High-Mobility Group A2 Protein Modulates hTERT Transcription To Promote Tumorigenesis

Abstract

The high-mobility group A2 gene (HMGA2) is one of the most frequently amplified genes in human cancers. However, functions of HMGA2 in tumorigenesis are not fully understood due to limited knowledge of its targets in tumor cells. Our study reveals a novel link between HMGA2 and the regulation of human telomerase reverse transcriptase (hTERT), the catalytic subunit of telomerase, which offers critical insight into how HMGA2 contributes to tumorigenesis. The expression of HMGA2 modulates the expression of hTERT, resulting in cells with enhanced telomerase activities and increased telomere length. Treatment with suberoylanilide hydroxamide (SAHA), a histone deacetylase (HDAC) inhibitor, causes dose-dependent hTERT reporter activation, mimicking HMGA2 overexpression. By interacting with Sp1, HMGA2 interferes with the recruitment of HDAC2 to the hTERT proximal promoter, enhancing localized histone H3-K9 acetylation and thereby stimulating hTERT expression and telomerase activity. Moreover, HMGA2 knockdown by short hairpin HMGA2 in HepG2 cells leads to progressive telomere shortening and a concurrent decrease of steady-state hTERT mRNA levels, attenuating their ability to form colonies in soft agar. Importantly, HMGA2 partially replaces the function of hTERT during the tumorigenic transformation of normal human fibroblasts. These findings are potentially clinically relevant, because HMGA2 expression is reported to be upregulated in a number of human cancers as telomere maintenance is essential for tumorigenesis.

INTRODUCTION

The human high-mobility group A2 gene (HMGA2) gene is located at chromosomal locus 12q14-15, a region that is frequently subjected to chromosomal amplification and rearrangements, resulting in the deregulation of its expression, truncation, or generation of fusion genes encoding chimeric transcripts containing the first three exons of HMGA2 and ectopic sequences from other genes (reviewed in references 5, 9, 15, and 52). In addition, HMGA2 translocation often results in a loss of its 3′ untranslated region (3′-UTR) that contains a binding site for the tumor suppressor Let-7, a microRNA specifically involved in the posttranscriptional repression of HMGA2 (19, 36, 39, 55). HMGA2 overexpression has been reported in colorectal, breast, pancreatic, ovarian, and lung carcinomas and in squamous carcinomas of the oral cavity (1, 9, 15, 52, 59). A direct correlation between high expression levels of HMGA2 and increasing degrees of neoplastic transformation has been suggested. Also, a mutation in the breast cancer susceptibility gene BRCA1 has been shown to induce HMGA2 derepression in breast cancer cells (14). Compelling evidence has accumulated on HMGA2-elicited oncogenic effects (9). Notably, HMGA2 has been reported to interact with pRb and shown to enhance E2F1 activity by displacing histone deacetylase 1 (HDAC1) (13). However, despite extensive studies of HMGA2, the underlying mechanism of how HMGA2 predisposes cells to transformation and/or confers a growth advantage to cells during tumorigenesis remains to be established.

The telomerase elongation of telomeres is a highly coordinated and tightly regulated process, so that the length of the telomeric repeats is kept within a cell type-specific narrow range from 3 to 20 kb in human cells (27). The human core telomerase enzyme consists of a catalytic protein subunit telomerase reverse transcriptase (hTERT) and an RNA moiety (hTR) that contains a short RNA template. In human cells, progressive telomere shortening eventually leads to the loss of telomere capping, resulting in recognition by a DNA damage response, the activation of p53, and the induction of cellular senescence (12, 23, 32), presumably serving as an antitumor mechanism to inhibit the progression of premalignant cells with mutations accumulated during their replicative life span. The expression of hTERT is reportedly the rate-limiting factor for the assembly of an active telomerase complex (6). In most somatic cells, telomerase activity is silenced or is present at very low levels, whereas cancer cells, germ line cells, and embryonic stem cells all show abundant hTERT expression. Notably, telomerase or hTERT reactivation is detected in up to 90% of human malignancies (54).

The critical roles of hTERT in tumor proliferation and stem cell behavior underscore the importance of understanding the regulatory mechanisms for hTERT transcriptional control. Accumulating evidence has suggested that histone modification-mediated chromatin remodeling, specifically histone acetylation or deacetylation at the hTERT promoter, leads to the transcriptional regulation of the hTERT gene (10, 26, 57, 61). However, the molecular mechanism responsible for increased histone acetylation at the hTERT promoter during tumorigenesis remains to be established. We now provide the first evidence that HMGA2 interacts with Sp1 and increases histone H3-K9 acetylation by interfering with the recruitment of HDAC2 to Sp1 at the hTERT proximal promoter, contributing to increased telomerase activity for telomere lengthening in HMGA2-expressing cells. These findings support the hypothesis that HMGA2 is critically involved in preventing the gradual shortening of telomeres in cancer cells. Given that the reactivation of hTERT is essential for tumor cell proliferation and self-renewal, we propose that HMGA2 plays a protumorigenic role by modulating hTERT expression.

MATERIALS AND METHODS

Cell culture, viral transduction, constructs, HMGA2 knockdown, and drugs.

