Transcriptional Regulation of the HMGA1 Gene by Octamer-Binding Proteins Oct-1 and Oct-2

The High-Mobility Group AT-Hook 1 (HMGA1) protein is an architectural transcription factor that binds to AT-rich sequences in the promoter region of DNA and functions as a specific cofactor for gene activation. Previously, we demonstrated that HMGA1 is a key regulator of the insulin receptor (INSR) gene and an important downstream target of the INSR signaling cascade. Moreover, from a pathogenic point of view, overexpression of HMGA1 has been associated with human cancer, whereas functional variants of the HMGA1 gene have been recently linked to type 2 diabetes mellitus and metabolic syndrome. However, despite of this biological and pathological relevance, the mechanisms that control HMGA1 gene expression remain unknown. In this study, to define the molecular mechanism(s) that regulate HMGA1 gene expression, the HMGA1 gene promoter was investigated by transient transfection of different cell lines, either before or after DNA and siRNA cotransfections. An octamer motif was identified as an important element of transcriptional regulation of this gene, the interaction of which with the octamer transcription factors Oct-1 and Oct-2 is crucial in modulating HMGA1 gene and protein expression. Additionally, we demonstrate that HMGA1 binds its own promoter and contributes to its transactivation by Oct-2 (but not Oct-1), supporting its role in an auto-regulatory circuit. Overall, our results provide insight into the transcriptional regulation of the HMGA1 gene, revealing a differential control exerted by both Oct-1 and Oct-2. Furthermore, they consistently support the hypothesis that a putative defect in Oct-1 and/or Oct-2, by affecting HMGA1 expression, may cause INSR dysfunction, leading to defects of the INSR signaling pathway.


Introduction
HMGA1 is an architectural transcription factor that specifically interacts with the narrow minor groove of AT-rich regions of DNA and regulate gene expression [1,2,3]. By modifying DNA conformation and by recruiting transcription factors to the transcription start site, it facilitates the assembly and stability of stereospecific DNA-protein complexes (so-called enhanceosomes), thus driving gene transcription in response to multiple extracellular and intracellular signals [4,5,6]. In addition to its role in gene transcription, HMGA1 is also involved in numerous other processes, including embryogenesis, differentiation, and neoplastic transformation [7,8,9]. Recent evidence assigns to HMGA1 an important role in the transcriptional regulation of glucose homeostasis [10,11,12,13,14]. Consistent with these latter findings, we have shown that defects in HMGA1 expression and/or function, by negatively affecting insulin receptor (INSR) signaling in insulin target tissues, may cause insulin resistance and increase susceptibility to type 2 diabetes mellitus [11,15]. Also, evidence has been provided implicating the HMGA1 locus as one conferring a high cross-race risk of development of type 2 diabetes and metabolic syndrome [16,17,18,19]. However, despite this variety of studies, the mechanisms that regulate HMGA1 gene expression are mostly unknown.
Cloning and characterization of the HMGA1 gene reveals a very complex structure [2,20], with a regulatory region comprising the untranslated exons 1-3, lacking TATA and CAAT, but rich in G and C. Its complexity includes at least two transcription start sites and an extensive alternative mRNA splicing, particularly in the 59untranslated region [2,20], resembling the genes that are regulated without the preferential utilization of any specific start site [21]. Instead, a preferential utilization of start site 2 has been demonstrated under certain experimental conditions, in which tight gene regulation is achieved, following the production of specific mRNA in response to different stimuli [20]. A mechanism of transcriptional regulation of the human HMGA1 gene has been postulated before both in embryonic and cancer cells highly expressing HMGA1 [22]. It has been proposed that elevated expression of HMGA1 in these cell models is mainly controlled by specificity protein 1 (Sp1) and activator protein-1 (AP-1) transcription factors, both of which seem to act as positive regulating elements, after binding to the transcription start site 1 and the transcription start site 2, respectively [22]. In the current study, we sought to further investigate the mechanisms that underlie HMGA1 gene expression in different human and rat cell lines widely used in biochemical and molecular genetic studies of gene function and regulation. We report herein the identification of a new mechanism of regulation of the HMGA1 gene, which involves the octamer DNA motif ''ATGCAAAT'' originally described in immunoglobulin (Ig) gene regulation [23]. As a cisacting transcriptional regulatory element found in both promoters and enhancers of a wide variety of ubiquitously expressed and cell type-specific genes, this octamer motif is specifically bound by a family of at least 11 distinct nuclear proteins, named octamerbinding (Oct) transcription factors, designated from Oct-1 to Oct-11 [24]. All these factors are characterized by two Pit-Oct-Unc (POU) domains, in which the N-terminal subunit is known as the POU-specific (POUS) domain, while the C-terminal subunit is a homeo domain (POUH). Both domains are required for highaffinity, site-specific binding to the octamer motif, and mediate specific protein-protein interactions with other transcription factors or co-factors, thus activating or inhibiting gene expression [24]. Here, we focused our attention on the regulation of HMGA1 gene transcription by the octamer-binding proteins Oct-1 and Oct-2, whose ubiquitous expression differ from other Oct proteins that exhibit, instead, a developmental and tissue-specific distribution [24]. Our data indicate a role for Oct-1 and Oct-2 in regulating HMGA1 expression through direct interaction with the upstream HMGA1 promoter region.

