A Novel Element and a TEF-2-like Element Activate the Major Histocompatibility Complex Class II Transactivator in B-lymphocytes*

Major histocompatibility complex (MHC) class II molecules play a central role in immune responses, and transcription of this family of genes requires the MHC class II transactivator (CIITA). CIITA has four promoters, which are transcribed in a tissue-specific manner. CIITA promoter III is constitutively active in mature B-lymphocytes. This report now describes the minimal 319-base pair promoter region necessary for maximal transcriptional activity in B-lymphocytes. Ultraviolet light and dimethylsulfate in vivo genomic footprinting analyses reveal five occupied DNA sequence elements present in intact B-lymphocytes. Functional analysis of these elements using promoter deletions and site-specific mutations demonstrates that at least two of the sites occupied in vivo are critical for transcriptional activity. In vitro protein/DNA analysis suggests that one of the sites is a TEF-2-like element and the other is occupied by a novel transcription activator. In addition, nuclear factor-1 associates with the promoter both in vivo and in vitro. In myeloma cell lines, loss of CIITA transcription correlates with a completely unoccupied CIITA promoter III. These findings suggest that CIITA transcription in B-lymphocytes is activated through at least two strong promoter elements, while loss of expression in myeloma cells is mediated through changes in promoter assembly.

Major histocompatibility complex (MHC) 1 class II molecules play a fundamental role in presenting exogenous antigenic peptides to CD4 ϩ helper T lymphocytes (1). These activated CD4 ϩ T cells release stimulatory cytokines that augment the response of CD8 ϩ cytotoxic T lymphocytes (2). Constitutive expression of MHC class II molecules is restricted to professional antigen-presenting cells, such as B lymphocytes and dendritic cells, and can be induced by interferon-␥ (IFN-␥) in macrophages, endothelial cells, and fibroblasts (3,4). The MHC class II family of genes are coordinately regulated at the level of transcription through a conserved trimeric promoter motif (3,5). A major component of this transcription complex was cloned by genetic complementation of an in vitro mutagenized MHC class II negative B cell line, RJ2.2.5. This component was named the MHC class II transactivator (CIITA) and restored MHC class II cell surface expression in one subgroup of bare lymphocyte syndrome patient cells (6). CIITA has since been shown to be a key regulatory molecule in the expression of all the known genes involved in the MHC class II antigen-presenting pathway (6 -9). CIITA is required for constitutive expression of MHC class II on B-lymphocytes (10) and with a few exceptions, such as in respiratory epithelial cells and dendritic cells (11,12), is required for cytokine-induced expression in other cell types. Mice in which functional CIITA protein was ablated clearly demonstrated an absolute requirement for CI-ITA in B cell expression of MHC class II (13). It has been shown that the CIITA gene has four distinct promoters each transcribing a unique first exon, which are located within approximately 13 kilobases of DNA (10). The promoters are utilized in a tissue-specific manner. CIITA expression in dendritic cells is primarily from promoter I. Promoter III is predominantly used in B-lymphocytes, while promoter IV is used predominantly in IFN-␥-induced expression. The first exon associated with the B cell and dendritic cell promoters encodes distinct amino-terminal residues of 24 and 101 amino acids, respectively, while mRNA from promoters II and IV initiates translation from within the common exon 2 (10). Promoter II has yet to be characterized.
The IFN-␥-responsive control elements at CIITA promoter IV were recently characterized (14,15). Both STAT-1 and IRF-1 are required at the proximal promoter for full response. Binding of the USF-1 transcription factor stabilizes STAT-1 binding (15). Initial mapping of the CIITA promoter III revealed that sequences that lie between Ϫ545 and Ϫ113 base pairs relative to the transcription start site are required for constitutive expression in B-lymphocytes (16). While this region of the promoter is not responsive to IFN-␥, IFN-␥ activation of CIITA promoter III can be observed when an additional 4000 base pairs of upstream DNA are present (16). However, nothing is known about the specific factors and promoter elements that control constitutive transcription from CIITA promoter III in B-lymphocytes.
