Separate promoters from proximal and medial control regions contribute to the natural killer cell-specific transcription of the human FcgammaRIII-A (CD16-A) receptor gene.

The molecular events governing the differentiation pathway of natural killer (NK) cells are not well understood. The phenotype of mature NK cells is specified by the expression of the low affinity Fc receptor for IgG (human FcγRIII, CD16) encoded by the FcγRIII-A gene. Here we report that the Pprox promoter (−198/−10) of FcγRIII-A stimulated by its own intron enhancer (+10/+712) was only one of the cis-elements that target the expression of a reporter gene in the immature NK cell line, YT. The transcription start sites of the FcγRIII-A a2/3 and a5/6 splice alternatives in NK cells were mapped to the medial −1817/−850 FcγRIII-A control region. Two promoters, Pmed1 (−942/−850) and Pmed2 (−1376/−1123) resided in this region and controlled for the initiation of these transcript classes encoding the known FcγRIII-A receptor protein. Deletion mapping studies demonstrated that the 93 base pairs −942/−850 Pmed1 sequence was sufficient to confer cell type-specific expression in YT cells. The 5′ end of Pmed1 (−942 to −921) was required for full promoter function indicating the presence of an important sequence motif recognized by a YT-specific factor. Our data suggest that this motif might be a useful tool for subsequent identification of putative transcription factors uniquely active in YT and NK cells.

Natural killer (NK) 1 lymphocytes are important effector cells in the first line of immunologic defense and play a major role in immunosurveillance. NK cells have the ability to lyse tumor cells and play a crucial role in the control of viral infections. NK cells, like cytolytic T cells (CTL), respond specifically to polymorphic determinants of MHC class I molecules. NK cells express receptors that bind to these molecules. However, in-stead of activating the cytolytic response as in CTLs the recognition of MHC class I turns off the NK cells. Therefore, virus infected and malignant transformed cells are lysed by NK cells as a consequence of loss of MHC class I surface expression (for reviews, see Refs. 1 and 2).
In addition to the allotype MHC class I specific receptors almost all human NK cells express the low-affinity receptor for the Fc portion of IgG, Fc␥RIII-A (CD16-A). Fc␥RIII-A belongs to the family of immunoglobulin G receptors (Fc␥R) involved in the clearance of immune complexes, phagocytosis of opsonized pathogens, and various forms of antibody-dependent cellular cytotoxicity (3). The Fc␥RIII-A ␣ chain forms a multimeric transmembrane receptor complex with homo-and heterodimers of the Fc⑀RI␥ subunit and/or CD3 subunit (4 -6). Fc␥RIII is also highly expressed in PMN but as a single chain receptor attached to the plasma membrane by a glycosylphosphatidylinositol anchor, Fc␥RIII-B. The transmembrane Fc␥RIII-A receptor on NK cells mediates antibody-dependent cellular cytotoxicity and all other antibody-dependent responses (7)(8)(9). The glycosylphosphatidylinositol-linked Fc␥RIII-B receptor on PMN is involved in cell activation but its detailed role is less clear (10 -12). Other surface molecules like Fc␥RII-A and CR3 receptors are likely to be involved in the activation process of PMN after Fc␥RIII-B crosslinking (13,14). The molecular basis for the expression of functionally distinct Fc␥RIII isoforms is given by the presence of the highly homologous Fc␥RIII-A and Fc␥RIII-B genes (15). Transfection experiments of reporter gene constructs indicate that differences in the proximal Ϫ198/Ϫ10 gene promoter regions might be crucial for directing the expression of the Fc␥RIII-A and B receptors to NK cells and PMN, respectively (16).
The characterization of the molecular events leading to NK cell-specific Fc␥RIII-A expression is complicated by the finding that transcripts initiating outside the Ϫ198/Ϫ10 Pprox promoter exist in NK cells (16,17). Cloning and sequencing of these Fc␥RIII-A transcripts, designated a2/3 and a5/6, demonstrated that they encode the known Fc␥RIII-A receptor. The coexpression of Fc␥RIII-A transcripts with alterations in the extracellular domain were also evident but could not be linked to the distinct a2/3 and a5/6 mRNA start sites. The medial Ϫ1817/Ϫ850 region of the Fc␥RIII-A gene containing the a2/3 and a5/6 initiation sites functioned as a transcriptional regulator in the immature NK cell line YT. This control region consisted of the two independent promoters Pmed1 (Ϫ942/ Ϫ850) and Pmed2 (Ϫ1376/Ϫ1123). The 93-bp Ϫ942/Ϫ850 Pmed1 promoter was further characterized. It contained a cisacting DNA element important in conferring optimal and YT cell-specific promoter activity within its first 21 bp. These results suggested that this DNA element might be a useful target for identification of YT and NK cell restricted transcription factors.

