Targeting Positive Regulatory Domain I-Binding Factor 1 and X Box-Binding Protein 1 Transcription Factors by Multiple Myeloma-Reactive CTL1
http://www.100md.com
免疫学杂志 2005年第14期
Abstract
Growing evidence indicates that multiple myeloma (MM) and other malignancies are susceptible to CTL-based immune interventions. We studied whether transcription factors inherently involved in the terminal differentiation of mature B lymphocytes into malignant and nonmalignant plasma cells provide MM-associated CTL epitopes. HLA-A*0201 (A2.1) transgenic mice were used to identify A2.1-presented peptide Ag derived from the plasma cell-associated transcriptional regulators, positive regulatory domain I-binding factor 1 (PRDI-BF1) and X box-binding protein 1 (XBP-1). A2.1-restricted CTL specific for PRDI-BF1 and XBP-1 epitopes efficiently killed a variety of MM targets. PRDI-BF1- and XBP-1-reactive CTL were able to recognize primary MM cells from A2.1+ patients. Consistent with the expression pattern of both transcription factors beyond malignant and nonmalignant plasma cells, PRDI-BF1- and XBP-1-specific CTL activity was not entirely limited to MM targets, but was also associated with lysis of certain other malignancies and, in defined instances, with low-to-intermediate level recognition of a few types of normal cells. Our results also indicate that the A2.1-restricted, PRDI-BF1- and XBP-1-specific human CD8+ T cell repertoire is affected by partial self tolerance and may thus require the transfer of high-affinity TCR to break tolerance. We conclude that transcription factors governing terminal cellular differentiation may provide MM- and tumor-associated CTL epitopes.
Introduction
The outcome of the majority of patients with multiple myeloma (MM)3 is unsatisfactory, although a proportion of patients benefit from high-dose therapy followed by autologous stem cell support (1). Allogeneic stem cell transplantation may result in higher complete response rates and, occasionally, in long-term survivors in true molecular remissions (1). This procedure, however, has yet been associated with substantial transplant-related mortality in MM patients (2). The advantages of allogeneic stem cell transplantation are the absence of MM cells in the graft and the existence of a CD8+ T cell-based graft-vs-myeloma effect (3, 4, 5). Clinical results, apart from donor lymphocyte infusions (3, 5, 6, 7), on a more specific CTL-mediated immunotherapy for MM are solely derived from vaccination trials using the Ig Id (8). The Id represents a private MM-associated Ag and thus cannot provide shared immunotherapy for various MM patients. Nonindividual MM-associated Ag, such as the cancer germline-specific MAGE/BAGE/GAGE/LAGE family gene products or mucin 1 (MUC1), have been identified (9). The expression pattern of MAGE-type genes, however, covers only a proportion (10–50%) of MM cells (9, 10). Although MUC1 has been reported to be expressed in 60–90% of MM samples (9, 11), only low-level cytotoxicity (20–30% specific lysis at E:T ratios of up to 100:1) in response to MM cells has been demonstrated for HLA-A*0201 (A2.1)-restricted CTL specific for the MUC1-derived peptide Ag M1.1 and M1.2 (11, 12). Hence, the identification of novel, MM-associated, MHC class I-presented CTL epitopes remains an important challenge. Moreover, Ag derived from proteins involved in the malignant phenotype are believed to be particularly useful for targeted anti-MM immune interventions, as the escape of neoplastic cells from immune recognition by the selection of Ag-loss variants would be less likely to occur (13, 14).
Naive B cells initiate a series of temporally and spatially regulated events that lead to the differentiation of both memory B cells and Ab-secreting plasma cells. Two transcription factors, the X box-binding protein 1 (XBP-1) and the mouse B lymphocyte-induced maturation protein 1 (Blimp-1) or its human homologue, the positive regulatory domain I-binding factor 1 (PRDI-BF1), have been recently demonstrated to be essential for terminal plasmacytic differentiation (15). XBP-1 was found to be required for the plasma cell phenotype, as no plasma cells develop in its absence (16). Blimp-1 has the unique ability to drive plasmacytic differentiation upon enforced expression in a B cell lymphoma cell line (17) or primary splenic B cells (18), and is found in all plasma cell types in vivo (19). More recently, an essential role for the transcription factor Blimp-1 in plasma cell differentiation and preplasma memory B cell formation has been defined (20). As with terminally differentiated nonmalignant plasma cells, MM cells have likewise been demonstrated to (over-)express mRNA and protein of XBP-1 and PRDI-BF1 (21, 22, 23, 24). Interestingly, however, PRDI-BF1 was found to be involved in additional terminal differentiation processes distinct from the B cell lineage (25), and XBP-1 expression did not appear to be exclusively specific for malignant and nonmalignant plasma cells (26).
In this study, we report on a novel concept for the immunotherapy of MM and other malignancies based on the induction of PRDI-BF1- and XBP-1-specific CTL. By circumventing self tolerance in A2.1 transgenic (Tg) mice (13, 14, 27), we identified several endogenously processed, A2.1-presented PRDI-BF1- and XBP-1-derived CTL epitopes. We show the entire response pattern of Tg mice-derived CTL, including the recognition of primary MM cells, malignant melanoma, breast, and hepatocellular cancer targets. Our results also indicate that the human T cell repertoire is affected by partial PRDI-BF-1- and XBP-1-specific self tolerance.
Materials and Methods
Tg mice (28) were kindly provided by L. A. Sherman (The Scripps Research Institute (TSRI), La Jolla, CA). Homozygous line cA2Kb, as compared with A2Kb mice, was derived from a different founder animal and had a 3-fold higher transgene expression. C57BL/6 mice were purchased from the breeding colony of the Johannes Gutenberg-University (JGU). Mice were maintained at the animal facility of JGU. Experimental procedures were performed according to German federal and state regulations and the National Institutes of Health Guide for the Care and Use of Laboratory Animals.
Peptides
PRDI-BF1 (377–386) (FLLPPYGMNC), (401–410) (FGLFPRLCPV), (402–410) (GLFPRLCPV), (406–414) (RLCPVYSNL), (406–415) (RLCPVYSNLL), and XBP-1 (18–27) (LLLSGQPASA), (19–27) (LLSGQPASA), influenza virus matrix protein (FluM1) (58–66) (GILGFVFTL), human tyrosinase (hTyr) (369–377) (YMDGTMSQV), hepatitis B virus core (128–140) (TPPAYRPPNAPIL) (29), human wild-type p53 (264–272) (LLGRNSFEV), vesicular stomatitis virus nucleoprotein (VSV-N) (52–59) (RGYVYQGL) (13), and HIV type 1 reverse transcriptase (HIV-RT) (476–484) (ILKEPVHGV) (30) peptides were synthesized by Affina Immuntechnik or Neosystem. Purity of synthetic peptides was ascertained by reversed phase high-performance liquid chromatography and mass spectrometry and was >80%. Purity of PRDI-BF1 (401–410) and (402–410) peptides was >97%.
