MHC class I expression is subject to both tissue-specific and hormonal regulatory mechanisms. Consequently, levels of expression vary widely among tissues, with the highest levels of class I occurring in the lymphoid compartment, in T cells and B cells. Although the high class I expression in B cells is known to involve the B cell enhanceosome, the molecular basis for high constitutive class I expression in T cells has not been explored. T cell-specific genes, such as TCR genes, are regulated by a T cell enhanceosome consisting of RUNX1, CBFβ, LEF1, and Aly. In this report, we demonstrate that MHC class I gene expression is enhanced by the T cell enhanceosome and results from a direct interaction of the RUNX1-containing complex with the class I gene in vivo. T cell enhanceosome activation of class I transcription is synergistic with CIITA-mediated activation and targets response elements distinct from those targeted by CIITA. These findings provide a molecular basis for the high levels of MHC class I in T cells.
Major histocompatibility complex class I gene expression is regulated by developmental, tissue-specific, and hormonal/cytokine-mediated signals. Class I genes are constitutively expressed in all adult tissues, although relative levels of class I expression vary dramatically. The highest levels of expression occur in tissues and cells of the immune system, in particular T and B lymphocytes, whereas the lowest levels are observed in the nervous system and germline cells (1, 2, 3, 4, 5). Within any cell type, the levels of MHC class I expression are actively modulated by extrinsic signals. In general, cytokines induce increased expression, whereas hormones decrease expression (6). For example, IFN-γ increases class I in nearly all tissues, whereas thyroid-stimulating hormone decreases it (6). The mechanisms that integrate these intrinsic and extrinsic regulatory signals are only beginning to be understood.
MHC class I expression is primarily transcriptionally regulated, and many of the DNA sequence elements that mediate both tissue-specific and hormonal/cytokine regulation have been identified (1, 3, 4). Tissue-specific expression is achieved through the combined effects of a promoter distal complex regulatory element and a series of promoter proximal elements. The promoter distal element, located between −700 and −800 bp, consists of overlapping enhancer and silencer elements (7). Silencer activity varies inversely with the level of class I expression in a given cell type. In cell types that express low levels of class I, such as liver or kidney, silencer activity dominates. In cell types that express high levels of class I, such as B lymphocytes, the silencer does not function (7). High class I expression in B lymphocytes is mediated by series of promoter proximal elements, located between −68 and −500 bp. These elements include enhancer A (8, 9, 10, 11, 12, 13), IFN-γ-stimulated response element (14, 15, 16, 17, 18, 19), and a composite RFX/cAMP response element (CRE)2 (16, 20, 21). These same elements are also the targets of hormone/cytokine signals (6). A variety of DNA-binding transcription factors have been identified that interact with these promoter proximal DNA sequence elements. For example, a B lymphocyte-specific enhanceosome consisting of the coactivator CIITA and DNA-bound transcription factors RFX, CREB/ATF, and NF-Y leads to high cell surface class I and II expression in B lymphocytes (22, 23, 24, 25). CIITA, a transcriptional coactivator that induces both MHC class I and class II genes does not bind DNA directly, but rather depends on factors, such as RFX and CREB/ATF, that bind to the CRE element (16, 20, 21). The constitutively high level of expression of MHC class I in B lymphocytes is due, in part, to the presence of a CIITA enhanceosome complex that maintains high transcription rates of class I genes.
T cells, like B cells, express high levels of class I. However, T cells, unlike B cells, do not normally express CIITA. The only cells in which CIITA is constitutively expressed are APC types, such as B lymphocytes, dendritic cells, and macrophages (1, 4, 26, 27, 28, 29, 30). Although activated human T cells can be induced to express CIITA (31, 32, 33), the constitutively high levels of expression in T cells cannot be due to CIITA. Indeed, little is known about how this high class I expression is achieved in T cells. What is known about cell-type-specific expression in T cells is that expression of TCRα and TCRβ genes is regulated by a T cell-specific enhanceosome that consists of the transacting factors RUNX1 (previously known as AML1), CBFβ, and LEF1 and the coactivator Aly (34, 35, 36, 37, 38, 39, 40).