MATERIALS AND METHODS
Cell Preparations and Culture Conditions-Polymorphonuclear granulocytes (PMN), peripheral blood mononuclar cells (adherent cell fraction, MO), and NK cells were prepared according to standard conditions described elsewhere (9,16,18,19). The ␥/␦ T cell (MK1) and cytotoxic T cell (1B3) clones derived from donors with distinct CD16 subsets from sorted fractions were generated by limiting dilution. Clones were plated at 1 cell/well onto a feeder layer with irradiated allogeneic PBL and Epstein-Barr virus-transformed B lymphoblastoid cells (Laz509) (8). Human tumor cell lines HL60 and YT used for transfection experiments were cultured in RPMI 1640 containing 10% fetal calf serum and supplements (16). HL60 is a promyelocytic cell line and YT a cell line with NK cell characteristics (20,21). In some experiments, HL60 cells were induced to express the Fc␥RIII-B receptor isoform upon culturing in the presence of 1.2% Me 2 SO (16).
Cloning of Fc␥RIII-A mRNA Start Sites by RACE PCR-The strategy to obtain cDNA clones for a1, a2/3, and a5/6 related transcription initiation sites was nearly identical to that described recently (16). Starting with 2 g of poly(A) ϩ RNA from NK cells, the reverse transcription reaction was performed using 20 pmol of a Fc␥RIII-A genespecific primer reverse complementary to EC1 sequences from exon V. The cDNA pools were subsequently tailed with 15 units of TdT (Life Technologies, Inc.) in the presence of 0.1 mM dATP for 10 min at 37°C. After purification of the reaction mixture, one-fifth was used for PCR amplification with 10 pmol of oligo(dT) 17 -adaptor, 25 pmol of adaptor, and 25 pmol of a second internal EC1 primer in a total volume of 100 l. 2 units of Taq DNA polymerase (Promega) was added and the mixture was annealed at 56°C for 2 min. The tailed cDNA was extended at 72°C for 30 min. PCR conditions were as described (16). Purified RACE PCR products were digested with SalI and BglII and cloned into SalI/ BamHI-digested pKSϩ, as outlined in Fig. 2. Miniprep plasmid DNA was sequenced using the 32 P-labeled second EC1 primer.
Primer Extension and RNase Protection Analysis-Analysis of the distinct Fc␥RIII transcript classes by primer extension and RNase protection assay was performed according to methods described previously (16). Total RNAs were prepared by the guanidinium thiocyanate method (22). The complementary oligonucleotide used for primer extension, 5Ј-CTTCCTCGTGTTACCCAGGTCCTGCGGATT-3Ј (Ϫ795 to Ϫ824), corresponds to the genomic Fc␥RIII-A sequence relative to the translation start codon (ATG). 10 -50 g of various RNAs were annealed to this 32 P-end-labeled oligonucleotide. The extension products were sized by electrophoresis on an 8% denaturing polyacrylamide gel and visualized by autoradiography. The dideoxynucleotide sequencing ladder of a plasmid containing the sequence with the transcription start sites was used as a molecular weight marker and run in parallel. For RNase protection analysis of a2/3 and b2 transcription initiation sites, the 236-bp HincII/ApaI restriction fragment from the Fc␥RIII-A gene (Ϫ942 to Ϫ707) was inserted into the pBluescript KS(ϩ) vector. For analysis of transcripts encoding for extracellular alterations an SalI/ PvuII fragment containing the EC1/EC2 exon border (position 258 to 406) of the pGP5 derived NA1 Fc␥RIII-B cDNA (23) was used and cloned into pKSϩ. The respective antisense RNA probes were synthesized using T7 RNA polymerase. All other conditions were the same as described previously (16).
RT-PCR Analysis of Fc␥RIII-A a2/3 and a5/6 Transcript Classes-1 g of total RNAs were reverse transcribed using 15 units of avian myeloblastosis virus reverse transcriptase (Promega). PCR for amplification of the cDNA pool was started following a denaturing step at 94°C with the addition of Taq DNA polymerase after 10 min at 85°C. 35 cycles were done each including a 40-s denaturing step at 94°C, a 30-s annealing step at 52°C or 64°C depending on the primer pair in use, a 45-s extension step at 72°C, followed by a final 5-min extension at 72°C. 5Ј primers specific for the distinct 5Ј-UTR of a2/3 (5Ј-AATC-CGCAGGACCTGGGTAACAC-3Ј, Ϫ824 to Ϫ802) and a5/6 (5Ј-TCCAC-CCCTAACAAGTATC-3Ј, Ϫ1157 to Ϫ1139) transcript classes were used with the same reverse complementary 3Ј primer specific for the 3Ј-UTR originally described for a1 (5Ј-CAGAGGCCTGAGGATGATGGGGT-TGC-3Ј, ϩ829 to ϩ854) (15). The PCR products were analyzed on a 1.2% agarose gel and visualized by ethidium bromide staining.
