|Home | About | Journals | Submit | Contact Us | Français|
Although expansion of CAG repeats in ATAXIN1 (ATXN1) causes Spinocerebellar ataxia type 1, the functions of ATXN1 and ATAXIN1-Like (ATXN1L) remain poorly understood. To investigate the function of these proteins, we generated and characterized Atxn1L−/− and Atxn1−/−; Atxn1L−/− mice. Atxn1L−/− mice have hydrocephalus, omphalocoele and lung alveolarization defects. These phenotypes are more penetrant and severe in Atxn1−/−; Atxn1L−/− mice, suggesting that Atxn1 and Atxn1L are functionally redundant. Upon pursuing the molecular mechanism, we discovered that several Matrix metalloproteinase (Mmp) genes are overexpressed and that the transcriptional repressor Capicua (Cic) is destabilized in Atxn1L−/− lungs. Consistent with this, Cic deficiency causes lung alveolarization defect. Loss of either Atxn1L or Cic derepresses Etv4, an activator for Mmp genes, thereby mediating Mmp9 overexpression. These findings demonstrate a critical role of ATXN1/ATXN1L-CIC complexes in extracellular matrix (ECM) remodeling during development and their potential roles in pathogenesis of disorders affecting ECM remodeling.
The abnormal expansion of a CAG-repeat encoding a polyglutamine (polyQ) tract in ATXN1 causes the neurodegenerative disease Spinocerebellar ataxia type 1 (SCA1) (Orr et al., 1993; Banfi et al., 1994). The resulting mutant ATXN1 protein acquires toxic functions that cause progressive degeneration of the cerebellum, brainstem, and spinocerebellar tracts (Zoghbi and Orr, 1995). Since ATXN1 was identified as the gene causing SCA1, the majority of studies on ATXN1 have been focused on uncovering the molecular mechanisms underlying the neurotoxicity of the mutant form of the protein. However, the function of wild-type ATXN1 remains unclear. In an effort to understand the in vivo function of ATXN1, we previously generated Atxn1-null mice (Matilla et al., 1998). These mice display deficits in spatial and motor learning, but are otherwise viable and fertile with no discernable defects in other tissues (Matilla et al., 1998), suggesting that ATXN1 may not be essential for normal development of peripheral tissues.
A mammalian paralog of ATXN1 was identified and named as ATAXIN1-Like (ATXN1L) (also called Brother of ATXN1, BOAT) (Bowman et al., 2007; Mizutani et al., 2005). ATXN1 and ATXN1L share two highly conserved domains, one in the amino terminal region and the other in the carboxyl terminal region termed Ataxin1 and HMG-box protein1 (AXH) domain (Mizutani et al., 2005), respectively. They are expressed ubiquitously and their tissue distributions are very similar (Mizutani et al., 2005), suggesting that they may be functionally redundant.
To date, most of the clues about the potential functions of these two proteins stem from knowledge of their interactors. ATXN1 and ATXN1L share many common binding partners, including Capicua (CIC), Silencing mediator of retinoid and thyroid hormone receptors (SMRT) and Histone deacetylase 3 (HDAC3) (Mizutani et al., 2005; Lam et al., 2006; Lim et al., 2006). Among the interactors, CIC is of particular interest, because so far it is the only binding partner whose protein levels are significantly reduced in Atxn1-null mice (Lam et al., 2006). CIC is an HMG box-containing DNA-binding protein that is evolutionarily conserved from Caenorhabditis elegans to human (Jiménez et al., 2000), and functions as a transcriptional repressor by preferential binding to TGAATGA/GA sequences in Drosophila and mammals (Ajuria et al., 2011; Kawamura-Saito et al., 2006). In Drosophila, Cic functions downstream of receptor tyrosine kinase (RTK) pathways including Torso, EGFR, Ras, Raf, and mitogen-associated protein kinases (MAPKs) to mediate specification of intervein areas in the wing, correct development of the head and tail, and dorsoventral patterning (Jiménez et al., 2000; Roch et al., 2002). In mammalian HEK293 and melanoma cells, MAPK signaling results in phosphorylation of CIC and subsequent loss of CIC-mediated transcriptional repression of PEA3 group genes (Dissanayake et al., 2010).
