A GATA factor mediates cell type-restricted induction of HLA-E gene transcription by gamma interferon

Abstract

The human major histocompatibility complex (MHC) class Ib gene, HLA-E, codes for the major ligand of the inhibitory receptor NK-G-2A, which is present on most natural killer (NK) cells and some CD8(+) cytotoxic T lymphocytes. We have previously shown that gamma interferon (IFN-gamma) induction of HLA-E gene transcription is mediated through a distinct IFN-gamma-responsive element, the IFN response region (IRR), in all cell types studied. We have now identified and characterized a cell type-restricted enhancer of IFN-gamma-mediated induction of HLA-E gene transcription, designated the upstream interferon response region (UIRR), which is located immediately upstream of the IRR. The UIRR mediates a three- to eightfold enhancement of IFN-gamma induction of HLA-E transcription in some cell lines but not in others, and it functions only in the presence of an adjacent IRR. The UIRR contains a variant GATA binding site (AGATAC) that is critical to both IFN-gamma responsiveness and to the formation of a specific binding complex containing GATA-1 in K562 cell nuclear extracts. The binding of GATA-1 to this site in response to IFN-gamma was confirmed in vivo in a chromatin immunoprecipitation assay. Forced expression of GATA-1 in nonexpressing U937 cells resulted in a four- to fivefold enhancement of the IFN-gamma response from HLA-E promoter constructs containing a wild-type but not a GATA-1 mutant UIRR sequence and increased the IFN-gamma response of the endogenous HLA-E gene. Knockdown of GATA-1 expression in K562 cells resulted in a approximately 4-fold decrease in the IFN-gamma response of the endogenous HLA-E gene, consistent with loss of the increase in IFN-gamma response of HLA-E promoter-driven constructs containing the UIRR in wild-type K562 cells. Coexpression of wild-type and mutant adenovirus E1a proteins that sequester p300/CBP eliminated IFN-gamma-mediated enhancement through the UIRR, but only partially reduced induction through the IRR, implicating p300/CBP binding to Stat-1alpha at the IRR in the recruitment of GATA-1 to mediate the cooperation between the UIRR and IRR. We propose that the GATA-1 transcription factor represents a cell type-restricted mediator of IFN-gamma induction of the HLA-E gene.

FIG. 1.

(A) Transient transfections in K562 cells with CAT reporter gene constructs containing the HLA-E promoter sequences indicated. The UIRR was localized to positions −231 to −194 relative to the transcription start site (n = 4). (B) Transient transfections of TK promoter-driven CAT constructs containing either no HLA-E sequences, the IRR sequence alone, the UIRR sequence alone, or both in K562 and U937 cells showed that the UIRR does not function in all cells and does not act as an IFN-γ response element by itself (n = 5). Data represent induction in response to IFN-γ and are displayed as the average ± the standard error of the mean.

FIG. 2.

(A) EMSA with ³²P-labeled UIRR as probe and nuclear extracts from untreated and IFN-γ-treated K562 cells. Numbers above lanes refer to micrograms of nuclear extract used in the reaction mixture illustrated in that lane. Two specific complexes formed between the UIRR and extract from K562 cells (shown by arrows on left-hand side), but only one specific complex formed in U937 extracts (arrow on right side), as determined by competition with cold UIRR probe (data not shown) (see Fig. 5). (B) UIRR sequence alignment of known transcription factor consensus binding sequences motifs derived from TRANSFAC analysis. Only the MyT-1 site is a perfect match for the published consensus.

FIG. 3.

Transient transfections in K562 cells with CAT constructs containing UIRR linker scanning mutants. These constructs consisted of a UIRR mutated at the sequences indicated, a 6-bp spacer, and the IRR in front of the TK promoter. Data represent induction in response to IFN-γ and are displayed as the average of five experiments ± the standard error of the mean. Only one functional domain, at position −223 to −215, was shown to cause a twofold or greater effect on induction. *, P < 0.05.

FIG. 4.

Transient transfections in K562 cells with CAT reporter constructs containing targeted mutations of the GATA and CEBP-β sites. Specific mutation of either the GA or TA half of the GATA core results in greater than twofold loss of function. Mutation of the CEBP-β core did not result in any significant change in inducibility in response to IFN-γ. Data represent induction in response to IFN-γ and are displayed as the average (n = 4) ± the standard error of the mean.

FIG. 5.