References
enerating PRDI-BF1- and XBP-1-specific CTL in Tg mice
Based on consensus motifs and computational score models for A2.1-binding peptides (35, 36), we selected a total of 61 and 16 synthetic peptides corresponding to PRDI-BF1 and human XBP-1 protein sequences, respectively. Almost all of the PRDI-BF1 (32 of 44) and XBP-1 (11 of 13) peptides with strong-to-intermediate A2.1-binding capacity were tested for their ability to induce A2.1-restricted and peptide-specific CTL in A2.1 Tg mice to bypass potential self tolerance (data not shown). Immunogenic peptides representing PRDI-BF1 (377–386,401–410,402–410,406–414,406–415) as well as XBP-1 (18–27) and (19–27) aa were nonhomologous to mouse self-PRDI-BF1/-XBP-1 sequences. These peptides had strong A2.1-binding efficiencies, except of PRDI-BF1 (406–414) that bound to A2.1 with intermediate activity (Fig. 1). Using human CD8 x A2.1 or CD8 x (c)A2Kb double Tg mice, we generated A2.1-restricted CTL lines specific for these immunogenic peptides. PRDI-BF1 (406–414)- and (406–415)-reactive CTL required as much as 0.1 μM Ag to induce lysis of peptide-pulsed T2 cells (data not shown). Provided that these Ag would have been processed endogenously, such CTL were likely of too low an avidity to allow recognition of natural Ag. In contrast, CTL lines specific for PRDI-BF1 (377–386,401–410), and (402–410) required <10 nM of either stimulatory Ag to induce killing of peptide-pulsed T2 targets (Fig. 2, A–C). A common property of immunogenic PRDI-BF1 peptides was the presence of a cysteine within the sequence. Due to its free S-H group, cysteine is the most chemically reactive amino acid under physiological conditions. Peptide titration in the presence of a reducing agent and the use of synthetic peptide variants, in which the cysteine side chain was converted into a nonreactive form, suggested that cysteine modification based on oxidization of the free S-H group (37) has an impact on PRDI-BF1-specific CTL recognition. Although we found a 10-fold increase in peptide sensitivity of PRDI-BF1 (401–410)- and (402–410)-specific CTL under reducing conditions, no effect was observed for PRDI-BF1 (377–386)-reactive CTL or control CTL recognizing a non-cysteine-containing Ag (data not shown). By loading PRDI-BF1 peptide substitutes in which cysteine was replaced by alanine onto T2 targets, we found that PRDI-BF1 (377–386)- and (402–410)-stimulated CTL were up to 1 log more sensitive in recognizing the alanine substitutes (Fig. 2, A and C) at various E:T ratios and peptide doses (data not shown). This observation was not due to differences in A2.1 binding between alanine- and nonsubstituted peptides (data not shown). In contrast to PRDI-BF1 (377–386)- and (402–410)-specific CTL, effector T cells propagated with PRDI-BF1 (401–410), demonstrating a more profound lytic activity in response to the 9-mer PRDI-BF1 (402–410) than to the 10-mer (401–410) (Fig. 2B), did not substantially cross-recognize the alanine variants of PRDI-BF1 (401–410) or (402–410) (data not shown). Therefore, our results suggest the possibility of a differential sensitivity of PRDI-BF1 (377–386)-, (401–410)-, and (402–410)-specific CTL to cysteine-modified Ag.
Effector T cells obtained after restimulation with XBP-1 (18–27) and (19–27) required 0.1 nM Ag for lysis of peptide-coated T2 targets (Fig. 2, D and E). We ascertained the Ag specificity and A2.1 restriction of all CTL lines based on their failure to lyse T2 cells loaded with the A2.1-binding FluM1 (58–66) peptide, and their lack of recognition of A2.1-negative EL4 targets coated with either stimulatory peptide (Fig. 2).
Targeting MM cells by PRDI-BF1- and XBP-1-specific CTL
MM cells have been demonstrated to (over-)express PRDI-BF1 and XBP-1 mRNA and protein (data not shown) (21, 22, 23, 24, 38). To assess whether the antigenic PRDI-BF1 and XBP-1 peptides represent naturally processed, A2.1-presented CTL epitopes, we explored MM targets for recognition by PRDI-BF1- and XBP-1-specific CTL. We found that A2.1+ as opposed to A2.1– (NCI-H929) MM cell lines were selectively killed by PRDI-BF1- and XBP-1-reactive CTL (Fig. 3, A–E). Effector T cells with specificity for PRDI-BF1 (402–410) were able to efficiently recognize the A2.1+ MM targets U266, OPM-2, L363, and DEPU (Fig. 3C). Comparable results were obtained for PRDI-BF1 (377–386)-specific CTL (Fig. 3A). Although PRDI-BF1 (401–410)- and (402–410)-reactive CTL were of almost equivalent avidity in response to the 9-mer (402–410) (Fig. 2, B and C), they differed substantially in their functional recognition of A2.1+ MM targets (Fig. 3, B and C). This finding would be consistent with the possibility of a differential sensitivity of these CTL to cysteine modified Ag (Fig. 2C and data not shown) (39).
BP-1 (27)- as compared with (18–27)-specific CTL gave rise to more robust A2.1-restrictedlytic responses to the majority of A2.1+ MM targets (Fig. 3, D and E). Although mutual cross-recognition of the 9- and 10-mer peptide by XBP-1 (18–27)-reactive CTL, respectively, occurred only with anti-XBP-1 (18–27) effector T cells, this ability vanished upon their prolonged in vitro maintenance and was accompanied by impaired recognition of natural Ag presented on MM targets (data not shown). However, their avidity in response to the stimulatory Ag XBP-1 (18–27) itself was not affected. These findings indicate that XBP-1 (19–27) as opposed to (18–27) is the naturally processed CTL epitope.
Lack of killing of A2.1– NCI-H929 targets by PRDI-BF1- and XBP-1-reactive CTL (Fig. 3, A–E) as well as inhibition of the cytolytic response to A2.1+ MM cells in the presence of an anti-A2 mAb (Fig. 3, A, C, and E) revealed that MM recognition by PRDI-BF1- and XBP-1-specific CTL was in fact A2.1-restricted. Positive and negative responder controls for MM cell recognition were allo-A2.1-reactive and FluM1 (58–66)-specific T cell lines, respectively (Fig. 3, F and G). We conclude that defined PRDI-BF1- and XBP-1-derived peptides provide naturally processed, MM-associated CTL epitopes.
Peptides presented by class I MHC molecules are most often derived from proteolytic processing of cellular proteins by the 20S proteasomal complex (40). To study whether the epitopes recognized by PRDI-BF1- and XBP-1-specific CTL had been generated by the proteasome, proteasomal Ag production by OPM-2 cells was modified by lactacystin (32). Naturally presented peptides were stripped off cell surface class I MHC molecules of OPM-2 cells by acid treatment to allow only novel peptides generated by proteolytic degradation to be presented for CTL recognition (14, 32). Production of the MM-associated model CTL epitopes PRDI-BF1 (402–410) and XBP-1 (19–27) was inhibited by exposing OPM2 to lactacystin (Fig. 4, A and B), as was lysis by peptide-dependent allo-A2.1-reactive control CTL (Fig. 4D). No effect was observed with CTL CD8 x A2Kb FluM1 (58–66) used as a negative control (Fig. 4C), indicating that the treatment of OPM-2 targets with the proteasome inhibitor per se did not render them susceptible to CTL-mediated cell death. These findings demonstrate that the natural PRDI-BF1 and XBP-1 peptides recognized by CTL were endogenously processed by the proteasome complex.
To extend the analysis of PRDI-BF1- and XBP-1-specific CTL to primary MM cells, BM samples from A2.1+ MM patients were separated into the CD138+ MM subset and normal CD138– BM mononuclear cells. PRDI-BF1- and XBP-1-specific CTL produced IFN- in response to freshly isolated MM as opposed to normal BM stimulators (Fig. 5A). Allo-A2.1- and FluM1 (58–66)-reactive CTL were used as positive and negative responder control, respectively. MM-specific cytolytic activity was only observed with PRDI-BF1 (402–410)- and (401–410)-specific T cells, albeit less efficiently (Fig. 5B). Comparable results were obtained with BM samples from four of six MM patients. Obviously, primary MM cells freshly isolated from A2.1+ patients are targeted by PRDI-BF1- and XBP-1-reactive effector T cells. Lack of recognition of CD138– BM targets also precludes the possibility that PRDI-BF1- and XBP-1-reactive T lymphocytes cross-recognized other non-MM cell-derived peptides or responded by peptide-dependent allorecognition.