Cells, protein extracts and Western Blot (WB)
HepG2 human hepatoma cells, and human cervix epithelioid carcinoma cells (HeLa) were maintained in DMEM (Invitrogen) supplemented with 10% fetal bovine serum (FBS). IM-9 cultured human lymphocytes (ATCC) were maintained at subconfluent densities in RPMI 1640 medium supplemented with 10% FBS, 2 mM glutamine, penicillin (100 U/ml), and streptomycin (100 mg/ml) in a humidified 5% CO 2 atmosphere at 37uC. L6 muscle rat cells were differentiated in DMEM containing 2% FBS, penicillin (100 U/ml), streptomycin (100 mg/ml), and 2 mM glutamine in a humidified 5% CO 2 atmosphere at 37uC. The medium was replaced every 3 days until myotube formation was reached. Total cellular lysates or nuclear extracts were prepared as previously [25] and final protein concentrations in the extracts were determined using the colorimetric assay of Bradford (Bio-Rad). WB analysis was performed on total cellular lysates or nuclear extracts as previously described [26]. The antibodies used for these studies were: anti-HMGA1 [26], anti-Oct-1 and anti-Oct-2 (Santa Cruz Biotech).

Chromatin immunoprecipitation (ChIP)
ChIP was performed in HepG2 cells, either untreated or pretreated with Oct-1 or Oct-2 small interfering RNA (siRNA), as described previously [13]. Formaldehyde-fixed DNA-protein complex was immunoprecipitated with either anti-Oct-1 or Oct-2 antibody. Sequence-specific primers for the HMGA1 gene promoter [human HMGA1 (NC_000006.11): HMGA1 forward 59-TGTACCAGGGAGGAAGGATAC-39 and HMGA1 reverse 59-ATGGCGAGGGGCAGGAAGCTGG-39] were used for PCR amplification of immunoprecipitated DNA (30 cycles), using PCR ready-to-go beads (GE Healthcare). PCR products were electrophoretically resolved on 1.5% agarose gel and visualized by ethidium bromide staining. Specifity of PCR product was confirmed by direct sequencing. Primers for quantitative reverse transcriptase PCR (qRT-PCR) of ChIP-ed samples are available upon request.

Transfection studies
The HMGA1-luciferase (HMGA1-luc) reporter plasmid was obtained by cloning the NheI/XhoI 1427-bp sequence fragment of the human HMGA1 gene promoter into pGL3-basic vector (Promega). Recombinant HMGA1-luc reporter construct, in the presence or absence of expression vectors for Oct-1 or Oct-2 [29], were transiently transfected into cultured cells using LipofectA-MINE 2000 reagent (Invitrogen), and luc activity was assayed 48 hrs later, using the dual-luciferase reporter assay system (Promega) [30]. Renilla control vector served as an internal control of transfection efficiency, together with measurements of protein expression levels. siRNA targeted to human HMGA1, Oct-1 or Oct-2, plus nonspecific siRNA controls with a similar GC content (Invitrogen) were transfected into cells at 50% to 60% confluency and cells were analyzed 48 to 96 hrs later, as described previously [31].