HeLa, CL48, and HepG2 cells were maintained in Dulbecco's modified essential medium (DMEM) (Mediatech) containing 10% fetal bovine serum (FBS; HyClone) and antibiotics. Stable HeLa/HMGA2 cells were cultured in the same medium in the presence of G418 (200 μg/ml). Normal human lung fibroblast IMR-90 (ATCC CCL-186) and WI-38 (ATCC CCL-75) cells were maintained according to the supplier's instructions. Lentiviral vectors pRRLsin.hCMV-HMGA2, pΔ8.7, and pVSV-G were constructed and used for lentiviral production in HEK 293T cells as previously described (44). Lentivirus (harboring HMGA2 or vector)-transduced WI-38 or IMR-90 cells were maintained for two passages (1:2 dilution) to avoid cloning bias prior to telomeric repeat amplification protocol (TRAP) assay. p-179-Luc and p-274-Luc were generated by the insertion of PCR amplification products of the hTERT promoter (GenBank accession no. AF097365) from HeLa genomic DNA and directionally cloned into BglII/HindIII-linearized pGL3 Basic luciferase reporter vector (Promega). pFlag-HDAC2 and pSin-SV40T-neo, as well as pSin-RasV12-neo and pSin-hTERT-neo, were kindly provided by Yanhong Shi and Jiing-Kuan Yee, respectively (City of Hope, CA). HepG2 cells were transduced with lentiviruses harboring the empty vector or short hairpin RNA (shRNA) against HMGA2 in lentivector pLKO.1, which was obtained from the RNAi Consortium at Academia Sinica (Taiwan). Transduced cells were selected in 2 μg/ml of puromycin. GRN163L and suberoylanilide hydroxamide (SAHA) were kindly provided by Sergei Gryaznov (Geron Corporation, CA) and Leo Kretzner (City of Hope, CA), respectively.

RNA extraction and real-time RT-PCR.

Total RNA was extracted from HeLa, HeLa/HMGA2, CL48, HepG2, and transduced IMR-90 cells using an RNeasy kit (Qiagen) according to the manufacturer's instructions. Subsequent cDNA synthesis was carried out with a iScript cDNA synthesis kit (Bio-Rad); a fraction of the reverse transcription (RT) reaction was amplified with IQ SYBR green supermix and specific primer pairs with a My IQ real-time PCR detection system (Bio-Rad). Specific primer pairs used in this study are listed in Table 1. Relative mRNA expression levels were calculated using the ΔΔC_T method against 18S rRNA. The expression of hTERT was determined by real-time RT-PCR throughout the study due to the lack of a specific hTERT antibody.

Table 1.

Primer pairs and sequences used in this study

Primer namea	Purpose	Sequence (5′ to 3′)
18S rRNA FP	RT and real-time PCR	CGGCGACGACCCATTCGAAC
18S rRNA RP	RT and real-time PCR	GAATCGAACCCTGATTCCCCGTC
hTERT FP	RT and real-time PCR	CACGCGAAAACCTTCCTCAG
hTERT RP	RT and real-time PCR	CAAGTTCACCACGCAGCCAT
HMGA2 FP	RT and real-time PCR	TCCCTTCACAGTCCCAGGTTTAG
HMGA2 RP	RT and real-time PCR	TTTTTCTCACCCGCCCACTC
hTERT −191 amplicon FP	ChIP (real-time PCR)	CCAGCTCCGCCTCCTCCGCG
hTERT −191 amplicon RP	ChIP (real-time PCR)	GGGGCCGCGGAAAGGAAGGGG
hTERT −276 amplicon FP	ChIP (real-time PCR)	ATTCGCGGGCACAGACGCC
hTERT −276 amplicon RP	ChIP (real-time PCR)	GCTGGAAGGTGAAGGGGCAG
hTERT −2929 amplicon FP	ChIP (real-time PCR)	GCTCTTGTTGCCCAGGCTGG
hTERT −2929 amplicon RP	ChIP (real-time PCR)	CAGGAGACGGAGGTTGCAGTG

^aFP, forward primer; RP, reverse primer.

Luciferase assay.

HeLa and HeLa/HMGA2 cells were cotransfected with p-179-Luc, p-274-Luc, or pFlag-HDAC2 and the internal control reporter pRL-TK. Cells were harvested at 24 h posttransfection, and firefly luciferase activity was measured with the dual-luciferase reporter assay system (Promega) and normalized against Renilla luciferase activity; n = 3 independent experiments in triplicate.

Immunoprecipitation and Western blot analyses.

For immunoprecipitation, cells were lysed in radioimmunoprecipitation assay (RIPA) buffer supplemented with phenylmethylsulfonyl fluoride (PMSF), aprotinin, and sodium orthovanadate. Equal amounts of protein lysates were subjected to immunoprecipitation using an anti-IgG control (Santa Cruz Biotech), antihemagglutinin (anti-HA) (Covance), anti-HDAC2 (Thermo Scientific), or anti-Sp1 (Abcam) antibody. For Western blot analyses, equal amounts of protein lysate (20 to 40 μg) were subjected to SDS-PAGE followed by Western blotting. Additional antibodies used were anti-HMGA2 (BioCheck), anti-simian virus 40 (anti-SV40) tag (Santa Cruz Biotech), anti-H-Ras (Santa Cruz Biotech), and antiactin (Chemicon). Images were visualized with an enhanced chemiluminescence detection kit (ECL-Plus; Amersham Pharmacia Biotech) and the Versadoc 5000 imaging system (Bio-Rad). Results of Western blot analyses are representative of two to four independent experiments.

Soft-agar assay.

The puromycin stably selected HepG2 vector, HepG2- shHMGA2-2, or HepG2-shHMGA2-4 cells were used for soft-agar assays. Briefly, 4 ml of complete medium containing 0.7% Difco agar (BD Biosciences) was used as the base layer of a 60-mm dish. This was overlaid with 3 ml of a second layer of 0.35% agar containing a suspension of 5,000 cells. After 20 days in culture, cells were stained (4 to 6 h, 37°C) with 0.2 mg/ml of 3-(4,5-dimethyl-2-thiazolyl)-2,5-diphenyl-2H-tetrazolium bromide (MTT) (Sigma). Combinations of SV40 large T-, hTERT-, HMGA2-, and RasV12-transduced IMR-90 cells were used for soft-agar assay using the method described above, except that 50,000 cells were used. Colonies of >0.2 mm in diameter were tabulated. All cultures were performed three times, each performed in triplicate.

Telomere length assay.