RUNX1 is a member of the RUNX family of transcription factors, which are essential for a number of cellular differentiation programs including osteogenesis, hematopoiesis, and gastric and neuronal development (41, 42, 43). RUNX1 expression is limited to hemopoietic tissues and is considered a master regulator of hematopoiesis (44, 45, 46). In addition to the TCR genes, RUNX1 regulates a number of genes involved in cell cycle and hemopoietic development, including the cytokine genes IL-3 (47), MIP-1α (48, 49), and GM-CSF (CSF2) (50), the macrophage CSF receptor (51, 52), and the tumor suppressor genes p21WAF-1 (53) and p14ARF (54). RUNX1 acts as both a transcriptional activator and repressor: RUNX1 represses CD4 gene expression by binding a silencer element in the CD4 locus (55, 56).
All RUNX family members contain the runt domain, which mediates DNA binding to a TG(T/C)GGT motif and interaction with its non-DNA binding partner CBFβ (57). CBFβ is ubiquitously expressed and dramatically increases the DNA-binding ability of runt-domain-containing transcription factors (46, 58). Genetic ablation of either RUNX1 or CBFβ results in embryonic lethality at E12.5 due to a complete lack of fetal liver hematopoiesis (41, 45, 46, 57).
The RUNX factors alone are relatively weak transcriptional regulators that interact with both coactivators and corepressor proteins. RUNX1 activation is increased by association with coactivators such as Aly, MOZ, and p300/CREB-binding protein (CBP); repression is mediated by recruitment of Groucho/transducin-like enhancer of split 1 and mSin3A corepressors (39, 40, 43, 48, 53, 59, 60). Whether activation or differentiation signals modify RUNX factors to specifically recruit certain cofactors and influence gene transcription patterns is not clear. In the T cell enhanceosome, RUNX1/CBFβ interact with the non-DNA binding coactivator Aly and LEF1 (36, 37, 38, 39, 40). LEF1 is a DNA binding protein expressed in T and early B lymphocytes that binds to the TTCAAAGG motif in the TCRα enhancer; LEF1 facilitates the assembly of the DNA-bound enhanceosome by inducing DNA bending through its high mobility group (HMG) domain (61, 62). In addition, the LEF1 activation domain contributes to enhanceosome-mediated enhancement of downstream TCRα promoter activity (37, 40, 62). Aly stabilizes the interactions of LEF1 and RUNX1/CBFβ with their respective DNA binding sites (40, 63). Whereas expression of LEF1 is restricted to T lymphocytes, Aly is ubiquitously expressed (40).
The existence of a T cell enhanceosome suggested the possibility that, like the B cell enhanceosome in B lymphocytes, it would be able to increase MHC class I transcription in T lymphocytes. In this report, we demonstrate that the T cell enhanceosome, consisting of RUNX1/CBFβ/LEF1/Aly, activates class I transcription both in T lymphocytes and when reconstituted in HeLa epithelial cells. Importantly, the T cell enhanceosome is associated with the class I promoter in vivo, suggesting that this activation directly affects the class I promoter. Furthermore, CIITA augmented the activation of the promoter by the T cell enhanceosome.
Materials and Methods
Cell lines and cultivation
HeLa epithelial cells were grown in DMEM supplemented with 10% FBS, 2 mM l-glutamine, 20 mM HEPES (pH 7.2), and gentamicin sulfate (10 μg/ml). The Jurkat (T cell) line was maintained in RPMI 1640 supplemented with 10% FBS, 2 mM l-glutamine, 55 μM 2-ME, 100 μM minimal essential amino acids, 1 mM sodium pyruvate, 20 mM HEPES (pH 7.2), and gentamicin sulfate (10 μg/ml). Cell lines were maintained in a humidified incubator at 37°C.
Plasmids and cloning strategies
The MHC class I promoter construct used in these studies derived from the swine class I gene PD1 (64, 65). The PD1 promoter truncation series, ligated to the cat gene (chloramphenicol acetyl transferase (CAT)) reporter, was previously described (66). The −209CAT and −209ΔCRE reporter constructs and CIITA expression vector were described previously (21). −416LUC was generated by ligating class I promoter sequences, extending from the 5′ XbaI site to the HindIII site at position +14 of −416WT (21), into the NheI/HindIII sites of the pGL2B luciferase expression vector (Promega). M10 and M12 mutants were generated by replacing wild-type core promoter sequences between unique NarI (−52) and HindIII (+14) restriction sites with double-stranded oligonucleotides containing the indicated nucleotide mismatches (indicated by italics) in the RUNX1 binding region: wild type, −43TGCGGTTCCC−34; M10, −43TGCGGTCGTA−34; M12, −43GCTAGTTCCC−34. The CBFβ, wild-type and mutant RUNX1 (53, 67), LEF1, wild-type Aly and mutant Alys (40), and mutant CBFβ-MYH11 expression constructs (68) were previously described.
