Deregulation of HMGA1 expression induces chromosome instability through regulation of spindle assembly checkpoint genes.

The mitotic spindle assembly checkpoint (SAC) is an essential control system of the cell cycle that contributes to mantain the genomic stability of eukaryotic cells. SAC genes expression is often deregulated in cancer cells, leading to checkpoint impairment and chromosome instability. The mechanisms responsible for the transcriptional regulation and deregulation of these genes are still largely unknown. Herein we identify the nonhistone architectural nuclear proteins High Mobility Group A1 (HMGA1), whose overexpression is a feature of several human malignancies and has a key role in cancer progression, as transcriptional regulators of SAC genes expression. In particular, we show that HMGA1 proteins are able to increase the expression of the SAC genes Ttk, Mad2l1, Bub1 and Bub1b, binding to their promoter regions. Consistently, HMGA1-depletion induces SAC genes downregulation associated to several mitotic defects. In particular, we observed a high number of unaligned chromosomes in metaphase, a reduction of prometaphase time, a delay of anaphase, a higher cytokinesis time and a higher percentage of cytokinesis failure by using live-cell microscopy. Finally, a significant direct correlation between HMGA1 and SAC genes expression was detected in human colon carcinomas indicating a novel mechanism by which HMGA1 contributes to cancer progression.


INTRODUCTION
The High Mobility Group A (HMGA) protein family is comprised of three proteins: HMGA1a and HMGA1b, which are encoded by the HMGA1 gene through alternative splicing [1], and the related HMGA2 protein, encoded by a distinct gene [2]. These proteins are nonhistone architectural nuclear factors, able to bind the minor groove of AT-rich DNA sequences through three "AT-hook" domains. The involvement of HMGA proteins in embryogenesis, cell proliferation, differentiation, apoptosis and, above all, cancer development has been extensively demonstrated [3]. In particular, HMGA proteins seem to play their major physiologic role during embryonic development where they are abundantly expressed, whereas their expression is low or negligible in normal adult tissues [4,5]. Conversely, HMGA overexpression is a feature of malignant neoplasias [3], where it represents a poor prognostic index, as it often correlates with a reduced survival [6]. It has been previously demonstrated that HMGA proteins have oncogenic activity since their overexpression leads to the transformation of rat fibroblasts [7] and human epithelial breast cells [8], and transgenic mice overexpressing the HMGA proteins develop multiple neoplasias [9][10][11][12][13].
The mechanisms accounting for cell transformation induced by the HMGA proteins are essentially based on their ability to activate or inhibit the expression of genes involved in the control of cellular proliferation, invasion and apoptosis [3,[14][15][16].
A hallmark of human cancer is genome instability, whose prominent form is represented by chromosomal instability (CIN). This can consist in gain or loss of whole chromosomes, translocation/deletion/duplication of chromosome segments or polyploidy. Alterations in chromosome number or aneuploidy are found in nearly all major human tumor types [17], generally associated with the most aggressive forms of cancer. Alterations of the genes coding for proteins involved in Spindle Assembly Checkpoint (SAC), frequently observed in human malignant neoplasias, account for aneuploidy of cancer cells [18]. However, mutations in SAC genes are quite rare in human cancers (reviewed in [19]), whereas their deregulation is more frequent. Indeed, downregulation of Bub1 expression has been detected in a subset of acute myeloid leukemia [20], whereas upregulation of Bub1 levels has been reported in lymphomas [21], breast [22] and gastric cancers [23,24]. Therefore, these studies demonstrate that both upregulation and downregulation of SAC genes can cause a checkpoint impairment leading to CIN.
Since HMGA expression is associated with a highly malignant phenotype that is characterized by CIN which accelerates tumor progression leading to a higher malignant state, the aim of this work has been to investigate whether HMGA1 proteins are able to regulate the expression of SAC genes, and, thereby, evaluate HMGA1 role in CIN.
Herein we report that HMGA1 increases the transcription of Bub1, Bub1b, Mad2l1 and Mps1/Ttk genes involved in the SAC. We have found that HMGA1 knock-down compromises the mitotic checkpoint activity, and consistently HMGA1-depleted cells show a higher percentage of metaphases with unaligned chromosomes and a reduced prometaphase time compared to control cells, indicating a SAC impairment. Finally, human colon carcinomas show high SAC gene expression that correlates with HMGA1 protein levels.