Reporter Plasmid Constructions-The constructs were generated by cloning Fc␥RIII-A and Fc␥RIII-B genomic sequences from Ϫ10 to Ϫ198, from Ϫ10 to Ϫ1817/Ϫ1821, from Ϫ850/Ϫ846 to Ϫ1817/Ϫ1821, from Ϫ850/Ϫ846 to Ϫ942/Ϫ947, and from Ϫ850/Ϫ846 to Ϫ921/Ϫ925 into the BamHI/BglII site of the promoterless luciferase expression vector pLuc. For promoter assays in the presence of the Fc␥RIII-A or the Fc␥RIII-B enhancer the sequence from position ϩ10/ϩ712 of both genes were cloned as a PvuII subfragment into the KpnI site 3Ј to the luciferase gene. 5Ј and 3Ј deletion mutants were prepared from the medial Ϫ1817/ Ϫ850 Fc␥RIII-A control region by the use of restriction sites at positions Ϫ1579, Ϫ1376, Ϫ1123, Ϫ942, and Ϫ921 for subcloning. Relevant deletion mutants from the corresponding Fc␥RIII-B region as well as some hybrid constructs were similarly generated. All constructs were sequenced by the dideoxy chain termination method and purified over two rounds of centrifugation in cesium chloride/ethidium bromide gradients.
DNA Transfection and Luciferase Assays-YT or HL60 cells, maintained at 1 or 2 ϫ 10 7 cells/ml, were electroporated with 10 or 80 g of the various reporter plasmids at 270 V and 750 or 2400 microfarads, using a Eurogentec gene pulser followed by transfer to 15 ml of prewarmed RPMI/10% fetal calf serum medium. Twenty hours after transfection cells were harvested and washed once in PBS. Cells were extracted in 100 l of hypotonic buffer (25 mM Tris phosphate, pH 7.8, 8 mM MgCl 2 , 1 mM EDTA, 10% glycerol) by two cycles of freezing and thawing. Luciferase activity was measured in a Berthold biolumat in 22.5 mM Tris phosphate, pH 7.8, 2 mM ATP, 10 mM MgSO 4 , and 0.2 mM Luciferin.

Mapping and Cloning of Novel Fc␥RIII-A Transcript
Classes-Functional analysis of the 5Ј-flanking region of the human Fc␥RIII-A gene and the respective region of the highly homologous Fc␥RIII-B gene led to the identification of the Pprox (Ϫ198/Ϫ10) promoter regions which display different tissuespecific transcriptional activities in NK cells and PMN (16). In addition, RACE-PCR cloning indicated that additional Fc␥RIII-A mRNA start sites occurring outside the Pprox region exist in NK cells. From a total analysis of 11 cloned RACE PCR products 3 cDNA clones assigned as a2/3 were identified to initiate at two sites from Ϫ860 and Ϫ849 in a separate exon spliced from position Ϫ795 to Ϫ44 (a2) or to Ϫ62 (a3) (16). We now extended the RACE PCR analysis and isolated 17 additional cDNA clones containing the 5Ј ends of Fc␥RIII-A transcripts from NK cells. 7 clones contained the 5Ј-UTR of transcript class Fc␥RIII-A a1 (start sites at Ϫ20, Ϫ27, and Ϫ33) and 7 other clones contained a2 but not a3 starting at Ϫ865, Ϫ871, Ϫ877, Ϫ881, and Ϫ891 (summarized in Figs. 1 and 2). The remaining 3 cDNA clones contained sequences from a further upstream region of the Fc␥RIII-A gene and represented novel transcript classes assigned as a5/6 ( Fig. 2). Fc␥RIII-Aa5 initiate at positions Ϫ1278 and Ϫ1254, whereas Fc␥RIII-Aa6 initiated at Ϫ1273. These 5Ј-untranslated ends of the a5/6 transcripts were encoded by exon I ending at position Ϫ900 (a5) or alternatively at Ϫ946 (a6) and spliced to Ϫ44 now recognized as the 5Ј-border of exon III. This splice site at Ϫ44 was used by a2 as well as a5/6 transcripts. That position is the strongest protected fragment in RNase protection experiments from Fc␥RIII-A expressing NK cells but not from Fc␥RIII-B expressing PMN (16).