Transient transfections were performed using a constant amount of DNA. Twenty-four hours before transfection, 1 × 106 HeLa were seeded in 100-mm tissue culture dishes. Transfections used standard calcium phosphate precipitation as previously described (69). The medium was replaced 24 h after transfection with fresh medium, and cells were harvested after an additional 24 h. Reporter activity was corrected by cotransfecting an internal control plasmid control, either pSV2LUC (200 ng) or CMV-β-galactosidase (100 ng). Jurkat cells were transfected by electroporation (250 V; 975 υF) using a BioRad Gene Pulser II electroporator. All CAT enzyme assays were measured in the linear range; control [14C]chloramphenicol values ranged between 20 and 80% among the different experiments. Luciferase determinations were made using a Monolight 2010 luminometer (Analytical Luminescence Laboratory).
Chromatin immunoprecipitation assay
Chromatin immunoprecipitation was performed on tissues from the B10.PD1 transgenic mouse (65) following the method of Oberley and Farnham (70). Precleared chromatin aliquots (average length 350–450 bp; ∼125 μg) were kept for “total” or were immunoprecipitated with 1 μg of specific Ab: antiPol II (Santa Cruz Biotechnology; no. sc899), anti-Acetylated H3 (Upstate Biotechnology; no. 06-599), anti-Acetylated H4 (Upstate Biotechnology; no. 06-598), anti-AML1 (Oncogene Research Products; no. PC284L), or rabbit IgG control (Jackson ImmunoResearch Laboratories; no. 011000003). For the final purification step after Proteinase K digestion, proteins and contaminants were removed by extraction and precipitation. Pellets were resuspended in 10 mM Tris-Cl (pH7.5), 30 μl per sample. Real-time PCR was performed using the Applied Biosystems Sybr-Green Kit (no. 4309155) on the Applied Biosystems 7900 machine. “Total” samples were diluted 1/100 and 1 μl was used. Immunoprecipitated samples were diluted 1/4 and 1 μl was used. PCR cycles: stage 1, 95°C for 15 min; stage 2, 95°C for 15 min, hybridization C 15, 72°C for 15 min, 40 times; stage 3, 95°C for 15 min, hybridization C 15, ramp 0.2 to 95°C. PCR primers: proximal promoter, hybridization temp 60°C, 5′, AGGCGTGGCTCTCAGGGTCTCAG, 3′, TTCCCGATCCCGCACTCACCCGCCTTGGT; control distal promoter, hybridization temp 50°C, 5′, CAATGTATTCGGTCTTAAAACTCTTAC, 3′, CTGTCTGGCTCATGGGAAAACCACT. Threshold values were corrected for rabbit IgG background and relative protein occupancy was determined relative to “total.”
RNA isolation and cytodot analysis
Total cellular RNA from transfected HeLa cells was isolated using RNAzol B (Tel-Test). RNA (20 μg) was blotted onto nytran nitrocellulose (Schleicher and Schuell) using a 96-well Minifold dot-blot arrray (Schleicher and Schuell), UV-cross-linked, and hybridized with MHC class I and control tubulin probes. The MHC class I probe was the 1.0-kb fragment of the pH 7 clone previously described (69), and the tubulin probe was obtained from Clontech Laboratories. Blots were hybridized at 42°C for 18 h; final washes were conducted at 65°C, 0.5× SSC, for the MHC class I probe or 42°C, 6× SSC for the tubulin probe. Hybridizations were analyzed using a Storm with ImageQuant software PhosphorImager (Molecular Dynamics) (Amersham Biosciences).
To examine the possibility that the T cell enhanceosome regulates MHC class I expression, we first investigated whether any of the T cell enhanceosome components individually affected the level of class I promoter activity. Each of the components—RUNX1, its binding partner CBFβ, and LEF1—was introduced by transient transfection, along with an MHC class I promoter construct consisting of 416 bp of 5′ regulatory sequences (−416WT) directing the expression of the CAT reporter into HeLa epithelial cells, which lack endogenous LEF1 and RUNX1. Although HeLa cells express basal levels of CBFβ, we also introduced an exogenous CBFβ expression vector to ensure that sufficient CBFβ protein was available to heterodimerize with RUNX1. Of the three individual components, only LEF1 showed a modest activation of MHC class I promoter (Fig. 1,A). Similarly, any two pairs of the three components displayed only small activation of class I promoter activity (Fig. 1,B). LEF1 and RUNX1 together displayed the largest effect, resulting in a 3-fold increase in promoter activity over either alone. A consistent 5-fold increase in promoter activity was observed when all three factors (RUNX1, CBFβ, and LEF1) were added (Fig. 1 B).