PRDI-BF1- and XBP-1-specific CTL activity beyond MM targets
PRDI-BF1-/XBP-1-targeted immunotherapy would ideally be associated with selective killing of malignant targets and, perhaps, nonmalignant plasma cells while preserving other nontransformed tissues. Recent observations, however, have indicated a broader expression pattern of PRDI-BF1 than has so far been anticipated, arguing for its involvement in additional terminal differentiation processes distinct from the B cell lineage (25). XBP-1, too, has been found expressed in other tissues, including the liver (16, 41). These findings led us to study PRDI-BF1- and XBP-1-specific CTL in response to other non-MM-related B cell malignancies and B-LCL, various solid tumor cells, and nontransformed targets (Fig. 6). FluM1 (58–66)- and allo-A2.1-reactive effector T cells were used as controls (data not shown).
PRDI-BF1-reactive CTL responded to the majority of EBV-transformed B-LCL, the PRDI-BF1+ osteosarcoma line U2OS (42), and the melanoma targets NA8-Mel and Malme 3M, whereas pre B-ALL were mostly ignored (Fig. 6, A–C). These results are in accordance with expression analyses of the mouse homologue of PRDI-BF1, Blimp-1 (17). The differential capacity of PRDI-BF1-specific CTL to respond to transformed A2.1+/PRDI-BF1+ cells, as shown in Figs. 3, A–C, and 6, A–C, was also observed with nontransformed targets. Whereas all normal cells were ignored by PRDI-BF1 (377–386)-reactive effector T cells (Fig. 6A), low level recognition of mature DC (25% lysis) by CTL PRDI-BF1 (401–410) in one of three experiments and of PBMC (20% lysis) by CTL PRDI-BF1 (402–410) occurred (Fig. 6, B and C). PRDI-BF1 (402–410)-specific CTL gave also rise to 45% specific lysis of purified B cells (Fig. 6C). Interestingly, formation of (preplasma) memory B cells has been demonstrated more recently to require the expression of Blimp-1 (20). Taken together, the overall pattern of cytolytic responses revealed for the different PRDI-BF1-specific CTL (Figs. 3, A–C, and 6, A–C) seemed to be cogoverned by their individual avidity (Fig. 2, A–C).
Effector T cells with specificity for XBP-1 (18–27) and (19–27) did not distinguish between different B cell malignancies and efficiently killed all A2.1+ EBV-transformed B-LCL and pre B-ALL lines while ignoring the A2.1– pre B-ALL targets, EU1 and UoC-B1 (Fig. 6D and data not shown). XBP-1-reactive CTL also recognized NA8-Mel and Malme 3M melanoma cells, whereas additional cytolytic activity in response to breast cancer (BT549) and hepatocellular carcinoma (HepG2) targets was only revealed for XBP-1 (19–27)-specific CTL (Fig. 6D and data not shown). Lysis of HepG2 cells by XBP-1 (19–27)-stimulated CTL is consistent with XBP-1 expression in hepatocellular carcinomas (43). Its level of expression in adult human liver (43), however, was not sufficient to allow XBP-1-specific killing of normal A2.1+ hepatocytes isolated from two different sources (Fig. 6D and data not shown). The differential avidity of XBP-1-reactive CTL in recognizing the (19–27) epitope became more apparent in their responses to other nontransformed cells. Although normal targets were ignored by XBP-1 (18–27)-reactive CTL, XBP-1 (19–27)-specific effector T cells demonstrated intermediate level killing of A2.1+ mature DC (40% lysis) (Fig. 6D and data not shown). Mature DC are believed to be protected from CTL-mediated apoptosis by expression of the granzyme inhibitor, serine protease inhibitor-6 (44). The observed anti-DC T cell responses could be due to the readout via the 51Cr-release as opposed to other assays, such as JAM, that are based on DNA fragmentation (44, 45).
Inhibition of the cytolytic response to the A2.1+ melanoma cell line Malme 3M and purified B cells in the presence of an anti-A2 mAb (Fig. 6, C and D) revealed that recognition of non-MM tumor targets and nontransformed B cells by PRDI-BF1- and XBP-1-specific CTL was indeed A2.1-restricted. Cold target inhibition of MM cell killing by PRDI-BF1- and XBP-1-reactive CTL (Fig. 6) as well as B cell and mature DC recognition by XBP-1-reactive CTL (Fig. 6D) again demonstrated the Ag specificity of these effector T cells and precludes the possibility that PRDI-BF1- and XBP-1-reactive CTL cross-recognized other non-PRDI-BF1- or non-XBP-1-derived peptides.
Based on these results and the expression patterns of PRDI-BF1 and XBP-1 (16, 17, 20, 25, 26, 38, 41), we conclude that PRDI-BF1- and XBP-1-reactive CTL activity is not entirely limited to MM targets, but is also associated with lysis of other malignancies and, in the case of PRDI-BF1 (401–410)-, (402–410)-, and XBP-1 (19–27)-, but not PRDI-BF1 (377–386)-specific CTL, with low-to-intermediate level recognition of a few types of nontransformed cells.
Consistent with these observations, we found expression of PRDI-BF1 and XBP-1 transcripts in a number of normal tissues (Fig. 7). However, expression levels were 4-fold lower as compared with a MM standard (NCI-H929) (Fig. 7). XBP-1 CTL-mediated lysis of BT549 and HepG2 as opposed to normal hepatocytes (Fig. 6D) seemed to correlate with elevated XBP-1 transcripts in breast and hepatocellular cancer relative to normal breast and liver (Fig. 7). The PRDI-BF1 and XBP-1 RNA expression in normal bone marrow (Fig. 7), however, was not associated with susceptibility to Ag-specific CTL recognition (Fig. 5).
Partial self tolerance to PRDI-BF1 and XBP-1 in the human CD8+ T cell repertoire
Supported by these findings, we wanted to confirm our prejudice that the human CD8+ T cell repertoire is affected by PRDI-BF1- and XBP-1-specific self tolerance. To this end, naive CD8+ T lymphocytes from at least two different A2.1+ healthy donors underwent in vitro stimulation with mature autologous DC that had been pulsed with either of the three relevant PRDI-BF1 Ag, XBP-1 (19–27), or the hTyr (369–377) control peptide. By using PRDI-BF1 (377–386) and (402–410) for CTL induction, the generated CD8+ T cell cultures were neither cytolytic nor did they secrete IFN- in response to peptide-coated T2 targets (data not shown). In contrast, CD8+ T cells able to specifically secrete IFN- upon challenge with peptide-loaded T2 targets were induced in all donors stimulated with PRDI-BF1 (401–410) and XBP-1 (19–27) (Fig. 8A). Interestingly, the PRDI-BF1 (401–410)-responding T cells cross-recognized T2 cells loaded with the 9-mer, PRDI-BF1 (402–410) (Fig. 8A). Ag specificity of effector T cells was demonstrated by their failure to secrete IFN- upon challenge with HIV-RT peptide-coated targets (Fig. 8A). Specific killing of Ag-pulsed T2 cells was only observed for PRDI-BF1 (401–410)- and XBP-1 (19–27)-reactive CTL obtained from donor 1 (100 and 30% specific lysis at an E:T of 20:1, respectively), and required higher peptide concentrations (10–5 M) as compared with Tg mouse CTL (Fig. 2, B and E). The failure to promote CTL in nonresponding cultures was due to the use of PRDI-BF1 and XBP-1 peptides as hTyr (369–377)-specific and malignant melanoma-reactive cytolytic effector T cells were generated in all donors (Fig. 8A and data not shown) (31). As the number of PRDI-BF1 (401–410)- and XBP-1-reactive, DC-induced human CD8+ effector T cells was limited and their avidity appeared to be low to intermediate, we tested their capability of recognizing MM targets in the IFN- ELISPOT assay. Likewise, only donor 1-derived CTL were capable of specifically secreting IFN- in response to the A2.1+ MM cell lines L363 and U266 (Fig. 8B). MM cell recognition was A2.1-restricted and not affected by NK cell activity as no response to A2.1– NCI-H929 and K562 targets occurred (Fig. 8B). Corresponding effector T cells stimulated with hTyr (369–377) served as negative control (Fig. 8B).
