Abstract
The HMGA1 protein is a major factor in chromatin architecture and gene control. It plays a critical role in neoplastic transformation. In fact, blockage of HMGA1 synthesis prevents rat thyroid cell transformation by murine transforming retroviruses, and an adenovirus carrying the HMGA1 gene in the antisense orientation induces apoptotic cell death in anaplastic human thyroid carcinoma cell lines, but not in normal thyroid cells. Moreover, both in vitro and in vivo studies have established the oncogenic role of the HMGA1 gene. In this study, to define HMGA1 function in vivo, we examined the consequences of disrupting the Hmga1 gene in mice. Both heterozygous and homozygous mice for the Hmga1-null allele show cardiac hypertrophy due to the direct role of HMGA1 on cardiomyocytic cell growth regulation. These mice also developed hematologic malignancies, including B cell lymphoma and myeloid granuloerythroblastic leukemia. The B cell expansion and the increased expression of the RAG1/2 endonuclease, observed in HMGA1-knockout spleen tissues, might be responsible for the high rate of abnormal IgH rearrangements observed in these neoplasias. Therefore, the data reported here indicate the critical role of HMGA1 in heart development and growth, and reveal an unsuspected antioncogenic potential for this gene in hematologic malignancies. (Cancer Res 2006; 66(5): 2536-43)
Introduction
The high-mobility group A (HMGA) protein family includes three members: HMGA1a, HMGA1b (both coded for by the same gene through an alternative splicing), and HMGA2 (1). These proteins bind the minor groove of AT-rich DNA sequences through three short basic repeats, called “AT-hooks”, located at the NH2-terminal region of the proteins. By interacting with the transcription machinery, HMGA1 proteins alter the chromatin structure and thereby regulate the transcriptional activity of several genes either enhancing or suppressing the ability of the more usual transcriptional activators and repressors (2). The HMGA proteins seem to be involved in embryonic development (3), adipocytic cell growth, and differentiation. In fact, the disruption of the murine Hmga2 gene induces a pygmy phenotype associated with a drastic reduction of fat tissue (4), whereas transgenic mice overexpressing the Hmga2 gene are giant and obese (5). Conversely, HMGA1 has an inhibitory effect on adipocytic cell growth and is required for differentiation (6). Both genes have a critical role in the process of carcinogenesis because they are overexpressed in most human malignant neoplasias (7), and the blockage of their expression has been shown to prevent thyroid cell transformation and lead malignant cells to death (8, 9). Moreover, in vitro and in vivo studies showed the oncogenic activity of overexpressed HMGA proteins. In fact, either HMGA1 or HMGA2 overexpression is able to transform mouse and rat fibroblasts (10, 11), and transgenic mice overexpressing either HMGA1 or HMGA2 develop natural killer T-cell lymphomas and pituitary adenomas (12–15). To define the physiologic role of the HMGA1 protein in vivo, we examined the consequences of disrupting the Hmga1 gene in mice. Here, we report that both heterozygous and homozygous Hmga1-null mice develop cardiac hypertrophy and myelo-lymphoid malignancies, including B-cell lymphomas and myeloid leukemia. In vitro experiments on rat cardiomyocytes showed a direct role for the HMGA1 protein on cardiac cell growth control. As far as the B-cell lymphomas are concerned, our data suggest a mechanism by which the reduced levels of HMGA1 lead to a dysregulation of T- and B-cell–specific cytokines, which in turn, causes expansion of the B-cell population and increased Rag1/2 levels which might account for the high rate of aberrant IgH rearrangements with consequent development of B-cell neoplasias.
Materials and Methods
Generation of HMGA1-KO mice. Several overlapping clones were isolated from a λΦIXII phage library of a 129SVJ mouse strain (Stratagene, La Jolla, CA) using standard procedures. Exons III, IV, and V of the murine Hmga1 gene were replaced by a neomycin-resistant (neo) cassette. The targeting construct was transfected by electroporation into embryonic stem (ES) AB2.1 cells. Genomic DNA from each G418-resistant clone was digested with EcoRI and hybridized by standard Southern blot to an external 5′ genomic fragment that detects 18 and 9 kb fragments corresponding to the wild-type (WT) and mutant alleles, respectively. Two correctly targeted ES cell lines were injected into C57Bl/6J blastocysts. Both gave rise to germ line chimeras that were backcrossed to C57Bl/6J females to obtain heterozygous HMGA1-knockout (KO) offspring. HMGA1-KO mice were crossed with each other to yield homozygous HMGA1-KO animals. All mice were maintained under specific pathogen-free conditions in our Laboratory Animal Facility (Istituto dei Tumori di Napoli, Naples, Italy) and all studies were conducted in accordance with Italian regulations for experiments on animals.
