HMG-I/Y Is a c-Jun/Activator Protein-1 Target Gene and Is Necessary for c-Jun–Induced Anchorage-Independent Growth in Rat1a Cells

c-Jun is a major component of the activator protein-1 (AP-1) transcription factor. It binds to numerous sites, including AP-1 (TGAG[C]TCA) and activating transcription factor (TGACGTCA) sites found in the promoters of a large number of genes. Interaction at these sites occurs as either a homodimer or a heterodimer with members of the Fos family (c-Fos, FosB, Fra1, and Fra2) or activating transcription factor family, thus regulating the expression of genes containing AP-1 binding sites in their promoters. c-Jun plays a critical role in several cellular processes including transformation (1, 2), cell cycle progression (3), differentiation (4), and apoptosis (5). We have previously shown that c-Jun transforms the immortalized rat fibroblast cell line Rat1a as a single gene (6), and overexpression of c-Jun allows the cells to grow in an anchorage-independent manner. Although we and others have identified several c-Jun target genes (7-12), the precise molecular mechanisms by which c-Jun/AP-1 mediates biological processes are yet to be determined.

The HMG-I/Y gene encodes the HMG-I and HMG-Y protein isoforms, which result from alternatively spliced mRNA (13, 14). HMG-I and HMG-Y proteins differ by an internal 11 amino acids present only in the HMG-I isoform (13, 14). Both HMG-I and HMG-Y proteins contain AT hook DNA binding domains that mediate binding to AT-rich sequences in the minor groove of chromosomal DNA (15-19). These proteins appear to play an important role in regulating gene expression (15-25) and have been called architectural transcription factors because they alter the conformation of DNA by modulating nuclear protein-DNA complexes. HMG-I/Y proteins also physically interact with transcription factors such as AP-1, nuclear factor-κB (20), Sp1, and CAAT/enhancer binding protein-β (21) to modulate their ability to activate gene expression.

HMG-I/Y is localized to the short arm of chromosome 6 in a region commonly involved in chromosomal abnormalities associated with human cancers (13, 14, 26). HMG-I/Y expression is increased in neoplastic transformation and proliferation (27-33). Increased expression of HMG-I/Y mRNA and/or protein has also been associated with numerous human malignancies (34-37) as well as mouse models of skin (38) and prostate carcinoma (39) and is also associated with metastasis (40, 41). In addition, HMG-I/Y may play a direct role in tumorigenesis by interfering with the ability of p53 to recognize damaged DNA (42). It also cooperates with AP-1 and CBP/p300 in transcription of HPV18 sequences (43).

A previous study has shown that HMG-I/Y is a c-Myc target gene important in transformation in Burkitt's lymphoma (33). HMG-I/Y also has several oncogenic properties (33, 36, 44, 45). Increased expression of HMG-I/Y in Rat1a cells, human lymphoid cells (33), or human breast cells (36, 44) results in a transformed phenotype. In addition, rat cells overexpressing HMG-I/Y form tumors in nude mice (33, 45). HMGI-I/Y was also shown to be induced by phorbol esters, although this induction was relatively modest (46). We used our previously reported Rat1a cell system in which c-Jun expression is regulated by doxycycline (9) to examine whether HMG-I/Y is a target gene involved in c-Jun–induced transformation and anchorage-independent growth. We report that c-Jun binds to the AP-1 site at position −1,091 in the HMG-I/Y promoter and enhances HMG-I/Y expression under nonadherent growth conditions. Inhibition of HMG-I/Y expression by an antisense construct inhibits transformation by c-Jun in soft agarose, and overexpression of HMG-I/Y alone allows cells to grow in an anchorage-independent manner. Taken together, these results indicate that HMG-I/Y is a c-Jun target gene that is necessary, and partially sufficient, for c-Jun–induced transformation and anchorage-independent growth of Rat1a cells.