To confirm these findings, human T cells from 10 different A2.1+ healthy donors were stimulated with CD154-CD40-activated autologous B cells that had been loaded with either of the PRDI-BF1 377 and 401 peptides or the XBP-1 19 Ag. Five of 10 donor T cells responded by IFN- release specifically to either the entire (donor 1) or a part of the MM peptide panel (donors 2–5) (Fig. 9). The majority of these IFN--secreting, MM Ag-specific human T cells recognized A2.1+ MM targets, but not A2.1– NCI-H929 MM cells (Fig. 9). However, all MM-reactive human T cells were ignorant to peptide-pulsed T2 and MM targets in 51Cr-release killing assays (data not shown). These results suggest that the A2.1-restricted, PRDI-BF1- and XBP-1-specific human CD8+ T cell repertoire is affected by partial self tolerance.
Discussion
Our studies demonstrate that key transcription factors which govern the terminal differentiation of B cells into plasma cells, such as PRDI-BF1 and XBP-1, provide naturally processed peptide epitopes for A2.1-restricted, MM- and tumor-reactive CTL. We cannot preclude that the differential pattern of target cell lysis obtained with PRDI-BF1 (377–386)-, (401–410)- and (402–410)-reactive CTL is due to their differential sensitivity to cysteine modification of natural PRDI-BF-1-derived peptide epitopes. It is likely that the specific CTL-based targeting of PRDI-BF1 and XBP-1 transcriptional regulators, which are essential for terminal B cell differentiation into all plasma cell types (16, 20), would not only allow to encompass almost all types of MM, but may also prevent the risk of malignant plasma cell escape from immune recognition by the selection of PRDI-BF1- or XBP-1-loss variants. This consideration is also emphasized by recent data demonstrating that the remarkable therapeutic efficacy of proteasome inhibitors in the treatment of MM (46) is coattributable to their effect on XBP-1 (47). By disrupting the unfolded protein response, a signaling pathway evoked by stress that is required for the correct folding and transport of proteins from secretory cells, such as plasma cells, they eventually lead to a functional deficiency of XBP-1 resulting in increased apoptosis of MM cells (47).
As expected by the broader expression pattern of PRDI-BF1 and XBP-1 (16, 17, 20, 25, 26, 38, 41), our studies revealed that CTL recognition was not limited to MM cells. We found that malignant melanoma targets were efficiently lysed by PRDI-BF1- and XBP-1-reactive CTL, and recognition by XBP-1-specific CTL was extended to breast cancer and hepatocellular carcinoma cells. However, normal BM cells were not recognized by the due CTL and low-to-intermediate level cytotoxicity to mature DC and purified B cells was only observed with a subset of PRDI-BF1- and XBP-1-reactive effector cells, but not with PRDI-BF1 (377–386)- or XBP-1 (18–27)-reactive CTL. Consistent with these findings, more recent results indicate that in lymphoid organs, high-level expression of Blimp-1 is restricted to Ab-secreting cells with a minority of cells expressing low level and the vast majority of cells expressing no Blimp-1 at all (48). Yet, the broader pattern of PRDI-BF1- and XBP-1-specific target cell recognition could nevertheless be accompanied by the risk of autoimmune T cell responses. Careful evaluation of the effect of Blimp-1- and mouse XBP-1-specific T cells in suitable preclinical mouse models is necessary to address this concern.
Our results suggest that the A2.1-restricted, PRDI-BF1- and XBP-1-specific human CD8+ T cell repe rtoire, in contrast to A2.1 Tg mice, is affected by partial self tolerance. It was by far more difficult to obtain a human T cell response to these peptides, consistent with an effect on the repertoire due to partial self tolerance. Provided that autoimmunity is controlled, limited to defined tissues, or absent (as indicated for PRDI-BF1 (377–386)), the findings reported here offer the appealing therapeutic possibility of turning human T lymphocytes into efficient MM- and tumor-reactive CTL by gene transfer of PRDI-BF1- and XBP-1-specific TCR (49), as has been recently demonstrated in our laboratory with A2.1-restricted, MDM2 (81–88)- and p53 (264–272)-specific TCR obtained by circumventing self tolerance in Tg mice (14, 50). Allo-MHC-restricted TCR (51), in vitro mutated and selected TCR of high affinity (52, 53), and TCR-like molecules (54) are due therapeutic instruments of human origin (55).
Finally, the presented experiments provide the basis to consider a new class of proteins, defined transcription factors, to tackle MM and other malignancies. Apart from PRDI-BF1 and XBP-1, another transcription factor, IFN regulatory factor 4 (IRF4), also known as MM oncogene 1 (MUM1), was shown to be inherently involved in plasmacytic differentiation (56). Recent studies reported on the recurrent genetic aberration in MM that juxtaposes the IgH locus to the IRF4 gene, resulting in the overexpression of IRF4 (57). Therefore, we propose that IRF4 is likely to serve as another candidate MM-associated Ag. Gene expression profiling in plasma cells revealed that these cells display a highly specialized genetic program including the expression of unique sets of known and novel transcription factors (24, 58) that may also provide novel target molecules for CTL-based immunotherapy of MM.
Disclosures
The authors have no financial conflict of interest.
Footnotes
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.
1 This work was supported in part by grants from the Deutsche Forschungsgemeinschaft (SFB 432 A3; to M.T.), the Laupitz Foundation (to M.T.), and the Mainzer Forschungsfoerderungsprogramm des Fachbereichs Medizin (MAIFOR) program (to C.L. and M.T.). C.L. is the recipient of a Lady Tata Memorial Trust Career Development Award. M.T. is a José Carreras Leukemia Foundation Professor.
2 Address correspondence and reprint requests to Dr. Carina Lotz at the current address: Department of Immunology (IMM1), The Scripps Research Institute, 10550 North Torrey Pines Road, La Jolla, CA 92037. E-mail address: clotz@scripps.edu
3 Abbreviations used in this paper: MM, multiple myeloma; MUC1, mucin 1; XBP-1, X box-binding protein 1; Blimp-1, B lymphocyte-induced maturation protein 1; PRDI-BF1, positive regulatory domain I-binding factor 1; Tg, transgenic; FluM1, influenza virus matrix protein; hTyr, human tyrosinase; VSV-N, vesicular stomatitis virus nucleoprotein; HIV-RT, HIV type 1 reverse transcriptase; B-LCL, B lymphoblastoid cell line; B-ALL, pre B acute leukemia cell line; DC, dendritic cell; BM, bone marrow; Ct, threshold cycle; IRF4, IFN regulatory factor 4.
Received for publication May 12, 2004. Accepted for publication May 2, 2005.
References
Blade, J., D. H. Vesole, M. Gertz. 2003. Controversy in hematology: transplantation for multiple myeloma: who, when, how often: high-dose therapy in multiple myeloma. Blood 102: 3469-3477.
Gahrton, G., H. Svensson, M. Cavo, J. Apperly, A. Bacigalupo, B. Bjorkstrand, J. Blade, J. Cornelissen, A. de Laurenzi, T. Facon, et al 2001. Progress in allogenic bone marrow and peripheral blood stem cell transplantation for multiple myeloma: a comparison between transplants performed 1983–93 and 1994–98 at European Group for Blood and Marrow Transplantation centres. Br. J. Haematol. 113: 209-216.
Tricot, G., D. H. Vesole, S. Jagannath, J. Hilton, N. Munshi, B. Barlogie. 1996. Graft-versus-myeloma effect: proof of principle. Blood 87: 1196-1198.