(A) EMSA with ³²P-labeled UIRR probe and K562 nuclear extracts. As shown, a 100-fold molar excess of unlabeled UIRR competes away two slow-moving complexes, indicated by the arrows, while excess GATA-1 sequence competes away only one of these complexes (arrow). Antibody to GATA-1 eliminates the complex, but antibodies to GATA-3, CEBP-β, or control mouse IgG have no effect. (B) EMSA with either a wild-type (WT) or a GATA mutant UIRR as probe with K562 nuclear extracts revealed that the specific complexes formed on the UIRR (indicated by arrows) require the GATA core consensus sequence. Specific competitors included wild-type UIRR and consensus GATA-1. In contrast, a UIRR probe containing a mutated GATA consensus sequence did not form any specific complex, as shown by the excess UIRR and GATA-1 cold competition at a 100-fold molar excess.

FIG. 6.

ChIP in K562 cells using GATA-1-specific antibody reveals that GATA-1 associates with the UIRR in vivo after treatment with IFN-γ. IN DNA was derived from total soluble formaldehyde-cross-linked nucleosomal DNA before addition of antibody. Anti-GATA-1 or control mouse IgG of the same isotype was added, and the UIRR sequence or control 18S gene sequence content of IP chromatin fractions was analyzed by real-time PCR. Data are expressed as enrichment relative to an IN DNA value of UIRR sequence abundance set at 1.0 and are presented as the average of three experiments ± the standard error of the mean.

FIG. 7.

GATA-1 gain of function confers enhancement of IFN-γ-mediated stimulation of HLA-E expression. (A) U937 tet-GATA-1 cells treated with IFN-γ and tetracycline were transfected with CAT reporter constructs driven by a minimal HLA-E promoter (pE-128) or HLA-E promoters that included only the IRR (pE-193), both the IRR and wild-type/UIRR (pE-231), or the IRR and a mutated UIRR (pE-231 GAmut). (B) Endogenous HLA-E expression in either untreated, IFN-γ-only treated, or IFN-γ-plus-tetracycline-treated cells was assayed by RNase protection assay and phosphorimaging quantitation. Results shown represent the average of three experiments with standard error indicated, with the control level set at 1. (C) RNase protection assay of GATA-1 expression from a representative experiment with the same control, IFN-γ-only, or IFN-γ-plus-tetracycline-treated U937 tet-GATA-1 cells shown in panel B.

FIG. 8.

GATA-1 knockdown in K562 cells diminishes IFN-γ response of the endogenous HLA-E gene. (A) RNase protection assay for HLA-E and triose phosphate isomerase expression in control, IFN-γ alone-, or IFN-γ-plus-tetracycline-treated K562 Si-GATA-1 cells, which are stably transfected with a tetracycline-inducible GATA-1 SiRNA expression vector. (B) Western blot analysis of aliquots of the same cell treatment samples shown in panel A. The Western blot transfer membrane was first developed with α-GATA-1 antibody followed by washing and reaction with a β-actin control antibody, as indicated.

FIG. 9.

Transient cotransfection of wild-type or mutant AdE1a inhibits IFN-γ responsiveness mediated through the UIRR. For K562 cells cotransfected with empty vector, pUC19, the UIRR significantly enhanced the IFN-γ response, as shown in Fig. 1 (*, P < 0.05; n = 3). Cotransfection with wild-type E1a construct significantly decreased the IFN-γ response mediated by both the pE-386 promoter, which contains both the UIRR and the IRR, and the pE-193 promoter, which contains the IRR alone, compared to the pUC19 vector control response (**, P < 0.05; n = 3). Cotransfection with the AdE1a mutant construct, Y/H 47/928, which binds to p300/CBP but not to Rb, inhibited any statistically significant IFN-γ induction through the UIRR-containing promoter, but only partially blocked induction through the IRR-only promoter. In contrast, cotransfection with the AdE1a mutant construct Δ2-36, which binds to Rb but not to p300/CBP, did not affect IFN-γ induction through the IRR, though the UIRR response was diminished. Data represent induction in response to IFN-γ and are displayed as the average ± the standard error of the mean of three independent experiments.

FIG. 10.

Hypothetical model of the interactions between the UIRR-bound GATA factor(s) and IRR-bound Stat-1α in the HLA-E promoter in the presence or absence of IFN-γ. A putative negative regulator that may either bind sequences adjacent to or overlapping the GATA site or bind indirectly through the GATA factor is indicated by an X in panel 1 and represents cells in which the UIRR mediates weak repression of transcription in untreated cells. The depicted factor binding and expression patterns in panels 1 and 3 represent GATA factor-expressing cells, while panels 2 and 4 represent cells that do not express GATA factors.