Cic exists in a large protein complex of about 2MDa in size together with Atxn1 and Atxn1L in mouse cerebellum (Bowman et al., 2007; Lam et al., 2006). Both ATXN1 and ATXN1L bind to CIC, and compete with each other for its binding (Bowman et al., 2007). In cerebella from Atxn1-null mice, Cic protein levels are decreased but mRNA levels are not changed, suggesting that Atxn1 stabilizes Cic through protein-protein interactions (Lam et al., 2006). Moreover, both ATXN1 and ATXN1L can enhance the transcriptional repressor activity of CIC in vitro (Lam et al., 2006; Crespo-Barreto et al., 2010), suggesting that some endogenous functions of the ATXN1 protein family must occur in cooperation with CIC.
Given that deletion of Atxn1 in mice yielded subtle learning and memory phenotypes but no insight into the cellular functions of this protein (Matilla et al., 1998), we hypothesized that ATXN1 and ATXN1L functionally substitute for each other and that we would need double mutant mice to understand the endogenous functions of the ATXN1 protein family. To this end, we generated Atxn1L−/− mice and characterized the phenotypes of either Atxn1L−/− or Atxn1−/−; Atxn1L−/− mice. We discovered that loss of Atxn1L destabilizes Cic and affects postnatal viability; that Atxn1 and Atxn1L are functionally redundant, as evident by the developmental defects and perinatal lethality of the double null mice; and that the Atxn1 protein family, together with Cic, regulates extracellular matrix (ECM) remodeling during development.
We generated Atxn1L knock-out mice using homologous recombination to target the gene in embryonic stem cells (Figure S1A). This strategy abolished the expression of Atxn1L (Figure S1B). In a mixed 129S6/SvEv and C57BL/6J background, all of the double null mice died before weaning age (P21) (Table 1). Interestingly, about 40% of Atxn1+/−; Atxn1L−/− mice also died before P21, whereas Atxn1−/−; Atxn1L+/− mice were viable at the weaning age, suggesting that Atxn1L is more crucial than Atxn1 for viability (Table 1). When we back-crossed Atxn1L mutant mice to an almost pure C57BL/6J background (more than 7 generations), we found that Atxn1L−/− mice were smaller than their littermates (Figures 1A and S1C) and that 50% of them died before P21 (Table S1). Moreover, about 31% of the surviving Atxn1L−/− animals developed hydrocephalus (Figure 1B). Symptomatic Atxn1L−/− mice developed a dome-shaped head, kyphosis, lethargy, and emaciation sometime between one and four weeks after birth (data not shown). In addition, Atxn1L+/− animals developed hydrocephalus at a very low frequency (< 1%) (Figure 1B). Deficiency of Atxn1L thus results in growth retardation, hydrocephalus, and perinatal lethality on a C57BL/6J background.
Given the perinatal lethality of the Atxn1−/−; Atxn1L−/− mice, we tried to determine if loss of these two proteins causes embryonic lethality. Of the 178 newborn pups from Atxn1+/−; Atxn1L+/− intercrosses, we observed 11 double null pups. This was consistent with the expected Mendelian ratio (one in sixteen), indicating that loss of Atxn1 and Atxn1L does not cause embryonic lethality (Table 1). However, about 73% (8/11) of the double null pups were cyanotic (Figure 1C) and died within three hrs after birth. This early lethality was also found in Atxn1+/−; Atxn1L−/−, Atxn1−/−; Atxn1L+/−, and even in Atxn1L−/− animals, with lower ratios (less than 20%; data not shown).