Tight regulation of the expression of MHC class II genes has been observed during B cell development. MHC class II gene transcription begins in the pre-B cell stage, is maintained until B-lymphocytes attain maturity (17,18), and is down-regulated when B-lymphocytes terminally differentiate into plasma cells (19,20). CIITA expression is also up-regulated in the pre-B cell stages and then lost upon terminal differentiation (21,22). The mechanisms of activation and repression are unknown. CIITA promoter III does not contain any cis-acting sequences with significant homology to B cell activators except a putative octamer binding site.
In this report, we now characterize the CIITA promoter III regulatory elements in B-lymphocytes and identify in vivo and in vitro five occupied sites. Importantly, two of the sites are absolutely required for CIITA transcription in B-lymphocytes. One is a TEF-2-like element, while the other represents a binding site for a novel transcription activator.

EXPERIMENTAL PROCEDURES
DNA Constructs-The CIITA promoter III reporter constructs CI-ITAp3.1140-Luc and CIITAp3.545-Luc contain the human CIITA promoter III sequences from Ϫ1140 to ϩ123 or Ϫ545 to ϩ123 base pairs, respectively, relative to the transcription initiation site. These two constructs were originally carried in the pGL2-Basic vector and described previously as p1300CIITA.Luc and p668CIITA.Luc (16), but for this study they have been moved into the pGL3-Basic vector (Promega). Each of the other deletion constructs (namely CIITAp3.319-Luc, CI-ITAp3.195-Luc, CIITAp3.151-Luc, and CIITAp3.113-Luc) were constructed similarly and contain CIITA promoter III sequences from ϩ123 up to the indicated number on the construct name. All of the mutations except as specifically noted were carried in the CIITAp3.545-Luc construct. Mutant constructs were derived by the unique site elimination method (CLONTECH) (23). In the construct CIITAp3.545mtSiteA, the sequence 5Ј-TTGGCGGGCTCCCA-3Ј was changed to 5Ј-TTGtCtaGaT-CCCA-3Ј (mutations in lowercase boldface type). In CIITAp3.545mtSit-eB, 5Ј-TTCTTTGCATGT-3Ј was changed to 5Ј-TTCgTcGacaGT-3Ј. In the construct CIITAp3.545mtARE-2, 5Ј-TGATGATCCCT-3Ј was changed to 5Ј-TGATctagaCT-3Ј. In the construct CIITAp3.545mtARE-1, 5Ј-TTAAGGGAGTGTGGTAA-3Ј was changed to 5Ј-TTAAGtctagaTGGTAA-3Ј. In the construct CIITAp3.545mtSiteC, 5Ј-GGAAGTGAAATT-3Ј was changed to 5Ј-GGAgtcGAcgTT-3Ј. Three additional constructs with mutations in the activation response element-1 (ARE-1) sequence were developed in the pGL2-Basic vector backbone. The first (pGL2.545mtARE1-2) is in the context of the Ϫ545 base pair promoter, while the second (pGL2.151mtARE1-3) and third (pGL2.151mtARE1-4) constructs are in the context of the Ϫ151 base pair promoter. The specific base changes are shown in Table I. All mutations were confirmed by sequencing. The BSAP expression vector was cloned by reverse transcription PCR into pCDNA3.1 (InVitrogen) and confirmed by sequencing.
Cell Lines-Raji is a human Epstein-Barr virus-transformed Burkitt's lymphoma cell line that constitutively expresses MHC class II antigens (5). NCI-H929 and U266 are human plasmacytoma cell lines and were grown in 10% heat-inactivated fetal bovine serum, 2 mM L-glutamine, and 100 units/ml penicillin and streptomycin. NCI-H929 cells were also supplemented with 0.05 mM 2-mercaptoethanol.
Transient Transfections-Cells were transfected by electroporation using the Bio-Rad Gene Pulser II apparatus. Cells (1 ϫ 10 7 ) were pulsed with 200 V at a capacitance of 1070 microfarads, and cells were harvested after 42 h for luciferase assay. All cell lysates were normalized to recovered protein concentration with the Bradford protein assay (Bio-Rad).