Telomere length was measured using the nonradioactive chemiluminescent Telo TAGGG telomere length assay (Roche) according to the manufacturer's protocol. Mean telomere restriction fragment (TRF) length was calculated based on the following formula: mean TRF = Σ(OD_i)/Σ(OD_i/L_i), where OD_i is the chemiluminescent signal and L_i is the length of the TRF at position i (22).

TRAP.

Telomerase activity was evaluated using a PCR-based TRAP-eze telomerase detection kit (Chemicon) according to the manufacturer's instructions. Using equivalent total protein amounts, telomerase activity in each extract was determined by TRAP assay. The telomerase PCR products were resolved on 12% native polyacrylamide gels and visualized by SYBR green staining.

ChIP and ChIP-ReIP assays.

Chromatin immunoprecipitation (ChIP) experiments were performed using a ChIP assay kit (Upstate Biotech) as described by the manufacturer. Cross-linked proteins were immunoprecipitated using an anti-Sp1 (Abcam), anti-HDAC2 (Thermo Scientific), or anti-acetyl-histone H3-K9 (Upstate Biotech) antibody. For ChIP-reimmunoprecipitation (ChIP-Re-IP) experiments, primary immunocomplexes obtained with anti-Sp1 were washed and eluted with 10 mM dithiothreitol (DTT) (37°C, 30 min, with occasional vortexing), and eluates were diluted (1:50) in IP dilution buffer to perform ReIP with anti-HA (Covance). The recovered DNA was amplified with specific primer pairs (Table 1) using a My IQ real-time PCR detection system and IQ SYBR green supermix (Bio-Rad). Normal IgG and input DNA values were used to subtract and normalize values from ChIP and ChIP-ReIP samples. The percent input method was used to quantitate the values of the immunoprecipitated DNA.

Statistical analysis.

Unpaired Student's t tests, assuming equal variances, were performed using Microsoft Excel to determine significant differences between groups, with P < 0.05 considered significant.

RESULTS

HMGA2 stimulates telomerase activity.

HeLa cells are known as HMGA2 low-expressing cells (Fig. 1 A, left) and have been used as a model to study the posttranscriptional modification of HMGA2 mRNA by the microRNA Let-7 (36). Gene expression profiles analyzed in two independent preparations of RNA samples obtained from HeLa and HeLa/HMGA2 cells indicated that the hTERT mRNA level was 1.6-fold upregulated (P < 0.005) in HeLa/HMGA2 cells (data not shown). Real-time RT-PCR analyses confirmed that there was a 3-fold increase in hTERT mRNA in HeLa/HMGA2 cells (Fig. 1A, left).

Fig. 1.

HMGA2 stimulates telomerase activity. (A) hTERT mRNA and HMGA2 protein levels were assessed by real-time RT-PCR and Western blot analyses, respectively (left). Serial dilutions of total protein lysates from HeLa and HeLa/HMGA2 cells were subjected to TRAP analysis to evaluate the basal levels of telomerase activity (right). IC, internal PCR control. (B) HeLa and HeLa/HMGA2 cells were incubated (for 72 h) with increasing concentrations of GRN163L, an antisense oligonucleotide against hTR. Cells then were subjected to TRAP assays. −, negative control (telomerase-positive cell extract heated at 85^°C for 10 min); +, telomerase-positive cell extract. (C) Equal amounts of total protein lysate from lentiviral transduced IMR-90 and WI-38 cells were subjected to Western blot analyses to confirm the expression of HMGA2 (left) and to TRAP assays (right). The quantitative analysis of telomerase activities was performed using Quantity One densitometry software to measure the intensity of all bands, except the IC bands, on the gels. Relative telomerase activities are shown in italics.

A telomeric repeat amplification protocol (TRAP) was used to test whether increased hTERT expression could lead to functional telomerase modulation. Due to the high endogenous levels of telomerase activity found in HeLa cells (7, 42) and the high sensitivity of TRAP assays, serial dilutions of protein lysates from HeLa and HeLa/HMGA2 cells were subjected to TRAP analysis to detect their differential telomerase activity levels. HeLa/HMGA2 protein lysate (10 ng) produced an intense distinct telomerase product ladder, whereas a relatively less intense ladder was produced with the same amount of HeLa lysate (Fig. 1A, right). Furthermore, when 1 ng of protein lysate was used, telomerase activity was readily detectable in HeLa/HMGA2 cells but barely detectable in HeLa cells. To further confirm the HMGA2-mediated enhancement of hTERT expression, the effect of GRN163L, a lipidated 13-mer oligonucleotide N3′-P5′-thio-phosphoramidate targeting the active site in the template region of hTR (2, 29), was evaluated in HeLa and HeLa/HMGA2 cells. Treatment with increasing concentrations of GRN163L showed that HeLa/HMGA2 cells were 4-fold more resistant to GRN163L than to HeLa cells at 72 h posttreatment (Fig. 1B, lane 13 versus lane 5).

To determine whether HMGA2 was capable of activating telomerase activity in normal cells, HMGA2 or an empty vector was ectopically expressed in normal human lung fibroblast IMR-90 and WI-38 cells, known as telomerase-negative cell strains with undetectable hTERT expression, via lentiviral transduction. Upon HMGA2 expression (Fig. 1C, left), telomerase activity was induced by 2- to 3-fold in IMR-90/HMGA2 and WI-38/HMGA2 cells but not in IMR-90/vector or WI-38/vector cells by TRAP assays (Fig. 1C, right), confirming that the ability of HMGA2 to modulate telomerase activity is not restricted to one cell type. As the steady-state hTERT level is the rate-limiting factor for the assembly of an active telomerase complex, our results indicated that HMGA2 stimulates hTERT expression.

hTERT promoter activity was enhanced by increased HMGA2 expression in a dose-dependent manner.