To investigate the cause of lethality in the double null mice, we analyzed neonates and E18.5 - E19 embryos by anatomical and histological approaches. Given the hydrocephalus in Atxn1L−/− mice, we checked brain morphology and found that the third and lateral ventricles were enlarged in 75% (3/4) of the double null mice at birth, whereas 30% (3/10) of Atxn1+/−; Atxn1L−/− and about 11% (1/9) of Atxn1L−/− neonates had this symptom (Figure 1D and Table 2). We also observed an omphalocoele (umbilical hernia), a mild type of abdominal wall closure defect, in about 45% of the double null embryos, and to a lesser degree in Atxn1+/−; Atxn1L−/− embryos (Figure 1E and Table 1). It is known that an omphalocoele results in the loss of internal organs, such as gut and liver, because the mother cannibalizes organs protruding into the umbilical ring in the process of removing the placenta after birth (Thumkeo et al., 2005). Anatomical analysis of organogenesis in the newborn pups revealed that about 45% of the double null pups lost the midgut region from the small intestine to the cecum after birth (Figure 1E and Table 1). The frequency of omphalocoeles and gut loss was higher in Atxn1+/−; Atxn1L−/− animals (17%) than in Atxn1−/−; Atxn1L+/− animals (8%), and these defects were sometimes observed in Atxn1L−/− (4%), but not in Atxn1−/− animals (Tables 1 and and2).2). Collectively, these data suggest that Atxn1L plays a more important role than Atxn1 for viability.
Histological analysis of Atxn1−/−; Atxn1L−/− neonates and E17.5-18.5 embryos revealed no abnormalities in most major organs. However, we found that the double null embryos had a lower proportion of empty space, representing saccules or canaliculi, in lung sections in comparison with double heterozygous littermates (Figures S2A and S2B). This finding suggested that lung development may be abnormal in the double null embryos. Mouse lung development proceeds in five distinct stages: primary budding (E9.5-11.5), pseudoglandular (E11.5-16.5), canalicular (E16.5-17.5), saccular (E17.5-P5), and alveolar (P5-P30) stages (Shi et al., 2007; Greenlee et al., 2007). Given that defects in saccular development and septation at embryonic stages eventually lead to alveolarization defects at postnatal stages (Oblander et al., 2005), we assessed lung morphology in WT, Atxn1−/−, and Atxn1L−/− mice at both early (P6) and late (P17-23) alveolar stages. Histological analysis of lungs showed that lung morphology was comparable among the three different genotypes at P6, whereas alveolarization defects causing air space enlargement were apparent in 65% (13/20) of Atxn1L−/− mice at late alveolar stage (P17-23) (Figure 2A and Table 2). Neither Atxn1−/− (0/13) nor wild-type mice (0/33) displayed defective alveolarization at this timepoint (Figure 2A and Table 2). Furthermore, the alveolarization defect was more severe in Atxn1+/−; Atxn1L−/− mice compared with Atxn1L−/− mice, suggesting that Atxn1 partially compensates for loss of Atxn1L function (Figures S2C and S2D). Early perinatal lethality precluded an analysis of lung alveolarization in the double null mice (Table 2). We also assessed epithelial cell formation in lung tissues by hematoxylin and eosin (H&E) staining and immunohistochemistry for lung epithelial cell specific markers (Clara Cell Secretory Protein (CCSP) for Clara cells and Surfactant Protein C (SFTPC) for type II lung epithelial cells) (Liu et al., 2003) and found that the lung epithelia form normally in Atxn1L−/− mice at both early and late alveolar stages (Figure S3A).
To elucidate the molecular mechanism mediating the alveolarization defect, we carried out microarray analyses using lung total RNA from WT, Atxn1−/− and Atxn1L−/− mice at P6, a timepoint prior to the manifestation of alveolarization defects. We found a total of 406 gene expression alterations (328 up-regulated and 78 down-regulated) with a fold change > 1.2 (P value < 0.05) in Atxn1L−/− lung tissues (Table S2B), whereas only 54 genes (42 up-regulated and 12 down-regulated) were altered in Atxn1−/− lungs (Table S2A). Eighteen genes are shared amongst the differentially expressed genes (DEGs) in Atxn1−/− and Atxn1L−/− lungs, and about 72% of them (13 out of 18) have a higher fold change in Atxn1L−/− mice compared to Atxn1−/− mice (Table S2C). Overall, these data suggest that the regulation of gene expression relies more on Atxn1L rather than Atxn1 in lungs at P6. To examine whether particular gene classes were enriched amongst the DEGs in Atxn1L-null lung tissue, we analyzed the DEGs for gene ontology (GO) terms related to cellular component using DAVID software (Dennis et al., 2003). We found that the GO terms related to cell surface, plasma membrane and extracellular region were significantly highly ranked among the top 10 categories (Table S2D), suggesting that compositions of the external area of cells might be altered in Atxn1L−/− lungs. Specifically, we found one set of genes whose expression changes may be directly responsible for the phenotypes