Enhanceosome formation over the beta interferon promoter underlies a remote-control mechanism mediated by YY1 and YY2

Abstract

The expression of beta interferon genes from humans and mice is under the immediate control of a virus-responsive element (VRE) that terminates 110 bp upstream from the transcriptional start site. Whereas a wealth of information is available for the enhanceosome that is formed on the VRE upon the signals generated by viral infection, early observations indicating the existence of other far-upstream control elements have so far remained without a molecular fundament. Guided by a computational analysis of DNA structures, we could locate three as-yet-unknown transcription factor-binding regions at -0.5, -2, and -3 kb. Our present study delineates the interplay of factors YY1 and YY2 as it occurs at the sites at -3 kb and -2 kb (otherwise called HS1 and HS2), consistent with the idea that the novel factor YY2 antagonizes the negative actions exerted by YY1. Differences between the human and murine control regions will be described.

FIG. 1.

Localization of all identified factor binding sites in the IFN-β upstream control region (SIDD profile). SIDD properties of the human IFN-β gene were calculated as part of the standard plasmid pTZ18R (5). The destabilized interferon beta promoter and terminator regions are marked P_I and T_I, respectively, whereas the coding sequence (IFN-β) is marked by a horizontal arrow. The strongly destabilized main upstream peaks of interest are termed HS1, HS2, and HS3. Transcription factor-binding sites identified by EMSAs are indicated as follows: arrowheads, YY1/YY2; dashed arrow, PURα, PURβ, and UF complex; small black arrows, Oct-1.

FIG. 2.

Luciferase reporter constructs used to demonstrate the existence of far-upstream control mechanisms. Construct P is restricted to IFN-β promoter sequences down to position −281. All other constructs contain the ∼4.2-kb EcoRI-C upstream fragment in addition. Construct P-C comprises the intact upstream region, whereas the bottom four constructs contain base pair substitutions in the marked positions, which coincide with the SIDD minima HS1, HS2, and HS3. Mutants P-C^HS1mut and P-C^HS2mut represent single mutants of either of the upstream YY1/YY2-binding sites. P-C^HS1/HS2mut is a construct with mutations at both positions. For P-C^HS3mut, the site for the undetermined factor (UF complex) (Fig. 1) in the promoter-proximal site HS3 has been modified to elucidate its potential contribution to the induced expression level of IFN-β.

FIG. 3.

The influence of upstream sequences on huIFN-β promoter induction in stable clone mixtures. (A) Consequences of a deletion (P) and a mutation in HS2 (P-C^HS2mut). The construct with the intact −4.5-kb control sequence (P-C) shows a maximum induced expression which is impaired significantly (fivefold) due to a 5-bp mutation (P-C^HS2mut; ttaaaATctcgag in place of ttaaaATGGtggt). For the minimal construct (P), the induction level is reduced only twofold, demonstrating that the extended upstream region contains both positive and negative regulatory elements. All values have been normalized to the basal construct, which was set at a value of 100; this procedure was maintained for all subsequent measurements. Standard deviations of at least three independent experiments are indicated. (B) Copy number control (Southern blots) for stable integrants with a luciferase probe. All signals in lanes 1 to 6 refer to equal amounts of genomic, high-molecular-weight DNA. Lanes 4 (P), 5 (P-C), and 6 (P-C^HS2mut) refer to genomic DNA from the stable clone mixtures tested under the experiments shown in panel A. Controls: lanes 1 and 2, genomic DNA from an independent experiment on cells previously transfected with either P or P-C, respectively; lane 3, genomic DNA from nontransfected cells; lane 7, plasmid DNA. The radioactive signals correspond to a Sca I-fragment of approximately 2.4 kb of the reportergene constructs. A DNA standard as well as relative densiometric values, as percentages of the integrated copies, are indicated.

FIG. 3.