Isolation of mRNA and RT-PCR. Total RNA was extracted using TRI-reagent solution (Molecular Research Center, Cincinnati, OH) according to the manufacturer's protocol, and treated with DNase I (Invitrogen, San Giuliano Milanese, Italy). cDNA was synthesized using random hexamers (100 mmol/L) and MuLV reverse transcriptase (Perkin-Elmer, Boston, MA). The PCR was run with a 25 μL volume for 20, 25, and 30 cycles (30 seconds at 94°C, 30 seconds at 55°C, and 30 seconds at 72°C, respectively) using the Gene Amp PCR System 9700 (Applied Biosystems, Foster City, CA). As a negative control, RNA samples that had not been reverse-transcribed before PCR were used. Primer sequences are available on request.
Histologic analysis and immunohistochemistry. For histologic examination, dissected tissues were fixed by immersion in 10% formalin and embedded in paraffin. Mounted sections (5 μm thick) were stained with H&E using routine procedures. To estimate myocyte size, slides containing abundant myocytes in a cross-sectional orientation were photomicrographed at ×40 magnification. The avidin-biotin-peroxidase LSAB+ kit (Dako, Glostrup, Denmark) and microwave antigen retrieval was used for immunoperoxidase staining. The antisera were directed to CD79a (Abcam, Cambridge, United Kingdom), CD3 (Dako), B220 (PharMingen, San Diego, CA), Mac-2 (Cedarlane, Tornby, Canada), and MPO (Dako).
Transthoracic echocardiography. Mice were anesthetized by i.p. injection of 2.5% Avertin (15 μL/g body weight). Echocardiography was done with a Hewlett Packard Sonos 5500 ultrasound system fitted with a 12 mHz transducer (described in detail in the Supplemental data).
Primary culture of neonatal rat ventricular myocytes and adenovirus infection. Primary cultures of cardiac ventricular myocytes from 1-day-old Crl:(WI)BR-Wistar rats (Charles River Laboratories, Calco, Italy) were prepared as described previously (16). Cells were cultured in the cardiac myocyte culture medium containing DMEM/F12 supplemented with 5% HS, 4 μg/mL transferrin, 0.7 ng/mL sodium selenite, 2 g/L bovine serum albumin (BSA; fraction V), 3 mmol/L pyruvic acid, 15 mmol/L HEPES, 100 μmol/L ascorbic acid, 100 μg/mL ampicillin, 5 μg/mL linoleic acid, and 100 μmol/L 5-bromo-2'-deoxyuridine (Sigma-Aldrich, Milan, Italy). We obtained myocyte cultures in which >95% were myocytes, as assessed by immunofluorescence staining with a monoclonal antibody against sarcomeric myosin (MF20). Myocytes were plated at a density of 1 × 106 per well in six-well plates, the culture medium was changed to serum-free medium at 24 hours and cultured in serum-free conditions for 48 hours prior to experiments. Cardiomyocytes were exposed to 10 and 100 pfu/cell of the specific adenovirus and further cultured for 48 hours.
Measurement of the total protein content. After insulin-like growth factor-I (IGF-I) stimulation (17) or adenovirus infection, cardiomyocytes were rinsed thrice with PBS. The cell layer was scraped with 1 mL of 1× SSC containing 0.25% (w/v) SDS and vortexed extensively. The total cell protein and the DNA content were determined by the Lowry method and the Hoechst dye method, as described previously (18). The protein content was normalized by the DNA content to correct for differences in the cell number.
Analysis of splenocyte cell surface antigens. Spleens removed from mice were dissociated into single cells and RBCs were lysed by hypotonic shock using Tris-NH4Cl buffer [NH4Cl, 0.14 mol/L; Tris, 0.017 mol/L (pH 7.2)]. Cells were washed in PBS supplemented with 2% FCS, 0.2% sodium azide, and labeled with FITC-, PE-, and Cy5-conjugated appropriate antibodies for 30 minutes at 4°C. Cells were then washed and analyzed on FACSCalibur flow cytometer (Becton Dickinson, Buccinasco, Italy). All the antibodies used were obtained from PharMingen/BD Biosciences.
Transient transfections and luciferase activity. WT and Hmga1−/− ES cells have been previously described (19). Before transfections, feeder fibroblasts were removed. WT or double knockout ES cells (4 × 105) were plated in six-well plates and transfected, 48 hours later, with 1 μg of reporter plasmid using FuGene6 (Roche, Monza, Italy). Where indicated, 5 μg of a plasmid expressing HMGA1b (6) was cotransfected. Cells were harvested 48 hours posttransfection and lysates were analyzed for luciferase activity using the Dual Light kit (Applied Biosystems, Monza, Italy). Transfection efficiency was normalized according to β-galactosidase activity. All the assays were done in triplicate and repeated in three independent experiments.