HMGA1 increases the transcription of Bub1, Bub1b, Mad2l1 and Ttk genes involved in the SAC
In order to identify a possible role of HMGA1 in the induction of chromosomal instability we have investigated, first, the ability of HMGA1 to regulate the expression of SAC genes, focusing our attention on Bub1, Bub1b, Mad2l1 and Ttk genes.
These genes have been chosen because their promoters contain AT-rich regions, which represent putative binding sites for HMGA1 proteins, flanking the consensus binding sites for the same transcription factors (Supplementary Figure S1), some of which have been described as interactors of HMGA1 proteins (Supplementary Figure S1). Therefore, we have evaluated the expression of Bub1, Bub1b, Mad2l1 and Ttk at mRNA level by quantitative PCR (qPCR) in mouse NIH3T3 cells transfected with the pCEFL-HA-Hmga1b expression vector encoding HMGA1b. As shown in Figure 1A, the expression of Bub1, Bub1b, Mad2l1 and Ttk genes is higher in the cells overexpressing HMGA1 in comparison with the empty vector-transfected NIH3T3 cells (CV). Parallel results were obtained when total cellular extracts from the same cells were analyzed by western blotting with the antibodies raised against BUBR1 and MAD2 proteins ( Figure 1B). Moreover, the same results have been obtained in human colorectal HCT116, SW620 and SW48 cells overexpressing or not the HMGA1b protein ( Figure 1C and 1D, and Supplementary Figure S2).
These results suggest a role of the HMGA1b protein in the activation of SAC gene expression.

HMGA1 binds Bub1, Bub1b, Mad2l1 and Ttk promoters in vivo
As described above, Bub1, Bub1b, Mad2l1 and Ttk promoter regions contain some AT-rich DNA sequences which are the preferred binding sites for HMGA1 proteins. Therefore, we analyzed the ability of the HMGA1 proteins to bind the promoter regions of these genes by chromatin immunoprecipitation (ChIP) assays. Thus, HCT116 cells were cross-linked with formaldehyde, and DNA-chromatin complexes were subjected to immunoprecipitation with anti-HMGA1 or anti-IgG antibodies used as control. The recovered DNA was subsequently analyzed by qPCR, using primers spanning specific AT-rich regions of the indicated promoters. Specifically, the following ATrich regions of these promoters: -760/-567 region of the Bub1 promoter, -740/-550 region of the Bub1b promoter, -442/-198 region of the Mad2l1 promoter, and -380/-194 region of the Ttk promoter, containing putative binding sites for HMGA1 proteins (see Supplementary Figure S1), were analyzed. The immunoprecipitated DNA has also been amplified using primers for the glyceraldehyde 3-phosphate dehydrogenase (Gapdh) gene promoter as negative control. As shown in Figure 2A, occupancy of Bub1, Bub1b, Mad2l1 and Ttk promoters by HMGA1 has been detected in the anti-HMGA1-precipitated chromatin from HCT116 cells. Conversely, no amplification was observed with anti-IgG precipitates and when primers for the Gapdh promoter were used, indicating the specificity of the binding of HMGA1 proteins to the Bub1, Bub1b, Mad2l1 and Ttk promoters.
Finally, to assess the functional consequences of the HMGA1 binding to these promoters, we investigated whether HMGA1 proteins were able to regulate these promoters performing luciferase activity assays in NIH3T3 cells. To this aim, NIH3T3 cells have been co-transfected with a reporter vector (pGL3-Basic) carrying the fireflyluciferase gene under the control of Bub1, Bub1b, Mad2l1 and Ttk promoters, with different amounts of pcDNA3.1-Hmga1b and renilla-luciferase reporter vector (used for normalization). Insert-less pcDNA3.1 vector has been used as control (CV). As shown in Figure 2B, HMGA1 increases Bub1, Bub1b, Mad2l1 and Ttk promoter activities.
Taken together, these results demonstrate that HMGA1 is able to bind to the Bub1, Bub1b, Mad2l1 and Ttk promoters, and increase their transcriptional activity strongly supporting a critical role of HMGA1 in the regulation of SAC gene expression.