Next, we performed primer extension and RNase protection to determine the a2/3 start sites more precisely. In the RNase protection experiments the Fc␥RIII-A specific riboprobe ranging from Ϫ942 to Ϫ707 covering the 3Ј-border of exon II at Ϫ795 from the ATG was used, as described under "Materials and Methods." This analysis was done using RNA from various cells types. Multiple bands were observed preferentially in NK cells and 1B3 T cells, in very reduced amounts in culture activated peripheral blood monocytes but not in U937, HL60, and YT cell lines (Fig. 3). As shown on the right side of Fig. 3 one major protected fragment which mapped to position Ϫ887, as well as several minor products ranging from Ϫ865 to Ϫ889 were observed in NK cells. Consistent with these data, primer extension experiments located the a2/3 transcription initiation sites to the same region, with the major site identified at position Ϫ865 and minor sites at Ϫ867, Ϫ871, Ϫ872, Ϫ873, Ϫ878, Ϫ880, and Ϫ884.
We also analyzed transcription initiation from the same region of the Fc␥RIII-B gene using RNA from Me 2 SO-treated HL60 cells and PMN (lanes 3, 9, and 13 in Fig. 3). Differentiation of HL60 cells by Me 2 SO resulted in no induction of Fc␥RIII-Bb2 transcript initiation. In contrast, induction of Fc␥RIII-Bb1 transcript initiation is evident (16). Both primer extension and RNase protection identified the major Fc␥RIII-Bb2 mRNA start site in PMN at position Ϫ875 from the ATG translation start codon. That the Ϫ875 start site in PMN is specifically driven by the Fc␥RIII-B and not the Fc␥RIII-A gene was assessed by RT-PCR cloning and sequence analysis from PMN of two NA-2 homozygous donors. Specificity was demonstrated for several independent cDNA clones by the presence of nucleotides T and C at positions 147 and 141 within the exon V encoding the EC1 domain, as shown in Fig. 4, left side.

The Alternative Fc␥RIII-A Transcripts a2/3 and a5/6 Encode the Same Fc␥RIII-A Receptor Isoform in NK Cells-
The heterogeneity and the level of transcription of the distinct Fc␥RIII-A as well as Fc␥RIII-B mRNAs suggested the presence of simultaneously active but separate transcriptional control regions within each Fc␥RIII gene in their respective cell types. Alternative Fc␥RIII promoters could regulate the tissue-specific expression of transcripts encoding for additional Fc␥RIII-A or Fc␥RIII-B gene derived receptor-related isoforms within a single cell. As a first step testing this possibility we performed RNase protection experiments with a riboprobe containing a partially overlapping sequence from exons V/VI encoding the two extracellular EC1 and EC2 domains of Fc␥RIIIA. For the synthesis of the riboprobe a 147-bp portion of the cDNA pGP5 from the SalI site in exon V to the PvuII site in exon VI was subcloned in sense orientation upstream of the phage T7 promoter within the pKSϩ plasmid to generate pEC1-2, as described under "Materials and Methods." Analysis was done by hybridizing the EC1/EC2 spanning riboprobe made from pEC1-2 to various Fc␥RIII-A positive and negative RNAs. As expected, in the negative HL60, U937, Jurkat, and YT cell lines and with the control yeast tRNA no Fc␥RIII-A specific fragments protected by the riboprobe were detected. Interestingly, NK cells, 1B3 cytotoxic T cells, MK1 ␥/␦ T cells, and culture activated monocytes, all of them encoding the Fc␥RIII-A receptor on the cell surface, demonstrated coexpression of full-length transcripts along with transcripts containing alterations within the EC1/EC2 domains. As shown in Fig. 4, two protected fragments of 73/74 and of 147 nucleotides can be distinguished. The 147-nucleotide fragment matched the used EC1/EC2 overlap. The 73/74-nucleotide fragment was mainly protected from exon V derived EC1 sequences. Using an exon IV/V overlapping riboprobe only one protected fragment was observed (data not shown). These results indicated the effective transcription of normal Fc␥RIII-A as well as splice alternatives beyond exon V. Whether such splice variants led to the production of Fc␥RIII-A related receptors remains to be addressed.