We have previously shown that overexpression of c-Jun in Rat1a cells results in the activation of target genes, many of which may have roles in proliferation and transformation (9, 10). HMG-I/Y is a likely candidate for regulation by c-Jun/AP-1 because several potential AP-1 binding sites have been identified in its promoter (33, 46, 47). To determine the effect of c-Jun on HMG-I/Y expression, we used Rat1a cells with doxycycline-inducible c-Jun expression (9). Northern blot analysis showed that HMG-I/Y transcription was considerably up-regulated following c-Jun induction in cells grown in nonadherent conditions for 48 hours, while a GFP-expressing control cell line showed no change (Fig. 1A). This regulation was observed in two independent c-Jun–expressing clones, Rat1a-J2 (data not shown) and Rat1a-J4. These results were confirmed by Western blot analysis showing a concomitant increase in HMG-I/Y protein following c-Jun induction (Fig. 1B). There was no change in the relative ratios of HMG-I and HMG-Y isoforms under c-jun induction (data not shown). The increase in basal HMG-I/Y expression in Rat1a-J4 cells (−dox) compared with GFP control cells is likely due to leakiness of c-Jun expression in the absence of doxycycline.

In addition to the E-box at −1,337 and AP-2 site at −1,321 in the mouse HMG-I/Y promoter (33, 47), three potential AP-1 binding sites are located at −1,457, −1,423, and −1,091 (Fig. 2A; refs. 46, 47). To determine if HMG-I/Y is a target for regulation by c-Jun/AP-1, we analyzed these potential AP-1 binding sites by in vitro gel mobility shift assays. All three oligomers showed DNA-protein interactions with nuclear extracts isolated from Rat1a-J4 cells grown in the absence and presence of doxycycline (Fig. 2B). The oligomers corresponding to the “AP-1 sites” at positions −1,457 and −1,091 had increased protein complex binding in the presence of doxycycline, suggesting that c-Jun/AP-1 may interact at these locations. The −1,091 oligomer, in particular, showed increased binding and migration of two complexes in the doxycycline-treated cell extracts. To determine if c-Jun is contained in the DNA-protein complexes at −1,457 and −1,091, we performed supershift analysis with an antibody against c-Jun. The complex binding at −1,457 was not supershifted, suggesting that c-Jun is not contained in the complex (Fig. 2C). Similarly, no supershifted complexes were detected at −1,423 (data not shown). Supershifted complexes were detected, however, at −1,091, and both complexes detected in doxycycline-induced extracts were supershifted (Fig. 2D). The complex binding at −1,091 was also supershifted by an antibody to JunD, suggesting that JunD, in addition to c-Jun, was a component of the complex (Fig. 2E). In the presence of doxycycline, the amount of the complex supershifted by the c-Jun antibody increased and that by the JunD antibody decreased; in addition, a supershifted band was detected with an antibody against Fra1 (Fig. 2E). An isotype control antibody did not result in any reproducibly supershifted complexes. These results suggest that c-Jun is a component of the complex interacting at −1,091 in the HMG-I/Y promoter, and in the presence of induced c-Jun, this complex is altered.

To further determine if AP-1 could activate HMG-I/Y transcription in cells, we performed transfection experiments and promoter analysis in Rat1a, Rat1a-GFP, and Rat1a-J4 cells. Transient transfection of Rat1a cells with a −1,895 to +75 HMG-I/Y promoter luciferase construct in combination with the doxycycline-inducible pLRT-c-Jun plasmid resulted in a significant increase in promoter activity (Fig. 3A). In contrast to HMG-I/Y promoter activation by c-Jun, the AP-1 dominant negative mutant Tam67 had no effect (Fig. 3A). Transfection of the −1,895 to +75 HMG-I/Y promoter construct into Rat1a-J4 cells resulted in a significant increase in promoter activity in the presence of doxycycline (Fig. 3B). This increase in promoter activity was prevented by mutation of the AP-1 site at −1,091. In addition, deletion of the region −1,895 to −974 containing the E-box, AP-1 and AP-2 sites, also prohibited promoter activation by c-Jun/AP-1 (Fig. 3B). The slight increase in induced promoter activity seen in the truncated promoter construct may reflect the loss of an inhibitory element for which we have preliminary evidence.³