Le Blanc, R., S. Montminy-Metivier, R. Belanger, L. Busque, D. Fish, D. C. Roy, J. Kassis, J. Boileau, R. Lavallee, D. Belanger, et al 2001. Allogeneic transplantation for multiple myeloma: further evidence for a GVHD-associated graft-versus-myeloma effect. Bone Marrow Transplant. 28: 841-842.(Carina Lotz2,*, Sarah Abd)
Growing evidence indicates that multiple myeloma (MM) and other malignancies are susceptible to CTL-based immune interventions. We studied whether transcription factors inherently involved in the terminal differentiation of mature B lymphocytes into malignant and nonmalignant plasma cells provide MM-associated CTL epitopes. HLA-A*0201 (A2.1) transgenic mice were used to identify A2.1-presented peptide Ag derived from the plasma cell-associated transcriptional regulators, positive regulatory domain I-binding factor 1 (PRDI-BF1) and X box-binding protein 1 (XBP-1). A2.1-restricted CTL specific for PRDI-BF1 and XBP-1 epitopes efficiently killed a variety of MM targets. PRDI-BF1- and XBP-1-reactive CTL were able to recognize primary MM cells from A2.1+ patients. Consistent with the expression pattern of both transcription factors beyond malignant and nonmalignant plasma cells, PRDI-BF1- and XBP-1-specific CTL activity was not entirely limited to MM targets, but was also associated with lysis of certain other malignancies and, in defined instances, with low-to-intermediate level recognition of a few types of normal cells. Our results also indicate that the A2.1-restricted, PRDI-BF1- and XBP-1-specific human CD8+ T cell repertoire is affected by partial self tolerance and may thus require the transfer of high-affinity TCR to break tolerance. We conclude that transcription factors governing terminal cellular differentiation may provide MM- and tumor-associated CTL epitopes.
Introduction
The outcome of the majority of patients with multiple myeloma (MM)3 is unsatisfactory, although a proportion of patients benefit from high-dose therapy followed by autologous stem cell support (1). Allogeneic stem cell transplantation may result in higher complete response rates and, occasionally, in long-term survivors in true molecular remissions (1). This procedure, however, has yet been associated with substantial transplant-related mortality in MM patients (2). The advantages of allogeneic stem cell transplantation are the absence of MM cells in the graft and the existence of a CD8+ T cell-based graft-vs-myeloma effect (3, 4, 5). Clinical results, apart from donor lymphocyte infusions (3, 5, 6, 7), on a more specific CTL-mediated immunotherapy for MM are solely derived from vaccination trials using the Ig Id (8). The Id represents a private MM-associated Ag and thus cannot provide shared immunotherapy for various MM patients. Nonindividual MM-associated Ag, such as the cancer germline-specific MAGE/BAGE/GAGE/LAGE family gene products or mucin 1 (MUC1), have been identified (9). The expression pattern of MAGE-type genes, however, covers only a proportion (10–50%) of MM cells (9, 10). Although MUC1 has been reported to be expressed in 60–90% of MM samples (9, 11), only low-level cytotoxicity (20–30% specific lysis at E:T ratios of up to 100:1) in response to MM cells has been demonstrated for HLA-A*0201 (A2.1)-restricted CTL specific for the MUC1-derived peptide Ag M1.1 and M1.2 (11, 12). Hence, the identification of novel, MM-associated, MHC class I-presented CTL epitopes remains an important challenge. Moreover, Ag derived from proteins involved in the malignant phenotype are believed to be particularly useful for targeted anti-MM immune interventions, as the escape of neoplastic cells from immune recognition by the selection of Ag-loss variants would be less likely to occur (13, 14).
Naive B cells initiate a series of temporally and spatially regulated events that lead to the differentiation of both memory B cells and Ab-secreting plasma cells. Two transcription factors, the X box-binding protein 1 (XBP-1) and the mouse B lymphocyte-induced maturation protein 1 (Blimp-1) or its human homologue, the positive regulatory domain I-binding factor 1 (PRDI-BF1), have been recently demonstrated to be essential for terminal plasmacytic differentiation (15). XBP-1 was found to be required for the plasma cell phenotype, as no plasma cells develop in its absence (16). Blimp-1 has the unique ability to drive plasmacytic differentiation upon enforced expression in a B cell lymphoma cell line (17) or primary splenic B cells (18), and is found in all plasma cell types in vivo (19). More recently, an essential role for the transcription factor Blimp-1 in plasma cell differentiation and preplasma memory B cell formation has been defined (20). As with terminally differentiated nonmalignant plasma cells, MM cells have likewise been demonstrated to (over-)express mRNA and protein of XBP-1 and PRDI-BF1 (21, 22, 23, 24). Interestingly, however, PRDI-BF1 was found to be involved in additional terminal differentiation processes distinct from the B cell lineage (25), and XBP-1 expression did not appear to be exclusively specific for malignant and nonmalignant plasma cells (26).
In this study, we report on a novel concept for the immunotherapy of MM and other malignancies based on the induction of PRDI-BF1- and XBP-1-specific CTL. By circumventing self tolerance in A2.1 transgenic (Tg) mice (13, 14, 27), we identified several endogenously processed, A2.1-presented PRDI-BF1- and XBP-1-derived CTL epitopes. We show the entire response pattern of Tg mice-derived CTL, including the recognition of primary MM cells, malignant melanoma, breast, and hepatocellular cancer targets. Our results also indicate that the human T cell repertoire is affected by partial PRDI-BF-1- and XBP-1-specific self tolerance.
Materials and Methods
Tg mice (28) were kindly provided by L. A. Sherman (The Scripps Research Institute (TSRI), La Jolla, CA). Homozygous line cA2Kb, as compared with A2Kb mice, was derived from a different founder animal and had a 3-fold higher transgene expression. C57BL/6 mice were purchased from the breeding colony of the Johannes Gutenberg-University (JGU). Mice were maintained at the animal facility of JGU. Experimental procedures were performed according to German federal and state regulations and the National Institutes of Health Guide for the Care and Use of Laboratory Animals.
Peptides
PRDI-BF1 (377–386) (FLLPPYGMNC), (401–410) (FGLFPRLCPV), (402–410) (GLFPRLCPV), (406–414) (RLCPVYSNL), (406–415) (RLCPVYSNLL), and XBP-1 (18–27) (LLLSGQPASA), (19–27) (LLSGQPASA), influenza virus matrix protein (FluM1) (58–66) (GILGFVFTL), human tyrosinase (hTyr) (369–377) (YMDGTMSQV), hepatitis B virus core (128–140) (TPPAYRPPNAPIL) (29), human wild-type p53 (264–272) (LLGRNSFEV), vesicular stomatitis virus nucleoprotein (VSV-N) (52–59) (RGYVYQGL) (13), and HIV type 1 reverse transcriptase (HIV-RT) (476–484) (ILKEPVHGV) (30) peptides were synthesized by Affina Immuntechnik or Neosystem. Purity of synthetic peptides was ascertained by reversed phase high-performance liquid chromatography and mass spectrometry and was >80%. Purity of PRDI-BF1 (401–410) and (402–410) peptides was >97%.