IgH gene rearrangements. Southern blot of DNA digested with StuI was prepared following conventional methods and hybridized with the 32P-labeled DNA probe PJ3 representing the JH4 region of the IgH locus. The probe was synthesized by PCR amplification from mouse DNA as previously described (20).
Protein extraction and Western blot. Tissues were lysed in 1% NP40 buffer, 1 mmol/L EDTA, 50 mmol/L Tris-HCl (pH 7.5), and 150 mmol/L NaCl, supplemented with complete protease inhibitors mixture (Roche). Total proteins were separated by SDS-PAGE and transferred to nitrocellulose membrane. Membrane was blocked with 5% BSA in TBS and incubated with the specific primary antibodies recognizing the following proteins: c-myc (C-19, Santa Cruz Biotechnology, Santa Cruz, CA), HMGA1 (polyclonal antibody raised against a synthetic peptide located in the NH2-terminal region), CaMKII (M-176, Santa Cruz Biotechnology), and α-tubulin (clone DM1A, Sigma-Aldrich). Bound antibody was detected by the appropriate secondary antibody, diluted in 5% nonfat milk in TTBS, and revealed with an enhanced chemiluminescence system (Amersham Biosciences, Cologno Monzese, Italy).
Statistics. The results are expressed as the mean ± SD. For the comparison of statistical significance between the two groups, Student's t test was used. P < 0.05 was considered statistically significant.
Results
Generation of HMGA1-KO mice. We used gene targeting techniques in ES cells to generate a null mutation at the murine Hmga1 genomic locus (Fig. 1A). Heterozygous progeny of chimeric animals were identified by Southern blot analysis of EcoRI-digested tail DNA (data not shown), and matings were established to produce mice heterozygous or homozygous for the Hmga1-null allele (Fig. 1B). Both heterozygous and homozygous mutant mice were viable and fertile. We verified Hmga1 gene inactivation by RT-PCR using total RNA from adult hearts and spleens. As expected, homozygous mutants did not express Hmga1 mRNA, whereas heterozygous mutants expressed an intermediate amount of Hmga1 (Fig. 1C). Western blot of cell lysates from embryonic fibroblasts confirmed the suppression of the protein expression in the Hmga1-null cells (data not shown).
Generation of a loss-of-function allele at the Hmga1 locus. A, schematic representation of the WT and mutant alleles and the targeting vector. Abbreviations: (RI) EcoRI, (N) NotI, (H) HindIII, (K) KpnI, (X) XbaI, (tk) thymidine kinase, (neo) neomycin. B, Southern blot of pups obtained from heterozygous intercrosses. C, RT-PCR analysis of total RNA from hearts and spleens of WT (+/+), heterozygous (+/−), and homozygous (−/−) mice.
Generation of a loss-of-function allele at the Hmga1 locus. A, schematic representation of the WT and mutant alleles and the targeting vector. Abbreviations: (RI) EcoRI, (N) NotI, (H) HindIII, (K) KpnI, (X) XbaI, (tk) thymidine kinase, (neo) neomycin. B, Southern blot of pups obtained from heterozygous intercrosses. C, RT-PCR analysis of total RNA from hearts and spleens of WT (+/+), heterozygous (+/−), and homozygous (−/−) mice.
HMGA1-KO mice display cardiac hypertrophy. Mice homozygous for the Hmga1-null mutation seem to develop normally, and histologic examination of serial sections confirms the normal structure and development of all major organ and tissues (data not shown). Cardiac hypertrophy was clearly evident in nearly all the adult homozygous KO animals (starting from 2 months of age and becoming more consistent at 12 months of age). The heart weight/body ratio increased by an average of 77% in homozygotes and 35% in heterozygotes compared with age-matched WT controls (P < 0.001; Fig. 2A and B). Body weight was the same in KO and WT mice or sometimes slightly less in homozygous KO mice (data not shown). Thus, the increased heart/body weight ratio in the KO mice was due to increased ventricular mass. Histologic examination of H&E stained heart sections showed marked myocardial hypertrophy and an abnormal myocyte architecture in heterozygous and homozygous KO mice (Fig. 2C; data not shown). To evaluate whether increased cardiac size in KO mice was due to enlarged myocytes, we examined cross-sections of myocytes at a higher magnification. Myocytes were larger in the hypertrophic hearts (outsets, Fig. 2C) indicating that cardiac hypertrophy was due to myocyte hypertrophy as opposed to increased interstitial content. To evaluate further abnormalities in cardiac structure (as observed in histologic sections) and to correlate them with cardiac function, we measured wall thickness, ventricular diameter, and cardiac function by transthoracic echocardiography in both 2- and 11- to 12-month-old KO mice and WT littermates (Fig. 2D; data not shown). End-diastolic and end-systolic left ventricular chamber diameters were increased 42% and 51%, respectively, in KO animals versus age-matched WT controls. Furthermore, left ventricular posterior wall thickness was increased 19%, and the calculated left ventricular mass was increased 70% in homozygous KO mice versus controls. Although cardiac function, measured by fractional shortening, was not significantly different in WT and KO mice, a trend toward decreased fractional shortening was observed in KO mice (Supplementary Table S1). The echocardiographic variables of heterozygous KO mice were intermediate between those from homozygous and WT mice (Fig. 2D; Supplementary Table S1). No hypertension was observed in KO mice (data not shown). This result rules out that the hypertrophy occurring in HMGA1-KO mice might be dependent on hemodynamic changes, but clearly suggests a direct role of the HMGA1 proteins in the regulation of the growth properties of cardiomyocytes.