In T lymphocytes, the RUNX1/CBFβ heterodimer and LEF1 interact with and recruit the coactivator Aly to form an enhancesome that activates T lymphocyte-specific genes, including the TCRαβ genes (36, 37, 38, 40). Aly alone increased MHC class I only modestly in the absence of either RUNX1 or LEF1 (Fig. 1,B). However, an Aly-dependent increase was observed in the presence of any two pairs of DNA-binding factors (Fig. 1,B). The Aly-dependent increase in class I promoter activity was greatest in the presence of LEF1 and RUNX1, whether CBFβ was present or not (Fig. 1,B). These data suggest that Aly, through the formation of a T cell enhanceosome, synergizes with LEF1, RUNX1, and CBFβ transcription factors to activate the class I promoter. A mutant Aly (Alys), which has previously been shown not to activate the TCRα enhancer (40), does not activate the class I promoter in the presence of RUNX1 and LEF1 (Fig. 1 C). Consistent with the constitutive expression of low levels of Aly in HeLa cells, class I promoter activity is somewhat reduced in the presence of Alys, which functions as a dominant negative.
To further examine the role of the T cell enhanceosome in class I promoter activity, we determined the effect of a mutant CBFβ protein (CBFβ-MYH11), which is known to disrupt RUNX1 activation (68), on RUNX1/LEF1-mediated activation of class I promoter activity. The class I promoter construct, −416WT, was cotransfected into HeLa cells with a constant amount of RUNX1/LEF1/CBFβ expression vectors and increasing amounts of either mutant CBFβ-MYH11 or control vectors. Increasing amounts of the mutant CBFβ-MYH11 increasingly inhibited class I promoter activity, demonstrating that it functions as a dominant negative of the RUNX1-mediated activation (Fig. 2 A). These results are consistent with the interpretation that introduction into HeLa cells of exogenous RUNX1/LEF1/CBFβ leads to formation of a T cell enhanceosome that activates the class I promoter.
Taken together, these results demonstrate that the T lymphocyte-specific transcription factors, RUNX1 and LEF1, CBFβ, and the coactivator Aly, which form the T cell enhanceosome, activate the class I promoter, and the results also suggest that they contribute to the high levels of class I expression in T lymphocytes.
To extend this conclusion, we next determined whether the endogenous T cell enhanceosome in Jurkat T cells similarly activated the class I promoter. If the endogenous T cell enhanceosome activates class I expression, then cotransfection of the class I promoter with the mutant CBFβ-MYH11 into Jurkat T cells would repress class I promoter activity. Indeed, the presence of the mutant CBFβ-MYH11 sharply decreased the activity of the −416WT class I promoter construct in Jurkat T cells to half of its endogenous level (Fig. 2 B). These data support the conclusion that the T cell enhanceosome plays an active role in establishing high levels of MHC class I expression in T lymphocytes.
These findings demonstrate that the T cell enhanceosome activates MHC class I expression, but they do not establish whether the effect is due to direct binding of the enhanceosome to the class I promoter or if it is due to an indirect effect. To determine whether the enhanceosome binds directly to the class I promoter, we performed chromatin immunoprecipitation assays on tissues from transgenic mice that contain a stably integrated MHC class I gene, PD1, transgene. Tissues that express low (brain), intermediate (kidney), and high (spleen) levels of MHC class I expression were included in the analysis (Fig. 3). The associations of RUNX1, RNA polymerase II (pol II), and acetylated histones H3 and H4 with both distal and proximal promoter sequences in each of the tissues were examined. As shown in Fig. 3, RNA pol II and the acetylated histones were detected bound to the proximal promoter region in spleen and kidney. Only acetylated H3 histone was clearly associated with the proximal promoter in the brain. The extent of binding correlates with the known levels of class I expression in each tissue: the greatest association with the class I promoter is in the spleen, which has the highest levels of class I expression, and the least association is with the brain, which expresses the lowest levels of class I. RUNX transcription factors were detected associated with proximal MHC class I promoter sequences in the spleen and kidney, but were not observed interacting with distal sequences. The extent of RUNX1 binding to the proximal promoter in spleen and kidney correlated with the extent of RNA pol II binding. (In contrast with RUNX1 binding, Oct-1 binding was detected on the distal, not the proximal, promoter (J. Weissman and D. Singer, unpublished observations).) The detection of RUNX binding to the proximal promoter in the kidney may reflect the presence of contaminating T cells from blood. Alternatively, because the anti-RUNX Ab used in these experiments does not discriminate between members of the RUNX family, it may have bound to a cross-reactive RUNX family member present in the kidney. Importantly, the present results indicate that 1) RUNX1 interacts with the MHC class I promoter in vivo and 2) the interaction maps to the promoter proximal sequences. These findings are consistent with the conclusion that the T cell enhanceosome directly activates the class I promoter in vivo.
