Abstract
The low-affinity receptor for IgG (human FcγRIIIA) is selectively expressed by a subset of T lymphocytes, NK cells, and macrophages. To understand the mechanisms underlying this pattern of cell type-specific expression, we initially identified alternative promoters, Pmed1/2 and Pprox, in the 5′ end of the FcγRIIIA gene. In this study, we focused on the Pmed1 promoter and demonstrated this 93-bp region to be highly specific in governing restriction to NK/T cell lines. This property of Pmed1 is context independent and can extend to a disparate promoter. Deletion analysis defined a contribution of two separate elements located to the 5′ 21-bp (−942/−922) and 3′ 72-bp (−921/−850) regions of Pmed1 in conferring NK/T cell specificity. The 5′ part of Pmed1 contains binding sites for Sp1 and NK element-recognizing factors and substitution mapping studies revealed a critical requirement of the Sp1-I site. The importance of Sp1 protein to regulate maximal Pmed1 promoter activity was further established by EMSAs and cotransfection experiments in Sp1-null Drosophila SL2 cells. Our data suggest that Sp1 can contribute, in part, to NK/T cell restriction and further indicate that the FcγRIIIA Pmed1 sequence might be useful to direct the NK/T cell-specific expression of heterologous genes.
Natural killer lymphocytes are important effector cells in natural immunity and play a major role in the killing of certain tumor cells and virally infected cells that lack some or all MHC class I molecules on their surface (1, 2, 3). This mode of NK cell reactivity has been termed missing self-recognition, and it is now well accepted that the lysis of MHC-I-deficient cells is based on the failure of NK cells to receive an inhibitory signal via MHC receptors such as mouse Ly49 or human KIR family receptors (4, 5, 6).
Very little is known about the molecular events that regulate differentiation and gene expression in NK cells. In the mouse, there are few reports on the transcriptional regulation of NK cell-specific genes encoding allotype MHC-specific NK cell receptors, Ly49A and 2B4 (7, 8), and perforin (9, 10, 11). Binding of the transcription factor TCF-1 to two sites in the Ly49A promoter is critical to regulate gene activity and to direct clonal acquisition of the Ly49A receptor during NK cell development (7). Sp1-and Ets-related transcription factors constitute a minimal promoter region responsible for NK- and CTL-specific expression of perforin (11). A killer cell-specific Ets-related factor, termed NF-P2, has been observed to interact with a 5′ upstream NK enhancer (NKE)3 motif of the perforin gene (9). In addition to the MHC-specific receptors and perforin, mouse NK cells are characterized by the expression of low-affinity FcRs, FcγRIII, through which they bind IgG and exert Ab-dependent cellular cytotoxicity (12, 13). High levels of FcγRIII are found on early fetal thymocytes containing progenitor cells of NK and T cells suggesting regulated FcγRIII expression during murine NK cell development (14).
Previous studies in humans show mutually exclusive expression of distinct FcγRIII isoforms that are encoded by the two homologous genes, FcγRIIIA and FcγRIIIB (15). The transmembrane FcγRIIIA isoform is present on macrophages, γδT lymphocytes, and NK cells, whereas FcγRIIIB is a GPI-anchored protein that is mainly expressed on granulocytes (13, 16, 17, 18, 19). Reconstitution studies of the different FcγRIII cell type specificities in transgenic mice suggest the 5′ end of each gene to be involved in macrophage and NK/T cell vs granulocyte restriction (20). Reporter gene assays confirm that the first 0.2 kb of 5′ FcγRIIIA/B sequences, the Pprox promoter regions (−198/−10), target gene expression to their respective cell types (21, 22). Interestingly, expression of FcγRIIIA in NK and γδ T cells, but not macrophages can also occur via Pprox-independent transcript initiation. Functional analyses of the FcγRIIIA gene identify alternative promoters, termed Pmed1 (−942/−850) and Pmed2 (−1376/−1123), demonstrated to be active in the NK-like YT cell line (23). The nucleotide sequence of Pmed1/2 differ at minor positions to their corresponding regions of the FcγRIIIB gene and these changes may determine that the transmembrane FcγRIIIA but not the GPI-linked FcγRIIIB receptor isoform is expressed in NK/T cells.
