The Role of High-Mobility Group I(Y) Proteins in Expression of IL-2 and T Cell Proliferation

Cell culture and transfection

The basic medium for Jurkat cell culture was RPMI 1640 medium containing 10% FCS, supplemented with l-glutamine and penicillin-gentamicin antibiotics (RPMIJ). The basic medium for PBL culture was as above but contained 20% FCS, 100 μM 2-ME, and 5% conditioned media (RPMIL). Mononuclear cells were isolated from peripheral blood using lymphoprep (Nycomed Pharma AS, Oslo, Norway) and cultured for 4 days in RPMIL and PHA (5 μg/ml) (Boehringer Mannheim, Mannheim, Germany). Cells were then stimulated with 5 ng/ml of IL-12 (PharMingen, San Diego, CA) for 4 h to promote transition to G₁ and panned to remove macrophages. Nonadherent PBLs were pelleted and resuspended in RPMIL at 1 × 10⁷ cells in 400 μl media. PBLs were assayed for CD3, CD4, and CD8 expression using the Cyto-Stat assay kit from Coulter (Palo Alto, CA). Ninety-four percent of the cells were CD3 positive, 74% were CD4 positive, and 20% were CD8 positive. Cells were transfected by electroporation using a Bio-Rad Gene Pulser II (Richmond, CA) at 290 V with a capacitance of 975 μF. Jurkat cells were resuspended at 5 × 10⁶ cells in 400 μl media and electroporated at 270 V, 975 μF capacitance. The efficiency of transfection was determined using the pCMVGFP expression plasmid and flow cytometry (Epics XL-MCL; Coulter) and ranged from 2 to 4% green fluorescence protein (GFP) positive in PBLs and 15 to 25% GFP positive in Jurkat cells. Transfected cells were incubated in RPMIL with 10% conditioned media (PBLs) or RPMIJ media (Jurkats) for 24 h. Cells were then sorted for high fluorescence using a FACStar^Plus cell sorter (Becton Dickinson, Mountain View, CA), and the top 1–2% for PBLs and 10–15% for Jurkats were collected and used in IL-2 ELISA and proliferation assays.

Reporter assay, ELISA, and proliferation assay

pRcCMV and pcDNA3.1/Zeo were obtained from Invitrogen (San Diego, CA), and pRcCMVIGMH, pIL-2luc, and pHIVluc have been previously described (14). Standard site-specific mutagenesis procedures, the details of which will be reported elsewhere (Li et al., unpublished observations), were used to create a non-DNA binding mutant form of the HMGI protein designated HMGI(mII,mIII) starting with the wild-type human HMGI cDNA, clone 7C (38). Briefly, HMGI(mII,mIII) had four proline to alanine substitutions introduced at amino acid residues 57, 61, 83, and 87 located in its second and third DNA-binding domains, the primary regions of the protein that interact with the minor groove of A:T-rich substrates (39). As a consequence, the recombinant HMGI(mII,mIII) protein lacks the ability to specifically bind with high affinity to A:T-rich DNA sequences in vitro but retains its ability for specific protein-protein interactions with other transcription factors (Li et al., unpublished observations). For expression in transfected mammalian cells, the mutant HMGI(mII,mIII) cDNA was subcloned into the pcDNA3.1/Zeo plasmid vector (Invitrogen) and used as described above. The HMGY cDNA was also subcloned into pcDNA3.1 to generate pcDNAHMGY for expression of the protein in cells. Jurkat cells used in the reporter assay were cotransfected with 5 μg of pIL-2luc or pHIVluc together with pRcCMVIGMH, pcDNAHMGI(mII,mIII), or pcDNAHMGY, in combination with the amount of parent plasmid needed to normalize DNA at 10 μg. At 24 h posttransfection, cells were pelleted and resuspended in phenol red free RPMIJ media at 1 × 10⁶ cells/ml. Then, 100 μl of cells/well were plated in 96-well culture plates and stimulated with 50 ng/ml PMA and 1 μM Ca²⁺ ionophore (P/I) (A23187; Boehringer Mannheim) or with the above and a 1/10,000 dilution of ascites fluid containing activating Ab to the CD28 receptor (α-CD28) (Bristol-Myers Squibb, New York, NY) for 9 h. Cells were then harvested and assayed for luciferase activity using a 96-well plate Luminometer (TopCount; Packard, Meriden, CT) as previously described (14). PBLs were cotransfected with 5 μg of pIL-2luc and either 10 μg of pRcCMVIGMH, pcDNAHMGI (mII,mIII), or pcDNAHMGY and the appropriate parent plasmid. At 24 h posttransfection, cells were stimulated with PHA (5 μg/ml) for 16 h. Cells were then harvested and cell extracts assayed for luciferase activity as previously described (40).