To determine the potential correlation between exon V/VI splice variants and separate clusters of transcription initiation we performed RT-PCR analysis using RNA from NK cells with distinct a2/3 or a5/6 specific 5Ј-UTR primers and a single 3Ј end primer complementary to known Fc␥RIII-A 3Ј-UTR sequences, as described under "Materials and Methods." Using this strategy of amplification, a2/3 and a5/6 main products of 928/946 and 1156/1110 bp were generated indicating no gross alteration within exons V and VI encoding for the extracellular domains (Fig. 5). This was also verified by sequence analysis (data not shown). Therefore, the transcript classes a2/3 and a5/6 encode for the same Fc␥RIII-A receptor in NK cells as originally demonstrated for a1 (15).
Proximal and Medial Fc␥RIII-A Control Regions Act as YT Cell-specific Transcriptional Regulators-Differences in the cell type specific activities of the proximal Ϫ198/Ϫ10 Fc␥RIII-A and Fc␥RIII-B Pprox gene promoters were most evident in combination with enhancer elements. Such elements might be provided by intronic sequences from ϩ10 to ϩ712 between exon III and IV or the more upstream regions from Ϫ1817 or Ϫ1821 to Ϫ198 in both genes, as described recently (16,17). The newly identified Fc␥RIII-A and -B transcription initiation clusters now suggest that differential promoter activities would not be restricted to the proximal Ϫ198/Ϫ10 regions but could also be located to the stimulatory Ϫ1817/Ϫ1821 to Ϫ198 upstream regions.
To examine this possibility we first linked the Ϫ1817/Ϫ850 and Ϫ1821/Ϫ846 fragments from the Fc␥RIII-A/B upstream regions covering with their most 3Ј ends the a2/3 or b2 mRNA start sites to the promoterless luciferase gene. The reporter plasmids pIII-A(Ϫ1817/Ϫ850) ϩ (intr.A)Luc and pIII-B(Ϫ1821/ Ϫ846) ϩ (intr.B)Luc which also contain their respective intron enhancers cloned downstream to the luciferase gene were then transfected into HL60 and YT cells. These two cell lines were used in all our functional studies. Although presenting a more immature phenotype they can serve as model systems for PMN and NK cells, as described earlier (16). As shown in Fig. 6, the Ϫ1817/Ϫ850 region of Fc␥RIIIA produced a strong luciferase activity in the immature NK cell line YT. The amount of activity observed in HL60 cells was strongly reduced and a weak expression was seen only in the presence of the intron enhancer. Compared with the complete Ϫ1817/Ϫ10 and the proximal Ϫ198/Ϫ10 regions the Ϫ1817/Ϫ850 sequence caused a significant higher promoter activity but a similar cell type specificity with preferential expression in YT cells. Such promoter activities were not detected in transfected Jurkat T cells and myeloid U937 cells (data not shown). From these data we conclude that in addition to the Ϫ198/Ϫ10 Fc␥RIII-A Pprox promoter the Ϫ1817/Ϫ850 upstream region acts as an YT specific transcriptional regulator. This region will be referred to as the medial control region of the Fc␥RIII-A gene.
Surprisingly, the same Ϫ1821/Ϫ846 medial region from the Fc␥RIIIB gene did not produce significant luciferase activity in HL60 cells even in the presence of its endogenous intron enhancer. Very low activity could be detected in YT cells. As shown in Fig. 6, this was in sharp contrast to the results when using the complete Ϫ1821/Ϫ10 and the proximal Ϫ198/Ϫ10 Fc␥RIII-B regions. Reporter plasmids containing these latter sequences produced very strong promoter activities specific for HL60 cells. Our data indicate that HL60 cells lack some factors necessary for proper function of the medial but not the proximal Fc␥RIII-B promoters. Based on these observations we focused in our subsequent analysis mainly on the Fc␥RIII-A medial control region.
The Fc␥RIII-A Medial Control Region Consists of the Two Separate Promoters Pmed1 and Pmed2-After establishing the contribution of the Fc␥RIII-A medial control region in YT cellspecific expression we addressed the question whether the cis-acting sequences from this region were sufficient and responsible for constitutive promoter activity and cell type specificity. Two series of 5Ј and 3Ј deletion mutants were generated and cloned upstream to the luciferase gene into reporter plasmids lacking the endogenous intron enhancer and tested by transfection into YT cells. This approach provided evidence that the Fc␥RIII-A medial control region consists of two separate promoters, termed Pmed1 and Pmed2. These two promoter activities appeared to be YT cell specific. All of the deletion mutants were almost negative for luciferase expression when transfected into HL60 cells (data not shown).