A major mechanism to regulate telomerase activity in human cells is the transcriptional control of the hTERT gene (40, 48). To test whether the observed enhanced telomerase activity by HMGA2 correlated with stimulated hTERT promoter activity, the hTERT minimal promoter containing 274 bp (61) and a shorter proximal region with 174 bp upstream of the ATG start were fused upstream of the luciferase reporter to generate p-274-Luc and p-174-Luc, respectively. The proximal 274-bp minimal promoter of hTERT is GC rich and has no detectable TATA or CAAT boxes but contains binding sites for several transcription factors (11, 34). Among these are two E-boxes and five Sp1 binding sites (Fig. 2 A).

Fig. 2.

hTERT promoter activity was enhanced by overexpressing HMGA2 in a dose-dependent manner. (A) Schematic diagram of the hTERT promoter with sequence of the 274-bp proximal promoter region showing putative Myc (in italics and underlined) and Sp1 (in boldface and underlined) binding sites. The locations of Myc and Sp1 binding sites are presented as previously described (34). A fragment containing 179 or 274 bp upstream of hTERT gene was cloned into pGL3 Basic vector to generate p-179-Luc and p-274-Luc, respectively. The A of the translation start codon (ATG) of hTERT gene is designated +1. (B) Relative luciferase activities (RLA) of extracts from HeLa and HeLa/HMGA2 cells cotransfected with p-179-Luc or p-274-Luc and pRL-TK (internal control vector) for 24 h were subjected to luciferase assay (left). HeLa cells were transiently cotransfected with p-179-Luc or p-274-Luc and pRL-TK with increasing amounts of HMGA2 expression vector (right). Twenty-four h posttransfection, cells were lysed and subjected to luciferase assay while aliquots of the same lysates were resolved by SDS-PAGE and immunoblotted with the indicated antibodies. (C) HeLa and HeLa/HMGA2 cells were cotransfected with p-179-Luc or p-274-Luc, vector or pFlag-HDAC2, and pRL-TK, followed by luciferase assay 24 h posttransfection. (D) At 8 h posttransfection of HeLa and HeLa/HMGA2 cells with p-179-Luc or p-274-Luc and pRL-TK, cells were treated with increasing concentrations of SAHA, and luciferase assays were performed after 24 h. Luciferase activity was normalized against pRL-TK to control for transfection efficiency variation. RLA of all transfectants was presented by setting the activity of p-179-Luc in vector-transfected cells to 1. Results represent means ± standard deviations from three independent experiments. *, P < 0.05; **, P < 0.01.

p-179-Luc or p-274-Luc and pRL-TK were cotransfected into HeLa and HeLa/HMGA2 cells; luciferase assays were performed 24 h later, and the transfection efficiency was normalized against Renilla luciferase activity. Both p-274-Luc and p-174-Luc directed more than 3-fold higher activity in HeLa/HMGA2 cells than HeLa cells (Fig. 2B, left), suggesting that stable HMGA2 expression leads to increased hTERT promoter activity. Notably, p-274-Luc demonstrated more robust reporter activation than p-179-Luc in both HeLa and HeLa/HMGA2 cells. Presumably, this increased promoter activity was due to the presence of Myc binding sites (E boxes) within p-274-Luc, as c-Myc has been reported to activate hTERT expression (34). To establish a correlation between HMGA2 expression and hTERT promoter activation, HeLa cells were transiently transfected with increasing amounts of HMGA2, and the reporter activity was assayed. HMGA2 stimulated hTERT promoter activity in an HMGA2 dose-dependent manner (right), indicating that increased HMGA2 expression leads to a correspondingly higher induction in hTERT promoter activity. A similar dose-dependent induction of p-274-Luc and p-174-Luc reporters by HMGA2 was observed in IMR-90 and WI-38 cells (see Fig. S1A in the supplemental material).

HDAC-mediated histone deacetylation is proposed as one of the transcriptional repression mechanisms of the hTERT gene (10, 57, 58, 61). Specifically, Won et al. have shown that Sp1 recruits HDAC2 to the hTERT promoter for its transcriptional silencing (61). To test whether HDAC2 was involved in regulating the hTERT promoter activity in our system, the HDAC2 expression vector was cotransfected with p-179-Luc or p-274-Luc followed by luciferase assays. As expected, HDAC2 significantly repressed hTERT promoter activity in both HeLa and HeLa/HMGA2 cells compared to expression by the control vector (Fig. 2C). Moreover, treatment with increasing concentrations of suberoylanilide hydroxamide (SAHA), a pan-inhibitor of class I and II HDAC proteins, significantly increased the promoter activity in a concentration-dependent manner (Fig. 2D), indicating the involvement of histone deacetylation in the repression of hTERT promoter.

HMGA2 competes with HDAC2 for binding to Sp1.

To investigate whether HMGA2 interferes with HDAC2 recruitment by Sp1 to modulate hTERT expression, equal amounts of protein lysate from HeLa and HeLa/HMGA2 cells were subjected to coimmunoprecipitation (co-IP) coupled with Western blot analyses. Notably, HMGA2 was present in the immunoprecipitated complexes pulled down by anti-HDAC2 or anti-Sp1 antibody (Fig. 3 A, lanes 6 and 8, top). Interestingly, HDAC2 and Sp1 were detected to be interacting partners in HeLa (lanes 5 and 7) but not in HeLa/HMGA2 cells (lanes 6, bottom, and 8, middle), suggesting that the protein-protein interaction between HMGA2 and HDAC2 or Sp1 impedes the formation of the Sp1/HDAC2 complex. Since HADC2 expression reduced hTERT promoter activity even in the presence of HMGA2 (Fig. 2C), we hypothesized that HDAC2 and HMGA2 compete for the binding of Sp1. To test this possibility, HDAC2 was overexpressed in HeLa/HMGA2 cells followed by co-IP coupled with Western blot analyses. Consistently with our prediction, the interaction between Sp1 and HDAC2 was restored in HeLa/HMGA2 cells upon the overexpression of HDAC2 (Fig. 3B, lanes 6, bottom, and 8, middle), supporting the idea that HMGA2 displaces HDAC2 from the Sp1/HDAC2 complex, hence interfering with the repressive role of SP1/HDAC2 on hTERT expression.