In the present study, we examined the capacity of Pmed1, in comparison to Pprox, to regulate FcγRIIIA expression in macrophages and/or NK/T cells by using transient transfections and reporter gene assays of different cell types. Moreover, we investigated the contribution of transcription factor binding to FcγRIIIA Pmed1 in cell type-specific transcriptional initiation. By using this strategy, we showed that Pmed1 represents a highly active and NK/T cell-specific promoter. Sp1 is one of the factors required for maximal Pmed1 promoter activity that contributes, in part, to NK/T cell-specific expression. Importantly, FcγRIIIA Pmed1 sequences can confer NK/T cell restriction at a disparate promoter. Analysis of this unique property should result in the development of new DNA-based vectors for targeting expression of heterologous genes specifically to NK/T cells.
Materials and Methods
Reporter plasmid constructions
Subcloning of the Pprox (−198/−10), Pmed1 (-942/-850), and Pmed1 Δ21bp (−921/−850) promoter fragments of the FcγRIIIA gene into the BamHI/BglII sites of the basic promoterless luciferase expression vector pLuc was described previously (21, 22, 23). For the experiments with heterologous promoters, the Pmed1 and Pmed1 Δ21bp fragments as well as three copies of the 5′ 21-bp (−942/−921) part were cloned in inverse orientation upstream to the minimal TATA box element of the thymidine kinase (TK) promoter into the pTATA luciferase plasmid. Relevant hybrid constructs from the corresponding FcγRIIIB gene region were similarly generated. Deletion and substitution mutants of the 21-bp 5′ part in FcγRIIIA Pmed1 were prepared from the −942/−850 FcγRIIIA gene region by a PCR-based approach using modified oligonucleotides containing BamHI restriction sites for subcloning into pLuc. A similar strategy was used to construct the mutations of the different Sp1 consensus binding sites in the FcγRIIIA Pmed1 luciferase clone. All subclones containing inserts were sequenced to confirm the presence of the deletion or mutation. Large-scale plasmid preparations were made either by two rounds of centrifugation in cesium chloride/ethidium bromide gradients or by a purification kit (Qiagen, Valencia, CA) for all reporter gene constructs, divided into aliquots, and stored at −20 to −80°C until used in transfection assays.
Tissue culture, transient transfections, and luciferase assay
YT (human cell line with NK-like characteristics (24)) cells were maintained in Iscove’s medium (Biochrom, Berlin, Germany) containing 15% FCS and supplements. U937 (human histiocytic lymphoma cell line) and Jurkat (human T cell leukemia cell line) cells were maintained in RPMI 1640 medium containing 10% FCS and supplements. RAW264.7 (mouse monocyte-macrophage cell line) cells were grown as monolayer culture in RPMI 1640 medium containing 10% FCS, 0.1 mmol/L 2-ME, and supplements. In some experiments, HL-60 cells were induced to express FcγRIIIB during differentiation to neutrophils upon treatment with 1.2% DMSO (21).
For transient transfection experiments, ∼1–2 × 107 cells were electroporated with 20–40 μg of the various reporter plasmids using an Eurogentec gene pulser Eurogentec, Brussels, Belgium). To monitor transfection efficiency, 2 μg of the internal pCMVβgal plasmid (a gift from B. Lüscher, Hannover Medical School, Hannover, Germany) was used. Transiently transfected cells were incubated in growth medium at 37°C in 5% CO2 for 20 h. Cells were harvested, washed twice with PBS, and analyzed for luciferase activity with an assay kit (Promega, Madison, WI) according to the manufacturer’s instructions. Luminescence was detected by using a Berthold Lumat LB 9501 (Berthold, Nashua, NH) and luciferase expression levels were normalized to the levels of β-galactosidase.