ELISA and proliferation assays were performed with PBL or Jurkat cells transfected with 10 μg of the appropriate expression plasmid together with 5 μg of pCMVGFP for cell sorting. Then, 1 × 10⁴ sorted cells in 100 μl were plated in 96-well tissue culture plates and stimulated with either PHA for 24 h (PBL) or P/I+α-CD28 for 8 h (Jurkat), and supernatants were assayed for IL-2 using a Quantikine ELISA kit (R&D Systems, Minneapolis, MN). Proliferation was assayed in similarly treated cells using the chemiluminescent bromodeoxyuridine incorporation kit from Boehringer Mannheim.

Production and purification of recombinant proteins

Recombinant hexahistidine-tagged full-length Fos and Jun proteins, as well as truncated Fos_116–211 and Jun_224–334 proteins, were prepared and purified by affinity chromatography using a Ni-NTA-agarose column (Qiagen, Chatsworth, CA) as described (41). Fos and Jun proteins were corenatured in vitro into the heterodimer transcription factor AP-1 by step-wise dialysis from 6 M urea with the final buffer containing 25 mM sodium phosphate, pH 7.6, 5% glycerol, and 5 mM DTT (42). Full-length recombinant human HMGI protein (i.e., the unspliced member of the HMGI(Y) protein family; Ref. 29) was produced using the expression vector pET7C carrying the wild-type human HMGI cDNA (38) as described (43, 44). NF-ATp and c-Rel were prepared as described (45, 46). The purity of each recombinant preparation was assessed by SDS-PAGE (47) Protein concentrations were determined spectrophotometrically employing either a Bio-Rad protein assay kit or using the extinction coefficient ε₂₂₀ = 74,000 L/mol · cm for wild-type HMGI protein (48).

Preparation of nuclear extracts

Nuclear extracts were prepared from Jurkat T cells stimulated with PMA (20 ng/ml), Ca²⁺ ionophore (1 μm), and CD28 Ab (1:10,000) for 1 or 6 h. Nuclear extract preparation was as previously described (14).

In vitro DNase I footprinting

A cloned 410 bp XhoI/HindIII restriction fragment encompassing nucleotides (nt) −360 to +50 of the human IL-2 gene proximal promoter (49) was the starting fragment for use in in vitro protein footprinting. Promoter subfragments were isolated by either selective restriction enzyme digestion or PCR amplification techniques (47). A fragment from −180 to −60 was used for some of the footprinting experiments. Standard gel electrophoretic procedures (47) were used to isolate all DNA fragments followed by purification on a Qiagen column as described by the manufacturer. Restriction enzyme fragments were 5′-end radiolabeled with T4 polynucleotide kinase and [γ-³²P]ATP. The ends of PCR fragments were selectively radiolabeled by incorporating, during the final few cycles of the amplification reaction, either one or the other of the two PCR primers that had been 5′ radiolabeled with T4 polynucleotide kinase and [γ-³²P]ATP.

Footprinting of both the HMGI and AP-1 recombinant proteins on promoter DNA fragments radiolabeled on one 5′ end, employing the nuclease DNase I, followed published protocols (35, 44). For each protein and DNA substrate, optimal conditions for footprinting were empirically determined. Single-stranded DNA cleavage products were then separated by electrophoresis on a 6% sequencing gel with Maxam-Gilbert “G-lane” chemical cleavage products of control DNA fragments serving as reference standards (47). Band intensities were quantitatively analyzed using a PhosphorImager machine and ImageQuant software (Molecular Dynamics, Sunnyvale, CA).

EMSA

Oligonucleotides for gel shift analysis had the following sequences: IL-2 CD28RR, 5′-TGGGGGTTTAAAGAAATTCCAGAGAGTCATCAG-3′; CD28APm, 5′-TGGGGGTTTAAAGAAATTCCAGAGAGagATCAG-3′; CD28REm, 5′-TGGGGGTTTAAAGccATTCCAGAGAGTCATCAG-3′; IL-2 NF-IL-2A, 5′-AAGTCTTTGAAAATATGTGTAATATGTAAAACATTTTGA-3′; GM170, 5′-GATCCTGTAGGAAACAGGGGCTTGAGTCACTCCAG-3′; TRE, 5′-GTGGAATTCGGGGAAGTGACTCAGCGCTCGGACG-3′. The lower case letters represent the mutations introduced into the CD28RR to generate the AP mutant or the RE mutant. Double-stranded oligonucleotides were end labeled using T4 polynucleotide kinase (40). Affinity-purified recombinant NF-ATp and truncated Fos and Jun proteins were used in binding reactions with the NF-IL-2A or GM170 radiolabeled oligonucleotides (0.1 ng) in a buffer containing 20 mM HEPES, 5 mM KCl, 20 mM NaCl, 2 mm MgCl₂, 2 mM DTT, and 5 μg purified BSA (New England Biolabs, Beverly, MA) (50). Protein:DNA complexes was resolved on a nondenaturing 5% polyacrylamide gel containing 0.5× TBE buffer (0.5 mM Tris, 42 mM boric acid, 1 mM EDTA, pH 8.3). Binding of recombinant proteins to the CD28RR was conducted in 10 mM Tris, pH 7.5, 10 mM MgCl₂, 5 mM EDTA, pH 7.5, 10 mM DTT, 0.2% Nonidet P-40, 1% glycerol, 0.4% sucrose, 0.5 mg/ml BSA, and 100 ng poly(dG:dC) in 20 μl reactions. A total of 0.2 ng of radiolabeled probe was generally used with the amounts of recombinant proteins indicated for individual experiments. Reactions were separated on 5% 0.5× TBE polyacrylamide gels. EMSAs for nuclear extract were conducted as previously described except that 5% glycerol was used instead of Ficoll and 0.5 mM PMSF was included. Competitor double-stranded oligonucleotides or Abs was added to the binding reactions at the concentrations indicated in individual experiments and incubated for 10 min before the addition of the radiolabeled probe. All gels were dried and exposed to x-ray film. Digital images were produced on a Fujifilm LAS1000 Plus CCD camera (Tokyo, Japan) and Image Gauge software (Fujifilm).