Cardiac hypertrophy in HMGA1-KO mice. A, representative gross morphology of hearts from Hmga1+/+ (left) and Hmga1−/− (right) mice at 50 weeks of age. B, heart weight/body weight ratio at 48 to 54 weeks of age. Columns, means of a cohort of four WT, four heterozygous, and four homozygous HMGA1-KO mice; bars, ± SD (*, P < 0.01 versus WT). C, representative histologic analysis of WT (left), Hmga1+/−, (middle), and Hmga1−/− (right) mice at 44 weeks of age, cut at the midsagittal level and parallel to the base. Outsets, hearts at a higher magnification, showing cardiomyocyte enlargement in heterozygous and homozygous KO hearts. lv, left ventricle; rv, right ventricle. D, representative M-mode tracings from transthoracic echocardiography in the mice from (C).
Cardiac hypertrophy in HMGA1-KO mice. A, representative gross morphology of hearts from Hmga1+/+ (left) and Hmga1−/− (right) mice at 50 weeks of age. B, heart weight/body weight ratio at 48 to 54 weeks of age. Columns, means of a cohort of four WT, four heterozygous, and four homozygous HMGA1-KO mice; bars, ± SD (*, P < 0.01 versus WT). C, representative histologic analysis of WT (left), Hmga1+/−, (middle), and Hmga1−/− (right) mice at 44 weeks of age, cut at the midsagittal level and parallel to the base. Outsets, hearts at a higher magnification, showing cardiomyocyte enlargement in heterozygous and homozygous KO hearts. lv, left ventricle; rv, right ventricle. D, representative M-mode tracings from transthoracic echocardiography in the mice from (C).
Subsequently, using semiquantitative RT-PCR, we evaluated the expression of a panel of hypertrophic genes in the hearts of both heterozygous and homozygous HMGA1-null mice. As shown in Supplementary Fig. S1, there was a modest but significant increase in myosin heavy chain-β, atrial natriuretic factor, brain natriuretic peptide, skeletal actin, and myosin light chain levels, whereas myosin heavy chain-α and cardiac α-actin levels were not changed. Interestingly, the class II calcium/calmodulin-dependent protein kinase (CaMKII), that has recently been shown to play a crucial role in inducing cardiac hypertrophy (21), showed a drastic increase. Also, mRNA levels of cardiotrophin-1 (Ctf1; ref. 22), GATA4, and the two AP1 genes, c-fos and junB, whose expression is associated with cardiac stress and hypertrophy (23), were significantly increased.
The reduction of HMGA1 expression levels is directly responsible for cardiomyocytic hypertrophy. To finally assess the direct role of HMGA1 reduction in cardiomyocytic hypertrophy, we infected neonatal rat cardiac myocytes (Supplementary Fig. S2), cultured in a serum-free condition, with an adenovirus carrying the Hmga1 cDNA in an antisense orientation (Ad-Yas-GFP; ref. 24) and that was able to drastically reduce the HMGA1 protein levels (Fig. 3A). Two days after the infection, we evaluated the occurrence of cardiomyocytic hypertrophy both by actin filament staining and protein/DNA content. Similar to what occurs for a well-known hypertrophic stimulus like IGF-I (16), we observed a significant increase in protein/DNA content in cardiomyocytes infected with the Ad-Yas-GFP virus (Fig. 3B). In fact, the protein/DNA content of the Ad-Yas-GFP–infected cardiomyocytes was 20.1 ± 1.04 versus 4.65 ± 0.413 of the Ad-GFP (mock) infected cells. As expected, the IGF-I-stimulated cardiomyocytes showed an increased protein/DNA content with respect to untreated cells. Interestingly, the ratio of protein/DNA content in IGF-I cells (20.7 ± 2.51) was similar to that observed in Ad-Yas-GFP–infected cardiomyocytes. Consistently, the Ad-Yas-GFP–infected cells showed reorganization of actin in the sarcomere similar to that observed in IGF-I-stimulated cardiomyocytes (data not shown). Conversely, untreated cells exhibited poorly organized actin (data not shown). These results strongly support a direct role for the lack of HMGA1 expression in the induction of cardiac hypertrophy. Interestingly, HMGA1 protein levels in IGF-I-treated cells were lower than controls (Fig. 3A), suggesting HMGA1 as a potential downstream molecule in the signal pathway activated by IGF-I. To better analyze whether the overexpression of CaMKII is a direct consequence of the hypo- or null-expression of HMGA1, we analyzed CaMKII protein expression in rat cardiomyocytes in which HMGA1 expression was knocked down by the Ad-Yas-GFP adenovirus. As shown in Fig. 3A, both Ad-Yas-GFP–infected and IGF-I-induced cells express increased levels of CaMKII compared with uninfected and Ad-GFP–infected cells. These results would suggest that up-regulation of CaMKII, in the absence of HMGA1, might have a role in cardiac hypertrophy developed by HMGA1-null mice.