Sp1-null Drosophila SL2 cells (ATCC CRL-1963) were grown in Schneider medium (Life Technologies, Grand Island, NY) at 25°C as described previously (25, 26). Briefly, 5 × 106 SL2 cells were electroporated at 270 V and 1500 μF with 40 μg of reporter plasmids and 5 μg of effector plasmid pPac-Sp1 (containing full-length Sp1 cDNA in the pPAC expression vector; kindly provided by R. Tjian, Berkeley, CA (27)). Luciferase activity was examined 24 h after transfection.
Preparation of nuclear extracts
Nuclear extracts were prepared according to a modified protocol by Dignam, allowing rapid isolation of nuclear proteins (28). Briefly, 1 × 107 cells were washed twice in ice-cold PBS and once in buffer A containing 10 mmol/L HEPES (pH 7.8), 1.5 mmol/L MgCl2, 10 mmol/L NaCl, 0.2 mmol/L EDTA, 0.5 mmol/L DTT, 0.5 mmol/L PMSF, and a mixture of proteinase inhibitors, including leupeptin (0.1 μg), aprotinin (1 μg), pepstatin A (1.5 μg), and soybean trypsin inhibitor (1 μg). Cells were swollen by 10 min of incubation on ice and nuclei were recovered by centrifugation at 4°C for 3 min at 13.000 × g. The nuclei were then lysed for 30 min in high salt buffer C (20 mmol/L HEPES (pH 7.8), 25% glycerol, 420 mmol/L NaCl, 1.5 mM MgCl2, 0.2 mM EDTA, and the same proteinase mixture as used in buffer A), and nuclear extracts were obtained by centrifugation at 4°C for 5 min. Total protein concentration in the nuclear extract preparations was measured by using a Bio-Rad kit (Hercules, CA). Most preparations contained 1.5–3 μg/μl protein, and these protein extracts were frozen immediately and stored at −80°C until used.
EMSAs
Complementary oligonucleotides were annealed and 5′ end-labeled using [γ-32P]ATP (3000 Ci/mmol; Amersham Biosciences, Braunschweig, Germany) and T4 polynucleotide kinase (MBI Fermentas, Vilnius, Lithuania). Radiolabeled double-stranded oligonucleotides were purified from unincorporated ATP by passage through a Sephadex G-25 spin column and stored at −20°C until used.
EMSAs were performed by incubating 3 μg extracted nuclear proteins with 10,000 cpm of double-stranded oligonucleotides in a total volume of 20 μl reaction mixture containing 50 mmol/L Tris-HCl buffer (pH 7.4), 250 mmol/L NaCl, 10 mmol/L DTT, 10 mmol/L PMSF, 25% glycerol, and 5% skim milk powder at 20°C for 30 min. For competition analysis, a 100-fold molar excess of unlabeled double-stranded oligonucleotides was preincubated with nuclear extracts. For the Sp1 shift experiment, a polyclonal Sp1 Ab preparation (Santa Cruz Biotechnology, Santa Cruz, CA) was incubated with nuclear extracts for 15 min after the addition of radiolabeled probe. The binding reactions were loaded on a 4% nondenaturing polyacrylamide gel containing 0.5× Tris-borate-EDTA buffer and electrophoresed at 200 V. Gels were exposed and analyzed by a phosphor imager (Fuji-Raytest, Straubenhardt, Germany).
Results
Transient transfection of FcγRIIIA Pprox and Pmed1 promoter plasmids in different cell types
The previous analysis on the regulation of FcγRIIIA expression in NK and γδ T cells had identified distinct a1, a2/3, and a5/6 transcript classes differing by their 5′ untranslated (UT) sequences but encoding the same FcγRIIIA receptor protein (23). As outlined in Fig. 1⇓, the mRNA start sites had been mapped to distinct promoter regions, namely, the Pprox (located from position −198 to −10 of the ATG translation start codon) for a1, Pmed1 (−942/−850) for a2/3, and Pmed2 (−1376/−1123) for a5/6. In contrast, FcγRIIIA expression in macrophages had been shown to utilize FcγRIIIAa1 transcript initiation only (21, 23), suggesting different regulatory capacities of the Pprox and Pmed1/2 promoters in directing FcγRIIIA expression to either macrophages and/or NK/T cells, respectively.