To define the region of MHC class I proximal promoter responsive to a T lymphocyte-specific enhancer complex, we assessed the effect of T cell enhanceosome components on a series of 5′ truncation reporter constructs of the PD1 MHC class I promoter in transiently transfected HeLa epithelial cells (Fig. 4,A). The combination of only CBFβ and LEF1 did not significantly affect promoter activity of any of the constructs, consistent with the findings described above for the construct extending 416 bp upstream (see Fig. 1). Therefore, all transfection responses were normalized to the level of activity generated by the various promoter constructs in the presence of CBFβ+LEF1. The addition of RUNX1 (in the presence of CBFβ+LEF1) activates the −313CAT, −177CAT, and −127CAT truncation constructs to approximately the same level, namely 3- to 5-fold above the level of the control (Fig. 4 A). In contrast, further truncation to −68CAT markedly reduces the ability of RUNX1 to activate the promoter. This maps a RUNX1-responsive region between −127 and −68 bp upstream of transcription initiation.
In the presence of both Aly and RUNX1, promoter constructs with 5′ upstream sequences between −313 and −127 bp are activated ∼10- to 15-fold (Fig. 4,A). Further truncation to 68 bp of upstream sequence dramatically ablates this activation (Fig. 4 A). Thus, the promoter region necessary for an Aly response maps to the same promoter segment as the RUNX1-responsive region, namely between −68 and −127 bp upstream, and defines these sequences as necessary for T cell enhanceosome activation.
Within the proximal class I promoter, there is a single consensus RUNX1 binding site, TGCGGT, at −38 to −43 bp, raising the possibility that it contributes to enhanceosome activation. We examined whether this site is critical for RUNX1-mediated activation by generating two mutations of this site within the context of the extended class I promoter and testing their ability to respond to RUNX1 in transient transfection assays in Jurkat T cells. As shown in Fig. 4 B, both of the binding site mutations reduced promoter basal activity, but neither mutation completely abrogated the ability of RUNX1 to activate the promoter. Furthermore, purified recombinant RUNX1 protein did not bind the RUNX1 DNA consensus site in the proximal promoter or to any promoter DNA fragments further upstream, as assessed by gel mobility shift assays (T. K. Howcroft and D. S. Singer, unpublished observations). Therefore, the consensus RUNX1 binding sequence in the promoter proximal region is not essential for enhanceosome activation.
The minimal TCRα enhancer consists of a CRE element upstream of LEF1 and RUNX1 binding sites (37, 40). As noted above, the DNA segment between −68 bp and −127 bp does not contain the RUNX1 consensus binding site. Furthermore, there is no consensus LEF1 binding site within this segment, nor has LEF1 been shown to bind in gel shift assays (data not shown). Indeed, the only well-characterized regulatory element in this DNA segment is a CRE that serves to mediate both tissue-specific and dynamic transcriptional signaling of class I gene expression (6, 71). The CRE, located between −100 and −107 bp, is known to interact with members of the ATF/CREB family and has been shown to both positively and negatively regulate MHC class I transcription (6, 16, 71, 72, 73, 74, 75, 76). Activation by CIITA, the IFN-γ mediator and B cell coactivator, depends on the CRE element: deletion of the CRE eliminates activation of the class I promoter by IFN-γ (16, 20, 21). Therefore, we considered the possibility that the class I CRE element participated in T cell, as well as B cell, enhanceosome function. To address this possibility, the ability of a CRE deletion construct to respond to T cell enhanceosome activation was examined. The CRE element (−100 to −107) was deleted from a construct with 209 bp of upstream sequence; the parental −209 construct (−209CAT), like the −127 truncation, is activated by the T cell enhanceosome (Fig. 4,C). Deletion of the CRE (−209ΔCRE) does not affect promoter activity in response to RUNX1 alone. Furthermore, deletion of the CRE (−209ΔCRE) does not eliminate promoter activation by the combination of RUNX1 and Aly, indicating that the CRE element is not essential for enhanceosome function. Indeed, deletion of the CRE modestly increases the response, suggesting that the CRE element functions as a silencer of T cell enhanceosome-mediated activation (Fig. 4 C). Although it has not been possible to identify a single element within the DNA segment between −68 bp and −127 bp that is either required for promoter activity or that binds RUNX1/CBFβ or LEF1 in gel shift assays (data not shown), these data indicate that the T cell enhanceosome, unlike the B cell enhanceosome, does not depend on the CRE to activate the class I promoter. Rather, the T cell enhanceosome targets a response element distinct from that used by IFN-γ/CIITA, providing a mechanism for T cells to further increase class I expression when CIITA is induced, either by IFN-γ or in activated human T cells.