HMGA1 directly regulates cardiomyocytic hypertrophy. A, immunoblots showing reduced HMGA1 and increased CaMKII protein levels in Ad-Yas-GFP–infected and IGF-I-stimulated myocytes compared with uninfected and mock-infected (Ad-GFP) cells. B, forty-eight hours after infection, myocytes were harvested, and the protein/DNA content was determined. The histogram shows the relative protein/DNA content with respect to uninfected cells (n = 3; *, P < 0.005 versus mock infected control).
HMGA1 directly regulates cardiomyocytic hypertrophy. A, immunoblots showing reduced HMGA1 and increased CaMKII protein levels in Ad-Yas-GFP–infected and IGF-I-stimulated myocytes compared with uninfected and mock-infected (Ad-GFP) cells. B, forty-eight hours after infection, myocytes were harvested, and the protein/DNA content was determined. The histogram shows the relative protein/DNA content with respect to uninfected cells (n = 3; *, P < 0.005 versus mock infected control).
HMGA1-null mice develop malignant myelo-lymphoid diseases. From 12 months of age, a high percentage of heterozygous (78% of 23 mice) and homozygous (80% of 10 mice) Hmga1-KO mice showed splenomegaly (data not shown). This was histologically diagnosed as lymphoid/myeloid hyperplasia (35% of heterozygotes; data not shown), B type lymphoma (43% of heterozygotes and 60% of homozygotes; Fig. 4A-N), and myeloid leukemia (30% of homozygotes; Fig. 4O and P). The B lymphoma subtypes, as defined by Morse et al. (25) were: (a) pre–B cell (36%; Fig. 4A-D), (b) follicular B cell (36%; Fig. 4E-H), (c) diffuse large B cell histiocyte-associated lymphomas (21%; Fig. 4I-N). Immunohistochemical analysis that showed positive staining with antibodies against the CD45 receptor antigen B220 and no significant reaction with antibodies against the T-cell receptor antigen CD3 showed that all these lymphomas were of the B cell type (Fig. 4B,, C, F, G, I, and M). We further characterized the myeloid disorders by flow cytometry on splenocytes. As shown in Fig. 5, in both heterozygotes and homozygotes bearing a myeloid hyperproliferation, we observed an increased proportion of CD19−/CD3− cells, likely representing myeloid cells, among which both granulocytes (Gr1+ cells) and erythroblasts (Cd45−/Gr1− cells) resulted significantly more abundant in HMGA1-KO than in WT samples. Cytomorphologic analyses confirmed the presence of many granulocytes and erythroblasts in these samples (data not shown).
Histologic analyses of representative samples of myelo-lymphoid malignancies from HMGA1-KO mice. A-D, spleen showing pre–B-cell lymphoma; E-H, spleen showing follicular B-cell lymphoma; I-N, spleen showing diffuse large B-cell lymphoma; O and P, liver in which the normal structure is replaced by a cellular proliferation of immature myeloid cells. H&E staining plus immunostaining with specific antibodies, as indicated above the panels, were done to make the differential diagnosis: B220 and CD3 to distinguish between B and T cells (note that a few infiltrating mature T cells are seen in some CD3-stained panels), respectively; cytoplasmic CD79a to make the differential diagnosis between pre-B (positive) and mature B-cell lymphomas (negative); MAC-2 to stain the histiocytes associated with diffuse large B-cell lymphomas; myeloperoxidase (MPO) to highlight the large percentage of the promyelocytic/blastic cells that replaced the normal liver structure (magnification, ×20).