To determine if HMG-I/Y is necessary for c-Jun–regulated anchorage-independent growth, we isolated pools of independent isolated clones of Rat1a-J4 cells with constitutive antisense HMG-I/Y expression. The antisense expression suppressed the c-Jun–induced nonadherent cell growth of J4 cells (Fig. 4A). Each pool of antisense clones was suppressed ∼80% compared with the cloning of the parental cell line. The colonies were smaller, and the number of colonies was significantly reduced compared with those of c-Jun–expressing Rat1a-J4 cells. No suppression of cloning was observed for a pool of Rat1a-J4 cells transfected with the pSG5 vector alone (Fig. 4A). In addition, the pools expressing antisense HMG-I/Y demonstrated ∼50% suppression of HMG-I/Y when determined by Western blot (Fig. 4B). Western blot analysis of HMG-C demonstrated no change in expression (data not shown). Constitutive expression of antisense HMG-I/Y did not affect the growth rate of the cells. The pools of Rat1a-J4 cells overexpressing antisense HMG-I/Y had a comparable doubling time to parental Rat1a-J4 cells both in the presence and in the absence of doxycycline (data not shown). Further, we isolated two independent clones of Rat1a-J4-asHMG-I/Y and tested them for their ability to clone in soft agarose. J4HMGas-9 and J4HMGas-23 were suppressed in their ability to grow in soft agarose (>90%; Fig. 4C), whereas two independent Rat1a-J4 clones transfected with the pSG5 vector alone were not. Western blot analysis demonstrated suppression of HMG-I/Y expression (Fig. 4D). The discrepancy in colony formation between the antisense HMG-I/Y-expressing pools and clones (compare Fig. 4A and D) was due to leakiness of HMG-I/Y expression in the pool cells. HMG-I/Y expression in the antisense clones was reduced by 98% (Fig. 4D). These results indicate that HMG-I/Y expression in Rat1a-J4 cells is necessary for anchorage-independent growth by c-Jun.

Because our antisense experiments indicated that HMG-I/Y is required for transformation by c-Jun in this system, we next sought to determine if HMG-I/Y was sufficient for transformation in our Rat1a cells. The overexpression of HMG-I/Y in Rat1a cells induced colony formation (Fig. 5B), although the colonies were slightly smaller, less compact, and fewer than those induced by c-Jun overexpression. This is similar to that reported previously (33, 45).

In this article, we show that c-Jun binds to the HMG-I/Y promoter and induces expression of the HMG-I/Y chromatin binding protein. Decreasing HMG-I/Y proteins using an antisense approach inhibits c-Jun/AP-1–induced transformation, while overexpression of HMG-I/Y confers anchorage-independent cell growth. The colonies in Rat1a cells overexpressing HMG-I/Y were smaller and fewer, suggesting that HMG-I/Y expression only partially recapitulates the c-Jun/AP-1 phenotype. HMG-I/Y is a chromatin remodeling protein with DNA binding domains called AT hooks, which bind to the minor groove of stretches of AT-rich DNA (13-21, 26). Several studies suggest an important role for HMG-I/Y in regulating gene transcription (13-26, 42-44). The HMG-I/Y gene is highly conserved among species; the transcribed regions of the human and mouse HMG-I/Y genes are ∼80% identical at the nucleotide sequence level and >90% identical when only the protein-coding exons are considered (48). Both mouse and human HMG-I/Y promoters have a consensus AP-1 site at position −1,091 (47). The conservation of this site reflects the importance of the transcription factor binding site in the regulation of this gene. Ogram and Reeves (46) demonstrated that the AP-1 complex regulates HMG-I/Y expression during differentiation of human K562 erythroleukemic cells after treatment with 12-O-tetradecanoylphorbol-13-acetate. However, promoter analysis in this case revealed that a variant AP-1 site with one-base mismatch (TGACACA) within the human HMG-I/Y promoter was responsible for 12-O-tetradecanoylphorbol-13-acetate induction. This site (which is conserved within the mouse promoter) was downstream to the promoter sequence used for our studies. The variant AP-1 site was not included in our studies because it is in a region of the HMG-I/Y promoter that is not responsive to serum or growth factor stimulation in murine fibroblasts. Thus, it may not contribute to the induction of HMG-I/Y in response to growth-regulatory signals. Our gel mobility shift and reporter assays demonstrated that the AP-1 site at position −1,091 is a novel and critical element for c-Jun/AP-1 up-regulation of HMG-I/Y. Likely, the usage of different AP-1 sites depends in part on the transcriptional milieu, which in turn reflects the biological system being tested.