Real-time PCR
For qRT-PCR, total cellular RNA was extracted from cells using the RNAqueous-4PCR kit and subjected to DNase treatment (Ambion). RNA levels were normalized against 18S ribosomal RNA in each sample, and cDNAs were synthesized from 2 mg of total RNA using the RETROscript first strand synthesis kit (Ambion). Primers for human and rat HMGA1, Oct-1, Oct-2, INSR, Sp1 and RPS9 were designed according to sequences from the GenBank database. A real-time thermocycler (Eppendorf Mastercycler ep realplex ES) was used to perform quantitative PCR. SYBR Green fluorescence was measured, and relative quantification was made against the RPS9 cDNA used as an internal standard [12]. All PCR reactions were done in triplicate.

Statistical analysis
All calculations were performed with the SPSS 20.0 software (SPSS Inc.). Mean values were compared with t-tests. A p-value ,0.05 (two-tailed) was considered significant.

Putative binding sites on the HMGA1 gene promoter
In an attempt to identify the mechanism(s) that regulate HMGA1 gene expression, we initially performed functional experiments aimed at identifying protein-DNA interactions within the P1 to P8 sequence elements of the HMGA1 gene promoter ( Figure 1). Consensus transcription factor-binding sites within these eight regions were explored by using the TRANSFAC database (containing a large library of predefined matrix descriptions for known transcription factor recognition sequences) searched with MatInspector professional software [32]. Putative binding sites for the octamer-binding proteins Oct-1 and Oct-2, and HMGA1 were identified and characterized by EMSA using homologous radiolabeled fragments. As shown in Table 1, the octamer motif ''ATGCAAAT'' and the flanking sequence at both sides appeared highly conserved in evolution and located in the proximal promoter region of the HMGA1 gene.
Nuclear protein-DNA interaction within the HMGA1 promoter region As shown in Figure 2A, DNA-protein binding activity was observed following incubation of either probe P3 or P5 with nuclear extracts from IM-9 cells, a human B lymphocyte cell line   Figure 2B). Moreover, mutations within the consensus sequences abolished DNA-protein binding activity of either probe P3 or P5 ( Figure 2B). As shown in Figure 3A, a 30-fold molar excess of 22-mer double stranded oligonucleotides representing the consensus binding sites for the nuclear factors AP-1, AP-2, AP-3, Sp1 and N-Myc did not compete for binding of nuclear proteins to P3 and P5 probes. Instead, unlabeled competitor DNA for the octamer motif ''ATGCAAAT'', com-pletely abolished protein-DNA complex formation with P3 probe ( Figure 3A), indicating that the POU domain is essential for the formation of a DNA-protein complex at this site. On the other hand, binding of HMGA1 to P5 probe was displaced by excess unlabeled double-stranded oligonucleotides containing the PRDII element motif (GGGAAATTCCGTGGGAAATTCCGAGCT) ( Figure 3A), which, as reported previously by us and others, specifically binds HMGA1 in the minor groove of the central ATrich region [4,26,28], thus suggesting a potential autoregulatory role of HMGA1 on its own promoter. Next, we showed that the protein-DNA complexes formed upon incubation of nuclear extracts from IM-9 cells with labeled P3 were recognized and supershifted to a slower-migrating form by the anti-Oct-1 or anti-Oct-2 antibody ( Figure 3B); instead, binding of HMGA1 to labeled P5 produced a protein-DNA complex that was recognized and supershifted by the anti-HMGA1 antibody ( Figure 3B). Binding of Oct-1 and Oct-2 to P3 sequence was substantiated further by ChIP coupled with qRT-PCR of ChIP-ed samples, showing that binding of both Oct-1 and Oct-2 to the endogenous HMGA1 chromosomal locus was present in living IM-9 cells naturally expressing Oct-1 and Oct-2, and was considerably decreased in cells exposed to siRNA against either Oct-1 or Oct-2 ( Figure 3C).