As shown in Fig. 7, different Fc␥RIII-A 5Ј deletion mutants originating at position Ϫ1817 produced variable amounts of luciferase activities in YT cells suggesting the presence of compensatory enhancer, repressor as well as promoter elements. The original reporter plasmid pIII-A(Ϫ1817/Ϫ850)Luc produced luciferase activities in the range of 15 to 23 ϫ10 3 relative light units (RLU) which were set as 100% relative activity. As an internal standard we used the pTK luciferase control plasmid. Deletion of the most upstream region from Ϫ1817 to Ϫ1579 drastically reduced activity indicating the presence of a stimulatory element. Further deletion to position Ϫ1376 Ϫ942 to Ϫ707 of the Fc␥RIII-A gene was hybridized to the various RNAs and then digested with RNase A and RNase T 1 . Primer extension was performed with a 32 P-labeled oligonucleotide complementary to nucleotides Ϫ795 to Ϫ824. The autoradiographs of RNase protected fragments (left) and reverse transcribed products (middle) analyzed on 8% polyacrylamide sequencing gels are shown. Sequencing ladders (ACGT) derived with the same primer from the Fc␥RIII-A gene were run in parallel. The localization of the RNA start sites in NK cells relative to the ATG codon is indicated (right). The numbering of the start sites (arrows) indicate their distance to the ATG codon designated as ϩ1.
caused an increase to full luciferase expression. This suggested that negative regulatory sequences reside in the Ϫ1579 to Ϫ1376 region which might be compensated by the enhancer element from the most upstream region. To analyze whether the Ϫ1579 to Ϫ1376 region acted as a silencer, this part of the Fc␥RIII-A gene was cloned upstream to the thymidine kinase (TK) promoter into the pTK luciferase control plasmid and transfected into YT cells. A reduction in the expression levels down to 26% was achieved indicating that this property of repression is transferable to a heterologous promoter (data not shown).
Luciferase activity continued in more extensive 5Ј deletion mutants containing the Ϫ942 to Ϫ850 sequence covering all of the mapped a2/3 mRNA start sites. The Ϫ942/Ϫ850 sequence within the Fc␥RIII-A medial control region will be referred to as the Pmed1 promoter. A further deletion by 21 bp with the pIII-A(Ϫ921/Ϫ850)Luc plasmid resulted in a strong decrease in Pmed1 promoter activity. This suggested that the sequence between Ϫ942 and Ϫ921 contains an YT cell-specific promoter element.
The importance of the 21-bp Ϫ942/Ϫ921 region for YT specific activity of the Pmed1 promoter was also observed in functional studies using 3Ј end deletion mutants. The deletion of the first 70 bp from the Pmed1 promoter within the pIII-A(Ϫ1817/Ϫ921)Luc plasmid retained almost complete luciferase expression. A further deletion of the remaining 21-bp Pmed1 sequences reduced the relative activity of the pIII-A (Ϫ1817/Ϫ942) 3Ј deletion mutant to levels of about 25-30%. Most importantly, these data also suggested that additional promoter sequences within the Fc␥RIII-A medial control region upstream from the Pmed1 promoter account for this residual luciferase expression. To define the region responsible for this remaining promoter activity we generated more extensive 3Ј end deletion mutants as well as constructs containing nonoverlapping fragments. As shown in Fig. 7, all the constructs containing the Ϫ1376 to Ϫ1123 sequence were able to confer activity to the luciferase reporter gene. In the absence of the upstream repressing element of the Ϫ1579 to Ϫ1376 region the pIII-A(Ϫ1376/Ϫ1123) reporter construct produced the highest amounts of luciferase activity comparable to what has been observed with the Pmed1 promoter. This region also contains the 5Ј ends of the cloned a5/6 RACE PCR products at Ϫ1278, Ϫ1270, and Ϫ1254 and will be referred to as the Fc␥RIII-A Pmed2 promoter.