There is substantial evidence supporting the role of HMG-I/Y in cellular transformation and human malignancy. HMG-I/Y is expressed at low levels in normal or differentiated cells while higher in response to a wider range of growth factors (30, 49-52). HMG-I/Y is a direct target of the c-myc oncogene. Moreover, HMG-I/Y is necessary for c-myc–mediated transformation in Burkitt's lymphoma cells, a c-myc–driven human malignancy. Further, expression of HMG-I/Y in nontumorigenic breast cells converts them to a tumorigenic state (44). Previous work has demonstrated that several malignant tumors have aberrant expression of HMG-I/Y proteins (26-41, 48, 53, 54). In addition, use of an adenovirus carrying the mouse HMG-I/Y gene in an antisense orientation suppressed HMG-I/Y protein synthesis and induced programmed cell death of a human thyroid anaplastic carcinoma cell line ARO (54). Interestingly, in the ARO cell line, c-Jun, JunD, Fra1, and AP-1 binding activities are all increased (55).

Critical to our understanding of how c-Jun/AP-1 mediates such diverse biological actions is to precisely identify its mechanisms of transcriptional activity and its relevant “downstream” targets. c-Jun/AP-1 regulates the expression of a myriad of genes (9, 10, 56). Many of these genes are direct transcriptional targets for c-Jun/AP-1 with AP-1 consensus sequences within their promoters. Determining the relevance of these genes for a given biological activity has proven more difficult. We have previously characterized the c-Jun–responsive genes in Rat1a cells using microarrays (9, 10) and demonstrated that several of these genes are necessary for nonadherent cell growth. HMG-I/Y is clearly another one and perhaps the most intriguing. Not only is HMG-I/Y expression necessary for c-Jun/AP-1–induced nonadherent cell growth, forced expression of the gene produces transformation of Rat1a cells. Thus, HMG-I/Y appears to partially recapitulate the biological system. The precise role of c-Jun/HMG-I/Y in the nonadherent growth of Rat1a cells remains to be defined. Because there is no effect on cellular proliferation of antisense HMG-I/Y, c-Jun/HMG-I/Y acts by a different mechanism(s). Rat1a cells undergo anoikis when in the nonadherent state, and c-Jun/HMG-I/Y may be inhibiting this process. Further, the role of HMG-I/Y in chromatin remodeling (15-19, 22-26) suggests a new and novel way in which c-Jun/AP-1 can regulate gene expression. c-Jun/AP-1–induced expression of HMG-I/Y may result in alterations in chromatin structure and in turn availability of c-Jun/AP-1–regulated targets. Of interest, genes that are induced by HMG-I/Y expression include many genes involved in the extracellular matrix including the matrix metalloproteinases and collagen genes (44, 57, 58), which are also AP-1–regulated genes (59-62). This suggests that HMG-I/Y may mediate some of the downstream AP-1 effects. In fact, direct interactions between HMG-I/Y and AP-1 components have been described (17, 20, 63-65). A systematic comparison of genes up-regulated by HMG-I/Y and AP-1 might identify genes with transcription dependent on c-Jun induction of HMG-I/Y.

In conclusion, the results in this article establish that HMG-I/Y is a direct target of c-Jun/AP-1. In Rat1a cells, up-regulation of HMG-I/Y is necessary and partially sufficient for c-Jun–induced nonadherent cell growth.