References
enerating PRDI-BF1- and XBP-1-specific CTL in Tg mice
Based on consensus motifs and computational score models for A2.1-binding peptides (35, 36), we selected a total of 61 and 16 synthetic peptides corresponding to PRDI-BF1 and human XBP-1 protein sequences, respectively. Almost all of the PRDI-BF1 (32 of 44) and XBP-1 (11 of 13) peptides with strong-to-intermediate A2.1-binding capacity were tested for their ability to induce A2.1-restricted and peptide-specific CTL in A2.1 Tg mice to bypass potential self tolerance (data not shown). Immunogenic peptides representing PRDI-BF1 (377–386,401–410,402–410,406–414,406–415) as well as XBP-1 (18–27) and (19–27) aa were nonhomologous to mouse self-PRDI-BF1/-XBP-1 sequences. These peptides had strong A2.1-binding efficiencies, except of PRDI-BF1 (406–414) that bound to A2.1 with intermediate activity (Fig. 1). Using human CD8 x A2.1 or CD8 x (c)A2Kb double Tg mice, we generated A2.1-restricted CTL lines specific for these immunogenic peptides. PRDI-BF1 (406–414)- and (406–415)-reactive CTL required as much as 0.1 μM Ag to induce lysis of peptide-pulsed T2 cells (data not shown). Provided that these Ag would have been processed endogenously, such CTL were likely of too low an avidity to allow recognition of natural Ag. In contrast, CTL lines specific for PRDI-BF1 (377–386,401–410), and (402–410) required <10 nM of either stimulatory Ag to induce killing of peptide-pulsed T2 targets (Fig. 2, A–C). A common property of immunogenic PRDI-BF1 peptides was the presence of a cysteine within the sequence. Due to its free S-H group, cysteine is the most chemically reactive amino acid under physiological conditions. Peptide titration in the presence of a reducing agent and the use of synthetic peptide variants, in which the cysteine side chain was converted into a nonreactive form, suggested that cysteine modification based on oxidization of the free S-H group (37) has an impact on PRDI-BF1-specific CTL recognition. Although we found a 10-fold increase in peptide sensitivity of PRDI-BF1 (401–410)- and (402–410)-specific CTL under reducing conditions, no effect was observed for PRDI-BF1 (377–386)-reactive CTL or control CTL recognizing a non-cysteine-containing Ag (data not shown). By loading PRDI-BF1 peptide substitutes in which cysteine was replaced by alanine onto T2 targets, we found that PRDI-BF1 (377–386)- and (402–410)-stimulated CTL were up to 1 log more sensitive in recognizing the alanine substitutes (Fig. 2, A and C) at various E:T ratios and peptide doses (data not shown). This observation was not due to differences in A2.1 binding between alanine- and nonsubstituted peptides (data not shown). In contrast to PRDI-BF1 (377–386)- and (402–410)-specific CTL, effector T cells propagated with PRDI-BF1 (401–410), demonstrating a more profound lytic activity in response to the 9-mer PRDI-BF1 (402–410) than to the 10-mer (401–410) (Fig. 2B), did not substantially cross-recognize the alanine variants of PRDI-BF1 (401–410) or (402–410) (data not shown). Therefore, our results suggest the possibility of a differential sensitivity of PRDI-BF1 (377–386)-, (401–410)-, and (402–410)-specific CTL to cysteine-modified Ag.
Effector T cells obtained after restimulation with XBP-1 (18–27) and (19–27) required 0.1 nM Ag for lysis of peptide-coated T2 targets (Fig. 2, D and E). We ascertained the Ag specificity and A2.1 restriction of all CTL lines based on their failure to lyse T2 cells loaded with the A2.1-binding FluM1 (58–66) peptide, and their lack of recognition of A2.1-negative EL4 targets coated with either stimulatory peptide (Fig. 2).
Targeting MM cells by PRDI-BF1- and XBP-1-specific CTL
MM cells have been demonstrated to (over-)express PRDI-BF1 and XBP-1 mRNA and protein (data not shown) (21, 22, 23, 24, 38). To assess whether the antigenic PRDI-BF1 and XBP-1 peptides represent naturally processed, A2.1-presented CTL epitopes, we explored MM targets for recognition by PRDI-BF1- and XBP-1-specific CTL. We found that A2.1+ as opposed to A2.1– (NCI-H929) MM cell lines were selectively killed by PRDI-BF1- and XBP-1-reactive CTL (Fig. 3, A–E). Effector T cells with specificity for PRDI-BF1 (402–410) were able to efficiently recognize the A2.1+ MM targets U266, OPM-2, L363, and DEPU (Fig. 3C). Comparable results were obtained for PRDI-BF1 (377–386)-specific CTL (Fig. 3A). Although PRDI-BF1 (401–410)- and (402–410)-reactive CTL were of almost equivalent avidity in response to the 9-mer (402–410) (Fig. 2, B and C), they differed substantially in their functional recognition of A2.1+ MM targets (Fig. 3, B and C). This finding would be consistent with the possibility of a differential sensitivity of these CTL to cysteine modified Ag (Fig. 2C and data not shown) (39).
BP-1 (27)- as compared with (18–27)-specific CTL gave rise to more robust A2.1-restrictedlytic responses to the majority of A2.1+ MM targets (Fig. 3, D and E). Although mutual cross-recognition of the 9- and 10-mer peptide by XBP-1 (18–27)-reactive CTL, respectively, occurred only with anti-XBP-1 (18–27) effector T cells, this ability vanished upon their prolonged in vitro maintenance and was accompanied by impaired recognition of natural Ag presented on MM targets (data not shown). However, their avidity in response to the stimulatory Ag XBP-1 (18–27) itself was not affected. These findings indicate that XBP-1 (19–27) as opposed to (18–27) is the naturally processed CTL epitope.
Lack of killing of A2.1– NCI-H929 targets by PRDI-BF1- and XBP-1-reactive CTL (Fig. 3, A–E) as well as inhibition of the cytolytic response to A2.1+ MM cells in the presence of an anti-A2 mAb (Fig. 3, A, C, and E) revealed that MM recognition by PRDI-BF1- and XBP-1-specific CTL was in fact A2.1-restricted. Positive and negative responder controls for MM cell recognition were allo-A2.1-reactive and FluM1 (58–66)-specific T cell lines, respectively (Fig. 3, F and G). We conclude that defined PRDI-BF1- and XBP-1-derived peptides provide naturally processed, MM-associated CTL epitopes.
Peptides presented by class I MHC molecules are most often derived from proteolytic processing of cellular proteins by the 20S proteasomal complex (40). To study whether the epitopes recognized by PRDI-BF1- and XBP-1-specific CTL had been generated by the proteasome, proteasomal Ag production by OPM-2 cells was modified by lactacystin (32). Naturally presented peptides were stripped off cell surface class I MHC molecules of OPM-2 cells by acid treatment to allow only novel peptides generated by proteolytic degradation to be presented for CTL recognition (14, 32). Production of the MM-associated model CTL epitopes PRDI-BF1 (402–410) and XBP-1 (19–27) was inhibited by exposing OPM2 to lactacystin (Fig. 4, A and B), as was lysis by peptide-dependent allo-A2.1-reactive control CTL (Fig. 4D). No effect was observed with CTL CD8 x A2Kb FluM1 (58–66) used as a negative control (Fig. 4C), indicating that the treatment of OPM-2 targets with the proteasome inhibitor per se did not render them susceptible to CTL-mediated cell death. These findings demonstrate that the natural PRDI-BF1 and XBP-1 peptides recognized by CTL were endogenously processed by the proteasome complex.
To extend the analysis of PRDI-BF1- and XBP-1-specific CTL to primary MM cells, BM samples from A2.1+ MM patients were separated into the CD138+ MM subset and normal CD138– BM mononuclear cells. PRDI-BF1- and XBP-1-specific CTL produced IFN- in response to freshly isolated MM as opposed to normal BM stimulators (Fig. 5A). Allo-A2.1- and FluM1 (58–66)-reactive CTL were used as positive and negative responder control, respectively. MM-specific cytolytic activity was only observed with PRDI-BF1 (402–410)- and (401–410)-specific T cells, albeit less efficiently (Fig. 5B). Comparable results were obtained with BM samples from four of six MM patients. Obviously, primary MM cells freshly isolated from A2.1+ patients are targeted by PRDI-BF1- and XBP-1-reactive effector T cells. Lack of recognition of CD138– BM targets also precludes the possibility that PRDI-BF1- and XBP-1-reactive T lymphocytes cross-recognized other non-MM cell-derived peptides or responded by peptide-dependent allorecognition.