Fig. 3.

HMGA2 competes with HDAC2 for binding to Sp1. (A) Equal amounts of protein lysates were subjected to coimmunoprecipitation (co-IP) assays, followed by Western blot analyses with the respective antibodies as indicated. (B) HDAC2 expression construct or its vector control was transiently transfected into HeLa/HMGA2 cells, and co-IP followed by Western blot analyses were performed at 48 h posttransfection as described for panel A. IB, immunoblot. α-IgG, α-HA, α-HDAC2, and α-SP1, antibodies to IgG, HA, HDAC2, and Sp1, respectively.

HMGA2 interferes with the recruitment of HDAC2 to the hTERT promoter, leading to increased histone H3-K9 acetylation at the proximal hTERT promoter.

To directly address whether HMGA2 interferes with the association of Sp1/HDAC2 with the hTERT promoter, chromatin immunoprecipitation (ChIP) and ChIP-reimmunoprecipitation (ChIP-Re-IP) assays were performed using −276 and −191 amplicons that are on the proximal hTERT promoter, which encompass one and four Sp1 binding sites, respectively, and the −2929 amplicon, which is on the distal hTERT promoter lacking any Sp1 binding site (Fig. 4 A, upper). The −2929 amplicon should serve as a control for background DNA that is nonspecifically immunoprecipitated. The occupancy of HDAC2 at the hTERT proximal promoter in HeLa/HMGA2 cells was decreased significantly compared to that in HeLa cells (Fig. 4A, middle left). Moreover, marked increases in H3-K9 acetylation at the hTERT proximal promoter were observed in HeLa/HMGA2 cells compared to that of HeLa cells (Fig. 4A, middle right). However, the association of Sp1 with the hTERT promoter was not significantly different between HeLa and HeLa/HMGA2 cells (Fig. 4A, lower left), which is consistent with a previous report that the occupancy of the hTERT promoter by Sp1 does not change between acetylation statuses (61). HMGA2 protein preferentially binds to AT-rich DNA; hence, the direct binding of HMGA2 to the hTERT proximal promoter (which is GC rich) was unlikely. However, the detected interaction of HMGA2 with HDAC2 or Sp1 (Fig. 3) raised the possibility that HMGA2 is indirectly bound to the promoter through protein-protein interactions. The ChIP-ReIP experiment, which first was performed with an anti-Sp1 antibody followed by an anti-HA antibody, revealed that the occupancy of stably integrated HMGA2 was notably detected at the −276 and −191 amplicons (Fig. 4A, lower right).

Fig. 4.

HMGA2 interferes with the recruitment of HDAC2 to the hTERT promoter, leading to increased histone H3-K9 acetylation at the proximal hTERT promoter. (A) A schematic diagram of hTERT promoter indicating the relative position of the transcription start site, Sp1 binding sites, and amplicons used in the ChIP assays (upper). The occupancy by HDAC2 at the hTERT proximal promoter and distal region (middle left) and the acetylation of histone H3-K9 within the hTERT promoter (middle right) were assessed by ChIP assays using equal numbers of HeLa and HeLa/HMGA2 cells. The occupancy by Sp1 at the hTERT promoter (lower left) and the recruitment of HA-HMGA2 by Sp1 to the hTERT promoter (lower right) were assessed in HeLa and HeLa/HMGA2 cells by ChIP and ChIP-ReIP experiments, respectively. (B) HeLa/HMGA2 cells were transiently transfected with HDAC2 expression construct or its vector control for 48 h, followed by the assessment of occupancy by HDAC2 (upper left), acetylated histone H3-K9 (upper right), and Sp1 at the hTERT proximal promoter and distal region (lower left), as well as the recruitment of HA-HMGA2 by Sp1 to the hTERT promoter (lower right) using ChIP and ChIP-ReIP assays. IgG controls were performed (not shown), and the values were subtracted from results for ChIP and ChIP-ReIP samples. Results represent means ± standard deviations from three independent immunoprecipitations. *, P < 0.05; **, P < 0.01.

We next investigated whether the abundance of HDAC2 also affects the occupancy of HMGA2 at the hTERT promoter. Results from ChIP assays revealed that the overexpression of HDAC2 in HeLa/HMGA2 cells significantly increased the occupancy of HDAC2 compared to that of the vector control at the hTERT proximal promoter (Fig. 4B, upper left), leading to a decrease in the histone H3K9 acetylation profile in HDAC2-overexpressing cells (Fig. 4B, upper right). Again, the association of Sp1 with the hTERT promoter was not significantly affected by HDAC2 overexpression (Fig. 4B, lower left). Finally, the abundance of HDAC2 in HeLa/HMGA2 cells displaced HMGA2 from the Sp1/HMGA2 complex to restore Sp1/HDAC2 interaction (Fig. 3B), leading to a reduction in the occupancy of HMGA2 at the hTERT proximal promoter (Fig. 4B, lower right). Taken together, our results indicated that HMGA2 interacted with Sp1 at the −276 and −191 amplicon regions and caused an impediment to HDAC2 being recruited to the hTERT proximal promoter, promoting the localized acetylation of histone H3-K9, hence modulating the hTERT transcriptional regulation.

A positive correlation between the expression levels of HMGA2 and hTERT.