Organization of the FcγRIIIA 5′ transcriptional control region. FcγRIIIA exons of the distinct 5′ untranslated regions (5UT1, 5UT2), the signal peptide (S1, S2), and the first extracellular Ig-like domain (EC1) are shown as boxes. Important cis-active elements (E, enhancer; S, silencer; P, promoter) are indicated. The position of alternative promoters Pprox (−198/−10), Pmed1 (−942/−850), and Pmed2 (−1376/−1123) associated with transcript initiation of FcγRIIIa1, a2/3, or a5/6 mRNA are shown below the FcγRIIIA gene as the distance to the ATG translational start codon (+1).
To test this hypothesis, the FcγRIIIA Pprox and Pmed1 promoter fragments cloned 5′ to the luciferase reporter gene were transiently transfected into various cell types. Transfected cells were harvested 20 h after transfection and luciferase activity was measured. As shown in Fig. 2⇓, the highest level of luciferase activity was found for the Pmed1 plasmid in the immature NK cell line YT set as 100% (activity measured in relative light units (RLU) was in the range of 90,000–125,000 RLU). A reduced level of ∼40% luciferase activity was also detected for Pmed1 Luc in Jurkat T cells but not in U937.1 and RAW267 cell lines. The Pprox promoter plasmid showed substantial luciferase activities of 40–50% in YT NK and Jurkat T cells, as well as of 20–30% in monocytic-macrophage U937.1 and RAW267 cells. These results are in good agreement with the previous FcγRIIIA transcript analysis (23) and show that the Pmed1 is a highly active promoter directing NK/T cell-specific expression. In addition, the data on the broader specificity of Pprox are similar to those observed for the entire −1817/−10 FcγRIIIA promoter sequences (Ref. 21 and data not shown), indicating Pprox but not Pmed1 to be responsible for the additional regulation of FcγRIIIA expression in macrophages. In our subsequent analysis, we focused mainly on the NK/T cell-specific activity of Pmed1.
Activity of FcγRIIIA Pmed1 and Pprox after transfection into different cell types. FcγRIIIA promoter fragments Pmed1 (−942/−850) and Pprox (−198/−10) were cloned into the basic pLuc vector and transiently transfected into the indicated cell lines. To monitor efficiency of transfection, the pCMVβgal plasmid was used as positive control (data not shown). Cells were harvested 20 h later and luciferase reporter gene activity was measured. Three to five independent experiments were performed in triplicate. Representative luciferase activities relative to the Pmed1 plasmid set as 100% in YT cells are shown.
The FcγRIIIA Pmed1 sequence contains binding sites for Sp1 and NKE-recognizing factors and is capable of conferring NK/T cell specificity on a TATA box
We first investigated whether FcγRIIIA Pmed1 could control NK/T cell-specific expression in the context of a heterologous promoter. Hereby, the 92-bp Pmed1 DNA fragment was cloned in inverse orientation upstream to the minimal TATA box promoter element of the TK gene in the pTATA vector. Cloning of the 92-bp sequence in inverse orientation to the promoterless pLuc served as control plasmid. When transfected into different cells, the pTATA/Pmed1-inv construct but not the pTATA and pPmed1-inv plasmids achieved high levels of specific expression in YT and Jurkat cell lines (Fig. 3⇓), indicating that the 92-bp Pmed1 segment is capable of conferring NK/T cell-specificity on a disparate promoter.
FcγRIIIA Pmed1 is sufficient for NK/T cell induction of a heterologous promoter. The structure of the reporter plasmids—pTATA, pPmed1, pPmed-inv, and pTATA/Pmed1-inv—are shown (left). These constructs were transfected in YT, Jurkat, U937.1, and RAW267 cells, and resultant luciferase activities expressed as RLU from a representative experiment are shown (right).