UV Irradiation and Dimethylsulfate in Vivo Genomic Footprinting-In vivo methylation of cells with dimethylsulfate and DNA preparation were as described previously (24). In vivo UV irradiation footprinting was done essentially as described by Pfeifer and Tornaletti (25). Briefly, cells were collected by centrifugation at 4°C and resuspended at 1 ϫ 10 7 cells/ml in Dulbecco's phosphate-buffered saline. Cells (1 ml) were spread on a 100-mm plastic dish and exposed to 500 -2000 J/m 2 UV light (UV Crosslinker; Fisher). After exposure (30 -120 s), the cells were immediately harvested by the addition of an equal volume of 2ϫ lysis buffer (200 mM EDTA, 120 mM Tris-HCl (pH7.5), 0.2% SDS, 1 mg/ml proteinase K). The DNA was isolated and digested with HindIII as described for the dimethylsulfate-treated samples. The DNA (20 g) was cleaved at cyclobutane pyrimidine dimers as described previously using 20 units T4 endonuclease V (Epicenter Technologies) and 5 g of Escherichia coli photolyase generously provided by A. Sancar (University of North Carolina). Control in vitro UV-irradiated DNA samples were obtained by first isolating and purifying the genomic DNA from untreated cells, dissolving 20 g of DNA in 100 l of buffer (10 mM Tris-HCl (pH 7.5), 0.1 mM EDTA) and spotting the DNA as 5-l aliquots on parafilm. UV treatment and processing was as described for the in vivo treated samples.
Semiquantitative Reverse Transcription PCR-Cytoplasmic RNA was prepared from 1-5 ϫ 10 7 cells using the RNeasy System (Qiagen). RNA (500 ng) was reverse transcribed using the cDNA Cycle kit (In-Vitrogen) and resuspended in 20 l of H 2 O. The number of PCR cycles that produced a linear phase of amplification for each product was empirically determined. Semiquantitative PCR was performed in the linear phase of amplification using 0.5, 1.0, and 2.0 l of the cDNA for each primer pair. Each reaction was done in the presence of 2.5 mM MgCl 2 , 0.5 mM each dNTP, 1 M each primer, 2 units of Taq DNA polymerase (Life Technologies), and 2 Ci of [␣-32 P]dCTP in a total volume of 20 l.

RESULTS
The First 319 Base Pairs of CIITA Promoter III Are Sufficient for Transcriptional Activity in B-lymphocytes-The CIITA promoter III has been shown to be constitutively active in Blymphocytes (10). To define the functional regions of the promoter, a series of progressive 5Ј deletions of CIITA promoter III fused to the luciferase reporter gene were constructed and transiently transfected into the B cell line Raji. Constructs containing 1140, 545, or 319 base pairs of the promoter including the first 123 base pairs of 5Ј-untranslated sequences exhibit similar high levels of luciferase activity (Fig. 1). Additional upstream sequences spanning 7000 base pairs of promoter III do not further enhance transcriptional activity (16). This indicates that the first 319 base pairs of CIITA promoter III are sufficient for transcriptional activation in B-lymphocytes.
Deletion of the region from base pair Ϫ319 to Ϫ195 leads to a 40% decrease in transcriptional activity, which suggests the presence of a positive acting element(s) in this 124-base pair region. Deletion of base pairs Ϫ195 to Ϫ151 does not alter activity. However, deletion of the region between base pairs Ϫ151 and Ϫ113 greatly diminishes activity, to only 20% of that observed with the full-length promoter. This suggests the presence of a second positive element in this 38-base pair region. Similar results were obtained by transient transfections into the Namalwa B cell line (data not shown).