Western blot analysis

For Western blot, Jurkat cells were transfected, sorted, and stimulated as above, except 1 × 10⁶ cells were recovered from the GFP-positive population. A similar analysis using PBLs was not feasible due to the low transfection efficiency of these cells and the large numbers of cells required for the assay. Nuclei were isolated from cells and HMG I(Y) proteins were extracted with 5% perchloric acid and precipitated with TCA (44). Proteins were separated on 18% denaturing polyacrylamide gels, and HMGI(Y) was detected by Western blot using a 1:250 dilution of a rabbit polyclonal anti-HMGI(Y) Ab (51). The enhanced chemiluminescence detection system (Amersham, Arlington Heights, IL) was used, and the resultant bands were detected by x-ray film and scanned by densitometry (Bio-Rad).

HMGI(Y) binds to multiple elements within the IL-2 promoter

We have previously shown that transcription from a transfected IL-2 promoter reporter construct was inhibited by coexpression of antisense RNA for HMGI(Y) proteins (14). A comparison of the sensitivity of the promoters for a number of T cell-expressed genes (IL-2, IL-3, GM-CSF, and the HIV long terminal repeat (LTR)) to HMGI(Y) depletion found that IL-2 was the most significantly affected by changes in HMGI(Y) levels (data not shown). The IL-2 proximal promoter contains ∼65% A:T base pairs and has many potential binding sites for HMGI(Y) with A:T sequences present within or adjacent to several known transcription factor binding sites. To identify the binding sites for HMGI(Y) on the IL-2 promoter, DNase I footprinting was performed on a fragment of the IL-2 promoter from nt −360 to +50 (Fig. 1⇓). DNase I footprinting was conducted on both DNA strands by labeling each end of the DNA fragment independently. Fig. 1⇓A represents footprints obtained with the DNA labeled at −360, and in Fig. 1⇓B the DNA was labeled at +50. As shown in Fig. 1⇓, A and B, footprints for HMGI(Y) were observed at a number of locations across the proximal promoter (indicated by vertical bars and filled arrowheads). Fig. 1⇓C shows the location of these HMGI(Y) binding sites in relation to the sequence and the known transcription factor binding sites on the promoter. Because different binding sites within the promoter have different affinities for HMGI(Y) (data not shown), only those sites that are consistently seen in independent replicate experiments are indicated. The HMGI(Y) footprints are within or adjacent to the −45 NF-AT/TATA-2 site (11), the NF-IL-2A region, the NF-IL-2B region, the CD28 response region (CD28RR), and the −285 NF-AT site in ARRE-2. All of these elements have been shown to be important in IL-2 promoter activity (reviewed in Refs. 3 and 5). Interestingly, at the concentrations used in this assay, HMGI(Y) was also observed to specifically footprint at the −30 TATA-1 sequence (Fig. 1⇓), a region of the IL-2 promoter that is known to be protected in vivo by proteins in both unstimulated and stimulated T cells (21). As a control for these experiments, a Fos/Jun (AP-1) heterodimer was also footprinted to the promoter DNA, both alone and in combination with the HMGI(Y) protein (Fig. 1⇓, A and B). AP-1 was used at two different concentrations, which were empirically determined, in Fig. 1⇓, A and B. As expected, at the high concentration of input protein used in Fig. 1⇓A, the Fos/Jun complex altered the footprinting pattern across the entire promoter but specific footprints were seen across known AP-1 sites in the promoter region (indicated by vertical dashed lines), including the low-affinity NF-IL-2A (ARRE-1) site located between nt −80 and −90. These specific footprints were confirmed in Fig. 1⇓B at lower concentrations of AP-1. However, when both HMGI(Y) and Fos/Jun were combined together in the reaction mixture, binding of Fos/Jun to the NF-IL-2A site and the CD28RR was considerably reduced, and this is most clearly seen at the higher concentrations of protein used in Fig. 1⇓A (see below).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