HMGA1 depletion impairs the mitotic checkpoint activity
Since the above described results indicate that HMGA1 is able to activate the expression of SAC genes acting at transcriptional level, we investigated whether HMGA1-induced SAC genes deregulation affects the mitotic checkpoint in mammalian cultured cells. To this aim, we used HeLa cells that have a functional SAC and are often used as a model for studying the checkpoint function. Since HeLa cells express HMGA1 protein at high levels, the further over-expression of HMGA1 failed to induce a consistent up-regulation of SAC genes. For this reason, considering that both up-regulation and down-regulation of one or more SAC genes impair the checkpoint, we evaluated whether HMGA1-silencing by RNAi is able to down-regulate SAC genes expression in HeLa cells. To this aim we transfected HeLa cells with siRNA targeting the HMGA1 gene (HMGA1i cells) or with control siRNA (Ctli cells). As expected, we observed a drastic downregulation of SAC gene expression (Bub1, Bub1b, Mad2l1 and Ttk) at mRNA ( Figure 3A) and protein level (Supplementary Figure S3). The analysis of mitotic cells by immunofluorescence showed the presence of a high number of unaligned chromosomes in metaphase in HMGA1i cells but not in the scrambled oligonucleotide transfected HeLa cells ( Figure 3B and 3C). Subsequently, to evaluate the role of HMGA1 in the mitotic checkpoint regulation, we used live-cell microscopy to follow mitotic progression in individual cells. Filming of mitoses in HeLa HMGA1i and Ctli cells revealed several mitotic errors in HMGA1i cells. The most prominent phenotype was a reduction of pro-metaphase time, i.e. from the round-up to the chromosome segregation, in HMGA1i cells (t = 70.21 ± 40.35 min (n = 32) versus 154.52 ± 61.46 min in Ctli (n = 32); Figure 3D and 3F and Supplementary Movies S1 and S2). We have observed that HMGA1i cells exhibited also an anaphase delay (t = 14.7 ± 8.89 min (n = 39) versus 8.05 ± 5.02 min in Ctli (n = 23); Figure 3D and 3F and Supplementary Movies S1 and S2) and that they showed a slightly higher cytokinesis time (from the early telophase to abscission; t = 199.5 ± 83.65 min n = 30) than that of Ctli cells (t = 161.28 ± 38.43 min n = 24), associated to a higher percentage of cytokinesis failure resulting in binucleated cells (21% versus 12% in Ctli cells; Figure 3D and 3E). Moreover, we observed a higher percentage of death in mitosis and of metaphase failure in the HMGA1i than in Ctli cells ( Figure 3E).
Altogether, these results suggest that the deregulation of SAC genes induced by HMGA1 depletion impairs the activity of the checkpoint.

HMGA1 expression levels directly correlate with the expression of genes involved in SAC
We evaluated the correlation between the expression of HMGA1 and of the SAC genes BUB1, BUB1B, MAD2L1 and TTK, in global mRNA profiles of colorectal cancer generated by The Cancer Genome Atlas (TCGA; http:// cancergenome.nih.gov/) for 365 and 222 samples profiled with, respectively, RNAseq and DNA microarrays. HMGA1 was found to be significantly correlated with each of the SAC genes both in RNAseq and microarray data (Pearson r between 0.22 and 0.38; p < 10 −6 in 6 comparisons, and < 2*10 −4 in the remaining two) ( Figure 4A). We extended the analysis also to other SAC genes (AURKB, CENPE, AURKA and TPX2) obtaining the same results for two of them. Interestingly, the correlation further raised when the expression of the four SAC genes was averaged (r = 0.34 in RNAseq and 0.35 in array) ( Figure 4A and 4B).
It is noteworthy that a stronger correlation between HMGA1 and SAC gene mRNA expression was found when liver metastases from colon carcinomas (n = 185) were analyzed by Illumina array ( Figure 4A). In particular, we obtained a Pearson r > 0.62 and a p value < 10 −6 for seven of the eight analyzed SAC genes, with a further increase for the averaged SAC genes expression (Pearson r = 0.72; p < 10 −6 ).
Therefore, we attempted to confirm the association of HMGA1 expression with that of two SAC genes (BUBR1 and TTK) at protein level by immunohistochemical analyses  Figure 5. Indeed, as previously reported [25,26], HMGA1 expression was abundant in the colon carcinoma samples, whereas it was not detectable in normal colonic mucosa. Interestingly, a weak immunoreactivity was observed in normal colon mucosa after staining with antibodies raised versus BUBR1 and TTK, whereas colon carcinoma samples showed a strong immunoreactivity that significantly correlated with HMGA1 staining (p = 0.006 for BUBR1 and p < 0.001 for TTK). In fact, 55.5% and 44.4% of tumor samples with a weak HMGA1 staining (1+) had negative or weak staining also for BUBR1 and TTK respectively, whereas 61.5% and 66.7% of tumor samples with a strong HMGA1 immunoreactivity (3+) had also a strong immunoreactivity for BUBR1 and TTK, respectively. Representative cases of normal colon mucosa (a, b, c), colon carcinoma samples with a weak HMGA1 immunoreactivity (d, e, f) and colon carcinoma samples with a strong HMGA1 reactivity (g, h, i) stained with antibodies versus BUBR1 and TTK are shown in Figure 5.
Therefore, human colon carcinoma show SAC gene overexpression that strongly correlates with that of HMGA1, likely contributing to cancer progression.