Because the 92-bp Pmed1 is a NK/T cell-specific promoter, we examined more closely the sequence for potential binding sites of known transcription factors. Three consensus binding sites for Sp1 (I, II, III) and a NKE motif were found (Fig. 4⇓A). To demonstrate binding of the Sp1 and NKE-recognizing factors to these sites, we performed EMSA analysis using five consecutive and overlapping 30-mer base pair segments of the FcγRIIIA Pmed1 promoter fragment, termed OL1 to OL5. Nuclear extracts from the YT cell line were used as a source of nuclear proteins. As shown in Fig. 4⇓B, three main complexes were formed between the OL1 and OL5 probes and YT-derived nuclear proteins. Complex I was seen with OL1, OL2, and OL5 containing the Sp1-I/II/III sites; complex II consisted of two bands and occurred with OL1 containing the NKE motif (a similar EMSA profile had been previously described by Koizumi et al. (9) for the NKE sequence of the mouse perforin gene); complex III formed a broad smear with OL4 which contains the region of the a2/3 mRNA initiation sites (22, 23), thus suggesting an interaction of the RNA polymerase II containing transcription initiation complex.
FcγRIIIA Pmed1 contains binding sites for Sp1 and NKE-recognizing factors. A, Nucleotide sequence of FcγRIIIA Pmed1. The main transcription start points (mtsp) of FcγRIIIa2/3 mRNA and putative binding sites for Sp1 protein and NKE-recognizing factors are indicated. The overlapping double-stranded 30-bp probes OL1 to OL5 used in DNA-protein binding assays are shown. B, 32P-labeled OL1 to OL5 were incubated with nuclear extracts of YT cells, and the samples were analyzed by EMSA.
Binding of Sp1 to the 5′ end Sp1-I site is critical for maximal FcγRIIIA Pmed1 activity as indicated by mutagenesis, EMSA, and cotransfection experiments
After establishing the importance of the FcγRIIIA Pmed1 region to regulate NK/T cell-specific expression in a context-independent manner, we addressed the question which of the identified cis-acting sequences (Sp1-I/II/III and NKE) are responsible for promoter activity and/or cell type specificity. A 5′ deletion by 21 bp of the −942 to −850 Pmed1 sequence with the (−921/−850) Pmed1Δ21bp Luc plasmid had been previously shown to lead to an almost complete abrogation of promoter activity (23). To define the relative contribution of the Sp1-I (position −937 to −928) and NKE (position −930 to −922) binding motifs within the 21-bp 5′ end region, we first generated ΔSp1-I and ΔNKE promoter mutants. As shown in Fig. 5⇓A, deletion of the −942/−931 sequences in the Pmed1 ΔSp1-I Luc plasmid resulted in a strong decrease of activity down to 20% RLU in both YT and Jurkat cells. In contrast, deletion of the −927/−922 sequences in the Pmed1 ΔNKE reporter construct showed only moderate effects with residual activities of 40–50% RLU in YT cells and 80–85% in Jurkat cells. These data indicate a prominent role of the Sp1-I recognition site, while the NKE motif appears to be more dispensable.
Mutation analysis of the 21-bp 5′ end region of the FcγRIIIA Pmed1 demonstrates functional binding of Sp1 to the Sp1-I site in the regulation of maximal promoter activity. A and B, Constructs with the indicated changes (inv21), deletions (Δ21bp, ΔSp1-I, and ΔNKE) and mutations (M1, M2, M3, M4, M5) were transfected into YT and Jurkat cells. Representative luciferase activities relative to the Pmed1 plasmid set as 100% for each cell type are shown. C, Left panel: changed sequences of triple-base mutated oligonucleotides (M1–M5 as compared with WT) used in the direct binding or competition assays. The location of the putative Sp1-I binding site is shown. Right panel: EMSA was performed with nuclear extracts from YT cells. The binding reaction mixtures contained 32P-end-labeled DNA probes (WT or M1–M5), and a 100-fold molar excess of competitors as indicated. Functional binding of Sp1 was examined by pretreatment with an anti-Sp1 polyclonal Ab.