of Atxn1 and Atxn1L double mutants. Matrix metalloproteinase (Mmp) genes, the critical players in ECM remodeling, were up-regulated in lung tissue from Atxn1L−/− mice (Table S2B). Either the overexpression of MMP9 in alveolar macrophages, the induction of MMP12 expression in lung epithelial cells or the loss of Tissue inhibitor of metalloproteinase 3 (Timp3) causes air space enlargement in mouse lungs (Foronjy et al., 2008; Qu et al., 2009; Leco et al., 2001), a prominent phenotype in mice lacking Atxn1L. It is also noteworthy that alterations in MMP levels and ECM formation are closely associated with pathogenesis of hydrocephalus and abdominal wall hernias (Wyss-Coray et al., 1995; Zechel et al., 2002; Antoniou et al., 2009; Suzuki et al., 1996), both highly penetrant in Atxn1L-null mice. Therefore, we set out to confirm the Mmp gene expression changes observed by transcriptional profiling using qRT-PCR. Among eight Mmp genes, we found that Mmp8, Mmp9, Mmp12 and Mmp13 levels were significantly up-regulated in Atxn1L−/− lung tissues obtained from an independent cohort of animals, consistent with the microarray results (Figure 2B and Table S2B). However, also consistent with the microarray results (Table S2A), the levels of most Mmp genes, except for Mmp12 (P value = 0.0542), were not significantly changed in Atxn1−/− lung tissues, suggesting that increased levels of multiple Mmp genes may be related with the alveolarization defect in Atxn1L−/− mice (Figure 2B). We confirmed the overexpression of Mmp9 proteins in lung and meningeal tissues from Atxn1L−/− mice compared with their WT littermates (Figure 2C).
Given that alveolar macrophages are the major source for secretion of Mmp9 and Mmp12 in lungs (Greenlee et al., 2007), we investigated whether the number of alveolar macrophages is increased in Atxn1L−/− mice compared with WT mice. We counted the number of alveolar macrophage cells in bronchoalveolar lavage fluid (BALF) and found that there is no difference in the alveolar macrophage cell numbers between WT and Atxn1L−/− mice (Figure S3B), suggesting that up-regulation of a subset of Mmp gene levels in Atxn1L−/− mice is not simply due to increase in the number of macrophage cells in lungs. Given that TIMPs are physiological inhibitors for MMPs, we checked whether levels of Timp genes are changed in Atxn1L−/− mice. Microarray and qRT-PCR analyses showed that the levels of Timp genes were not significantly changed in lungs from Atxn1L−/− mice (Figure 2D and Table S2B).
We further assessed the integrity of ECM formation in Atxn1L−/− lung tissues by Verhoeff staining for elastin to investigate whether the overexpression of Mmps indeed affects ECM formation in Atxn1L−/− mice. Reduction in the area occupied by elastic fibrils within the alveolar walls was evident in symptomatic 7-9 month old Atxn1L−/− mice compared with WT mice (Figures 2E and 2F). These data suggest that loss of Atxn1L causes the alveolarization defects associated with the overexpression of Mmps and the decrease in elastic fiber formation in the lungs.
To gain further insight into the molecular mechanism underlying the phenotypes and the overexpression of a subset of Mmp genes in the mutant animals, we focused on a common interactor of ATXN1 and ATXN1L. Both Atxn1 and Atxn1L bind the transcriptional repressor Cic to form endogenous protein complexes in mouse cerebella (Bowman et al., 2007; Lim et al., 2006). Co-expression of either ATXN1 or ATXN1L with CIC synergistically enhanced the transcriptional repressor activity of CIC in vitro (Lam et al., 2006 ; Crespo-Barreto et al., 2010), suggesting that ATXN1/ATXN1L and CIC could cooperatively function to regulate expression of CIC target genes. To investigate whether this functional relationship between ATXN1/ATXN1L and CIC is conserved in other peripheral tissues or restricted only in brain, we checked expression profiles of Atxn1, Atxn1L and Cic in various tissues from E18.5 embryos. We found that their expression patterns were very similar, suggesting that Atxn1/Atxn1L and Cic might function together in peripheral tissues during embryogenesis (Figure 3A). Previous work revealed that loss of Atxn1 reduces levels of Cic protein, without change in Cic mRNA levels, in adult mouse cerebella (Lam et al., 2006). Accordingly, we checked Cic levels in brain and lung tissues from E18.5 embryos of five different genotypes. We found that both Atxn1 and Atxn1L are required for maintaining the steady state level of Cic but that Cic levels are more dramatically decreased in Atxn1+/−; Atxn1L−/− than in Atxn1−/−; Atxn1L+/− mice, suggesting that Atxn1L plays a critical role for stabilizing Cic during embryogenesis (Figures 3B and 3C). The predominant role of Atxn1L for Cic stabilization was also observed in lung tissue during postnatal stages. Reduced levels of Cic protein were more apparent in Atxn1L−/− mice compared with Atxn1−/− mice at P6 (Figures 3D and 3E), whereas Cic mRNA levels were not significantly changed in Atxn1L−/− mice (data not shown).