In Vivo Genomic Footprinting Analysis in B-lymphocytes Detects Multiple Protein/DNA Interactions-In vivo genomic footprint analysis of promoters has been invaluable in the visualization of protein/DNA interactions within the physiologically relevant setting of the intact cell. This technique was exploited to directly identify occupied DNA elements present in the CI-ITA promoter. In Raji B cells both strands of the promoter were analyzed from ϩ69 to Ϫ319 base pairs using dimethylsulfate (DMS) as the modifying agent ( Fig. 2A, summarized in Fig. 3). Close association of transcription factors in vivo with DNA can inhibit or enhance DMS methylation of guanine residues. The resulting pattern of methylation is resolved and compared with the pattern observed when the proteins are removed prior to DMS treatment. Analysis of the upper strand reveals seven very strong protections clustered between base pairs Ϫ142 and Ϫ133 ( Fig. 2A, lanes 2 and 4 compared with lanes 1 and 3). These contacts map within the strong activation domain functionally defined in Fig. 1 between base pairs Ϫ151 and Ϫ113. For ease of discussion, this element will be defined as ARE-1. ARE-1 does not contain guanine residues on the lower strand; thus, DMS cannot reveal interactions on the lower strand (Fig.  2B, lanes 5 and 6). A second region of in vivo protein/DNA interaction is observed between base pairs Ϫ64 and Ϫ56, and is designated activation response element-2 (ARE-2). A single strong enhancement and a partial protection are detected on the upper strand (lanes 1 and 2), while the lower strand displays three very strong protections (lanes 7 and 8). In addition to these two obvious domains of in vivo binding, three weaker sites of interaction were detectable. A cluster of five contacts designated site A is found between Ϫ18 and Ϫ27, corresponding to a sequence with homology to the NF-1 transcription factor binding site. Two partial protections are observed on the upper strand (lanes 1 and 2), while the lower strand displays two weak enhancements and a partial protection (lanes 7 and 8). A second cluster of contacts, site B, is located at a putative octamer-binding site homology (base pairs Ϫ45 to Ϫ38) and is visualized as a single protected residue on the upper strand flanked by a partial protection and weak enhancement on the lower stand. Finally, a single protection is observed at position Ϫ181 on the upper strand (lanes 3 and 4) and is designated site C. Protections were not detected on either strand upstream of base pair Ϫ195, and only two isolated weak enhancements were observed at base pairs Ϫ226 and Ϫ291.
In vivo footprinting with DMS is limited to analysis of guanine residues, which may allow some interactions to escape detection. In order to expand the effective analysis of interactions on CIITA promoter III, we established in vivo genomic footprinting using UV light as the modifying agent in place of DMS. Ultraviolet light induces the formation of pyrimidine dimers that can be subsequently converted into double strand DNA breaks. Studies on the PKG1, JUN, PCNA, and FOS gene promoters have demonstrated suppression or enhancement of UV photoproduct formation associated specifically with known sites of in vivo protein/DNA interaction (34,35). Analysis of CIITA promoter III demonstrates two strong photoproduct enhancements on the lower DNA strand at the ARE-1 element (Fig. 2B, lane 6 versus lane 4). This is revealed by the more intense signal in vivo at these two pyrimidine pairs compared with the neighboring pyrimidine pairs at similar or lower doses of UV light. This confirms the strong in vivo protections that were detected by DMS on the upper strand and validates the usefulness of the technique. A second cluster of three very strong photoproduct enhancements is located between base pairs Ϫ183 and Ϫ172, corresponding to the single DMS enhancement detected on the upper strand at site C. This confirms that site C is a site of in vivo protein binding. In summary, the in vivo footprinting analyses identified five sites of protein/DNA interaction within the first 183 base pairs of the promoter (Fig. 3).
Differentiation of B-lymphocytes into plasma cells is accompanied by suppression of MHC class II gene expression (36). In the single plasmacytoma cell line examined to date, CIITA expression was down-regulated, and introduction of CIITA rescued MHC class II expression (37). However, studies of other plasmacytoma cell lines have identified some lines that weakly express MHC class II (38). This may be due to the state of differentiation of the transformed cell lines. Two different plasmacytoma cell lines were examined to determine the level of CIITA expression and if in vivo protein/DNA interactions at the CIITA promoter were altered. One cell line, NCI-H929, has no detectable MHC class II DRA or CIITA mRNA (Fig. 4A, lanes  7-9). In contrast, the U266 plasmacytoma cell line has low levels of both (lanes 10 -12) compared with the B cell lines Raji and IM9 (lanes 1-6). When examined by DMS in vivo genomic footprinting, CIITA promoter III was completely bare of all in vivo protein/DNA interactions in the CIITA-negative NCI-H929 cell line (Fig. 4B). In contrast, in the low CIITA-expressing cell line U266, interactions were detected at the ARE-1, ARE-2, site A, and site B elements, which were indistinguishable from those observed in the Raji B cell line. Only site C binding was diminished in the U266 cells when examined by either DMS (Fig. 4B) or UV in vivo footprinting (data not shown). These findings indicate that suppression of CIITA transcription in U266 cells is mediated through the downregulation of activity of a preformed promoter complex. However, complete transcriptional repression, which occurs late in plasma cell differentiation and was observed in NCI-H929 cells is accompanied by complete disassembly of the CIITA promoter complex.