A and B, DNase I footprinting of HMGI(Y) and AP-1 on IL-2 promoter DNA. The DNA fragment was labeled at either −360 (A) or +50 (B). The proteins added are indicated at the top of the gels with a Maxam-Gilbert “G” chemical cleavage of the naked DNA serving as a sequence reference marker. A+H represents the simultaneous addition of HMGI(Y) and AP-1. Lanes 1–5 and 6–10 are two different loadings of the same samples on the sequencing gel that have been electrophoresed for different lengths of time. The nucleotide numbers relative to the major transcriptional start site are indicated next to the G lanes. The recognition sequences for known protein transcription factors and vertical lines next to the nucleotide numbers indicate functional elements. A vertical thick solid line and a filled arrowhead indicate the footprinted areas protected by HMGI, and a vertical broken line indicates those protected by AP-1. C, Sequence of the IL-2 proximal promoter showing HMGI(Y) footprints relative to known transcription factor binding sites. The HMGI(Y) footprints that are consistently observed are marked by lower case letters in the sequence and filled dots underneath. The sequence is numbered relative to the transcription start site of +1. Horizontal lines above the sequence indicate transcription factor binding sites and control elements.

HMGI(Y) modulates AP-1 and NF-AT binding to the IL-2 promoter

HMGI(Y) has previously been shown to promote or inhibit transcription factor binding to their regulatory elements overlapping or adjacent to the HMGI(Y) binding sites (reviewed in Ref. 29). We have previously shown that HMGI(Y) can modulate the binding of c-Rel to the CD28RR on the IL-2 promoter. We wished to determine whether HMGI(Y) affected the binding if other transcription factors to the IL-2 promoter. Several of the regions on the IL-2 promoter to which HMGI(Y) binds have been described as NF-AT and AP-1 binding sites (reviewed in Refs. 3 and 4). We chose to examine the effect of HMGI(Y) on NF-AT and AP-1 binding to two of these sites because of their distinct properties. The NF-IL-2A (ARRE-1) region has previously been reported to cooperatively bind NF-AT and AP-1 to form a higher order NF-AT complex on the NF-IL-2A region (11). The second site we examined, the CD28RR, has also been reported to bind NF-ATp (11, 46) and AP-1, but the binding sites are in a distinct configuration and cooperative binding is not observed (46).

At low protein concentrations, recombinant truncated Fos/Jun alone does not stably bind to the low-affinity AP-1 site in the NF-IL-2A site in EMSA (Fig. 2⇓A, lane 1). However, the addition of recombinant HMGI(Y) resulted in the appearance of an AP-1 band (Fig. 2⇓A, lanes 2–5). The intensity of the AP-1 band initially increased (up to 2 ng HMGI(Y)) and then decreased with the addition of increasing amounts of HMGI(Y) (Fig. 2⇓A). These results demonstrating that high concentrations of HMGI(Y) inhibit Fos/Jun binding are consistent with the previously described footprinting results obtained with varying concentrations of these proteins (Fig. 1⇑A). Recombinant NF-ATp protein was added to determine the effect of HMGI(Y) proteins on formation of the NF-AT/AP-1 complex. Under the binding conditions used here, NF-ATp did not bind alone to the probe and the formation of the higher-order NF-AT complex did not occur (Fig. 2⇓A, lane 6). Addition of HMGI(Y) protein resulted in a dose-dependent increase in formation of the NF-AT complex as well as the previously observed increase in AP-1 binding (Fig. 2⇓A, lanes 7–10). At higher levels of HMGI(Y), the intensity of the NF-AT complex did not significantly decline as was seen for AP-1 alone (Fig. 2⇓A, lanes 7–10). The amount of AP-1 and NF-ATp used in these experiments was sufficient for strong binding to the consensus sites contained in an oligonucleotide from the GM-CSF enhancer (GM170) that can bind NF-ATp or AP-1 alone and also form a higher-order complex (Fig. 2⇓A, lanes 11–15). This oligonucleotide showed no binding of HMGI(Y), and neither the binding of AP-1 or NF-ATp alone nor the formation of the NF-AT higher-order complex were affected by addition of HMGI(Y) at any concentration (Fig. 2⇓A, lanes 11–15).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

A, HMGI(Y) promotes AP-1 and NF-ATp binding to the NF-IL-2A region of the IL-2 promoter. EMSAs were performed using radiolabeled oligonucleotides spanning the NF-IL-2A region of IL-2 (lanes 1–10) or the GM170 region of the GM-CSF enhancer (lanes 11–15) and recombinant truncated NF-ATp (0.4 ng), truncated c-Fos/c-Jun (tAP-1) (2 ng), and HMGI(Y) (1, 2, 4, and 8 ng) proteins. The results were visualized using a Molecular Dynamics PhosphorImager. The proteins and probes used in the individual lanes are indicated above the lanes, and the positions of the free DNA and the individual protein:DNA complexes are indicated. B, HMGI(Y) promotes AP-1 but inhibits NF-ATp binding to the IL-2 CD28RR. EMSAs were performed as above using full-length recombinant c-Fos and c-Jun (AP-1) at 12 ng (lanes 1, 4, and 7), 6 ng (lanes 2, 5, and 8), or 3 ng (lanes 3, 6, and 9) and the CD28RR radiolabeled oligonucleotide. In lanes 10–13, 0.1 ng NF-ATp was added with 3 ng AP-1. HMGI protein was added to the binding reactions at 2 ng (lanes 4–6), 20 ng (lanes 7–9), or 5 (lane 11), 10 (lane 12), or 20 ng (lane 13). EMSAs were processed, and the diagram was labeled as in A.