To further explore the role of HMGA2 in regulating hTERT expression in a physiological context, the expression profiles of endogenous HMGA2 protein and hTERT mRNA were assessed in immortalized liver CL48 cells and hepatoma HepG2 cells. Consistently with a previous report (47), HepG2 cells expressed a much higher level of endogenous HMGA2 than CL48 cells. As predicted, the upregulation of HMGA2 expression was associated with increased hTERT mRNA in HepG2 cells (Fig. 5 A). To confirm whether HMGA2 was, at least in part, responsible for the increased hTERT expression, endogenous HMGA2 was knocked down in HepG2 cells using shRNA; upon knockdown, both shHMGA2-2 and shHMGA2-4 conferred a decrease in hTERT mRNA compared to the level of the vector control (Fig. 5B). To assess the biological importance of HMGA2 in the induction of hTERT expression, soft-agar assays were performed using HepG2 cells with distinct levels of HMGA2. Efficient colony formation was observed in soft agar with HepG2/vector cells; however, both the number and size of colonies formed were greatly reduced upon endogenous HMGA2 knockdown (Fig. 5C), demonstrating the ability of HMGA2 in promoting cell transformation. It has been extensively demonstrated that HMGA2 is involved in the pleiotropic regulation of the transcription of many different genes; hence, the influence of HMGA2 on some of these genes, rather than its effect on hTERT per se, may explain the reduction in colony-forming ability. To distinguish these possibilities, hTERT was transduced in HepG2/shHMGA2-2 cells and then subjected to soft-agar assays. When HepG2/shHMGA2-2 cells were supplemented with hTERT (Fig. 5D, left), significantly more colonies were observed in hTERT-transduced than in vector-transduced HepG2/shHMGA2-2 cells (Fig. 5D, right), recapitulating the biological consequence of the HMGA2-mediated hTERT modulation. The greater number of colonies formed in HepG2/vector cells (Fig. 5C) than that in HepG2/shHMGA2-2/hTERT cells (Fig. 5D, right) most likely is due to the dysregulation of HMGA2 on genes besides hTERT.

Fig. 5.

Positive correlation between HMGA2 and hTERT expression levels in hepatoma cell lines. (A) hTERT mRNA and HMGA2 and HDAC2 protein levels were assessed by real-time RT-PCR and Western blotting, respectively. (B) HepG2 cells were transduced with lentiviruses containing an empty vector, shHMGA2-2, or shHMGA2-4, followed by puromycin selection. The relative levels of HMGA2 protein normalized against actin are italicized. **, P < 0.01. (C) Effect of HMGA2 knockdown on the anchorage-independent growth of HepG2 cells in soft agar. After 20 days, colonies greater than 0.2 mm in diameter were tabulated. Results represent means ± standard deviations from three independent experiments. (D) HepG2/shHMGA2-2 cells were transduced with lentiviruses harboring vector or hTERT. The cells then were subjected to real-time RT-PCR and Western blotting to assess the hTERT mRNA and HMGA2 and HDAC2 protein levels, respectively (left). Soft-agar assays (right) were performed on these cells in the same manner as that described for panel C.

HMGA2 expression results in telomere elongation.

Telomere length then was assessed to investigate the cellular consequences of HMGA2-mediated hTERT modulation. The mean telomere restriction fragment (TRF) length was significantly longer in HeLa/HMGA2 cells (7.8 kb) than in HeLa cells (5.0 kb) (Fig. 6 A, lanes 1 and 2). Telomere lengths in non-HMGA2-expressing CL48 and HMGA2-expressing HepG2 cells also were assessed. As expected, a much shorter mean TRF was observed in CL48 cells (3.2 kb) compared to that of HepG2 (6.1 kb) cells (Fig. 6A, lanes 3 and 4). These results strongly suggested that HMGA2-dependent hTERT modulation was able to elongate telomeres. Conversely, after HepG2/vector, HepG2/shHMGA2-2, and HepG2/shHMGA2-4 cells were propagated in culture for approximately 100 population doublings (PD), the mean TRF length shortened progressively by more than 25%, from 6.1 to 4.5 and 4.2 kb in HepG2/shHMGA2-2 and HepG2/shHMGA2-4 cells, respectively (Fig. 6B, lane 1 versus lanes 2 and 3). In contrast, HMGA2-expressing HepG2/vector cells maintained a stable mean TRF length (Fig. 6B, lane 1). Our results suggested that the knockdown of endogenous HMGA2 led to telomere shortening, supporting the role of HMGA2 in modulating telomerase activity.

Fig. 6.

Telomere length was increased in HMGA2-expressing cells and shortened upon HMGA2 knockdown. (A) Genomic DNA isolated from various cell lines was digested with HinfI and RsaI and subjected to Southern blot analysis by hybridization with a telomere-specific probe. White dots indicate the mean telomere restriction fragment (TRF) length of various samples: HeLa, 5.0 kb; HeLa/HMGA2, 7.8 kb; CL48, 3.2 kb; HepG2, 6.1 kb. (B) HepG2 cells were infected with vector control, shHMGA2-2, or shHMGA2-4, followed by puromycin selection. After approximately 100 population doublings, genomic DNA prepared from the puromycin-selected HepG2/vector, HepG2/shHMGA2-2, or HepG2/shHMGA2-4 cells was assessed for TRF size. White dots indicate the mean TRF length: HepG2/vector, 6.1 kb; HepG2/shHMGA2-2, 4.5 kb; HepG2/shHMGA2-4, 4.2 kb. MWM, molecular size marker.

Ectopic expression of HMGA2 with a cooperating niche provided by SV40 large T and RasV12 result in tumorigenic conversion of normal human fibroblast cells.