HMGA1 is necessary for recruitment and binding of Oct-2 to the HMGA1 promoter
Physical association between HMGA1 and Oct-1 and Oct-2 has been reported previously in pull-down assays indicating that direct contact of Oct-1 and Oct-2 with HMGA1 occurs through their conserved POU homeodomains [33]. In the light of the abovementioned data indicating that a relationship may exist between these factors, to see whether HMGA1 had any influence on the binding of Oct-1 and Oct-2 to the HMGA1 promoter, we carried out ChIP experiments in living IM-9 cells, either untreated or pretreated with siRNA against HMGA1. As detected by qRT-PCR of ChIP-ed samples, whereas binding of Oct-1 to the HMGA1 gene promoter was not affected in IM-9 cells expressing reduced levels of endogenous HMGA1 (Figure 4A), the binding of Oct-2 was significantly decreased in cells exposed to siRNA against HMGA1 ( Figure 4B), thus indicating that HMGA1 is required for binding of Oct-2, but not Oct-1, to the HMGA1 gene promoter in vivo, in intact cells.

Oct-1 and Oct-2 differentially regulate HMGA1 gene promoter
To test whether Oct-1 and Oct-2 had a functional role in regulating the HMGA1 gene at the transcriptional level, human HeLa and HepG2 cells and rat skeletal muscle cells (differentiated from L6 myoblasts), three cell lines naturally expressing HMGA1, were cotransfected transiently with the HMGA1-luc reporter plasmid, plus increasing amounts of Oct-1 or Oct-2 expression vector. As shown in Figure 5A, overexpression of Oct-1 significantly reduced HMGA1-luc activity in all cell lines and this effect occurred in a dose-dependent manner. In contrast, overexpression of Oct-2 caused a significant increase of HMGA1-luc activity in all these cells ( Figure 5B), supporting the notion that, by binding the same consensus site in the HMGA1 promoter, Oct-1 and Oct-2 exert an opposite effect on the HMGA1 gene: while the former decreases the transcriptional activity and protein level of HMGA1, the latter increases them.

Oct-1 and Oct-2 may affect INSR expression
Based on the above data indicating that interaction of Oct-1 and Oct-2 with the HMGA1 gene promoter may play a role in HMGA1 expression, it was important to ask whether this interaction could have an effect on the expression of the INSR gene, a well-characterized molecular target for HMGA1 which , or HMGA1 (lane 11), before addition of the probe. Supershifting of the Oct-1-, Oct-2-and HMGA1-DNA complexes are shown with an arrowhead. Control (unrelated rabbit serum IgG) antibody did not alter the mobility of the complexes (lanes 2, 6, 10). Lane 3, IM-9 NE plus an excess of unlabeled octamer motif. Free (DNA) and bound (DNA-P) probes are indicated. A representative of three separate assays is shown for each condition. (C) ChIP of the HMGA1 gene promoter in IM-9 cells, untreated or pretreated with either Oct-1 (left) or Oct-2 (right) siRNA, using an anti-Oct-1 or anti-Oct-2 specific antibody (Ab), respectively. Representative assays are shown, together with qRT-PCR of ChIP-ed samples. *p,0.05 versus control (white bar). doi:10.1371/journal.pone.0083969.g003 causes an increase in intracellular INSR in cells and tissues [10,11,26]. For this purpose, the ability of Oct-1 and Oct-2 to drive INSR gene expression was measured in HepG2 cells, in which protein levels of Oct-1 or Oct-2 were increased by transfecting cells with plasmids for expression of Oct-1 or Oct-2. As shown in Figure 6A, overexpression of Oct-1 in HepG2 cells caused a decrease in endogenous HMGA1 mRNA and protein levels and this was followed by a decline in INSR mRNA and protein expression. The opposite was observed when HepG2 cells were transfected with an expression vector coding for Oct-2: in this case, a significant increase was observed in the levels of HMGA1 mRNA and protein abundances, which were paralleled by increased levels of mRNA and protein for the INSR ( Figure 6A). These effects were specific for HMGA1, as both mRNA and protein expression for the ubiquitously expressed transcription factor Sp1 were unaffected under both conditions. Similar results were obtained in differentiated L6 skeletal muscle cells naturally expressing INSR ( Figure 6B). Thus, these data indicate that Oct-1 and Oct-2, by differentially modulating HMGA1 gene transcrip-tion, may affect INSR gene and protein expression and perhaps the expression of other HMGA1-target genes.