Rat1a cells expressing either c-Jun (Rat1a-J2 and Rat1a-J4) or green fluorescent protein (Rat1a-GFP) in a doxycycline-controlled manner have been described previously (9). All cells were maintained in DMEM (Invitrogen Life Technologies, Inc., Carlsbad, CA) supplemented with 10% fetal bovine serum (Gemini Bio-Products, Woodland, CA) and 1% l-glutamine (Invitrogen Life Technologies). For nonadherent growth, cells were plated in PolyHeme-coated dishes in the presence or absence of doxycycline (2 μg/mL) to induce c-Jun expression.

A 2-kb HindIII/BamHI fragment from the 4.6-kb HMG-I/Y-GH plasmid (33) was subcloned into pBluescript II KS. The resulting plasmid was digested with KpnI and BamHI, and the released fragment was cloned into pGL2-basic (Promega, Madison, WI). This plasmid was designated pGL2 HMG-I/Y−1,895/+75. A 5′ deletion fragment of the HMG-I/Y promoter from −974 to +75 was cloned into pBluescript II KS using HindIII and BamHI. A 1.1-kb KpnI/BamHI fragment from HMG-I/Y−974/+75 pBluescript II KS was subcloned into pGL2-basic and designated pGL2 HMG-I/Y−974/+75. The plasmid pGL2 HMG-I/Y−1,895/+75 AP-1 mutant was created by mutating the AP-1 consensus site in the HMG promoter in plasmid pGL2 HMG-I/Y−1,895/+75. The site was mutated using the GeneEditor in vitro site-directed mutagenesis system (Promega) according to the manufacturer's recommendations with the following oligonucleotide with the mutated bases italicized: GGGGAACAGAGTTATCTCGAGCAGTCGTGTGTCACT. The plasmid expressing murine HMG-I in the antisense orientation (pSG5-AS-HMG-I) was made by cloning an HMG-I BamHI fragment (a gift from J. Maher, Jackson, MS) in the antisense orientation into pSG5 at the same site. pLRT-c-Jun, pLRT-Tam67, pSG5-HMG-I, and pSG5-HMG-Y have been described previously (9, 16).

Total cellular RNA from Rat1a-J2, Rat1a-J4, and Rat1a-GFP cells was extracted using the RNeasy Mini Kit (Qiagen, Valencia, CA) according to the manufacturer's instructions. Total RNA (1 μg) was separated on 1% agarose gels and transferred to nitrocellulose membranes. The membranes were probed with [α-³²P]dCTP-labeled cDNA probes, and the signal was detected by autoradiography. The cDNA probe for HMG-I was obtained from pSG5-HMG-I (33), and that for c-jun was obtained from pBlue-c-jun (10).

Cell lysates from Rat1a-J4 and Rat1a-GFP cells grown in the absence or presence of doxycycline (2 μg/mL) were prepared by lysing the cells in radioimmunoprecipitation assay buffer [150 mmol/L NaCl, 1% Triton X-100, 1% deoxycholate, 0.1% SDS, 10 mmol/L Tris (pH 7.4)] supplemented with 0.1 mmol/L phenylmethylsulfonyl fluoride, 100 μg/mL aprotinin, and 100 μg/mL leupeptin. The cell lysates were sonicated and centrifuged to remove debris, and protein concentrations were determined using the Bio-Rad Protein Assay (Bio-Rad, Hercules, CA). Equal amounts of protein were separated on 12% or 15% SDS gels, transferred to nitrocellulose membranes (PROTRAN, Schleicher & Schuell, Keene, NH), and incubated with the following antibodies: anti-HMG-I/Y antibody (29), anti-c-Jun (sc45x), and anti-β-tubulin (sc9104, Santa Cruz Biotechnology, Santa Cruz, CA). The signal was detected by enhanced chemiluminescence.