PRDI-BF1- and XBP-1-specific CTL activity beyond MM targets
PRDI-BF1-/XBP-1-targeted immunotherapy would ideally be associated with selective killing of malignant targets and, perhaps, nonmalignant plasma cells while preserving other nontransformed tissues. Recent observations, however, have indicated a broader expression pattern of PRDI-BF1 than has so far been anticipated, arguing for its involvement in additional terminal differentiation processes distinct from the B cell lineage (25). XBP-1, too, has been found expressed in other tissues, including the liver (16, 41). These findings led us to study PRDI-BF1- and XBP-1-specific CTL in response to other non-MM-related B cell malignancies and B-LCL, various solid tumor cells, and nontransformed targets (Fig. 6). FluM1 (58–66)- and allo-A2.1-reactive effector T cells were used as controls (data not shown).
PRDI-BF1-reactive CTL responded to the majority of EBV-transformed B-LCL, the PRDI-BF1+ osteosarcoma line U2OS (42), and the melanoma targets NA8-Mel and Malme 3M, whereas pre B-ALL were mostly ignored (Fig. 6, A–C). These results are in accordance with expression analyses of the mouse homologue of PRDI-BF1, Blimp-1 (17). The differential capacity of PRDI-BF1-specific CTL to respond to transformed A2.1+/PRDI-BF1+ cells, as shown in Figs. 3, A–C, and 6, A–C, was also observed with nontransformed targets. Whereas all normal cells were ignored by PRDI-BF1 (377–386)-reactive effector T cells (Fig. 6A), low level recognition of mature DC (25% lysis) by CTL PRDI-BF1 (401–410) in one of three experiments and of PBMC (20% lysis) by CTL PRDI-BF1 (402–410) occurred (Fig. 6, B and C). PRDI-BF1 (402–410)-specific CTL gave also rise to 45% specific lysis of purified B cells (Fig. 6C). Interestingly, formation of (preplasma) memory B cells has been demonstrated more recently to require the expression of Blimp-1 (20). Taken together, the overall pattern of cytolytic responses revealed for the different PRDI-BF1-specific CTL (Figs. 3, A–C, and 6, A–C) seemed to be cogoverned by their individual avidity (Fig. 2, A–C).
Effector T cells with specificity for XBP-1 (18–27) and (19–27) did not distinguish between different B cell malignancies and efficiently killed all A2.1+ EBV-transformed B-LCL and pre B-ALL lines while ignoring the A2.1– pre B-ALL targets, EU1 and UoC-B1 (Fig. 6D and data not shown). XBP-1-reactive CTL also recognized NA8-Mel and Malme 3M melanoma cells, whereas additional cytolytic activity in response to breast cancer (BT549) and hepatocellular carcinoma (HepG2) targets was only revealed for XBP-1 (19–27)-specific CTL (Fig. 6D and data not shown). Lysis of HepG2 cells by XBP-1 (19–27)-stimulated CTL is consistent with XBP-1 expression in hepatocellular carcinomas (43). Its level of expression in adult human liver (43), however, was not sufficient to allow XBP-1-specific killing of normal A2.1+ hepatocytes isolated from two different sources (Fig. 6D and data not shown). The differential avidity of XBP-1-reactive CTL in recognizing the (19–27) epitope became more apparent in their responses to other nontransformed cells. Although normal targets were ignored by XBP-1 (18–27)-reactive CTL, XBP-1 (19–27)-specific effector T cells demonstrated intermediate level killing of A2.1+ mature DC (40% lysis) (Fig. 6D and data not shown). Mature DC are believed to be protected from CTL-mediated apoptosis by expression of the granzyme inhibitor, serine protease inhibitor-6 (44). The observed anti-DC T cell responses could be due to the readout via the 51Cr-release as opposed to other assays, such as JAM, that are based on DNA fragmentation (44, 45).
Inhibition of the cytolytic response to the A2.1+ melanoma cell line Malme 3M and purified B cells in the presence of an anti-A2 mAb (Fig. 6, C and D) revealed that recognition of non-MM tumor targets and nontransformed B cells by PRDI-BF1- and XBP-1-specific CTL was indeed A2.1-restricted. Cold target inhibition of MM cell killing by PRDI-BF1- and XBP-1-reactive CTL (Fig. 6) as well as B cell and mature DC recognition by XBP-1-reactive CTL (Fig. 6D) again demonstrated the Ag specificity of these effector T cells and precludes the possibility that PRDI-BF1- and XBP-1-reactive CTL cross-recognized other non-PRDI-BF1- or non-XBP-1-derived peptides.
Based on these results and the expression patterns of PRDI-BF1 and XBP-1 (16, 17, 20, 25, 26, 38, 41), we conclude that PRDI-BF1- and XBP-1-reactive CTL activity is not entirely limited to MM targets, but is also associated with lysis of other malignancies and, in the case of PRDI-BF1 (401–410)-, (402–410)-, and XBP-1 (19–27)-, but not PRDI-BF1 (377–386)-specific CTL, with low-to-intermediate level recognition of a few types of nontransformed cells.
Consistent with these observations, we found expression of PRDI-BF1 and XBP-1 transcripts in a number of normal tissues (Fig. 7). However, expression levels were 4-fold lower as compared with a MM standard (NCI-H929) (Fig. 7). XBP-1 CTL-mediated lysis of BT549 and HepG2 as opposed to normal hepatocytes (Fig. 6D) seemed to correlate with elevated XBP-1 transcripts in breast and hepatocellular cancer relative to normal breast and liver (Fig. 7). The PRDI-BF1 and XBP-1 RNA expression in normal bone marrow (Fig. 7), however, was not associated with susceptibility to Ag-specific CTL recognition (Fig. 5).
Partial self tolerance to PRDI-BF1 and XBP-1 in the human CD8+ T cell repertoire
Supported by these findings, we wanted to confirm our prejudice that the human CD8+ T cell repertoire is affected by PRDI-BF1- and XBP-1-specific self tolerance. To this end, naive CD8+ T lymphocytes from at least two different A2.1+ healthy donors underwent in vitro stimulation with mature autologous DC that had been pulsed with either of the three relevant PRDI-BF1 Ag, XBP-1 (19–27), or the hTyr (369–377) control peptide. By using PRDI-BF1 (377–386) and (402–410) for CTL induction, the generated CD8+ T cell cultures were neither cytolytic nor did they secrete IFN- in response to peptide-coated T2 targets (data not shown). In contrast, CD8+ T cells able to specifically secrete IFN- upon challenge with peptide-loaded T2 targets were induced in all donors stimulated with PRDI-BF1 (401–410) and XBP-1 (19–27) (Fig. 8A). Interestingly, the PRDI-BF1 (401–410)-responding T cells cross-recognized T2 cells loaded with the 9-mer, PRDI-BF1 (402–410) (Fig. 8A). Ag specificity of effector T cells was demonstrated by their failure to secrete IFN- upon challenge with HIV-RT peptide-coated targets (Fig. 8A). Specific killing of Ag-pulsed T2 cells was only observed for PRDI-BF1 (401–410)- and XBP-1 (19–27)-reactive CTL obtained from donor 1 (100 and 30% specific lysis at an E:T of 20:1, respectively), and required higher peptide concentrations (10–5 M) as compared with Tg mouse CTL (Fig. 2, B and E). The failure to promote CTL in nonresponding cultures was due to the use of PRDI-BF1 and XBP-1 peptides as hTyr (369–377)-specific and malignant melanoma-reactive cytolytic effector T cells were generated in all donors (Fig. 8A and data not shown) (31). As the number of PRDI-BF1 (401–410)- and XBP-1-reactive, DC-induced human CD8+ effector T cells was limited and their avidity appeared to be low to intermediate, we tested their capability of recognizing MM targets in the IFN- ELISPOT assay. Likewise, only donor 1-derived CTL were capable of specifically secreting IFN- in response to the A2.1+ MM cell lines L363 and U266 (Fig. 8B). MM cell recognition was A2.1-restricted and not affected by NK cell activity as no response to A2.1– NCI-H929 and K562 targets occurred (Fig. 8B). Corresponding effector T cells stimulated with hTyr (369–377) served as negative control (Fig. 8B).