Given the critical roles of Atxn1L in both the phenotypes found in mutant animals and the stabilization of Cic, we hypothesized that the phenotypes might be related to reduced Cic levels. Recently, we generated Cic mutant (Cic-L−/−) mice using ES cells in which the Cic gene is targeted by a β-geo genetrap cassette (Baygenomics). The expression of the longer form of Cic (Cic-L) is completely abolished in these mice, whereas about 15% of the short form of Cic (Cic-S) is still expressed in the cerebella (Fryer and Zoghbi, unpublished data). The majority of Cic-L−/− mice died before weaning age (P21), and some of the survivors were smaller than their littermates on the C57BL/6J background (Fryer and Zoghbi, unpublished data). To test the hypothesis that Cic deficiency causes alveolarization defects comparable to the defects found in Atxn1L-null mice, we examined the lung morphology and Mmp9 levels in the Cic-L−/− survivors. The Cic-L−/− mice indeed have increased levels of Mmp9 and severe alveolarization defects, compared with Cic-L+/− mice (Figures 3F and 3G). These data suggest that the lung alveolarization defect in Atxn1 and Atxn1L double mutant mice are likely due to deficiency of Atxn1/Atxn1L-Cic complexes.
Given the possibility that deficiency of Atxn1/Atxn1L-Cic complexes is responsible for the alveolarization defect, we turned our attention to the Cic target genes that may contribute to the alteration of Mmp gene expression in Atxn1L−/− mice. Among the significantly up-regulated genes (fold increase >1.2, P value <0.05) in Atxn1L−/− lung tissues, we searched for CIC binding motifs (TGAATGA/GA) within a 1 kb region upstream from the transcriptional start site. We also considered conservation scores of the 30 vertebrate species from the phastCons track of the UCSC Genome Browser (http://genome.ucsc.edu/) (Siepel et al., 2005). We discovered that four genes, Etv4 (also known as Pea3), Prkcb, Nfkbie, and Runx3, had between one to four CIC binding motifs and were significantly up-regulated by 1.2-1.4 fold (P value < 0.01) (Table S2B). Among these genes, Etv4 got our attention, not only because it has the highest fold increase and significance (Table S2B), but also because it is known as a transcriptional activator for many different types of MMP genes (Gum et al., 1996; Yan and Boyd, 2007). Moreover, more than one purine rich PEA3 element (A/CGGAA/T) exists in promoter regions (within 5 kb upstream from the transcription start site) of most mouse Mmp genes (Table S3). Etv4 is one of the members in Pea3 group transcription factors, which include Etv1/Er81 and Etv5/Erm. All Pea3 group members have at least one Cic binding motif in their promoter regions (Figure 4A). Notably, previous studies have also verified that their expression is regulated by CIC (Kawamura-Saito et al., 2006; Dissanayake et al., 2011). To examine the possibility that expression of Pea3 group genes could be regulated by Cic in lung, we investigated Cic promoter occupancy of Pea3 group genes in lung by chromatin immunoprecipitation (ChIP) using anti-Cic antibodies followed by PCR of the promoter regions containing Cic binding motifs (Figures 4A and 4B). We found that Cic is indeed bound to the promoters of Pea3 group genes in lung cells (Figure 4B). Next, we checked the levels of Pea3 group genes in lung tissues from 6 day-old WT, Atxn1−/−, and Atxn1L−/− mice by qRT-PCR. The levels of Etv1 and Etv4 were significantly up-regulated in Atxn1L−/− mice, whereas Atxn1−/− mice did not have significant alterations in the levels of these three genes (Figure 4C), suggesting that Atxn1L plays a more important role than Atxn1 in regulation of Pea3 group gene expression and consistent with the predominant effect of Atxn1L deficiency on destabilization of Cic (Figure 3E). An increase in Etv4 protein levels in Atxn1L−/− mice was also confirmed by Western blot analysis (Figure 4D). We then compared the levels of Pea3 group genes among Atxn1 and Atxn1L double mutants. The levels of Pea3 group genes were most dramatically and significantly up-regulated in the double null embryos at E18.5 (Figure 4E). Moreover, the levels of Pea3 group genes were inversely correlated with the levels of Cic proteins, demonstrating that mice with the most significant reduction in Cic levels had the highest fold change in Pea3 group gene expression (Figures 3C and and4E).4E). These data provide further support to the idea that the loss of Atxn1/Atxn1L leads to derepression of Cic target genes due to destabilization of Cic.