Discussion
For the first time in this work we identified and characterized a novel molecular mechanism that contribute to the regulation of HMGA1 gene expression. Using DNA-protein interaction studies combined with ChIP coupled with qRT-PCR, we demonstrated a differential role for Oct-1 and Oct-2 transcription factors in the transcriptional regulation of the HMGA1 gene which, in addition to the regulation of HMGA1 mRNA and protein expression, may contribute to the modulation of HMGA1-target genes. The observation that Oct-1 and Oct-2 differentially affect HMGA1 gene expression by binding to the same octamer motif ''ATG-CAAAT'' is consistent with previous reports, in which it has been demonstrated that both Oct-1 and Oct-2 bind to the same DNA sequence but may differ in their activation potential [29,34,35,36]. Furthermore, here we show that, by facilitating the interaction of Oct-2 (but not Oct-1) with the HMGA1 gene, HMGA1 enhances transactivation of its own promoter by this octamer transcription factor, providing evidence for the existence of an auto-regulatory circuit in which HMGA1 activates its own transcription. In relation to Oct-1, it needs to be considered that several isoforms of this transcription factor were described previously, each of which is active in a tissue-specific manner, being able to mediate different and even opposing effects [24]. Thus, we cannot exclude that in other cells the effect of Oct-1 on HMGA1 may be different from that reported here in our work. In this regard, results have been provided indicating that Oct-1, NF-kappa B and HMGA1 cooperate for transactivation of the iNOS promoter in pancreatic b-cells [37].
As we previously reported, HMGA1 is a main activator of the INSR gene [10,26], and individuals with defects in HMGA1 have decreased INSR expression and increased susceptibility to type 2 diabetes mellitus [11,15,16,18]. Also, we showed that, as a novel downstream nuclear target of the INSR signaling pathway [13], HMGA1 represents an important novel mediator of glucose disposal and glucose homeostasis [12,14]. Therefore, based on these observations and the results from the present study indicating that a functional relation can be established between HMGA1, Oct-1 and Oct-2 in the context of the INSR gene, it is tempting to hypothesize that a putative defect in these nuclear proteins may cause INSR dysfunction with subsequent impairment of insulin signaling and action. In this regard, an association between Oct-1 gene variants and type 2 diabetes mellitus, which may support this intriguing hypothesis, has been reported [38]. Moreover, recent evidences suggest that Oct-1 may play a role in metabolic homeostasis by acting as a sensor for cAMP (cyclic AMP). In particular, elevation of cAMP in pancreatic and intestinal endocrine cells reduces nuclear Oct-1 content [39,40], and this is consistent with previous observations indicating that expression of HMGA1 increases in response to cAMP [12,41], thus suggesting that increased HMGA1 expression could be due, at least in part, to the reduction of Oct-1 content.
In conclusion, our findings from this study reveal new mechanistic insight into the transcriptional mechanisms governing HMGA1 gene expression. While additional investigations will be needed to further define the roles of Oct-1 and Oct-2 in HMGA1 expression in other cell systems and models, these results provide a paradigm to suggest other types of molecule, Oct-1 and Oct-2, in addition to HMGA1, which should be searched for in investigations designed to elucidate the causes of impairment of INSR signaling in humans with INSR dysfunction and insulin resistance.