Recombinant full-length c-Fos/c-Jun (AP-1) bound to the CD28RR in a dose-dependent manner, without the addition of HMGI(Y) (Fig. 2⇑B, lanes 1–3). An increase in AP-1 binding was observed with the addition of HMGI(Y) (Fig. 2⇑B, compare lanes 2, 5, and 8 or lanes 3, 6, and 9), but relatively high levels of HMGI(Y) were required to observe this effect (20 ng compared with 2 ng for the promotion of AP-1 binding to the ARRE-1 region). We have previously shown that HMGI(Y) promotes the binding of c-Rel to an adjacent site (CD28RE) in the CD28RR (14). NF-AT proteins can also bind to the CD28RR, and we have previously shown that HMGI(Y) can promote truncated recombinant NF-ATp binding to the CD28RR (46). Therefore, we examined whether HMGI(Y) could promote the formation of an NF-ATp/AP-1 complex on this site. At the concentrations used to promote AP-1 binding here (20 ng), the binding of NF-ATp was in fact inhibited (Fig. 2⇑B, lanes 10–13). Similar results were observed whether full-length AP-1 (Fig. 2⇑C) or truncated AP-1 (data not shown) was used in the binding reactions.

These results show that HMGI(Y) can modulate the binding of transcription factors that are required for IL-2 promoter function to at least two major control regions of the IL-2 proximal promoter. However, it appears that the ratio of HMGI(Y) to AP-1 or NF-ATp may be important in determining the functional outcome of the interaction.

HMGI(Y) modulates nuclear AP-1 binding to the CD28RR

The CD28RR consists of an AP-1 binding site as well as a c-Rel binding site (CD28RE). We have previously shown that the c-Rel-containing complexes from activated T cell nuclear extracts require HMGI(Y) to bind to the CD28RE (14). To determine whether nuclear AP-1 was also dependent on HMGI(Y), we examined binding of AP-1 from nuclear extracts of P/I/CD28-activated Jurkat T cells. Because the Rel and AP-1 complexes migrate at the same position on EMSA gels using the intact CD28RR (17), it was necessary to generate EMSA probes with mutations in either the CD28RE (RE mutant) or the AP-1 site (AP mutant). Binding of nuclear extracts to these probes showed that an inducible complex could bind to the RE mutant in extracts made both 1 and 6 h following stimulation with P/I/CD28 (Fig. 3⇓A, lanes 4–6). These complexes migrated at the same position as the complexes binding to the wild-type probe or the Rel-containing complexes binding to the AP mutant (Fig. 3⇓A, lanes 1–3 and lanes 7–9). The identity of the complexes binding to the RE mutant was confirmed by competition experiments where the RE mutant, the wild-type CD28RR, and a consensus AP-1 site (TRE) were able to compete for complex formation but the AP mutant was not (Fig. 3⇓B). Thus, these complexes contain AP-1-like proteins. To determine whether the formation of the AP-1-like complex was dependent on HMGI(Y), an anti-HMGI(Y) Ab was added to the binding reactions. The addition of the HMGI(Y) Ab to the binding reactions reduced AP-1 binding to the RE mutant (Fig. 3⇓C, lanes 1–3), as it did the c-Rel complexes binding to the AP mutant (Fig. 3⇓C, lanes 5–7). A control Ab (C) did not affect complex formation (Fig. 3⇓C, lanes 4, 8, and 11). We have previously shown that HMGI(Y) Ab removes the inducible complexes containing c-Rel from the CD28RR but does not affect NF-κB complexes binding to a distinct NF-κB site (14). AP-1 binding to a TRE was not significantly reduced by the addition of the HMGI(Y) Ab (Fig. 3⇓C, lanes 9 and 10).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3.

Binding of AP-1 to the CD28RR from Jurkat T cell extracts requires HMGI(Y). A, EMSAs were performed with nuclear extracts prepared from Jurkat T cells either unstimulated (lanes 1, 4, and 7) or activated with P/I/CD28 for 1 h (lanes 2, 5, and 8) or 6 h (lanes 3, 6, and 9) using radiolabeled CD28RR (lanes 1–3), RE mutant (lanes 4–6), or AP mutant lanes 7–9) oligonucleotides. The inducible protein DNA complexes are indicated by an arrow. B, Competition experiments using the RE mutant as the radiolabeled probe and extracts from cells activated for 6 h. The CD28RR wild-type sequence (lanes 2–4), AP mutant (lanes 5–7), RE mutant (lanes 8–10), or a TRE (AP-1 consensus binding site) (lanes 11–13) were used as competitors at 1, 10, and 20 ng per reaction as indicated. C, The effect of addition of HMGI(Y) Ab on complexes binding to the RE mutant (lanes 1–4), the AP mutant (lanes 5–8), and the TRE (lanes 9–11). The HMGI(Y) Ab (α-H; 1 μl (lanes 2 and 6) or 2 μl (lanes 3, 7, and 10)) or a nonspecific serum (C; 2 μl) was added to the binding reactions before the addition of probe. The AP-1 complexes are indicated. In A–C, only those parts of the gel containing the inducible bands are shown. D, HMGI promotes recombinant AP-1 binding to the RE mutant. EMSAs containing 3 ng recombinant AP-1 and increasing concentrations of HMGI (0.5, 1, 2.5, 5, 10, and 20 ng) were performed with either the wild-type CD28RR probe (lanes 1–8) or the RE mutant (lanes 9–16). Lanes 1 and 9 had no protein added, and lanes 2 and 10 had only AP-1. The positions of the HMGI and AP-1 bands are indicated.