To examine whether nucleotide differences were responsible for repressed Fc␥RIII-B medial promoter activity, some of the Fc␥RIII-A deletion mutants were compared with their corresponding Fc␥RIII-B sequences in the functional studies. To test the contribution of each Pmed1 and Pmed2 region, analysis were also done using A/B hybrid constructs. All constructs lacked the intron structures to avoid possible neutralizing effects due to promoter/enhancer competition. As shown in Fig. 8, the complete Ϫ1821/Ϫ846 medial control region of the Fc␥RIII-B gene was inactive in its respective HL60 cells and produced some residual activity about 8-fold lower compared to the Fc␥RIII-A derived sequences in YT cells. Slightly higher levels of luciferase expression were produced by the pIII-B(Ϫ947/Ϫ846)Luc plasmid, 4-fold lower than observed for the corresponding Fc␥RIII-A Pmed1 promoter region. The deletion of the Ϫ947/Ϫ846 region in the pIII-B(Ϫ1821/Ϫ947)Luc construct resulted in a complete loss of promoter activity, whereas the pIII-A(Ϫ1817/Ϫ942)Luc containing the Fc␥RIII-A Pmed2 promoter was still active (data not shown). Similar results were obtained using Pmed1/Pmed2 hybrid constructs from both genes (Fig. 8). Therefore, the residual Fc␥RIII-B activity resides within the Pmed1 but not the Pmed2 region. Total repression or strong reduction in the Fc␥RIII-B Pmed2 and Pmed1 activities in YT cells appeared to be mediated by differences in the nucleotide sequence. Comparison of the Ϫ1380/ Ϫ1127 Fc␥RIII-B and Ϫ1376/Ϫ1123 Fc␥RIII-A Pmed2 sequences revealed a single nucleotide exchange of A to G at position Ϫ1341 in Fc␥RIII-B creating a Sp1 consensus site in case of the Fc␥RIII-A gene (boxed in Fig. 1). As outlined in Fig.  9, the sequence of the Fc␥RIII-A Pmed1 promoter differed through a truncation of the 8-bp motif GGAGCCCT which is three times repeated in the respective Fc␥RIII-B region. For both genes the main a2/3 or b2 mRNA start sites were mapped to positions Ϫ887 and Ϫ865 in NK cells or Ϫ875 in PMN directly downstream to this motif (Fig. 3). DISCUSSION The human Fc receptors with low affinity for IgG (Fc␥RIII, CD16) are encoded by two genes (III-A and III-B) resulting in differential tissue-specific expression of alternative transmembrane or glycosylphosphatidylinositol-anchored isoforms in NK cells and PMN, respectively. Sequence conservation of about 97% identity have been described between both coding (15) and flanking (16) regions of each gene. Reconstitution studies of the distinct Fc␥RIII cell type specificities in transgenic mice have indicated that the cis-elements sufficient for NK or PMN restriction might locate to the same 5.8-kilobase fragment containing about 4.5 kilobase pairs of the 5Ј-flanking sequence and the first intron in each gene (24). In vitro transfection analyses have demonstrated that the first 0.2-kilobase (Ϫ198/Ϫ10) of 5Ј-flanking Fc␥RIII sequences (16) enhanced by their first introns contribute in directing the expression of a reporter gene to their respective YT (NK-like) or HL60 (PMN-like) cell types (17). The data presented here define transcription initiation upstream of the Ϫ198/Ϫ10 sequences from both genes, suggesting additional control through alternative promoters. In the case of Fc␥RIII-B the respective upstream sequences are rather inactive, most likely due to the premature phenotype of HL60 cells used as PMN substitutes (Fig. 6). Our deletion studies indicate that the YT/NK cell-specific expression of Fc␥RIII-A is dependent on a complex arrangement of transcriptional regu-latory regions including a putative enhancer from Ϫ1817 to Ϫ1579, a suppressor from Ϫ1579 to Ϫ1376 as well as the two adjacent promoter elements Pmed2 from Ϫ1376 to Ϫ1123 and Pmed1 from Ϫ942 to Ϫ850 (Fig. 7).
The engagement of transcription initiation factors by both Pmed1 and Pmed2 regions is supported by the finding that the mRNA start sites of transcript classes a2/3 and a5/6 mapped to both elements in mature NK cells. In all functional studies both Pmed1 and Pmed2 promoters are much more active than the recently characterized (Ϫ198/Ϫ10) Pprox promoter in YT cells. This is in accordance with RNase protection experiments which identifies the Ϫ44 splice site present in a2 and a5/6 as the prominent protected fragment (16). An important question addressed in these studies is whether the distinct transcript classes a2/3 and a5/6 compared to a1 are controlled by alternative promoters and are involved in the tissue-specific expression of more than a single Fc␥RIII-A receptor isoform in NK cells. The coexpression of mRNAs with or without alterations in exons V/VI encoding the extracellular EC1/EC2 domains (Fig. 4) strongly inferred that alternative splicing participate in Fc␥RIII-A transcript heterogeneity. On the other hand, RT-PCR amplification with NK cell-derived mRNA yields a pattern of a1, a2/3, and a5/6 products indicative for a single Fc␥RIII-A receptor ( Fig. 5 and data not shown). In addition, characterization of nine cDNA clones isolated from two independent libraries constructed to size-fractionated lymphokine-activated killer or NK cell mRNA identified two clones containing exon II sequences specific for a2 which are otherwise identical to the Fc␥RIII-A coding sequence (data not shown). We propose that evolvement of the separate Pprox, Pmed1, and Pmed2 promoters does not necessarily correlate with the events of alternative splicing within the extracellular domain of the Fc␥RIII-A receptor. Whether other transcript classes like Fc␥RIII-Aa4 (16) express for modified Fc␥RIII-A products remains to be addressed. Based on earlier experiments it is most likely that a4 contains additional 5Ј end sequences distinct from a1, a2/3, and a5/6, suggesting that the Fc␥RIII-Aa4 initiate by a more upstream transcriptional regulator different from Pmed1 and Pmed2 (16,17).