ARE-1 and ARE-2 Are Critical for Transcriptional Activi- ty-To determine the functional importance of the newly identified in vivo binding sites, we prepared site-specific mutations of the sequences where protein/DNA interactions were detected and assayed these mutations for effects on transcriptional activity (Fig. 5). All of the mutations were analyzed in the context of the 545-base pair CIITA promoter transiently transfected into the Raji B cell line.
Mutation of the ARE-1 region significantly lowered transcriptional activity to 25% of wild type activity (lane 1 versus lane 2). The reduction in activity closely parallels the loss of activity observed when the entire 38-base pair region was deleted (Fig. 1, CIITAp3.151-Luc versus CIITAp3.113-Luc). Together, these findings indicate that the ARE-1 element is a critical transcriptional activator in B cells. Mutation of the in vivo occupied ARE-2 element from base pair Ϫ57 to Ϫ61 nearly abolished transcriptional activity, indicating that it is absolutely critical for transcription (lane 3). The dramatic effect of this mutation was confirmed by testing the ARE-2-mutated promoter in another reporter vector. The significance of the weak in vivo binding activities detected at site A, site B, and site C were also tested by altering 4 or 5 base pairs at the center of each footprint. Transcriptional activity was not effected by mutation of any of the three sites (lanes 4 -6). In order to determine if site A, site B, or ARE-2 functioned cooperatively or synergistically, combinations of mutations were constructed and tested. In each construct, only mutation of ARE-2 altered transcription (lanes 7-9). Site C was also mutated in combination with the neighboring ARE-1 element. Loss of the site C binding site did not further decrease activity compared with the mutation of ARE-1 alone (lane 10 versus lane 2). These findings indicate that the primary elements responsible for CIITA promoter III activity are ARE-1 and ARE-2. containing potentially homologous known transcription factor binding sites. Incubation of an oligonucleotide containing the ARE-1 DNA element with nuclear extracts from B-lymphocytes revealed two specific complexes (Fig. 6A). Formation of each of the protein-DNA complexes was eliminated by the addition of a 50 -100-fold molar excess of unlabeled ARE-1 oligonucleotide (lane 1 versus lane 2). The upper two complexes are not B cell-specific, since they are detected in multiple cell types (lanes [7][8][9][10][11][12]. Competition with a mutated ARE-1 element or an oligonucleotide containing the site C element did not alter the binding pattern (lanes 3 and 6). A weak homology between ARE-1 and the B cell-specific activator protein, BSAP (32), as well as the Pu.1 transcription factor (39), was suggested by comparison with the available transcription factor data bases (40). A high affinity BSAP binding site from the CD19 promoter (32) did not compete for the upper two complexes, but it weakly blocked formation of the lower complex (lane 4). However, co-transfection of a BSAP expression vector along with CIITA promoter III reporter constructs containing a wild type or mutated ARE-1 element did not alter the activity of the CIITA promoter, while control promoter constructs were responsive to BSAP overexpression. 2 This finding indicates that BSAP is unlikely to be the important factor bound at the ARE-1 element in vivo. A consensus Pu.1 binding site oligonucleotide did not compete for any of the ARE-1 complexes (lane 5). Homology searches also indicated a 9 of 11 match with a region of the SV40 promoter that contains overlapping binding sites for AP3 and TEF-2 (41,42). In addition, an earlier study suggested homology with a c-Myb or c-Myc binding sequence (10). To directly test the significance of these binding site homologies, a series of site-specific ARE-1 mutations were engineered in the CIITA promoter III-luciferase construct and examined for the loss of transcriptional activation in B cells (Table I). Elimination of the c-Myb/c-Myc homology (Mt4) or partial mutation of the AP3 and TEF-2 sites (Mt3) did not significantly alter transcriptional activity when compared with the parent constructs. However, mutations (Mt1 and Mt2) that disrupt the central sequence of 5Ј-GGGAG-3Ј resulted in a loss of activity similar to deletion of the entire domain (CIITAp3.113-Luc, Fig. 1). This 2 G. Wright, personal communication.