Similar results were obtained in a second set of reporter gene assays using the five consecutive substitution mutants Pmed1-M1 (−939/−937; AAC changed into CCA), Pmed1-M2 (−936/−934; CCT changed into AAG), Pmed1-M3 (−933/−931; CCC changed into TTT), Pmed1-M4 (−930/−928; ACT changed into CTC), and Pmed1-M5 (−927/−925; TCC changed into AGT). The triplet-base pair exchanges in M1 and M5 were of minor importance, whereas major effects were obtained with the M2 to M4 Pmed1 mutant constructs resulting in strongly decreased promoter activity in YT and Jurkat cells (Fig. 5⇑B). This correlated with the disappearance of a dominant retarded band in EMSA analysis when using YT nuclear proteins and M2- to M4-labeled DNA sequences (Fig. 5⇑C). Moreover, the same DNA-protein complex was specifically eliminated either after preincubation with a polyclonal anti-Sp1 Ab or competitively by the addition of a 100-fold molar excess of unlabeled wild-type (WT), M1, and M5 but not M2 to M4 probes (Fig. 5⇑C).
To confirm the specificity of Sp1 binding to the Sp1-I site within the 21-bp 5′ end of FcγRIIIA Pmed1, we also performed cotransfection experiments of Sp1-deficient SL2 Drosophila Schneider cells using an Sp1 expression vector (pPac-Sp1) and three copies of the 21-bp (−942/−922) region fused in both orientationsupstream to the TK gene promoter. The results shown in Fig. 6⇓ indicate that cotransfection of 3 × 21bp-TK-Luc and 3 × 21bpinv-TK-Luc plasmids with pPac-Sp1 in SL2 cells resulted in an ∼19-fold increase of luciferase activity, thus providing evidence for functional binding of Sp1 to the Sp1-I recognition site of FcγRIIIA Pmed1. Similar results were obtained when using cotransfection of Pmed1-Luc with pPac-Sp1 (data not shown).
Sp1 binding to the 5′ Sp1-I site regulates transcriptional initiation at a heterologous promoter in Drosophila SL2 cells upon cotransfection with an Sp1 expression plasmid. Sp1-negative Drosophila SL2 cells were transiently cotransfected with pTKLuc and pTKLuc containing three copies of the Sp1-I regulatory sequence of the 21-bp 5′ end of FcγRIIIA Pmed1 either in sense or antisense orientation, with or without an expression plasmid for Sp1 (pPAC-Sp1). Cells were harvested 24 h after transfection and reporter gene activity was assessed. Results are expressed as severalfold induction of luciferase activity through pPAC-Sp1. One representative experiment of four performed in triplicate is shown.
Both the 5′ Sp1-I site and the inactive 71-bp 3′ region contribute to the NK/T cell specificity of FcγRIIIA Pmed1
The results of the mutagenesis, EMSA, and cotransfection studies show the importance of the Sp1 transcription factor to interact with its cognate Sp1-I sequence in the regulation of maximal FcγRIIIA Pmed1 promoter activity. To further characterize the role of the Sp1-I site in cell-type specificity, a heterologous promoter construct, pTATA/Pmed1Δ21bp-inv, was made in which the 5′ 21-bp region containing the Sp1-I site was deleted from Pmed1 inversely fused to the TATA element of the TK gene linked to the luciferase reporter. There was reduced expression of pTATA/Pmed1Δ21bp-inv when transfected into either YT or Jurkat cells (Fig. 7⇓), confirming that Sp1-I contributes to NK/T cell restriction.
The 3′ 71-bp Pmed1 subfragment of FcγRIIIA but not FcγRIIIB retains a reduced capacity in conferring transcriptional initiation at a heterologous promoter in NK/T cells. A, The structure of the reporter constructs pTATA/Pmed1 Δ21bp-inv along with sequence differences of FcγRIIIA and FcγRIIIB are shown. B, Plasmids were transfected into YT, Jurkat, U937.1, and RAW267 cells. Results are expressed as x-fold induction of luciferase activity to pTATA luciferase activity by representing the mean ± SE value from four independent experiments, each performed in triplicate.