To confirm that HMGI(Y) could promote AP-1 binding to the RE mutant, recombinant HMGI was titrated into binding reactions with a fixed amount of AP-1 using either the wild-type CD28RR or the RE mutant. As described above, AP-1 binding to the CD28RR was increased only at high concentrations of HMGI (Fig. 3⇑D, lanes 7 and 8). In contrast, a dose-dependent increase in AP-1 binding was observed with increasing levels of HMGI starting at 1 ng on the RE mutant (Fig. 3⇑D, lanes 9–16). We also observed that HMGI(Y) no longer bound to the RE mutant (Fig. 3⇑D), probably because the RE mutant affects the A:T stretch within the CD28RE to which HMGI(Y) most likely binds. These results 1) show that AP-1 binding in nuclear extracts is influenced by HMGI(Y) and 2) imply that the ability of HMGI(Y) to promote AP-1 binding is not dependent on DNA binding by HMGI(Y).

HMGI(Y) binding to DNA is not required for promotion of transcription factor binding to the CD28RR

The results above imply that the promotion of transcription factor binding may not always require DNA:HMGI(Y) interactions. To test this further, site-specific proline to alanine mutations were introduced into the second and third DNA binding domains of HMGI (two substitutions in each binding domain) to create a form of the protein (designated HMGI(mII,mIII)) that could not specifically bind to A:T-DNA under the assay conditions employed. We tested the ability of HMGI(mII,mIII) to bind to the IL-2 promoter both in footprinting and in gel shift assays. At concentrations of protein where recombinant HMGI or HMGY bound specifically to the CD28RR, HMGI(mII,mIII) protein did not bind (Fig. 4⇓A). The same result was obtained in footprinting assays across the IL-2 promoter from −180 to −60 (Fig. 4⇓B). Four footprints for HMGI were observed across this region of the promoter (Fig. 4⇓B, lanes 2–5). In contrast, HMGI(mII,mIII) did not form specific footprints across any of these regions (Fig. 4⇓B, lanes 6–9).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 4.

A and B, HMGI(mII,mIII) does not bind specifically to the IL-2 promoter. A, EMSAs were performed using the IL-2CD28RR oligonucleotide as a probe and 1, 10, and 20 ng of HMGI (lanes 1–3), HMGY (lanes 4–6), and HMGI(mII,mIII) (lanes 7–9). The HMGI/Y bands and the free DNA are indicated. B, DNase I footprinting assay using the −180 to −60 region of the IL-2 promoter. HMGI (lanes 2–5) or HMGI(mII,mIII) (lanes 6–9) at 10, 20, 50, or 100 ng were used in the footprinting reactions. Solid lines indicate the HMGI footprints, and the names of the regions are shown. C and D, HMGI(mII,mIII) can promote but not inhibit transcription factor binding to the IL-2 CD28RR. Increasing amounts (0.5, 1, 2.5, 5, 10, and 20 ng) of HMGI (lanes 4–9) or HMGI(mII,mIII) (lanes 10–15) were added to binding reactions containing the IL-2CD28RR probe and 3 ng AP-1 and 0.1 ng NF-ATp. Reactions containing NF-ATp alone (lane 1), AP-1 alone (lane 2), or AP-1 and NF-ATp together (lane 3) were also analyzed. The positions of the transcription factor:DNA complexes are indicated. D, Similar binding reactions as described in C were conducted using bacterial extract containing the Rel homology domain of c-Rel and increasing amounts (1, 5, 10, and 20 ng) of HMGI (lanes 2–5) or HMGI(mII,mIII) (lanes 6–9). Lane 1 contains c-Rel alone.

We also tested whether HMGI(mII,mIII) could interact with transcription factors by generating an affinity matrix for both HMGI and HMGI(mII,mIII). The results of binding experiments using these affinity matrices in binding experiments demonstrated that both the wild-type HMGI and mutant HMGI(mII,mII) proteins could bind specifically to the AP-1 and NF-AT proteins but could not interact with themselves or with each other (data not shown).