DISCUSSION
Overexpression of HMGA proteins, a general feature of human malignancies [3], has a critical role in cancer progression since it inhibits p53-dependent apoptosis by modulating transcription of p53 [27] and p53-target genes (reviewed in [14]), impairs DNA repair [28] and induces the epithelial-mesenchymal transition (EMT) [29,8]. Consistently, HMGA protein expression is associated with a highly malignant phenotype being a marker of poor prognosis for cancer patients as their overexpression correlates with the resistance to anti-cancer therapies [28] and a reduced survival [6].
Also alterations in chromosome number or aneuploidy are a feature of highly malignant neoplasias and it is due to defects in the mitotic checkpoint that controls chromosome segregation during mitosis [17]. Alterations of the spindle assembly checkpoint genes levels likely account for aneuploidy in several human cancer (reviewed in [30,31]) since mitotic checkpoint impairment and aneuploidy in human tumour cells are often associated with changes in the SAC gene-encoded protein levels [30]. It has also been reported that in some tumour cells these  changes occur through altered transcriptional regulation by tumour suppressors or oncogene products [32,33].
Therefore, in order to investigate the mechanisms accounting for the association of HMGA1 overexpression and chromosome instability, we have analyzed the ability of the HMGA1 proteins to transcriptionally regulate the genes involved in SAC.
Here, we report that HMGA1 proteins are able to activate the expression of the SAC genes Ttk, Mad2l1, Bub1 and Bub1b binding to their promoter regions. Consistently, HMGA1i-interfered HeLa cells (HMGA1i) show downregulation of these genes, whereas HMGA1overexpression in several human colon cancer and NIH3T3 cells induces their upregulation. Moreover, the analysis of mitotic cells by immunofluorescence showed the presence of a high number of unaligned chromosomes in metaphase in HMGA1i cells associated to a reduction of prometaphase time. This alteration suggests that the HMGA1i cells start the anaphase without waiting for the proper chromosome alignment due to SAC impairment. Interestingly, the HMGA1-interfered HeLa cells resemble the cells in which the SAC gene Bub1b was depleted [34], indicating that HMGA1 depletion affects the mitotic checkpoint in cultured cells.
Notably, we have found a strong correlation between HMGA1 and BUBR1 and TTK protein levels in human colon carcinomas (Figure 4). Moreover, analysis of the TCGA expression data generated on CRC showed a statistically significant correlation, with the highest Pearson index of 0.54, between HMGA1 and SAC genes expression, evaluated either by microarray analysis or by RNAseq. This correlation strongly increased in liver metastases from colon carcinomas (pearson r = 0.72), supporting the idea that HMGA1-induced SAC genes deregulation may be important during tumour progression.
Altogether, these findings suggest that SAC gene overexpression induced by HMGA1 may have a role in colorectal cancer progression causing chromosome instability. Indeed, although it has also been demonstrated that mice carrying conditional Bub1 mutation develop severe defects ranging from early lethality to tumorigenesis [35], even SAC gene overexpression has been reported to impair the spindle assembly checkpoint playing a critical role in cancer progression. In fact, MAD2 overexpression is a common event in many human tumors [36,37], and its overexpression increases nondisjunction events and aneuploidy in human fibroblasts and cell lines [32]. Moreover, Bub1 and Bub1b expression levels are drastically upregulated in gastric cancer associated with tumor cell proliferation [23]. Furthermore, colorectal cancer patients showing Bub1 and Bub1b upregulation have a shorter relapse-free survival with respect to the groups showing normal expression of these genes [38].
In conclusion, based on the results presented here, we propose a novel mechanism by which HMGA1 overexpression contributes to cancer progression activating SAC gene expression, thereby inducing chromosomal instability that eventually leads to a more advanced cancer status.