Hahn et al. reported that the overexpression of hTERT in combination with simian virus 40 (SV40) large T and an oncogenic allele of Ras led to the tumorigenic cell transformation of normal human primary fibroblasts (18); as such, we investigated whether HMGA2-mediated hTERT modulation can replace hTERT during the cellular transformation process. First, SV40 large T, whose expression inactivates p53 and pRB, was introduced into IMR-90 cells, followed by combinations of HMGA2, hTERT, and RasV12 using lentiviruses. For each infection, parallel cultures were infected with vector control lentiviruses. The ectopic expression of these transgenes and HMGA2-modulated hTERT expression were confirmed by Western blot or real-time RT-PCR analyses (Fig. 7 A). The ability of IMR-90 cells with different combinations of transgenes to grow in an anchorage-independent manner, one of the hallmarks of the tumorigenic state, then was assessed. Cells expressing vector control, large T, large T with RasV12, hTERT, or HMGA2 failed to form colonies in soft agar (Fig. 7B). Consistently with Weinberg et al. (18), efficient colony formation was observed in cells expressing the combination of large T, hTERT, and RasV12 (Fig. 7B). When hTERT was replaced by HMGA2, colonies also were seen, albeit in lower numbers (Fig. 7B), suggesting a direct effect by the HMGA2-mediated hTERT modulation. Furthermore, a slight increase in the number of colonies was observed in cells expressing large T, hTERT, Ras, and HMGA2 (Fig. 7B), implicating an additive effect by the expression of HMGA2. Although the effect of HMGA2, in combination with SV40 large T and RasV12, on IMR-90 transformation was less robust than that of hTERT in the same context, these results indicate that the identified HMGA2-mediated hTERT alteration contributes, at least in part, to the progress of tumorigenesis.

Fig. 7.

Ectopic expression of HMGA2 in combination with SV40 large T and RasV12 lead to the tumorigenic conversion of normal human fibroblasts. (A) hTERT mRNA (lower) and SV40 large T, RasV12, and HMGA2 protein levels (upper) were confirmed by real-time RT-PCR and Western blot analyses, respectively. The anti-H-Ras antibody recognizes both wild-type and mutated Ras. (B) Soft-agar assays were performed to assess the anchorage-independent growth of IMR-90 cells transduced with combinations of large T, hTERT, HMGA2, and RasV12. After 20 days, colonies greater than 0.2 mm in diameter were tabulated. Results represent means ± standard deviations from three independent experiments. (C) Model of hTERT activation by HMGA2. In the absence of HMGA2, HDAC2 is recruited by Sp1 to the hTERT proximal promoter to repress hTERT transcription. However, during HMGA2 overexpression, HMGA2 interacts with Sp1 and interferes with the recruitment of HDAC2 to the hTERT proximal promoter, leading to localized histone H3-K9 hyperacetylation, hence the transcription activation of hTERT.

DISCUSSION

To date, the mechanism by which the hTERT gene is modulated in cancers still represents a fundamental unsolved question, and telomere maintenance is essential for tumorigenesis. HMGA2 expression is reported to be upregulated in a number of human cancers, and HMGA2 is a multifunctional architectural transcriptional factor that plays a role in transcriptional regulation throughout the genome (15). In this report, we investigated the link between HMGA2 protein and the modulation of telomerase through hTERT transcriptional regulation in multiple paradigms. Our findings are 2-fold: first, HMGA2 plays a direct role in stimulating telomerase activities; second, HMGA2 modulates hTERT expression through a protein-protein interaction with Sp1, which interferes with the occupancy of HDAC2 at the hTERT proximal promoter, leading to increased histone H3-K9 hyperacetylation, a hallmark of transcriptional activation. The overexpression of HMGA2 protein was reported to be associated with progression and/or metastasis in several cancers and also served as a diagnostic molecular marker (4, 13, 14, 36, 39, 55, 62). A recent HMGA interactome study has revealed TBP-associated factor 3 (TAF3) and chromatin assembly factor 1 p150/CAF-1 as new protein partners of HMGA2, suggesting the influence of HMGA2 on the general aspects of transcription and involvement in chromatin remodeling and dynamics through protein-protein interaction (38). Our results disclosed a novel role of HMGA2 in modulating hTERT transcription to avoid telomere attrition and crisis as a potential contributor to the multistep process of the tumorigenesis pathway.

We propose a model for how HMGA2 modulates hTERT transcription. In the absence of HMGA2, HDAC2 is recruited to the hTERT proximal promoter via Sp1 to repress hTERT transcription. In contrast, when HMGA2 is overexpressed and recruited to Sp1 located at the −276/−83 hTERT promoter region, it results in a decreased occupancy of HDAC2 at the −276/−83 hTERT promoter region (Fig. 4A). Consequently, the decreased recruitment of HDAC2 leads to a localized increase of histone H3-K9 acetylation and eventual transcriptional modulation of hTERT expression (Fig. 7C). Given the prominent role of Sp1 in the transcriptional control of hTERT (25, 61, 63) and the frequent overexpression of HMGA2 in various cancers (reviewed in references 9, 15, and 52), it is likely that HMGA2-dependent pathways represent one of the means to modulate the regulation of hTERT expression in cancer cells. A role for HMGA2-induced changes of histone modification in the regulation of hTERT expression also is consistent with results from previous studies in which hTERT was repressed in most somatic cells through mSin3A/histone deacetylase (64) and in which the inhibition of HDAC activity in telomerase-negative cells altered chromatin structure and induced hTERT expression (10, 57, 58). Our results further demonstrated that the interference of HDAC2 recruitment caused by HMGA2 to the hTERT promoter was the key in leading to the H3-K9 hyperacetylation and modulation of hTERT in cancer cells. Recently, it has become apparent that the epigenetic regulation of the telomeric chromatin template critically affects telomere function and telomere length homeostasis from yeast to humans (3, 33). Telomeric repeats across all species carry features of repressive chromatin, and the disruption of this silent chromatin environment results in the loss of telomere-length control and increased telomere recombination. Therefore, it remains to be investigated whether the observed telomere lengthening in HMGA2-expressing cells also is contributed by HMGA2-mediated chromatin modification at telomeres in addition to its induction of hTERT transcription.

