HMGA1 Induces Intestinal Polyposis in Transgenic Mice and Drives Tumor Progression and Stem Cell Properties in Colon Cancer Cells

Background Although metastatic colon cancer is a leading cause of cancer death worldwide, the molecular mechanisms that enable colon cancer cells to metastasize remain unclear. Emerging evidence suggests that metastatic cells develop by usurping transcriptional networks from embryonic stem (ES) cells to facilitate an epithelial-mesenchymal transition (EMT), invasion, and metastatic progression. Previous studies identified HMGA1 as a key transcription factor enriched in ES cells, colon cancer, and other aggressive tumors, although its role in these settings is poorly understood. Methods/Principal Findings To determine how HMGA1 functions in metastatic colon cancer, we manipulated HMGA1 expression in transgenic mice and colon cancer cells. We discovered that HMGA1 drives proliferative changes, aberrant crypt formation, and intestinal polyposis in transgenic mice. In colon cancer cell lines from poorly differentiated, metastatic tumors, knock-down of HMGA1 blocks anchorage-independent cell growth, migration, invasion, xenograft tumorigenesis and three-dimensional colonosphere formation. Inhibiting HMGA1 expression blocks tumorigenesis at limiting dilutions, consistent with depletion of tumor-initiator cells in the knock-down cells. Knock-down of HMGA1 also inhibits metastatic progression to the liver in vivo. In metastatic colon cancer cells, HMGA1 induces expression of Twist1, a gene involved in embryogenesis, EMT, and tumor progression, while HMGA1 represses E-cadherin, a gene that is down-regulated during EMT and metastatic progression. In addition, HMGA1 is among the most enriched genes in colon cancer compared to normal mucosa. Conclusions Our findings demonstrate for the first time that HMGA1 drives proliferative changes and polyp formation in the intestines of transgenic mice and induces metastatic progression and stem-like properties in colon cancer cells. These findings indicate that HMGA1 is a key regulator, both in metastatic progression and in the maintenance of a stem-like state. Our results also suggest that HMGA1 or downstream pathways could be rational therapeutic targets in metastatic, poorly differentiated colon cancer.

Here, we report for the first time that the HMGA1 drives proliferative changes and polyp formation in the intestines of transgenic mice and directs molecular pathways important in tumor progression and stem cell properties in human colon cancer cells. Taken together, these findings suggest that HMGA1 promotes tumor progression in colon cancer by reprogramming colonic epithelium to a stem-like state.

Ethics Statement
All animal experiments were conducted in accordance with a protocol approved by the Johns Hopkins University Animal Care and Use Committee (protocol# MO08M263). All mice were housed in a sterile environment where they had free access to food and water as outlined in our institutional guidelines.
Anchorage-independent cell growth in soft agar, migration, and invasion assays. These assays were performed as we described [19][20][26][27][28], except that the migration assay was done with 60,000 cells/well and invasion assays were done with 20 ul of growth factor reduced matrigel.
Vectors. The short hairpin RNA (shRNA) interference plasmid for HMGA1 has been described [36]. The empty shRNA vector was used as a control.
Immunohistochemical analysis. Tissue sections (4 mm thick) from small and large intestines of transgenic and control mice were generated and deparaffinized in xylene and then hydrated in a graded alcohol series. Heat-induced epitope retrieval was performed in an autoclave for 10 minute in sodium citrate buffer (pH 6.0). Endogenous peroxidase activity was inactivated by incubation of the sections in 4% H 2 O 2 for 5 minutes. After rinsing in phosphate-buffered saline, the sections were incubated with rabbit monoclonal anti-Ki-67 antibody (clone 30-9, Ventana, Tucson, AZ, catalogue #790-4286) for 1 hour at room temperature. Positive staining was visualized with the DAKO EnVision Plus-HRP detection kit (DAKO, Carpinteria, CA) according to the manufacturer's instructions. Ki-67 positive cells were counted in 5 randomly selected fields from comparable areas in the small and large intestine from transgenic and control mice at 206 magnification. A total of 100 crypts for each mouse were counted from stained sections from transgenic and control mice (2 mice/group).
Cell morphology and size determinations. Cells were plated and allowed to grow to 80% confluency. Photographs for morphologic assessment and diameter measurements were obtained using the Zeiss Inverted M-scope with the Olympus DP72 Camera and Software (JHMI Confocal Facility). At least five random fields were photographed for each cell line with or without HMGA1 knock-down. The greatest diameter (mm) was measured for 700-1000 cells from each group. The greatest diameters were represented as the mean 6 the standard deviation; significance was determined using the student's t-test.