The influence of upstream sequences on huIFN-β promoter induction in stable clone mixtures. (A) Consequences of a deletion (P) and a mutation in HS2 (P-C^HS2mut). The construct with the intact −4.5-kb control sequence (P-C) shows a maximum induced expression which is impaired significantly (fivefold) due to a 5-bp mutation (P-C^HS2mut; ttaaaATctcgag in place of ttaaaATGGtggt). For the minimal construct (P), the induction level is reduced only twofold, demonstrating that the extended upstream region contains both positive and negative regulatory elements. All values have been normalized to the basal construct, which was set at a value of 100; this procedure was maintained for all subsequent measurements. Standard deviations of at least three independent experiments are indicated. (B) Copy number control (Southern blots) for stable integrants with a luciferase probe. All signals in lanes 1 to 6 refer to equal amounts of genomic, high-molecular-weight DNA. Lanes 4 (P), 5 (P-C), and 6 (P-C^HS2mut) refer to genomic DNA from the stable clone mixtures tested under the experiments shown in panel A. Controls: lanes 1 and 2, genomic DNA from an independent experiment on cells previously transfected with either P or P-C, respectively; lane 3, genomic DNA from nontransfected cells; lane 7, plasmid DNA. The radioactive signals correspond to a Sca I-fragment of approximately 2.4 kb of the reportergene constructs. A DNA standard as well as relative densiometric values, as percentages of the integrated copies, are indicated.

FIG. 4.

Relative influence of sites HS1, HS2, and HS3 on the induction process. All experiments refer to a series of transfections similar to but independent from the results shown in Fig. 3. The three left-hand experiments reproduce the above results. The subsequent two experiments demonstrate an effect similar to P-C^HS2mut either for mutations in HS1 (P-C^HS1mut) or for a simultaneous double mutation in HS1 and HS2 (P-C^HS1/HS2mut). Construct P-C^HS3mut demonstrates that the factor underlying the UF-complex in Fig. 1 is not essential for the induction of the IFN-β promoter.

FIG. 5.

Interaction of YY2 with HS1 and HS2: EMSA and yeast one-hybrid experiments. (A) The effect of in vitro-translated FLAGYY2 on HS1- and HS2-derived DNA sequences (E0c and sE7,) in an EMSA. Lanes 2 and 4 show a distinct DNA-FLAGYY2 complex, whereas lanes 1 and 3 (negative control; in vitro-translated luciferase) are blank. Lanes 5 and 6 show blocking/supershift assays with common polyclonal anti-YY1 antibodies sc-281 (Santa Cruz) (5) and sc-1703 (6) in experiments corresponding to the results shown in lane 4. The SIDD profile marks the positions of the EMSA probes (arrowheads), which are seen to coincide with the flanks of the destabilized regions. (B) Yeast cells were stably transformed with the sE7 30mer (YY1/YY2-binding site of HS2). YY2 or YY1 binding was tested by transient transformation with pGAD424YY2 (a) and pGAD424YY1 (b). (c) Negative control (pGAD424). The figure reflects the results of a lift filter assay where the β-galactosidase signals (dark dots) in panels a and b indicate interaction of the 30mer with both YY proteins.

FIG. 5.

Interaction of YY2 with HS1 and HS2: EMSA and yeast one-hybrid experiments. (A) The effect of in vitro-translated FLAGYY2 on HS1- and HS2-derived DNA sequences (E0c and sE7,) in an EMSA. Lanes 2 and 4 show a distinct DNA-FLAGYY2 complex, whereas lanes 1 and 3 (negative control; in vitro-translated luciferase) are blank. Lanes 5 and 6 show blocking/supershift assays with common polyclonal anti-YY1 antibodies sc-281 (Santa Cruz) (5) and sc-1703 (6) in experiments corresponding to the results shown in lane 4. The SIDD profile marks the positions of the EMSA probes (arrowheads), which are seen to coincide with the flanks of the destabilized regions. (B) Yeast cells were stably transformed with the sE7 30mer (YY1/YY2-binding site of HS2). YY2 or YY1 binding was tested by transient transformation with pGAD424YY2 (a) and pGAD424YY1 (b). (c) Negative control (pGAD424). The figure reflects the results of a lift filter assay where the β-galactosidase signals (dark dots) in panels a and b indicate interaction of the 30mer with both YY proteins.

