express HLA (Figure 4). To unequivocally demonstrate the co-expression of PML and HLA, we carried out two-color immunohistochemistry on prostate sections with different patterns of HLA and PML expression. As shown in Figure 5, PML and HLA staining were found in distinct location within the same cells regardless of the stage of the cancer. More strikingly, in cases where HLA+ and HLA- cells exist as a micro-chimerism, the correlation between HLA and PML expression also holds. Thus, as shown in Figure 6, a cluster of HLA+ cells surrounds a patch of HLA- cells. All of the HLA+ cells with defined nuclei also stained strongly for PML. In contrast, essentially all of the HLA- cells were devoid of PML. These results strongly suggest that PML down-regulation may be involved in HLA down-regulation in prostate cancer.
In order to determine whether HLA loss in human cancer is associated with loss of PML expression, we examined 37 cases of high-grade prostate adenocarcinoma with different histological patterns. Immunohistochemical analysis indicates co-expression of PML and HLA in prostate cancer, suggesting that PML down-regulation may be implicated in HLA down-regulation in prostate cancer. Nevertheless, it should be pointed out that we found 4 cases of prostate cancer that showed partial HLA loss without down regulation of PML. Therefore malfunction of PML cannot be responsible for HLA loss in these 4 cases. It is worth investigating whether other molecular mechanisms which have previously been described, such as mutations in TAP-1, TAP-2 and beta2M, are responsible for HLA loss (21, 22, 23, 24).
Although down regulation of PML in neoplastic tissues, especially in malignant ones, has been demonstrated previously (34, 35), our study appears to be the first that shows concordant down-regulation of PML and HLA. Our results support a critical role of PML defects in HLA loss in prostate cancer. Since HLA loss is a general phenomenon in many cancer types (36), it is worth studying whether this correlation can be extended to other tumor types. Since HLA plays an essential function in the presentation of antigens to CTLs, down-regulation of HLA may confer tumor resistance to host immunity and adoptive CTL therapy. Understanding the molecular defects involved may help us to develop molecular approaches to correct these defects.
Materials and methods
Antibodies and specimens
Thirty-seven surgically removed prostate adenocarcinoma samples from radical prostatectomy interventions were analyzed in this study. The specimens were fixed in 10% formalin, embedded in paraffin and processed for routine pathological examination. All cases were diagnosed and classified according to the Gleason grade. Histological patterns included small acinar, microacinar, cribriform and mixed (cribriform with small acinar and microacinar).
Two antibodies were used for the immunohistochemical analysis. PML-specific monoclonal antibody PG-M3 was purchased from Santa Cruz Biotechnology (Santa Cruz, CA), while HC10 (31), which reacts with HLA-B and C in paraffin sections was kindly provided by Dr. Hidde Ploegh (Harvard Medical School, Charles Town, MA) and Dr. Saldano Ferrone (Roswell Park Cancer Institute, Buffalo, NY).
Paraffin sections (5 µm thick) were deparaffinized through graded ethanol solutions. After an antigen retrieval procedure of 30 min using the target retrieval solution (DAKO, Carpintera, CA), the sections were stained using the avidin-biotin complex system (Vector Laboratory, Burlingame, CA). Step-one reagents were mouse antibodies specific for either HLA or PML. The biotinylated horse anti-mouse antibodies and horseradish peroxidase-ABC system (Vector) were used as second and third-step reagents. For single-color staining, 3, 3'-diaminobenzidine (DAB) was used as substrate. We also carried out two-color immunohistochemistry using different substrates. Briefly, the tissue sections were stained with anti-PML mAb, whose binding was detected by the ABC system using the VIP kit (Vector) as the substrate. The cell bound antibodies and enzymes were removed by acid treatment (1N hydrochloride acid for 10 min). After extensive washes, the HC10 antibody was added, and the bound antibodies were detected again using the ABC system using DAB as substrate.
To evaluate the extent of HLA or PML loss, we examined all tumor cells under the light microscope and determined the approximate percentage of tumor cells that were positive for HLA or PML expression. Expression was scored as follows: >90%, +++; 70-90%, ++; 50-70%, +; 20-50%, ±; and <20%, -.
We thank Lynde Shaw for editorial assistance. This work was supported by grants from the Cancer Research Institute in New York, the National Cancer Institute (CA82355), the Department of Defense (Grant DAMD17-00-1-0041), and a seed grant from The Ohio Cancer Research Associates.
- Received November 29, 2002.
- Accepted January 25, 2003.
- Copyright © 2003 by Pan Zheng