To confirm these findings, human T cells from 10 different A2.1+ healthy donors were stimulated with CD154-CD40-activated autologous B cells that had been loaded with either of the PRDI-BF1 377 and 401 peptides or the XBP-1 19 Ag. Five of 10 donor T cells responded by IFN- release specifically to either the entire (donor 1) or a part of the MM peptide panel (donors 2–5) (Fig. 9). The majority of these IFN--secreting, MM Ag-specific human T cells recognized A2.1+ MM targets, but not A2.1– NCI-H929 MM cells (Fig. 9). However, all MM-reactive human T cells were ignorant to peptide-pulsed T2 and MM targets in 51Cr-release killing assays (data not shown). These results suggest that the A2.1-restricted, PRDI-BF1- and XBP-1-specific human CD8+ T cell repertoire is affected by partial self tolerance.
Discussion
Our studies demonstrate that key transcription factors which govern the terminal differentiation of B cells into plasma cells, such as PRDI-BF1 and XBP-1, provide naturally processed peptide epitopes for A2.1-restricted, MM- and tumor-reactive CTL. We cannot preclude that the differential pattern of target cell lysis obtained with PRDI-BF1 (377–386)-, (401–410)- and (402–410)-reactive CTL is due to their differential sensitivity to cysteine modification of natural PRDI-BF-1-derived peptide epitopes. It is likely that the specific CTL-based targeting of PRDI-BF1 and XBP-1 transcriptional regulators, which are essential for terminal B cell differentiation into all plasma cell types (16, 20), would not only allow to encompass almost all types of MM, but may also prevent the risk of malignant plasma cell escape from immune recognition by the selection of PRDI-BF1- or XBP-1-loss variants. This consideration is also emphasized by recent data demonstrating that the remarkable therapeutic efficacy of proteasome inhibitors in the treatment of MM (46) is coattributable to their effect on XBP-1 (47). By disrupting the unfolded protein response, a signaling pathway evoked by stress that is required for the correct folding and transport of proteins from secretory cells, such as plasma cells, they eventually lead to a functional deficiency of XBP-1 resulting in increased apoptosis of MM cells (47).
As expected by the broader expression pattern of PRDI-BF1 and XBP-1 (16, 17, 20, 25, 26, 38, 41), our studies revealed that CTL recognition was not limited to MM cells. We found that malignant melanoma targets were efficiently lysed by PRDI-BF1- and XBP-1-reactive CTL, and recognition by XBP-1-specific CTL was extended to breast cancer and hepatocellular carcinoma cells. However, normal BM cells were not recognized by the due CTL and low-to-intermediate level cytotoxicity to mature DC and purified B cells was only observed with a subset of PRDI-BF1- and XBP-1-reactive effector cells, but not with PRDI-BF1 (377–386)- or XBP-1 (18–27)-reactive CTL. Consistent with these findings, more recent results indicate that in lymphoid organs, high-level expression of Blimp-1 is restricted to Ab-secreting cells with a minority of cells expressing low level and the vast majority of cells expressing no Blimp-1 at all (48). Yet, the broader pattern of PRDI-BF1- and XBP-1-specific target cell recognition could nevertheless be accompanied by the risk of autoimmune T cell responses. Careful evaluation of the effect of Blimp-1- and mouse XBP-1-specific T cells in suitable preclinical mouse models is necessary to address this concern.
Our results suggest that the A2.1-restricted, PRDI-BF1- and XBP-1-specific human CD8+ T cell repe rtoire, in contrast to A2.1 Tg mice, is affected by partial self tolerance. It was by far more difficult to obtain a human T cell response to these peptides, consistent with an effect on the repertoire due to partial self tolerance. Provided that autoimmunity is controlled, limited to defined tissues, or absent (as indicated for PRDI-BF1 (377–386)), the findings reported here offer the appealing therapeutic possibility of turning human T lymphocytes into efficient MM- and tumor-reactive CTL by gene transfer of PRDI-BF1- and XBP-1-specific TCR (49), as has been recently demonstrated in our laboratory with A2.1-restricted, MDM2 (81–88)- and p53 (264–272)-specific TCR obtained by circumventing self tolerance in Tg mice (14, 50). Allo-MHC-restricted TCR (51), in vitro mutated and selected TCR of high affinity (52, 53), and TCR-like molecules (54) are due therapeutic instruments of human origin (55).
Finally, the presented experiments provide the basis to consider a new class of proteins, defined transcription factors, to tackle MM and other malignancies. Apart from PRDI-BF1 and XBP-1, another transcription factor, IFN regulatory factor 4 (IRF4), also known as MM oncogene 1 (MUM1), was shown to be inherently involved in plasmacytic differentiation (56). Recent studies reported on the recurrent genetic aberration in MM that juxtaposes the IgH locus to the IRF4 gene, resulting in the overexpression of IRF4 (57). Therefore, we propose that IRF4 is likely to serve as another candidate MM-associated Ag. Gene expression profiling in plasma cells revealed that these cells display a highly specialized genetic program including the expression of unique sets of known and novel transcription factors (24, 58) that may also provide novel target molecules for CTL-based immunotherapy of MM.
Disclosures
The authors have no financial conflict of interest.
Footnotes
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.
1 This work was supported in part by grants from the Deutsche Forschungsgemeinschaft (SFB 432 A3; to M.T.), the Laupitz Foundation (to M.T.), and the Mainzer Forschungsfoerderungsprogramm des Fachbereichs Medizin (MAIFOR) program (to C.L. and M.T.). C.L. is the recipient of a Lady Tata Memorial Trust Career Development Award. M.T. is a José Carreras Leukemia Foundation Professor.
2 Address correspondence and reprint requests to Dr. Carina Lotz at the current address: Department of Immunology (IMM1), The Scripps Research Institute, 10550 North Torrey Pines Road, La Jolla, CA 92037. E-mail address: clotz@scripps.edu
3 Abbreviations used in this paper: MM, multiple myeloma; MUC1, mucin 1; XBP-1, X box-binding protein 1; Blimp-1, B lymphocyte-induced maturation protein 1; PRDI-BF1, positive regulatory domain I-binding factor 1; Tg, transgenic; FluM1, influenza virus matrix protein; hTyr, human tyrosinase; VSV-N, vesicular stomatitis virus nucleoprotein; HIV-RT, HIV type 1 reverse transcriptase; B-LCL, B lymphoblastoid cell line; B-ALL, pre B acute leukemia cell line; DC, dendritic cell; BM, bone marrow; Ct, threshold cycle; IRF4, IFN regulatory factor 4.
Received for publication May 12, 2004. Accepted for publication May 2, 2005.
References
Blade, J., D. H. Vesole, M. Gertz. 2003. Controversy in hematology: transplantation for multiple myeloma: who, when, how often: high-dose therapy in multiple myeloma. Blood 102: 3469-3477.
Gahrton, G., H. Svensson, M. Cavo, J. Apperly, A. Bacigalupo, B. Bjorkstrand, J. Blade, J. Cornelissen, A. de Laurenzi, T. Facon, et al 2001. Progress in allogenic bone marrow and peripheral blood stem cell transplantation for multiple myeloma: a comparison between transplants performed 1983–93 and 1994–98 at European Group for Blood and Marrow Transplantation centres. Br. J. Haematol. 113: 209-216.
Tricot, G., D. H. Vesole, S. Jagannath, J. Hilton, N. Munshi, B. Barlogie. 1996. Graft-versus-myeloma effect: proof of principle. Blood 87: 1196-1198.
Le Blanc, R., S. Montminy-Metivier, R. Belanger, L. Busque, D. Fish, D. C. Roy, J. Kassis, J. Boileau, R. Lavallee, D. Belanger, et al 2001. Allogeneic transplantation for multiple myeloma: further evidence for a GVHD-associated graft-versus-myeloma effect. Bone Marrow Transplant. 28: 841-842.(Carina Lotz2,*, Sarah Abd)