FIG. 6.

Effects of YY1 and YY2 overexpression on the induction of minimal (P) and extended (P-C) constructs stably transfected into LM^Tk− cells. (A) Both factors do not act upon the minimal promoter. Clone mixtures were supertransfected with the indicated amounts of either YY1 or YY2 expression plasmid, respectively. Neither YY1 nor YY2 overexpression influences the promoter activity of the minimal construct. All values have been normalized to the control experiment, which was set at a value of 100; this procedure was maintained for all subsequent measurements. Standard deviations of at least three independent experiments are indicated. (B) Mutual influences of YY1 and YY2 on the induction of reporter genes under the control of an extended upstream region (P-C). While the maximum effect of YY1 is already attained at 0.3 μg of transgene, YY2 alone appears to be without an effect throughout the concentration range (0.1 to 1 μg or 2.5 μg, respectively, expression plasmid). In a coexpression situation (YY1 plus YY2) (experiment shown in a separate box), however, YY2 clearly reduces the negative effect of YY1, indicating the competition of both factors for a single binding site.

FIG. 7.

Cloning and functional verification of the murine YY2 protein. (A) PCR fragments of the murine YY2 cDNA with different sets of primer pairs. RT-PCR generated single-stranded cDNAs derived from total RNA of murine NIH 3T3 cells were used as templates to amplify YY2-specific PCR products. Four distinct combinations of primer pairs were analyzed to ensure the correct amplification of YY2 DNA sequences corresponding to the published data of a cDNA originally designated as membrane-bound transcription factor protease, site 2 (MBTPS2, accession no. NP_839997 or NM_178266). 5′ primer (mYY2fw) plus lane 2, 3′ primer mYY2r1; lane 3, mYY2r2; lane 4, mYY2r3; and lane 5 (full length), mYY2r4. Lane 1 contains a 1-kb DNA ladder (Invitrogen). (B) DNA-binding capacity of in vitro-translated murine YY2, tested by EMSA. A nuclear extract of human MG63 cells (lane 1) was tested for sE7 binding, together with in vitro-translated YY2 from humans (lane 2; FLAGYY2) and mice (lane 3). Lane 4 (negative control) contains the in vitro translation mixture with the empty pcDNA3.1(−) plasmid.

FIG. 8.

The negative effect of YY1 is also valid for the endogenous IFN-β gene. For human MG63 cells, the IFN-β gene is induced by NDV subsequent to transfection of 0, 0.3, and 0.5 μg of YY1 expression plasmid (right-hand columns). Interferon levels have been determined via the protection of an indicator cell line (Vero) from the cytopathogenic effects of VSV.

FIG. 9.

Interaction of YY2 with an IFN-β promoter-associated binding site in the murine gene (32). A radioactively labeled murine IFN-β probe (P^mu) including the YY1-binding site around position −90 (32) binds FLAGYY2 (lane 4); its position in the SIDD profile has been marked by an arrow. If compared with the signal from the HS2-specific probe (lane 1), its affinity appears largely reduced. Lane 2 shows an experiment performed with the ds7-mutant M4; and lane 3 represents an EMSA analogous to the data shown in lane 4 but using the corresponding sequence from the human IFN-β promoter (P^hu).

FIG. 10.

The established proximal IFN-β promoters from both humans and mice contain additional (but different) factor-binding sites in the PRD IV element (base pairs are indicated relative to the transcriptional start site). For the murine promoter, the ATF-2/c-Jun DNA element overlaps with a YY1/YY2-binding site (for a discussion of the underlined letters, see reference 32) For the human promoter, the corresponding ATF-2/c-Jun element overlaps with an Oct-1 site (underlined letters). The presence of a functional Oct-1 site within the human IFN-β promoter sequence has been confirmed by EMSA-supershift analysis (see small inset image) using P^hu as a probe. +, the addition of an appropriate monoclonal anti-Oct-1 antibody (sc-8024; Santa Cruz); the arrow indicates the Oct-1-specific shift; the asterisk marks the corresponding supershifted DNA-protein complex.