We then investigated whether HMGI(mII,mIII) could alter transcription factor binding when added to binding assays using the CD28RR together with AP-1 and NF-ATp. The addition of HMGI(mII,mIII) lead to a dose-dependent increase in AP-1 binding but did not inhibit NF-ATp binding as did wild-type HMGI (Fig. 4⇑C, lanes 10–15). We observed that HMGI(mII,mIII) generated a more consistent dose dependence compared with the wild-type HMGI (Fig. 4⇑C, compare lanes 4–9 and 10–15). c-Rel binding to the CD28RR was also increased by HMGI(mII,mIII) (Fig. 4⇑D, lanes 1 and 6–9). In addition, inhibition of transcription factor binding sometimes seen at high concentrations of HMGI was never observed with HMGI(mII,mIII). In summary, these results show that HMGI binding to DNA is involved in the inhibition of binding but is not required for the promotion of transcription factor binding.

Alteration of HMGI(Y) expression results in the modulation of IL-2 promoter activity in both Jurkat and primary T cells

To determine whether HMGI(Y) had a functional role in IL-2 gene transcription, HMGI(Y) levels were altered in the cell by antisense or overexpression studies. The pRcCMVIGMH plasmid expressing antisense HMGI was cotransfected into Jurkat T cells with pIL2luc or pHIVluc. The HIV LTR responds to P/I and CD28 stimulation through NF-κB sites that closely match consensus sites and contain no potential binding sites for HMGI(Y) (our unpublished observations).

As previously reported (14), the expression of antisense HMGI RNA resulted in a 72% decrease in reporter activity in P/I-treated Jurkat cells and a 75% inhibition in P/I+α-CD28-treated cells at the maximum dose of antisense expression plasmid (10 μg; Fig. 5⇓A). The pHIVluc reporter transfections showed only a small reduction in promoter activity at the maximum dose of antisense plasmid (Fig. 5⇓A). To confirm that the antisense HMGI RNA was reducing the level of HMGI(Y) protein in the cells, a Western blot, using an anti-HMGI polyclonal Ab, was performed on nuclear extracts from transfected cells either unstimulated or P/I-stimulated following transfection. Densitometry scanning of the blot showed that there was a significant reduction in the levels of HMGI(Y) protein in the antisense-expressing cells (Fig. 5⇓B). This experiment also showed that the amount of HMGI(Y) in the nuclei of Jurkat T cells was increased almost 3-fold by P/I activation (Fig. 5⇓B).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 5.

Inhibition of HMGI(Y) production reduces IL-2 promoter activity in Jurkat T cells and PBLs. A, Jurkat T cells were transfected with increasing amounts of pRcCMVIGMH and the pIL-2luc or pHIVluc reporter plasmids. Transfected cells were treated with P/I (□) or P/I+α-CD28 (▪) before luciferase assays were performed. Luciferase activity is expressed as counts per second (CPS) readings from the Packard TopCount luminometer. B, Western blot showing the decrease in HMGI(Y) levels in cells expressing antisense HMGI. Cells were transfected with either pRcCMV (lanes 1 and 3) or pRcCMVIGMH antisense expression plasmid (lanes 2 and 4). Protein was extracted from nuclei of either unstimulated (lanes 1 and 2) or P/I-stimulated (lanes 3 and 4) cells, resolved on SDS polyacrylamide gels, and HMGI(Y) was detected by Western blot. Recombinant HMGI protein (lane 5) was also loaded as a reference. Numbers below represent the densitometry quantitation of HMGI(Y) protein bands. C, PBLs, pretreated to generate blasts, were transfected with the pRcCMV or pRcCMVIGMH plasmids, and the pIL-2luc reporter plasmid and luciferase activity was measured either before (□) or following PHA treatment (▪). Columns represent the means of five replicate assays, and the bars show the SEMs.

HMGI(Y) expression in transformed cell lines is generally high (29), hence the Jurkat T cell leukemia may display an aberrant role for HMGI(Y) proteins in IL-2 promoter activity. In contrast, HMGI(Y) levels are quite low or not detectable in most normal cell types. Therefore, the dependence of the IL-2 promoter on HMGI(Y) proteins for activation of transcription was also examined in normal PBL. PBLs isolated from blood were cultured for 4 days as described to generate blasts, transfected by electroporation, rested for 24 h, and restimulated with PHA. The IL-2 promoter transfected into PBLs had activity in cells that were not stimulated posttransfection most likely because of the primary stimulation with PHA before transfection to generate T cell blasts. However, these cells showed a 2- to 3-fold increase in promoter activity when stimulated with PHA posttransfection (Fig. 5⇑C). Transfection with pRcCMVIGMH resulted in an 84% inhibition of the reporter activity in unstimulated PBLs and an 81% inhibition in PHA-stimulated cells (Fig. 5⇑C). These results show that the activity of a transfected IL-2 promoter is dependent on HMGI(Y) in either a T cell line or in primary T cells.