The above studies demonstrated the association of RUNX1 with the class I promoter in vivo and mapped the targeted DNA segment of the class I promoter. However, they did not demonstrate the necessity of DNA binding by RUNX1 for promoter activity. To determine whether the RUNX1 DNA binding domain is required for activity, we examined the ability of various RUNX1 mutants to activate the MHC class I promoter (Fig. 5). Regulation of RUNX1 activities is complex and several overlapping regulatory domains have been described. DNA binding and association of RUNX1 with CBFβ are mediated by the runt domain; the L148D runt domain mutant can bind CBFβ but cannot bind DNA (53). Two corepressors known to associate with RUNX1 bind to C-terminal regions: the transducin-like enhancer of split proteins, which are mammalian homologues of the Groucho family of corepressors, bind the C-terminal 5 residues (VWRPY) of RUNX1, whereas the corepressor mSin3A interacts with residues 208–237 (39, 43, 53). A broad nuclear matrix targeting sequence and a transactivation domain also map to the C-terminal region and partially overlap the repression domains (77, 78, 79). As shown in Fig. 5, the DNA binding domain mutant (L148D) was unable to support activated transcription, providing evidence that RUNX1 DNA binding is critical to enhancing class I promoter activity. In addition, RUNX1 mutants with either internal deletions or truncations in the C-terminal domain between 208 and 480 aa demonstrated reduced ability to activate the class I promoter, indicating that the activation domain also is required for optimal activation.
The finding that T cell and B cell enhanceosomes target distinct regulatory elements within the class I promoter raised the question of whether these effectors would act independently or synergize with each other. To address this issue, the class I promoter construct was cotransfected into HeLa cells with various combinations of LEF1 and RUNX1 in the presence or absence of CIITA (Fig. 6, left panel). Under these conditions, LEF1 and RUNX1 together activated the class I promoter 7.3-fold. CIITA alone enhances class I promoter activity 6.6-fold, as we and others have demonstrated previously (16, 20, 21). However, in the presence of both LEF1 and RUNX1, CIITA dramatically increased promoter activity to 40.2-fold, suggesting that CIITA and the T cell enhanceosome act synergistically. This effect is in sharp contrast with the effect observed on the MHC class II DRα promoter. Unlike the class I promoter, the DRα promoter was transcriptionally silent in the absence of CIITA. CIITA induces the DRα promoter (Fig. 6, right panel). Importantly, the magnitude of CIITA activation of the DRα promoter was not increased by the presence of the T cell enhanceosome (Fig. 6, right panel). Thus, the synergy observed with the class I promoter is not universal.
The above studies all suggest that the T cell enhanceosome activates the class I promoter. Because these conclusions were based on the use of reporter assays, it was important to verify that RUNX1, CBFβ, and LEF1 could lead to increased transcription of endogenous class I genes. Therefore, we assessed the levels of endogenous class I mRNA in HeLa cells transiently transfected with RUNX1/CBFβ/LEF1 in the absence and presence of Aly (Fig. 7). In the presence of RUNX1, CBFβ, and LEF1, endogenous class I RNA levels were increased 1.4-fold. Full reconstitution of the T cell enhanceosome with CBFβ, RUNX1, LEF1, and Aly further increased class I RNA levels to 2.4-fold. In control CIITA-transfected HeLa cells, endogenous class I message was increased 3.8-fold over mock-transfected control. Taken together, the above data suggest that RUNX1, CBFβ, and LEF1 cooperate with Aly to increase class I expression in T lymphocytes in vivo.