Results
HMGA1a transgenics develop polyps and expansion in the stem cell compartment To define the role of HMGA1 in tumorigenesis, we engineered transgenic mice with the murine hmga1a transgene driven by the H2K promoter and immunoglobulin mu enhancer [9,[27][28][29]. As previously reported, all mice succumb to aggressive lymphoid malignancy [9] and the females develop uterine sarcomas [28][29].
To determine if HMGA1 promotes neoplastic transformation in intestinal epithelium, we investigated the intestines of these mice.
The transgene is expressed in the small and large intestines by 4 to 7-fold above that found in controls (Fig. S1). At necropsy, the intestines are hyperemic with a thickened mucosa and increased weight compared to controls (Fig. 1A-D). Histologically, both large and small intestines exhibit marked proliferative changes, ectopic crypt formation, and polyp formation (Fig. 1C-D). There was a significant increase in Ki-67 in both the small and large intestine, a marker of proliferation ( Fig. 2A-B). The increase in crypt number suggests that these mice could have expansion of the intestinal stem cell compartment [37].

HMGA1 is enriched in human colon carcinomas
High levels of HMGA1 protein or mRNA were reported in colon cancer cell lines or primary tumors in a small pilot study [38]. To determine if HMGA1 is overexpressed in larger studies of primary colon cancers, we queried gene expression profile analysis of primary colon cancers from six independent studies through the Oncomine TM public database (Compendia Bioscience, Ann Arbor, MI). We found that HMGA1 is highly enriched in colon adenocarcinoma compared to normal tissue in 5/5 prior studies. In all of these studies, HMGA1 was among the top 10% of enriched genes, and in 4 studies, it was among the top 1-3% of enriched genes. In a pilot of 10 primary, high-grade (grade III-IV) colon cancers (a generous gift from Bert Vogelstein), we found that HMGA1 expression was increased in most colon cancer samples (8/11) compared to adjacent normal tissue from the same patient. Moreover, the mean HMGA1 expression from all tumors was increased by 3-fold compared to the adjacent control tissue (Fig. 2C). We also assessed protein expression in a subset of primary colon tumors with sufficient material and found the highest levels of HMGA1 protein in the high grade, metastatic tumors with undetectable levels in the adjacent normal tissue (Fig. 2D).
HMGA1 is required for anchorage-independent cell growth, migration, invasion, and tumorigenesis in colon cancer cells To investigate the functional role of HMGA1 in colon cancer, we inhibited HMGA1 expression in two colon cancer cell lines derived from poorly differentiated (grade 4), metastatic colon cancers (HCT116 and SW480) using an shRNA approach. Transfection of cells with an HMGA1 shRNA vector resulted in a significant, stable decrease in HMGA1 protein (Fig. 3A) in the shRNA cells compared to cells transfected with the control vector. We found that knock-down of HMGA1 blocked anchorageindependent cell growth or foci formation in soft agar, migration, and invasion in both HCT116 and SW480 cells (Fig. 3B-C). Moreover, no tumors formed in nude mice after injection of HCT116 cells with HMGA1 knock-down (10 5 or 10 4 cells), while tumors formed from the cells transfected with control vectors (Fig. 3D). There was no significant difference in growth rates in the cell lines with or without HMGA1 knock-down in vitro (Fig. S2), indicating that the HMGA1 shRNA was not toxic to the colon cancer cells. These results demonstrate that HMGA1 drives cellular properties required for both tumor initiation (anchorageindependent cell growth, tumorigenesis) and tumor progression (migration, invasion).

HMGA1-dependent stem cell properties in colon cancer cells
Because HMGA1 is enriched in stem cells [5][6][7][8] and causes changes in the mouse intestine that could be consistent with expansion in the intestinal stem cell compartment, we sought to determine if HMGA1 is involved in stem cell properties in colon cancer cells. To this end, we assessed three-dimensional colonosphere formation in the colon cancer cell lines with or without knock-down of HMGA1. We discovered a significant decrease in the number of colonospheres in both the HCT116 and SW460 cells with HMGA1 knock-down compared to controls (Fig. 4A).
These findings indicate that HMGA1 is necessary for this stem cell phenotype (three-dimensional growth) in colon cancer cells. In limiting dilution tumorigenicity experiments, knock-down of HMGA1 blocked tumor formation when 104 or 105 cells were injected, whereas tumors formed in the control cells. Tumors formed in both control and knock-down groups if 106 cells were injected (Fig. 3D). These results indicate that knock-down of HMGA1 depletes the tumor-initiator cells.