Results from our functional studies showed high promoter activity within the 92-bp segment (residues Ϫ942 to Ϫ850) of the Fc␥RIII-A gene, termed the Pmed1 promoter. Comparison with other parts of the III-A and III-B genes in YT and HL60 cells revealed that differential cell type specificities are due to this region (Fig. 8). An 8-bp repeat motif differed between Pmed1 and the relative Fc␥RIII-B gene region. This motif, GGAGCCCT, is repeated three times in Fc␥RIII-B but affected in Fc␥RIII-A Pmed1 by a C to T exchange in the second repeat and absent in the third repeat (Fig. 9). In YT cells the truncated version present in Pmed1 shows strong improved activity over the Fc␥RIII-B sequence. Attempts to identify YT-specific DNAbinding proteins in gel retardation assays recognizing this sequence difference were not successful. The same pattern of binding were observed for both sequences (data not shown). Therefore, the different organization of the repeat motif might influence binding affinity or oligomerization of a transcription factor rather than destroying binding capability. Possibly, a functional cooperation between the distinct repeat motifs and a further upstream element common to both genes could render only the Fc␥RIII-A Pmed1 to be specifically active in YT cells. The 21-bp sequence from residue Ϫ942 to Ϫ921 on the 5Ј end of the Pmed1 is required for full promoter function. Deleting this region resulted in almost inactive Pmed1 activity. It is reasonable to assume a cis-element within this region to cooperate with the repeat motif. Sequence comparison revealed a consensus site for Ets proteins, GGAA/T, within this 21-bp segment. Several Ets family members have been shown to be involved in the differential expression of T-cell specific genes, such as the TCR, interleukin-2, and perforin (25)(26)(27). The 15-bp NKE motif, CCCACTTCCTGGCCA, bearing the core Ets site is nearly identical to the mPfp CTL-specific element (residues Ϫ508 to Ϫ494) (Fig. 9). The trans-acting factor NF-P2 exclusively expressed in cytolytic lymphocytes and specifically modulated FIG. 7. Extensive 5 and 3 end deletion mapping studies from the medial control region of the Fc␥RIII-A gene to identify two separate promoters Pmed1 (؊942/؊850) and Pmed2 (؊1376/؊1123). Constructs with the indicated 5Ј and 3Ј deletions (left) were transiently transfected into the immature NK cell line YT. Representative luciferase activities relative to the pIII-A(Ϫ1817/Ϫ850)Luc plasmid set as 100% are shown (right). Promoter activity measured in RLU after transfection of this plasmid into YT ranged between 15,000 and 23,000 RLU. Pmed1 is the 93-bp region from Ϫ942 to Ϫ850 and Pmed2 is the 254-bp region from Ϫ1376 to Ϫ1123 defined by this analysis. The main most upstream RNA start sites initiated by these two regions as defined in Figs. 1-3 are shown by arrows. upon activation has been identified to interact with the mPfp 15-mer sequence (24). The coexpression of perforin and Fc␥RIII-A in some subsets of CTLs, ␥/␦ T cells as well as NK cells, suggests that NF-P2 or related proteins may contribute to the Fc␥RIII-A Pmed1 specificity.
The molecular basis for NK and YT cell-specific transcription is completely unknown. Our data describe for the first time some of the relevant Fc␥RIII-A cis-acting gene elements includ- ing Pmed1 and Pmed2. A YT cell-specific sequence motif (NKE) located within the first 21 bp in the Pmed1 promoter element. We propose that a functional cooperation between the NKE and the so-called repeat motif contributes to the YT-specificity of the Fc␥RIII-A Pmed1 promoter. The identification and characterization of NK cell-specific DNA elements will allow targeting gene expression to NK cells clarifying the role of these important effector cells in the first line of immunologic defense. In addition, NK and YT cell-specific sequence motifs can be used to isolate transcription factors uniquely active in these cell types. Finally, elucidation of the factors involved in the expression of NK cell-specific genes as Fc␥RIII-A will give insight into the control of NK cell differentiation and development.