Exponentially growing Rat1a-J4 or Rat1a cells were seeded at a density of 3 × 10⁵ cells per 60 mm dish. The cells were transfected with 1 μg HMG-I/Y promoter reporter constructs using FuGene6 Transfection Reagent (Roche Diagnostics Corp., Indianapolis, IN). Forty nanograms of Renilla luciferase (Promega) were cotransfected to control for transfection efficiency. After overnight incubation, cells were trypsinized and plated under nonadherent growth conditions in PolyHeme-coated 12-well dishes in the presence or absence of doxycycline (2 μg/mL). Luciferase activity was measured 2 days after induction with doxycycline using the Dual-Luciferase Reporter Assay System (Promega).

Nuclear proteins were prepared from Rat1a-J4 cells grown for 3 days under nonadherent conditions in the presence or absence of doxycycline (2 μg/mL) as described previously (66). Briefly, cells were pelleted and washed once with cold PBS. The cell pellet was resuspended in five packed volumes of 10 mmol/L HEPES (pH 7.9), 1.5 mmol/L MgCl₂, 10 mmol/L KCl, and 0.5 mmol/L DTT and allowed to swell on ice for 10 minutes. The cells were lysed by slow uptake and rapid ejection (5×) through a 25 gauge needle. Nuclei were pelleted and resuspended in one-third volume of 20 mmol/L HEPES (pH 7.9), 25% glycerol, 0.45 mol/L NaCl, 1.5 mmol/L MgCl₂, 0.2 mmol/L EDTA, 0.5 mmol/L phenylmethylsulfonyl fluoride, 100 μg/mL aprotinin, and 1 μg/mL leupeptin and rotated for 30 minutes at 4°C. The resulting nuclear proteins were dialyzed against 50 volumes of 20 mmol/L HEPES (pH 7.9), 25% glycerol, 0.1 mol/L KCl, 0.2 mmol/L EDTA, 0.5 mmol/L phenylmethylsulfonyl fluoride, and 1 μg/mL each of aprotinin and leupeptin for 2 hours at 4°C. Supernatants were stored at −70°C, and protein concentrations were determined as described above. Five micrograms of nuclear extract were incubated with 2 μg of poly-deoxyinosinic-deoxycytidylic acid and 4 μL of incubation buffer [100 mmol/L HEPES (pH 7.9), 250 mmol/L KCl, 2.5 mmol/L EDTA, 5 mmol/L MgCl₂, and 20% Ficoll 400] in a final volume of 20 μL for 10 minutes on ice prior to the addition of double-stranded oligonucleotides labeled with [γ-³²P]ATP. The protein-DNA mixtures were incubated on ice for a further 20 minutes and separated on nondenaturing 5% polyacrylamide gels at 180 volts for 2 to 3 hours at 4°C in 0.5× Tris-borate EDTA. Unlabeled (100 ng) wild-type or mutated oligonucleotides were used as competitors to confirm the specificity of the complexes formed. The following oligonucleotides corresponding to the murine HMG-I/Y promoter were used as probes: AP-1–like (1), wild-type GCCCCTCCATGACTTCCTCCTTCTCCA and mutated GCCCCTCCATGGCTTGCTCCTTCTCCA; AP-1–like (2), wild-type CCCTGGAACTGAGGGTCACCTGGACC and mutated CCCTGGAACTGGGTGTCACCTGGACC; and AP-1, wild-type GGAACAGAGTTATGAGTCACAGTCGTGTGT and mutated GGAACAGAGTTATGGGTTGCAGTCGTGTGT.

For supershift assays, antibodies (sc45x, sc46x, sc74x, sc52x, and sc605x, Santa Cruz Biotechnology) were added to the DNA-protein mixture and incubated on ice for 20 minutes prior to the addition of the probe.

Cells (1 × 10⁴) were plated in triplicate in 6 mL of 0.35% agarose (Sea Plaque, FMC Bioproducts, Rockland, ME) in complete growth medium in the presence or absence of doxycycline (2 μg/mL) overlaid on a 0.7% agarose base, also in complete growth medium, and incubated for 2 to 3 weeks at 37°C. Colonies were stained with p-iodonitrotetrazolium violet (1 mg/mL) and counted using a GS710 Calibrated Imaging Densitometer and Quantity One software (Bio-Rad).