Histologic analyses of representative samples of myelo-lymphoid malignancies from HMGA1-KO mice. A-D, spleen showing pre–B-cell lymphoma; E-H, spleen showing follicular B-cell lymphoma; I-N, spleen showing diffuse large B-cell lymphoma; O and P, liver in which the normal structure is replaced by a cellular proliferation of immature myeloid cells. H&E staining plus immunostaining with specific antibodies, as indicated above the panels, were done to make the differential diagnosis: B220 and CD3 to distinguish between B and T cells (note that a few infiltrating mature T cells are seen in some CD3-stained panels), respectively; cytoplasmic CD79a to make the differential diagnosis between pre-B (positive) and mature B-cell lymphomas (negative); MAC-2 to stain the histiocytes associated with diffuse large B-cell lymphomas; myeloperoxidase (MPO) to highlight the large percentage of the promyelocytic/blastic cells that replaced the normal liver structure (magnification, ×20).
Flow cytometry of myeloid disorders. A, splenocytes from heterozygous (Hmga+/−) and homozygous (Hmga1−/−) mice diagnosed with splenic hyperplasia and/or leukemia were stained for various cell surface antigens and compared with asymptomatic WT littermate splenocytes. The absolute number of lymphocytes in each spleen is reported on the top (n). For each fluorescence-activated cell sorting analysis, 10,000 events were counted. B, the relative percentage of Gr1+ cells (granulocytes) and CD45− cells (erythroblasts) were plotted as histograms, showing their increase versus control (Hmga1+/+) samples.
Flow cytometry of myeloid disorders. A, splenocytes from heterozygous (Hmga+/−) and homozygous (Hmga1−/−) mice diagnosed with splenic hyperplasia and/or leukemia were stained for various cell surface antigens and compared with asymptomatic WT littermate splenocytes. The absolute number of lymphocytes in each spleen is reported on the top (n). For each fluorescence-activated cell sorting analysis, 10,000 events were counted. B, the relative percentage of Gr1+ cells (granulocytes) and CD45− cells (erythroblasts) were plotted as histograms, showing their increase versus control (Hmga1+/+) samples.
B cell expansion due to the IL gene dysregulation and RAG1/2 overexpression in HMGA1-null mice as possible mechanisms for the development of B-cell lymphomas. In an attempt to find out the mechanisms underlying the frequent occurrence of the lymphoproliferative malignancies in the HMGA1-null mice, we have analyzed the expression of interleukin (IL) genes in WT and Hmga1−/− spleens. As shown in Fig. 6A, HMGA1-KO spleens express lower levels of IL-2 and the T cell–specific gene, Thy, although they show higher levels of IL-6 and the B cell–specific gene V-preB in comparison with WT spleens. These results are consistent with those previously obtained analyzing the ES cell system (19). The dysregulation of IL expression might be responsible for the impaired T-cell differentiation with consequent B cell commitment and clonal expansion that would lead to B-cell hyperplasias and lymphomas in Hmga1+/− and Hmga1−/− mice. We also analyzed the expression of Rag1 and Rag2 genes, which are both overexpressed in HMGA1-KO spleens (Fig. 6A). Next, we showed that HMGA1 is directly involved in Rag transcription control because it is able to down-regulate the transcriptional activity of a Rag2 promoter driving a luciferase gene (Fig. 6B) in ES cells. RAG proteins mediate the V-to-DJ rearrangement at the IgH locus that leads to the formation of the BCR complex expressed in mature B cells (26). An increased RAG1/2 expression may induce a higher and aberrant V(D)J recombination that could account for the increased susceptibility to develop neoplasias of the lymphoid tissue observed in Hmga1+/− and Hmga1−/− mice. To check if the RAG1/2 overexpression in HMGA1-KO spleen samples could be associated with abnormal V(D)J recombination, we did Southern blot analyses using a probe representing the JH4 region of the IgH locus. As shown in Fig. 6C, DNA from the spleens of littermate controls showed the IgH gene in its germ line configuration, whereas most DNA from pathologic splenocytes presented extra-rearranged bands, indicating the presence of aberrant V(D)J rearrangements.
Mechanism of B-cell lymphomagenesis in HMGA1-KO mice. A, RT-PCR analysis showing the expression of IL 2, IL-6, Thy, V-preB, Rag1, and Rag2 genes in Hmga1+/+, Hmga1+/−, and Hmga1−/− spleens. Gapdh expression was evaluated as an internal control of RNA used. B, RT-PCR analysis showing the expression of the Rag2 gene in Hmga1+/+, Hmga1+/−, and Hmga1−/− ES cells (left). β-Actin expression was evaluated as an internal control of RNA used. Luciferase activity assay using the Rag2 promoter region transfected alone or with 5 μg of HMGA1 expression vectors in the ES cells (right). Columns, mean results of three different experiments; bars, ± SE. C, analysis of IgH gene configuration. IgH gene rearrangements were analyzed by Southern blot on StuI-digested spleen DNA. Control (+/+) is WT mouse with the genomic 4.7 kb StuI fragment representing the gene in its germ line configuration. Most KO mice show rearranged extra-bands (asterisks on the side). The histologic diagnoses of the tissues are indicated above. D, c-Myc protein expression analysis by Western blot in WT (+/+) and HMGA1-KO (−/−) spleen tissues. As a control for equal protein loading, the blotted proteins were incubated with tubulin-specific antibodies. The histologic diagnoses of the tissues are indicated above.