On the other hand, expression was not completely abolished in the deletion mutant, which indicates that elements of the almost inactive 71-bp 3′ end (−921/−850) of FcγRIIIA Pmed1 also serve to control NK/T cell-specific expression. To test this, a second heterologous promoter construct was generated using a similar deletion fragment (−925/−846), but derived from the FcγRIIIB gene. Consistent with previous findings, the Pmed1 DNA fragment of FcγRIIIB specifically induced luciferase expression in HL-60 cells upon 1.2% DMSO treatment, but was inactive in NK/T cells (Ref. 23 and data not shown). Sequence comparison between Pmed1 of FcγRIIIA and FcγRIIIB revealed identical Sp1-II/III sites, but differences in an 8-bp motif (5′-GGAGCCCT-3′; located around the start site of transcription) which is repeated 3-fold in FcγRIIIB and affected in FcγRIIIA by a C to T exchange in the second repeat and lack of the third repeat (Fig. 7⇑A). When transfected into YT and Jurkat cells, pTATA expression was ∼10-fold induced by the Pmed1Δ21bp-inv fragment of FcγRIIIA, but not FcγRIIIB (Fig. 7⇑B). These results suggest that, in addition to the 5′ Sp1-I region, the 3′ part of FcγRIIIA Pmed1 is involved in NK/T cell restriction.
Discussion
NK cells play an important role in innate immunity and tumor defense mechanisms (1, 2, 3). They are regulated by opposing signals from distinct receptor systems that either induce or inhibit NK activity (4, 5, 6). In addition, activation of NK cells can occur via FcγRs, leading to ADCC (12, 13). The most important FcγR on human NK cells is the transmembrane FcγRIIIA isoform. FcγRIIIA expression on NK cells, γδ T cells, and macrophages is under the control of separate promoters, including Pprox and Pmed1/2 (see Fig. 1⇑). The Pprox of FcγRIIIA contains a PU.1/Spi1 recognition site, the PU box, conserved among several FcγR genes, including FcγRI (29), FcγRIIA (30), and FcγRIII (31). PU.1 is expressed in myeloid cell lines (32) and can interact with the PU box of the human FcγRIIIA Pprox promoter (21), a finding that is consistent with Pprox-driven reporter gene expression in U937.1 and RAW267 cells (see Fig. 2⇑). PU.1 also determines the myeloid-specific activity of the murine FcγRIIIA promoter (31). Together, these results may indicate that similar mechanisms are operative in the transcriptional control of macrophage-specific FcγRIIIA expression in both mice and humans. In contrast, however, human but not mouse FcγRIIIA genes use Pmed1/2 as alternative control regions for NK/T cell-specific expression (23, 31). In this study, we thus focused mainly on the Pmed1 region. This region possesses a minimal structure of 93 bp with a 5′ NKE-like motif and three Sp1-I/II/III recognition sites and confers activity on the luciferase reporter gene exclusively in YT (NK) and Jurkat (T) cell lines (see Figs. 2⇑ and 3⇑). EMSA studies using NK cell extracts demonstrate that all of the Sp1-I/II/III sites and the NKE sequence participate in the binding by nuclear factors (see Fig. 4⇑). Previous results indicate a functional interaction of Ets-related proteins, NF-P1/2, with the NKE element of the mouse perforin gene and NKE multimers can enhance reporter gene expression driven by the minimal perforin promoter in killer cells (9). However, our initial attempts to demonstrate a similar capacity of the NKE-like motif of FcγRIIIA Pmed1 were not successful, since three copies of NKE sequences cloned upstream of the heterologous TK promoter do not induce reporter gene activity in a NK cell-specific manner (data not shown). Moreover, deletion of the NKE-like sequence has a minor effect on Pmed1 activity (see Fig. 5⇑). This may suggest that NKE is not the major regulatory element in the FcγRIIIA Pmed1 promoter. Competition, supershift, and cotransfection studies now proved that the Sp1 transcription factor can interact with FcγRIIIA Pmed1 and mutations of either Sp1-I or Sp1-III sites abolish promoter function (see Figs. 5⇑ and 6⇑). It is well established that self-association of Sp1 and subsequent DNA bending or looping between adjacent DNA-Sp1 activation sites is one of the general mechanisms of transcription initiation (33, 34). Since the two Sp1-I and Sp1-III sites of FcγRIIIA Pmed1 are located upstream and downstream of the major transcription start site, it might be possible that optimal Pmed1 activity requires such an interaction between Sp1 binding to promote transcription.