Enhanceosomes are complexes of interacting transcription factors and DNA sequence elements, assembled to integrate various transcriptional signals at the promoter (80, 81). Enhanceosome assembly generally has been associated with the integration of extracellular signaling, because it allows combinatorial association of various proteins and sequence elements. One of the best-characterized enhanceosomes is that governing TCR gene expression (34, 35, 36, 37, 38, 39, 40). T cell maturation is accompanied by the activation of TCR genes through an enhanceosome consisting of the component proteins RUNX1, CBFβ, LEF1, and Aly. In the present study, we report that the TCR enhanceosome also regulates MHC class I gene expression, providing a molecular basis for the constitutively high levels of MHC class I expression in T lymphocytes.
Like the TCR genes, full activation of the MHC class I promoter requires all four enhanceosome proteins. Both RUNX1 and LEF1 are DNA binding proteins, whose consensus binding sites have been determined (67, 82, 83). RUNX1 binding to DNA is through its central runt domain, and activation of the class I promoter depends on both its DNA binding and activation domains. LEF1 is an HMG protein whose binding to DNA causes bending of the backbone (34, 62, 84, 85). Neither RUNX1 nor LEF1 alone significantly affected MHC class I expression in the nonlymphocyte HeLa epithelial cell line, which does not express either of these factors (data not shown). However, the combination of RUNX1 and LEF1 (along with excess CBFβ to ensure RUNX1 binding) resulted in an ∼5-fold activation of class I expression. Both RUNX1 and LEF1 have been reported to bind to the coactivator Aly (40). Addition of Aly further augmented class I promoter activity in the presence of cotransfected RUNX1 and LEF1. Overexpression of a mutant Aly (Alys) not only failed to augment but actually repressed RUNX1 and LEF1-activated class I promoter activity in HeLa cells. Furthermore, abrogating RUNX1-dependent transcriptional activity by overexpressing the mutant CBFβ-MYH11 inhibited RUNX1 and LEF1-activated class I promoter activity in HeLa cells. Importantly, CBFβ-MYH11 reduced class I transcription in Jurkat T cells by interfering with endogenous RUNX1 transcription complexes. Taken together, these data indicate that high levels of MHC class I expression in T lymphocytes are due to a T lymphocyte-specific enhanceosome complex minimally composed of RUNX1, LEF1, CBFβ, and Aly.
Despite the similarities between MHC class I and TCR gene regulation by the TCR enhanceosome, there are surprising differences. In the TCRα gene, the organization of the enhancer that nucleates the enhanceosome consists of three DNA sequence elements, CRE, LEF1, and RUNX1 binding sites (34, 35, 36, 37, 38, 39, 40). All three elements are critical for TCR enhanceosome assembly and function. It is not clear how these elements participate in the activation of the class I promoter through the TCR enhanceosome. Although a LEF1 binding site is found within the extended MHC class I promoter (T. K. Howcroft and D. S. Singer, unpublished observations), it is not within the target region, −68 to −127 bp, of the promoter that is responsive to the TCR enhanceosome. A consensus RUNX1 sequence occurs within the core promoter region (−1 to −68 bp), but mutating it does not abrogate enhanceosome activation. Finally, the CRE element does not contribute to the enhanceosome activation of the MHC class I gene, although it is contained within the target region. Nevertheless, the RUNX1-containing complex is bound to the class I promoter in vivo. Thus, the nucleation of the enhanceosome must depend on novel sequence elements in the class I promoter.
High MHC class I expression in B lymphocytes has been shown to be due, at least in part, to the expression of the coactivator CIITA (16, 20, 21). Although constitutive CIITA expression is restricted to B lymphocytes and professional APCs, CIITA is also induced by the inflammatory cytokine IFN-γ in most other cell types, including T cells (16, 20, 26, 86). CIITA expression is also induced in activated human T cells (31, 32, 33). CIITA interacts with the constitutively expressed DNA-binding transcription factors RFX5, NF-Y, and CREB to assemble a B lymphocyte-specific enhanceosome that regulates MHC class II gene expression (22, 23, 24, 25, 26, 86). Spilianakis et al. (23) have recently described the temporal recruitment of transcription factors, cofactors, and basal transcriptional components leading to the expression of the MHC class II DRα gene by IFN-γ. CIITA is also a potent activator of MHC class I genes, where it is also thought to assemble a B cell enhanceosome. However, CIITA, which has intrinsic acetyl transferase activity, has been shown to bypass the requirement for the acetyl transferase activity of the TFIID constituent TAF1 to activate MHC class I expression (21). This raises the possibility that CIITA can assemble an alternative complex on the class I core promoter.