HMGA1 is required for metastatic progression in colon cancer
To determine if HMGA1 is required for tumor progression, we used a preclinical model for metastatic progression in colon cancer [35]. HCT116 colon cancer cells were injected directly into the spleens of immunodeficient mice and liver metastases were assessed after 5 weeks. Strikingly, we found a marked decrease in metastatic foci in the livers from mice injected with HMGA1 knock-down cells (Fig. 4B-C). These results underscore the critical role of HMGA1 in metastatic progression in this model. Because recent studies link EMT to epithelial stem cell properties [3][4], we sought to determine if HMGA1 regulates EMT genes in colon cancer. To this end, we assessed expression levels of 11 genes previously shown to play an important role in EMT and cancer initiator cells [4]. In the HCT116 cells, we found that both Twist1 and Vimentin were significantly repressed in cells with HMGA1 knock-down (Fig. 5A), indicating that HMGA1 induces their expression. In SW480 cells, E-cadherin is up-regulated in the knock-down cells, suggesting that HMGA1 normally represses its expression (Fig. 5B). Twist1 was also significantly repressed in the SW480 knock-down cells, although less than what we observed in the HCT116 cells. Taken together, our studies suggest that HMGA1 promotes tumor progression through transcriptional networks that facilitate metastatic progression and a stem-like state, at least in part, by inducing the Twist1 and Vimentin, while repressing E-cadherin. Of note, there were no changes in cell morphology or size in the knock-down cells (Fig.  S3), suggesting that down-regulation of HMGA1 in these cells is not sufficient to induce morphologic changes consistent with a reversion to a more epithelial state when grown as a monolayer. As noted previously, there were significant alterations in growth characteristics when grown in an anchorage-independent fashion or after implantation into mice. The variation in EMT genes modulated by HMGA1 in these two different cell lines likely reflects the differing milieu and reprogramming potential in the cancer cells based on underlying genetic and epigenetic changes.

Discussion
Metastatic colon cancer is highly lethal and the incidence is rising, particularly in younger individuals [1][2]. Current therapies are limited by the emergence of metastatic cancer cells that are resistant to treatment. Recent evidence suggests that these refractory cells develop because they co-opt the cellular networks involved in embryonic development and become capable of metastasizing and evading therapy [3][4]. The HMGA1 oncogene was recently identified as a key transcription factor enriched in ES cells and high grade tumors with poor outcomes [3], although its function in these settings has remained elusive. This gene is a member of the HMGA gene family, that also includes HMGA1 [11][12][13][14][15][16][17][18] and HMGA2 [31,[39][40]. All HMGA proteins are small, low molecular weight (thus high mobility group proteins) with an AT-hook DNA binding domain that mediates binding to AT-rich regions in the minor groove of chromatin. HMGA1 and other AT-hook proteins are thought to function by modulating chromatin structure and gene expression [15][16][17][18][41][42]. Here, we show for the first time that HMGA1 induces proliferative changes and polyp formation in the intestines of transgenic mice. Moreover, HMGA1 is required for limiting dilution tumorigenesis in vivo along with 3-dimensional colonosphere formation in vitro, both of which are phenotypes characteristic of epithelial stem cells. Furthermore, inhibiting HMGA1 expression blocks not only transformation properties involved in tumor initiation (anchorage-independent cell growth, tumorigenicity), but also cellular characteristics that promote metastatic progression (invasion and migration). We also discovered that HMGA1 is required for metastatic progression to the liver in vivo. Notably, several recent studies have also shown that HMGA1 is enriched in normal stem cells, including embryonic and hematopoietic stem cells [3][4][5][6][7][8], in addition to poorly differentiated, or refractory stem-like cancers , suggesting that HMGA1 helps to drive a stem-like state, both in normal development and cancer. HMGA1 is also highly expressed during embryogenesis, with low or undetectable expression in most differentiated, adult tissues [43].
To determine how HMGA1 orchestrates tumor progression and stem cell phenotypes, we investigated expression of genes that have been previously shown to be important in EMT, development, and tumor initiator cells [4]. In HCT116 cells, we found that HMGA1 up-regulates expression of both Twist1 and Vimentin. In SW480 cells, HMGA1 also up-regulates Twist1, while it represses E-cadherin. E-cadherin did not change in the HCT116 colon cancer cells with knock-down of HMGA1, which could reflect stable silencing of E-cadherin in these mesenchymal, highly metastatic colon cancer cell lines. The transcriptional networks regulated by HMGA1 likely depends upon the cellular milieu and underlying genetic and epigenetic characteristics.
In summary, our studies provide the first evidence linking HMGA1 to cellular properties and transcriptional networks important in stem cells, EMT, and metastatic progression in colon cancer. Although further work is needed, these results underscore the role of HMGA1 as a key regulator in tumor progression and a stem-like state in colon cancer and suggest that targeting HMGA1 pathways could be beneficial in therapy for colon cancer. Because HMGA1 is enriched in embryonic stem cells, tissue-specific stem cells, and virtually all aggressive tumors studied to date, our findings are likely to relevant not only to diverse human cancers, but also to normal development. Figure S1 HMGA1 transgene expression throughout the small and large intestines in the transgenic (TG) mice compared to wildtype (WT) controls. qRT-PCR for HMGA1 and the control gene, GAPDH, was performed from tissue obtained throughout the small intestine (duodenum, jejunum, and ileum) and large intestine (ascending colon, designated A colon, and descending colon, designated D colon) from the TG and WT mice. HMGA1 expression in the TG intestines is increased by 4 to 8-fold above that observed in the WT mice, which was arbitrarily assigned a value of 1. All qRT-PCR reactions were done in triplicate and repeated at least once.