Emerging evidence has suggested that hTERT has additional functions beyond its major role in telomere maintenance (reviewed in reference 21). For example, the expression of hTERT is involved in accelerating cell proliferation and tumor development (46, 56), enhancing cell protection from various cytotoxic stresses (35, 66) or oxidative insult (50), reducing basal cellular reactive oxygen species (ROS) levels as well as inhibiting endogenous ROS production in response to stimuli (28), maintaining pluripotency and human embryonic stem cell differentiation to extraembryonic and embryonic lineages (50, 65), and altering gene expression profiles (16). A significant correlation was identified between HMGA2 and hTERT expression levels in 34 breast cancer patient samples (see Fig. S1B in the supplemental material), indicating the clinical significance of HMGA2-mediated hTERT modulation in the contribution of tumor progression. Consistently with our report, Johnson et al. have identified a correlation between the chromosomal amplification of 12q14.3, where HMGA2 is localized, and the activation of hTERT to stabilize the telomeres in liposarcomas (30). Moreover, a number of studies have demonstrated that higher hTERT mRNA levels in breast cancers were associated with a high tumor grade, increased invasiveness, and poor prognosis, such as recurrent disease or decrease in overall survival rate (24, 41, 49). A recent study also has shown that telomerase inhibition using the potent telomerase inhibitor GRN163L depletes cancer stem cells, rare drug-resistant cancer cell subsets proposed to be responsible for the maintenance and recurrence of cancer and metastasis, in breast and pancreatic cancer cell lines (31). These reports further support the importance for connecting HMGA2 expression to hTERT modulation (30).

Recently, Narita et al. reported that HMGA2 proteins accumulate on chromatin in senescent fibroblasts and are essential structural components of senescence-associated heterochromatic foci (43). It seems contradictory that, being an activator of the immortalizing enzyme telomerase, HMGA2 proteins also induce senescence. Human fibroblasts are known to undergo senescence in response to oncogenic proteins, such as Ras^V12 and E2F1, so-called oncogene-induced senescence (OIS) (8, 51). Up to now, OIS has been considered a potential antitumor barrier, functioning mainly at the premalignant stage to prevent the proliferation of cells with oncogenic mutations (8, 17, 37). In contrast, the high-grade preneoplastic or cancerous cells might execute another pathway(s) to overcome the counterbalance by the OIS to stimulate proliferation or escape senescence. Furthermore, while the study by Narita et al. demonstrates that HMGA2 and p16^Ink4a induce senescence-associated heterochromatin foci in a mutually reinforcing manner (43), another report by Nishino et al. has shown an important role for HMGA2 in promoting neural stem cell self renewal in age-related mice by reducing the expression of p16^Ink4a/p19Arf (45). Given that the positive impacts by hTERT expression on preventing telomere erosion (6, 27, and this report), stimulating cell proliferation (46, 56), and protecting cells against cytotoxic or oxidative stress (35, 50, 66) highlight the possibility that HMGA2-triggered hTERT modulation serves as a catalyst by creating a permissive environment to promote oncogenic transformation when it occurs in the appropriate context.

Based on our results and the above discussion, we propose that among its pleiotrophic transcriptional regulation of many genes, the role for HMGA2-mediated hTERT modulation permits cells with existing mutations in replicative senescence effector genes, such as p53 or RB (53), to escape from the replication arrest imposed by telomere shortening or the senescence response by oncogenes. This notion is supported by the ability of IMR-90 cells expressing large T, HMGA2, and RasV12 to form colonies in soft agar (Fig. 7B). While forced HMGA2 expression in presenescent IMR-90 cells increased telomerase activities (Fig. 1C), these cells did not become immortalized and displayed features of senescence, such as elevated senescence-associated β-galactosidase activity compared to that of the vector control (see Fig. S2A in the supplemental material), after only about two population doublings, which is consistent with the results reported by Narita et al. (43). However, when SV40 large T was first introduced into IMR-90 cells followed by the transduction of HMGA2, the senescence phenotype was rescued (see Fig. S2B in the supplemental material). When such a permissive niche was created, further oncogenic stress induced by RasV12 enabled these cells to form colonies in soft agar (Fig. 7B), suggesting the ability of HMGA2 to promote oncogenesis when it is in the correct context. The HMGA2-mediated hTERT modulation accounts for only about 20% of the hTERT expression level of that of the hTERT transgene (Fig. 7A); however, this was sufficient to confer on cells the ability to form colonies in soft agar (Fig. 7B). The sustained proliferation of cells with oncogenic mutation and/or abnormalities is a major step toward tumorigenicity (20), and by modulating their telomerase activity, these cells likely become immortalized and achieve the ability to propagate their accumulated mutations. This observation is consistent with a report that the frequent amplification of hTERT in chromosome 5p15.33 in lung cancer is associated with the upregulation of telomerase activity in these cells (60). Our observation that the knockdown of HMGA2 by shHMGA2 leads to the attenuated ability of HepG2 cells to form colonies in soft agar (Fig. 5C) and the erosion of telomeres (Fig. 6B) further corroborate previous reports that the inhibition of telomerase in tumor cells results in telomere shorting and eventual cell proliferation failure.

In summary, our results indicate that (i) the overexpression of HMGA2 promotes functional telomerase activity, (ii) the knockdown of HMGA2 shortens telomere length in HepG2 cells and attenuates their ability to form colonies in soft agar, (iii) by coexpressing cooperating oncogenes, such as SV 40 large T and RasV12, HMGA2 partially replaces the function of hTERT during the tumorigenic transformation of normal human fibroblasts, and (iv) HMGA2 binds the hTERT proximal promoter region via protein-protein interaction with Sp1, interfering with the recruitment of HDAC2, and this leads to a localized histone H3-K9 hyperacetylation. Taken together, our data provide mechanistic insight into the novel role of HMGA2 in promoting tumorigenesis together with therapeutic potential.