FIG. 3. Schematic of the sequence of CIITA promoter III and in vivo protein/DNA interactions in B-lymphocytes. Open head arrows (protections) and closed head arrows (enhancements) indicate in vivo protein/DNA interactions observed after DMS treatment of cells.
Sites of ultraviolet light-induced enhancements of in vivo pyrimidine dimer formation are denoted by V. The 5Ј-end of each deletion construct tested in Fig. 1 is indicated by a vertical bar and deletion number. The transcription initiation site is indicated by ϩ1.

FIG. 4. DNA/protein interactions at the CIITA promoter III correlates with CIITA and MHC class II expression in myeloma cell lines.
A, semiquantitative reverse transcription PCR demonstrates that NCI-H929 does not express CIITA or MHC class II mRNA, while U266 expresses low levels of both. B cell lines, Raji and IM9, are shown as positive controls. To ensure the reactions are in the linear range, each group of three lanes represents PCR done with 0.5, 1.0, or 2.0 g of starting cDNA material. GAPDH was also amplified independently to control for the cDNA concentration and quality (data not shown). B, DMS in vivo footprint analysis of CIITA promoter III in myeloma cell lines. This experiment is the same as in Fig. 2A except lanes 1-4 are of the promoter in the U266 cells, and lanes 5-8 are of the promoter in the NCI-H929 cells. A nearly identical pattern of contacts is observed in the U266 line, while the NCI-H929 line is completely devoid of in vivo contacts. Lane markings are as described for Fig. 2A. sequence is a 4 of 5 match to the TEF-2 core pentanucleotide. To determine the relationship between the TEF-2 and ARE-1, in vitro binding competitions were performed. The upper ARE-1 protein-DNA complexes are competed with a TEF-2 consensus oligonucleotide (Fig. 6B). Upon using the TEF-2 consensus oligonucleotide as a DNA binding site probe, a strong specific protein-DNA complex was observed (Fig. 6B). Increasing the titration of competitor oligonucleotides demonstrated that ARE-1 was only 20% as effective as the consensus TEF-2 site in competing the TEF-2 complex. This indicates that ARE-1 binding is related to the TEF-2 binding activity but is a highly divergent and weak site. Interestingly, the presence of strong in vivo protections in this region compared with the weak binding observed in vitro suggests that the stability of the complex may depend on its interactions with other proteins at the CIITA promoter.
ARE-2 Binding Factor Is a Novel Transcriptional Activator-A similar series of experiments were performed to characterize the ARE-2 binding factor(s) (Fig. 7). A single specific complex is detected when an oligonucleotide containing the ARE-2 element is incubated with B cell nuclear extracts. The complex is abolished by the addition of a 50 -100-fold molar excess of an oligonucleotide containing the wild type ARE-2 sequence (lane 3). The addition of an oligonucleotide that contains the ARE-2 element with the mutation, which dramatically reduced transcriptional activity, did not compete for binding to this complex (lane 2). Because the ARE-2 element does not have significant homology to any known transcription factors, a panel of potentially related and unrelated DNA binding elements were used as competitors to test the specificity of the complex (lanes 4 -7). None of the oligonucleotides tested thus far have any affinity for the ARE-2 binding complexes. The analogous region from the mouse CIITA promoter III matches 8 of 11 nucleotides in the ARE-2 element; however, oligonucleotides containing the murine sequence did not have any affinity for the human ARE-2 complex (lane 4). The formation of ARE-2 complexes is not limited to the B cell extracts and can be detected in multiple cell types including lung, renal, and colon (data not shown).