We also tested whether increased levels of HMGI(Y) protein affected IL-2 promoter activity. Transfection of the pcDNAHMGY expression plasmid into Jurkat T cells together with the pIL-2luc or the pHIVluc increased luciferase activity ∼2-fold for IL-2 but had no effect on HIV LTR activity (Fig. 6⇓A). Because the mutant HMGI(mII,mIII) protein also promoted transcription factor binding, its effect in transfection assays was also tested. Cotransfection of an expression plasmid for the HMGI(mII,mIII) protein resulted in significant inhibition of the IL-2 promoter (60% in P/I and 73% in P/I+α-CD28-treated cells; Fig. 6⇓B) at the highest amounts of transfected plasmid. The pHIVluc reporter plasmid showed no loss of reporter activity with expression of the HMGI(mII,mIII) protein (Fig. 6⇓B). Thus, this non-DNA binding mutant appears to act as a dominant negative protein for IL-2 gene transcription.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 6.

A, Expression of HMGY in Jurkat T cells increases IL-2 promoter activity. Jurkat cells were transfected with either the pIL-2luc or the pHIVluc reporter plasmids together with the indicated amounts of the pcDNAHMGY expression plasmid or the control pcDNA3.1 plasmid. Cells were stimulated with P/I (□) or P/I+α-CD28 (▪) for 8 h before harvesting for luciferase assays. pHIVluc transfected cells were only stimulated with P/I+α-CD28. Luciferase activity is expressed as counts per second (CPS) as measured by a Packard TopCount scintillation counter. B, Expression of HMGI(mII,mIII) in Jurkat cells inhibits IL-2 promoter activity. The pIL-2luc or pHIVluc reporter plasmids were transfected into Jurkat T cells together with the indicated amounts of the pcDNAHMGI(mII.mIII) expression plasmid or the control pcDNA plasmid. Cells were activated and analyzed, and results are presented as described in A.

Taken together these results show that modulating HMGI(Y) levels or function in either Jurkat T cells or PBLs affects IL-2 promoter activity without a general effect on transcription as measured by HIV LTR function.

IL-2 production and primary T cell proliferation are dependent on HMGI(Y)

It is possible that the transfected IL-2 promoter may be more highly dependent on HMGI(Y) than its endogenous counterpart because of the likely differences in chromatin configuration on chromosomal and plasmid DNA. To monitor the effect of HMGI(Y) on expression from the endogenous IL-2 gene, the amount of IL-2 protein secreted into the supernatant of cells transfected with pRcCMVIGMH and pcDNAHMGI(mII,mIII) was assayed by ELISA. To enrich for transfected cells, 5 μg of pCMVGFP was cotransfected with the HMGI plasmids above, and the cells were subsequently sorted for high level GFP expression by FACS. The sorted cells were then stimulated with either PHA for PBLs or P/I+α-CD28 for Jurkat cells to induce IL-2 expression. In PBLs and Jurkats, significant inhibition of IL-2 production was detected in cells transfected with RcCMVIMGH (78 and 88%, respectively) or pcDNAHMGI(mII,mIII) (47 and 55%, respectively) (Table I⇓). These results imply that the role of HMGI(Y) may be extended to endogenous IL-2 gene transcription in both Jurkat T cells and primary T cells.

If IL-2 production in PBLs is dependent on HMGI(Y), then PBL proliferation may also be affected by changes in HMGI(Y) levels in the cells. To test the effect of inhibition of HMGI(Y) on cell growth, the IL-2-dependent PBLs and the IL-2-independent Jurkat cell line were transfected with either the parent plasmids or the expression plasmids RcCMVIGMH and pcDNAHMGI(mII,mIII) together with the pCMVGFP plasmid to allow sorting of transfected cells by FACS. Proliferation of PBLs was inhibited by RcCMVIGMH (62%) or pcDNAHMGI(mII,mIII) (29%), whereas the effect of these plasmids on growth of the Jurkat cell line was not significant (Fig. 7⇓, A and B). To test whether increased HMGI(Y) levels also altered proliferation, PBLs were transfected with the pcDNAHMGY plasmid and pCMVGFP to allow for sorting of transfected cells. Overexpression of HMGY in PBLs increased proliferation of these cells by ∼2-fold (Fig. 7⇓C).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 7.

Altering HMGI(Y) levels or function affects T cell proliferation. PBLs and Jurkat cells were transfected with either 10 μg of the antisense expression plasmid pRcCMVIGMH or the parent plasmid pRcCMV (A), and pcDNAHMGI(mII,mIII) or the parent plasmid, pcDNA3.1 (B) PBLs were also transfected with the pcDNAHMGY plasmid or the parent plasmid (C). All cells were cotransfected with 5 μg of pCMVGFP and subsequently sorted by FACS for GFP expression. Cells were then stimulated with PHA in the presence of bromodeoxyuridine and assayed for proliferation using enzyme-conjugated Ab to bromodeoxyuridine and a chemiluminescent substrate. Values are given as counts per second measured by a 96-well plate luminometer, and the columns represent the mean, and error bars the SEM, of four replicate assays.

These results show that IL-2 production in both PBLs and Jurkats is dependent on the correct level of HMGI(Y). This dependence can also be seen at the level of proliferation in PBLs that are IL-2 dependent but not in Jurkats that are IL-2 independent.