Mechanism of B-cell lymphomagenesis in HMGA1-KO mice. A, RT-PCR analysis showing the expression of IL 2, IL-6, Thy, V-preB, Rag1, and Rag2 genes in Hmga1+/+, Hmga1+/−, and Hmga1−/− spleens. Gapdh expression was evaluated as an internal control of RNA used. B, RT-PCR analysis showing the expression of the Rag2 gene in Hmga1+/+, Hmga1+/−, and Hmga1−/− ES cells (left). β-Actin expression was evaluated as an internal control of RNA used. Luciferase activity assay using the Rag2 promoter region transfected alone or with 5 μg of HMGA1 expression vectors in the ES cells (right). Columns, mean results of three different experiments; bars, ± SE. C, analysis of IgH gene configuration. IgH gene rearrangements were analyzed by Southern blot on StuI-digested spleen DNA. Control (+/+) is WT mouse with the genomic 4.7 kb StuI fragment representing the gene in its germ line configuration. Most KO mice show rearranged extra-bands (asterisks on the side). The histologic diagnoses of the tissues are indicated above. D, c-Myc protein expression analysis by Western blot in WT (+/+) and HMGA1-KO (−/−) spleen tissues. As a control for equal protein loading, the blotted proteins were incubated with tubulin-specific antibodies. The histologic diagnoses of the tissues are indicated above.
Because deregulated expression of the Myc cellular oncogene is a critical factor in B-cell neoplasms, we also looked at the expression of c-Myc in pathologic spleen tissues. As shown in Fig. 6D, we found increased levels of the c-myc protein in all the spleens with either myeloid or lymphoid disorders analyzed, providing another consistent mechanism by which these cells are the favorites in both proliferation and neoplastic transformation.
Discussion
Here, we report the generation of Hmga1-null mice. These mice, even at the heterozygous state, developed cardiac hypertrophy, B-cell lymphomas, and myeloid malignancies, in keeping with the dosage-dependency of the HMGA1-related defects.
HMGA1-KO mice develop concentric cardiac hypertrophy. Cardiac hypertrophy is generally characterized by either concentric or eccentric remodeling. Eccentric hypertrophy is distinguished by increases in both wall thickness and chamber size, with no relative change in the wall-to-chamber ratio. This type of remodeling is typically associated with increased volume loads, such as that occurring in patients with chronic valvular insufficiency. However, this is also the typical pattern of growth induced by physiologic stimuli, such as aerobic exercise training. On the other hand, concentric hypertrophy, which occurs in the HMGA1-KO mice described in this study, is classified by increases in ventricular wall thickness at the expense of the chamber volume, producing a relative increase in the wall-to-chamber ratio. Although concentric growth can be uniform across the ventricles, it most often occurs as an exaggerated growth of the interventricular septum or posterior wall in response to increased pressure loads, as from hypertension or valvular stenosis (27). Because we did not find any valvular defect in HMGA1-KO mice affected by cardiac hypertrophy, and their blood pressure levels were comparable to WT controls, we excluded that this phenotype occurred as a secondary event in response to stress and suggest that the lack or down-regulation of the Hmga1 gene could directly regulate the cardiomyocytic growth. This hypothesis seems to be confirmed by our in vitro experiments on neonatal rat cardiomyocytic cells, demonstrating that the down-regulation of the HMGA1 protein is by itself responsible for the hypertrophic phenotype, and suggesting that this architectural factor could somehow be involved in the negative control of pathologic cardiac signaling. Interestingly, analysis of the hypertrophic gene expression in hypertrophic hearts reveals a drastic increase of the CaMKII gene. Consistently, we observed increased levels of the CaMKII protein in cardiomyocytes in which HMGA1 expression was knocked down compared with control cells. These results suggest a direct down-regulation of CaMKII gene transcription by HMGA1 and a critical role of the CaMKII increased levels in the cardiac hypertrophic process developed by HMGA1-KO mice. Moreover, because cardiomyocytes stimulated by IGF-I showed decreased levels of HMGA1 and increased levels of CaMKII, we suggest that both HMGA1 and IGF-I are involved in a common pathway regulating cardiomyocytic growth. However, both IGF-I and CaMKII pathways are complex and require further experiments to better elucidate the exact cascade of molecular events that lead to either HMGA1 up-regulation, following IGF-I stimulation, or to CaMKII up-regulation, following HMGA2 inhibition.