Luciferase assay
Cells were plated in 6 wells plates and after 24 hours were transfected with 100 ng of pLuc-Bub1, pLuc-Bub1b, pLuc-Mad2l1 or pLuc-Ttk expression vectors together with 50 ng of pCMV-Renilla plasmid and with different amount of pcDNA3.1-Hmga1b expression vector. Luciferase and Renilla activities were assessed with the Dual-Light Luciferase system (Promega), 48 h after the transfection. Luciferase activity was normalized for the Renilla activity. All the experiments were performed three times in duplicate and the mean ± SD was reported.

Chromatin immunoprecipitation (ChIP) assay
48 hours after transfection, cells have been treated with formaldehyde 1%, washed and then lysed isolating the nuclei. Then the nuclei have been in turn lysed and chromatin has been sonicated. Chromatin has been immunoprecipitated using anti-HMGA1 antibody [39] or normal rabbit IgG as negative control. After the immunoprecipitation, the chromatin has been incubated overnight at 65°C with DNAse-free RNAse (Roche). Next day, the samples have been incubated 3 hours at 50°C with 0, 5% SDS and 0, 5 mg/ml Proteinase K (Roche) and after the DNA has been purified with a phenolchloroform extraction. For qPCR analysis, 2 μl out of 150 μl immunoprecipitated DNA was used with primers described below. Gapdh promoter amplicon was used as negative control in all the experiments. Input DNA was used as positive control.
Primers used were:

Statistical analysis
Student's t test was used to determine the significance for all the quantitative experiments. Error bars represent the standard deviation (SD) of the average. A p-value < 0.05 was considered statistically significant.
The χ 2 test was used to test the relationship between the immunoreactivity of HMGA1 and that of BUBR1 and TTK in colon samples. The test indicates how much of the association is accounted for linear trend. Statistical significance for all the tests was assessed by calculating the p-value.
For correlation analysis, TCGA expression data for HMGA1, BUB1, BUB1B, MAD2L1 and TTK in colorectal cancer were downloaded as z-scores from the cBioPortal (http://www.cbioportal.org). Pearson correlation between HMGA1 and SAC gene expression was calculated with Microsoft Excel. Statistical significance of the observed values was assessed on the online portal "Simple Interactive Statistical Analysis" (http://www.quantitativeskills.com/ sisa/statistics/correl.htm).

Immunofluorescence
Cells plated on cover-slides in 12 wells plates were fixed in 4% formaldehyde in PBS and permeabilized in a solution of 0.25% Triton X-100 in PBS. Anti β-tubulin antibody conjugated to CY3 (Sigma) was used to identify mitotic spindle in mitotic cells, and DAPI staining was used to identify chromosomes. Cells were observed with a fluorescent microscope (Zeiss, magnification 63X or 100X).

Live-cell imaging
Cells were seeded in slides 8-well (80826, ibiTreat, Ibidi) and observed under an Eclipse Ti inverted microscope (Nikon) using a 40x objective (Nikon). During the observation, cells were kept in a microscope stage incubator (Basic WJ, Okolab) at 37°C and 5% CO2. DIC images were acquired over a 24 hr period by using a DS-Qi1Mc camera. Image and video processing were performed with NIS-Elements AR 3.22.

Immunohistochemistry
Immunohistochemical analysis of paraffinembedded tissues was performed with antibodies against HMGA1, BUBR1 and TTK, using polyclonal antibodies above mentioned, as previously described [47]. No staining was observed when normal colon samples were stained without the primary antibodies (data not shown).