The regulatory DNA sequence elements and pathways targeted by CIITA and the TCR enhanceosome are largely distinct. CIITA activation of the class I promoter is dependent on the CRE element located at −100 bp to −107 bp within the TCR enhanceosome responsive region (16, 20, 21). Mutation of the CRE element, which eliminates CIITA-mediated activation (21), reduced but did not inhibit RUNX1 transactivation. Furthermore, the TCR enhanceosome recruits a CBP/pol II complex to effect an increased rate of transcription from the TCRα promoter (59, 60). In contrast, CIITA activation of the class I promoter is independent of the presence of CBP (21). These distinct regulatory pathways enable CIITA to synergize with the TCR enhanceosome in increasing MHC class I transcription, providing a molecular mechanism for CIITA-mediated activation of class I expression in T cells and for modulation of class I in T cells in response to infection.
The regulation of MHC class I expression by RUNX1 raises the interesting possibility that aberrant class I expression may contribute to the pathophysiology of RUNX1-related diseases. Aberrations in RUNX1 expression and function have been linked to leukemias, which result as a consequence of somatic chromosomal translocations involving transcription factor genes (43). A number of translocation breakpoint products have been identified that block the normal function of the RUNX1/CBFβ enhanceosome (27, 68, 87, 88). The (8;21) translocation, found in ∼10–12% of acute myeloid leukemias, gives rise to AML1/ETO (new nomenclature: RUNX1-CBF2T1), consisting of the runt domain of RUNX1 in frame with almost the entire ETO gene (77, 89). AML1/ETO knock-in mice, similar to RUNX1−/− and CBFβ−/− null transgenic mice, fail to develop any hemopoietic lineages and die in utero (90, 91). AML1/ETO is thought to act as a dominant transcriptional repressor that deregulates the expression of RUNX1-responsive cytokine and/or tumor suppressor genes, severely impairing the normal differentiation process. Similarly, a mutated binding partner of RUNX1 identified in leukemias associated with Inv(16)(p13q22), CBFβ-MYH11, acts as a dominant negative inhibitor, although the mechanism of action is not clear (68, 87). As we have shown, CBFβ-MYH11 inhibits MHC class I promoter activity in Jurkat T cells (which do not express CIITA constitutively). Furthermore, AML-ETO is a potent repressor of MHC class I transcription (T.K. Howcroft and D.S. Singer, unpublished observations). Thus, it is likely that the AML1/ETO fusion protein, in addition to its effects on tumor suppressor genes, represses class I expression, thereby rendering the tumor cells invisible to immune surveillance.
Aberrant RUNX1 expression recently has also been correlated with the autoimmune diseases systemic lupus erythematosus, rheumatoid arthritis, and psoriasis (92, 93). A genome-wide analysis identified single nucleotide polymorphisms associated with each of these diseases which, when analyzed, mapped either to a RUNX1 binding site or to a variation in the RUNX1 gene expression. Based on these findings, it has been proposed that the RUNX family plays a critical role in the development of autoimmunity. Interestingly, previous studies from our laboratory have demonstrated that susceptibility to experimental systemic lupus erythematosus is dependent upon MHC class I expression: mice deficient in MHC class I expression are resistant to disease (6, 94, 95). Taken together with the present finding that RUNX1 (in the context of the enhanceosome) is a potent activator of MHC class I, these observations raise the possibility that the RUNX1 variants associated with autoimmune disease may display increased activity, leading to inappropriately high levels of MHC class I expression, which may contribute to the development of autoimmunity.
We thank Drs. Susan McCarthy, Alfred Singer, and Helen Sabzevari for helpful discussions and critical review of the manuscript. We also acknowledge the generosity of Drs. Scott W. Hiebert, Rudolf Grosschedl, Pu Paul Liu, and Jenny Ting for providing expression plasmids.
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.
Abbreviations used in this paper: CRE, cAMP response element; CBP, CREB-binding protein; CAT, chloramphenicol acetyl transferase; RNA pol II, RNA polymerase II; HMG, high mobility group.