High Mobility Group A1 Is a Molecular Target for MYCN in Human Neuroblastoma Giuseppe Giannini, 1 Fabio Cerignoli, 1 Massimiliano Mellone, 1 Isabella Massimi, 1 Cinzia Ambrosi, 1 Christian Rinaldi, 3 Carlo Dominici, 2 Luigi Frati, 1,3 Isabella Screpanti, 1 and Alberto Gulino 1,3 Departments of 1 Experimental Medicine and Pathology and 2 Pediatrics, University ‘‘La Sapienza,’’ Rome, Italy and 3 Neuromed Institute, Pozzilli, Italy Abstract High mobility group A1 (HMGA1) is an architectural tran- scription factor and a putative protoncogene. Deregulation of its expression has been shown in most human cancers. We have previously shown that the expression of the HMGA family members is deregulated in neuroblastoma cell lines and primary tumors. On retinoic acid (RA) treatment of MYCN- amplified neuroblastoma cell lines, HMGA1 decreases with a kinetics that strictly follows MYCN repression. In addition, MYCN constitutive expression abolishes HMGA1 repression by RA. Here we explored the possibility that HMGA1 expression might be sustained by MYCN in amplified cells. Indeed, MYCN transfection induced HMGA1 expression in several neuroblas- toma cell lines. HMGA1 expression increased in a transgene dose–dependent fashion in neuroblastoma-like tumors of MYCN transgenic mice. In addition, it was significantly more expressed in MYCN-amplified compared with MYCN single- copy primary human neuroblastomas. MYCN cotransfection activated a promoter/luciferase reporter containing a 1,600 bp region surrounding the first three transcription start sites of the human HMGA1 and eight imperfect E-boxes. By heterodimerizing with its partner MAX, MYCN could bind to multiple DNA fragments within the 1,600 bp. Either 5V or 3V deletion variants of the 1,600 bp promoter/luciferase reporter strongly decreased luciferase activity, suggesting that, more than a single site, the cooperative function of multiple cis - acting elements mediates direct HMGA1 transactivation by MYCN. Finally, HMGA1 repression by RNA interference reduced neuroblastoma cell proliferation, indicating that HMGA1 is a novel MYCN target gene relevant for neuroblas- toma tumorigenesis. (Cancer Res 2005; 65(18): 8308-16) Introduction The high mobility group (HMG) proteins of the A type belong to a larger family of nonhistone DNA binding factors that play important architectural functions in the organization of active chromatin (1). The HMGA family is composed of three members: HMGA1a and HMGA1b, encoded by the same gene through an alternative splicing of a short exon, and HMGA2 (2). By means of specific DNA binding domains, called AT-hooks, HMGA proteins bind to AT-rich sequences in the minor groove of the DNA helix (3) and coordinate the assembly of higher-order multiprotein complexes (called enhanceosomes; refs. 4–6) involved in the regulation of the expression of a growing number of genes (reviewed in ref. 7). HMGA proteins are strongly and almost ubiquitously expressed in early mammalian development (8). Their expression declines through development and is scant or completely absent in adult differentiated tissues (9, 10). They are thought to be important regulators of cell growth and differenti- ation. Indeed, the HMGA2 knockout mouse shows a ‘‘pigmy’’ phenotype, characterized by reduced size and weight of most body organs, reduced body fat, and a cell-autonomous defect in cell growth (8). Conversely, a mouse overexpressing a shortened HMGA2 develops a ‘‘giant’’ phenotype with diffused lipomatosis, underlying the role of this molecule in the regulation of (adipocytic) cell growth and differentiation (11, 12). Adipocytic differentiation also requires HMGA1 up-regulation (13). Apart from their role in physiologic processes, deregulated HMGA1 and/or HMGA2 ex- pression has been described in most tumors of epithelial and mesenchymal origin and is considered a hallmark of cancer (14). In particular, HMGA1 expression was proposed to be a diagnostic indicator of carcinoma in thyroid and colorectal cancer, and higher levels of expression associate with more malignant and metastatic phenotypes in epithelial cancers (15–17). In experimental models, HMGA1 behaves as a transforming protein and has been indicated as a putative oncogene. Consistently, its repression strongly impairs the malignant phenotype of Burkitt’s lymphoma cells (18) and the tumorigenicity of several human cancer cell lines (19). Despite its established role in tumor biology, very little is known about the molecular mechanisms involved in its deregulation in cancer cells. We have reported on the expression of HMGA genes in neuroblastoma (20, 21). Neuroblastoma is a heterogeneous neoplasm in which different clinical, molecular, and genetic variables may contribute to prognostic stratification of the patients. Indeed, the expression of TRK receptor family members, gain of chromosome 17q, and loss of 1p and 11q were shown to affect tumor behavior (reviewed in ref. 22). The most powerful prognostic factor of neuroblastoma is MYCN amplification, which occurs in 20% to 25% of the tumors and predicts a negative outcome (23–25). Transgenic mice with the neural crest targeted expression of this oncogene develop neuroblastoma-like tumors (26). MYCN inactivation via an antisense strategy leads to decreased neuroblastoma cell proliferation and reduced anchor- age-independent growth in vitro (27, 28) and decreased mouse neuroblastoma tumorigenesis in vivo (29), suggesting that it also plays an important pathogenic role. MYCN belongs to the large family of helix-loop-helix proteins and operates in a complex network of transcription factors, including MAX and MAD. In an oversimplified view, MYC proteins (including c-MYC, MYCN, and L-MYC) might be recruited in transcriptionally active or inhibitory complexes by binding to MAX and MAD, respectively. Binding of these molecules to specific elements, called E-boxes, regulates Requests for reprints: Giuseppe Giannini, Department of Experimental Medicine and Pathology, University La Sapienza, Policlinico Umberto I, Viale Regina Elena, 32400161 Rome, Italy. Phone: 39-06-4958-637; Fax: 39-06-4461-974; E-mail: giuseppe.giannini@uniroma1.it. I2005 American Association for Cancer Research. doi:10.1158/0008-5472.CAN-05-0607 Cancer Res 2005; 65: (18). September 15, 2005 8308 www.aacrjournals.org Research Article