Potential oncosuppressor role of HMGA1 in lymphoproliferative malignancies. Heterozygous and homozygous Hmga1-null mice develop age-dependent splenomegaly associated with lymphoid cell expansions, resembling various human B-cell lymphomas. However, unlike human lymphoid malignancies that are always characterized by a monoclonal expansion of lymphocytes, the lymphomas developed by Hmga1-ko mice appear oligoclonal, as previously reported in other engineered mice (20, 28, 29). The onset of a B cell–specific malignant phenotype in HMGA1-KO mice, as opposed to the T cell–specific malignant phenotype in Hmga1-overexpressing mice, may be due to the direct role of the HMGA1 protein on the regulation of specific cytokines, that in turn promote the B or T cell expansion. In fact, it is known that HMGA1 is a modulator of the function of many of the transcription factors that control cytokine gene transcription (30). In particular, it positively regulates both IL-2 and IL-2Rα expression (31, 32), required for T cell expansion, accounting for the impaired T cell differentiation in HMGA1-KO ES cells (19). HMGA1-KO adult splenocytes show decreased expression of IL-2 and increased expression of IL-6, required for the B cell expansion, compared with WT controls. Moreover, they also show overexpression of both Rag1 and Rag2 genes due to a direct negative regulation by the HMGA1 protein. Because Rag genes code for a protein complex involved in the correct rearrangement of the immunoglobulin chains, we analyzed the IgH locus and found aberrant rearrangements in most pathologic spleen samples from HMGA1-KO mice. These data strongly suggest that RAG-mediated double-strand breaks (DSB) are joined, in the presence of reduced or null amounts of HMGA1, by nonclassical means that lead to the formation of aberrant chromosomal rearrangements. Interestingly, it has been very recently shown that RAG proteins are generally implicated in the maintenance of genomic stability (33). In fact, it has been proposed that the continued presence of RAG1 and RAG2 might protect DNA nicks from repair, allowing cell cycle progression and formation of a replication-induced DSB. Accordingly, a study from the Alt laboratory previously provided convincing evidence that V(D)J recombination intermediates are involved in generating oncogenic translocations (34).
The phenotype of the HMGA1-KO mice further confirms that HMGA1 and HMGA2 genes have distinct physiologic functions. In fact, the phenotype of Hmga1-null mice is vastly different from that of Hmga2-null mice previously reported (4). In fact, at odds with the “pygmy” phenotype of Hmga2-null mice, homozygous Hmga1-KO mice are the same size as their WT littermates. Equally, fat tissue is drastically reduced in Hmga2-null mice, whereas it does not seem to be modified in Hmga1-null mice. Moreover, neither heart changes nor hematologic alterations have been observed in the “pygmy” mice. In this study, we show the unexpected appearance of a neoplastic phenotype in Hmga1-KO mice. This indicates that the HMGA1 gene plays a hitherto unsuspected tumor suppressor role. Moreover, because the neoplastic tissues continue to express WT HMGA1 mRNA and protein in heterozygous animals (data not shown), HMGA1 seems to act as a haploinsufficient tumor suppressor gene. This seems to contrast with the oncogenic potential of HMGA1 that several studies, both in vitro and in vivo, have clearly established (10, 13–15). Therefore, the results reported here indicate the intriguing effect of a gene that is endowed with both oncogenic and antioncogenic properties. This might not be surprising given the mechanism of action of HMGA proteins. In fact, they do not exert transcriptional activity per se but they regulate the expression of several genes by interacting with diverse partners (2). Because the protein partners of the HMGA proteins might change depending on the cell type, HMGA1-dependent gene regulation may also differ from cell to cell, depending on the cellular context (35), which would also account for the dual effect of the HMGA1 protein on cell transformation.
Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).
Acknowledgments
Grant support: Associazione Italiana per la Ricerca sul Cancro, BioGem, and Ministero dell' Universita e della Ricerca Scientifica e Tecnologica, projects “Terapie antineoplastiche innovative” and “Piani di Potenziamento della Rete Scientifica e Tecnologica,” the Programma Italia-USA sulla Terapia dei Tumori coordinated by Prof. Cesare Peschle, by the MIUR project “Terapie antineoplastiche innovative,” and the “Ministero della Sanità.”
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
We are grateful to A. Affuso for advice on mice management, to A. Luciano for animal care, to J. Martinez Hoyos for adenovirus amplification, and to B. Rotoli and A. Risitano for cytomorphologic analysis of spleen spreads; to D. Palmieri for his help and the Associazione Partenopea per le Ricerche Oncologiche for its support; and to Jean Ann Gilder for editing the text.