An important feature of the human FcγRIIIA Pmed1 control region is its ability to extend NK/T cell restriction at a heterologous promoter (see Figs. 3⇑ and 7⇑). Two different regions of Pmed1 are required for this gene context-independent specificity, one of which is located to the 21-bp 5′ end containing the Sp1-I site, which may indicate a contribution of Sp1. In the absence of the 21-bp 5′ region, the remaining 72-bp 3′ part of FcγRIIIA Pmed1 displays no substantial promoter activity (see Fig. 5⇑), while retaining a residual capacity to confer NK/T cell restriction at a TATA box element (see Fig. 7⇑). Importantly, the natural insertion of an 8-bp repeat element in the 3′ end subregion of the related FcγRIIIB Pmed1 completely inactivates this type of specificity (see Fig. 7⇑). The functional differences of NK/T cell restriction thus seem to directly correlate with the structural differences in the two FcγRIII Pmed1 promoters and help to explain why FcγRIIIA but not FcγRIIIB is expressed in γδ T cells and NK cells (23).
In summary, we showed that the two alternative Pprox and Pmed1 promoters of FcγRIIIA have different specificities, with Pprox regulating expression in both macrophages and NK cells, while Pmed1 is uniquely active in NK/T cells. Moreover, the results presented here provide the first evidence that the Pmed1 control region of human FcγRIIIA can function as a context independent NK/T cell restriction element in vitro. This specificity is determined, in part, by binding of Sp1 and also involves a second component located to the inactive 3′ part of Pmed1. Nucleotide changes in this latter region which result in abrogated NK/T cell specificity are found in the FcγRIIIB gene and may account for the observed mutual exclusive expression of FcγRIIIA but not FcγRIIIB in γδT and NK cells. Our findings also imply for the potential use of FcγRIIIA Pmed1 sequences to direct experimental NK/T cell-specific expression of heterologous genes. To test for this possibility, an in vivo-transgenic approach is required and currently under investigation. Targeting specific expression of candidate genes will certainly provide novel insights on the molecular events leading to the differentiation and development of NK/T cells and, in addition, may help to modify NK cell-mediated effects important in bone marrow transplantation and the treatment of cancers (35, 36).
Acknowledgments
We thank B. Lüscher and R. Tjian for providing plasmids and members of our laboratory for helpful comments on this manuscript.
Footnotes
-
↵1 This work was supported in part by Grant Ge892/4-2 of the Deutsche Forschungs-gemeinschaft.
-
↵2 Address correspondence and reprint requests to Dr. J. Engelbert Gessner, Abteilung für Klinische Immunologie, Medizinische Hochschule Hannover, Carl-Neuberg-Strasse 1, 30625 Hannover, Germany. E-mail address: Gessner.Johannes{at}MH-Hannover.de
-
↵3 Abbreviations used in this paper: NKE, NK element; Pprox, proximal promoter region of FcγRIII genes; Pmed1, medial promoter 1 region of FcγRIII genes; RLU, relative light units; SL2, Drosophila Schneider line 2 cells; TK, thymidine kinase; WT, wild type.
- Received November 14, 2001.
- Accepted January 16, 2002.
- Copyright © 2002 by The American Association of Immunologists