Site A Is Specifically Bound by NF-1, while Site B Weakly Interacts with OTF-1-Site A and site B are located within the first 45 base pairs upstream of the transcription start site and were defined based on two clusters of weak in vivo interactions. These sites were tested for specific factors binding in vitro. The site A sequence formed one intense slow migrating complex when incubated with a B cell nuclear extract (Fig. 8). This complex was specifically eliminated by competition with a molar excess of site A oligonucleotide (lane 2) but not with the mutated site A sequence or the ARE-1 oligonucleotide (lanes 3 and 5). Site A contains sequences with homology to the binding site for NF-1 (30) and a weaker homology to the Sp1 binding site (43). An oligonucleotide containing the consensus NF-1 binding sequence also eliminated the complex (lane 4) when used as a competitor, but a consensus sequence for Sp1 binding had no effect (data not shown). The consensus NF-1 oligonucleotide forms a similar complex when it is used as the probe (lane 6). Importantly, the putative NF-1 site in CIITA promoter III can also effectively compete for this complex (lane 7). This provides strong evidence that NF-1 is the factor bound at this in vivo identified site of protein/DNA interaction.
Site B was previously suggested to be an octamer factor binding site based on a 7 of 8 base pair homology to the consensus sequence (10). However, only barely detectable levels of octamer factor OTF-1 could be observed at site B (data not shown). In comparison, strong binding of both OTF-1 and OTF-2 octamer factors was observed when the previously characterized MHC DRA octamer site was used as a probe in the same assay. This indicates that site B has very low affinity for octamer factors and is consistent with the weak in vivo interactions detected and the lack of transcriptional activation from this site. DISCUSSION In this report, we have characterized CIITA promoter III, which has previously been demonstrated to be the transcriptionally active promoter in B-lymphocytes (10,16,44). Our results show that the first 319 base pairs of CIITA promoter III are sufficient for basal transcriptional activity in B cells. Importantly, we have now utilized in vivo genomic footprinting to map the occupied promoter elements in intact B cells. Two strong AREs have been mapped in the first 151 base pairs of the promoter. In addition, the region between Ϫ195 and Ϫ319 base pairs conferred a small increase in activity (Ͻ2-fold), but only two weak and isolated in vivo enhancements were detected in this area. ARE-1 is located between Ϫ143 to Ϫ132 base pairs and contains a protein/DNA interaction identified by both in vivo footprint analysis and in vitro binding studies. The ARE-1 element has partial homology to several previously described transcription factor binding sites. BSAP, a factor that is expressed specifically in B cells and has a developmental expression pattern similar to CIITA, was present in one of the three complexes capable of binding to ARE-1 in vitro. However, overexpression of the BSAP protein in B cells did not alter CIITA promoter III activity. Thus, BSAP is unlikely to play an important role at the ARE-1 site or in CIITA transcription. ARE-1 also contains partial homology to the binding site for the ubiquitously expressed TEF-2 protein. Mutation of the pentanucleotide core sequence but not the less well conserved flanking region resulted in a loss of transcriptional activity. TEF-2-like binding activity was detected in vitro at the ARE-1 sequence, but only with weak affinity compared with the consensus TEF-2 site. Like ARE-1, TEF-2 has been shown to be a transcriptional activator, present in multiple cell types (33). TEF-2 belongs to the Kruppel family of transcription factors, which include the erythroid factor EKLF and bind to CTCCC motifs. CIITA may be activated specifically by TEF-2 or another as yet unknown family member. Interestingly, this sequence motif is also conserved in the murine CIITA promoter III sequence at an analogous position relative to the transcription start site (10).
Another region critical for basal transcriptional activity is the ARE-2 region located from Ϫ64 to Ϫ56 base pairs relative to the transcription initiation site. The ARE-2 element does not bear any homology to the binding site of any known transcription factor. Mutation of this site nearly abolishes transcription;

TABLE I
The ARE-1 sequence element and four mutations Transcriptional activity is the average luciferase activity per g of protein lysate from three independent experiments. Each mutation was tested independently with the appropriate wild type control, and the ratio is the value of activity of the mutant construct divided by the activity of the parent wild type construct. Mt1 and Mt2 are in the context of the 545-base pair promoter while Mt3 and Mt4 are in the context of the 151-base pair promoter. Partial sequence homologies are as follows: Myc/Myb (bold), TEF-2 (underline), and AP3 (italic). therefore, the binding of a transcriptional activator to this site is essential for basal CIITA transcription. Interestingly, the sequence of the murine CIITA promoter III is not well conserved compared with the human sequence in this area. Indeed, the murine sequence did not compete with the human sequence for the ARE-2 binding factor. This indicates that in contrast to ARE-1, a role for the ARE-2 binding factor may only be present at the human CIITA promoter III. Undoubtedly, the factors bound at ARE-1 and ARE-2 are critical for CIITA transcription and thus MHC class II expression in B-lymphocytes. However, their role in determining the B cell specificity of the promoter is not clear. Both of these factors are present in extracts derived from all cell types tested such as lung, colon, and kidney. However, this finding does not exclude them from being functionally regulated in a B cellspecific manner. It is possible that these factors need to be activated by coactivators, which may be present only during certain stages of B cell development analogous to the interaction of OTF-1 and its coactivator BOB-1 (45). The simplest mechanism of establishing B cell specificity through binding of OTF-2/BOB1 or OTF-1/BOB1 (46) is unlikely to play a role in CIITA B cell specificity. The only putative octamer binding site in the promoter is present at site B and does not activate transcription or bind either OTF-1 or OTF-2 with significant affinity.

GAGGGCTTAAGGGAGTGTGGTAAAA
Site A, located 12 base pairs upstream of the transcription start site, has been demonstrated to bind the transcription factor NF-1. NF-1 is involved in both transcriptional activation and replication (47,48). It binds as a dimer to the recognition site TGG(C/A)N 5 GCCAA. We observe partial protection of site A in vivo, although there is a two-base pair mismatch (lowercase type) with the consensus sequence (TGGg(C/A)N 5 cCCAA). In vitro DNA binding as examined by electrophoretic mobility shift assay demonstrated strong specific binding of NF-1 to site A. Surprisingly, mutation of this site did not alter transcriptional activity. However this finding does not exclude a role for NF-1 in CIITA transcription in vivo. NF-1 has been previously demonstrated to bind near the initiation site of several TATAless promoters such as the mouse complement C4 promoter and the AP2 promoter (49,50). NF-1 may play an important role in transcription start site selection. The CIITA promoter III also lacks a TATA box and yet predominantly initiates transcription from a single site 12 base pairs downstream of the NF-1 site. Interestingly, a previous report proposed an initiator element located at approximately 60 base pairs upstream of the initiation site; however, the site was not tested by the authors (44). Their proposed initiator overlaps with the ARE-2 transcriptional activator element. The functional significance of NF-1 binding and the mechanism of start site selection at CIITA promoter III remains to be determined.
When B-lymphocytes terminally differentiate to plasma cells, MHC class II expression is down-regulated. However, in plasmacytoma cell lines, the level of MHC class II on the cell surface varies from low to absent. Our examination of two representative plasmacytoma cell lines indicates, as predicted, that the level of MHC class II expression is correlated with the level of CIITA expression. In vivo genomic footprinting reveals that in CIITA-negative plasmacytoma cell lines the entire CI-ITA promoter III is unoccupied, while in CIITA-low cell lines full promoter occupancy is maintained. This suggests that partial suppression of CIITA transcription potentially in the early stages of differentiation may be mediated through down-regulation of the activity of established promoter complexes. In contrast, complete disassembly of the CIITA promoter and an alteration of the chromatin structure accompany complete repression of CIITA transcription in terminally differentiated plasma cells.
In conclusion, we have characterized the CIITA promoter in B-lymphocytes and identified two predominant activation elements. Importantly, ARE-1 appears to bind a TEF-2-like factor, and ARE-2 binds a novel transcription factor, both of which are likely to play a significant role in MHC class II expression. Knowledge of these activators is key to efforts directed at manipulating MHC class II expression such as in reactivation of expression in multiple myeloma and other MHC class II negative cells in order to increase their immunogenicity.