Previous reports demonstrated that ETV4 activates transcription of MMP9 in various cancer cell lines (Hida et al., 1997; Qin et al., 2008). Enhanced expression of MMP9 has been observed in alveolar macrophages from chronic obstructive pulmonary disease (COPD) patients (Finlay et al., 1997). Furthermore, transgenic overexpression of MMP9 in alveolar macrophages causes emphysema in mice and is associated with the loss of alveolar elastin (Foronjy et al., 2008). Therefore, we hypothesized that increased levels of Etv4 may lead to the overexpression of Mmp9 in either Atxn1L−/− or Cic-L−/− mice. To test this hypothesis, we first investigated whether Etv4 regulates expression of Mmp9 in mouse alveolar macrophage cell line (MH-S) by knocking-down Etv4. The levels of Mmp9 are reduced upon treatment with two different siRNAs against Etv4 in MH-S cells, suggesting that Etv4 indeed regulates expression of Mmp9 in alveolar macrophages (Figure 5A). However, knocking-down ETV4 did not change the levels of MMP9 in immortalized lung epithelial cells (A549) (Figure S4), suggesting that regulation of MMP9 expression by ETV4 might be cell type-specific. Next, we treated MH-S cells with siRNAs against Cic to examine whether knocking-down Cic up-regulates the levels of Mmp9 and Etv4. Expression of both Mmp9 and Etv4 is indeed highly induced in the cells treated with Cic siRNAs compared with the cells treated with negative control siRNAs (Figures 5B-5D). Finally, we transfected Etv4 siRNAs in cells with reduced Cic. The addition of Etv4 siRNAs rescued the levels of both Etv4 and Mmp9 to almost normal levels without affecting Cic knock-down efficiency, suggesting that loss of Cic mediates overexpression of Mmp9 in alveolar macrophages by derepression of Etv4 (Figures 5B-5D).
In this study, we set out to understand the in vivo function of the ATXN1 protein family by generating and characterizing both Atxn1L−/− and Atxn1−/−; Atxn1L−/− mutant animals. We uncovered several developmental deficits potentially caused by misregulation of ECM remodeling in the mutant mice. These phenotypes appear to be caused by a deficiency of Atxn1/Atxn1L-Cic complexes because we found that stability of Cic proteins relies on both Atxn1 and Atxn1L and that Cic-L−/− mice have the same developmental abnormalities found in the Atxn1 and Atxn1L double mutant animals. As a consequence, loss of Atxn1/Atxn1L-Cic complexes derepresses Cic target genes including Pea3 group genes, leading to overexpression of Mmps and defects in ECM remodeling.
Given that ATXN1 and ATXN1L share conserved domains and several interactors and that their expression profiles are very similar (Mizutani et al., 2005), we hypothesized that functional redundancy of Atxn1L might compensate for the deficiency of Atxn1 function in Atxn1-null mice. Loss of both Atxn1 and Atxn1L results in early perinatal lethality and several developmental abnormalities during embryogenesis. Moreover, a complete removal of Atxn1 from Atxn1L-null mice augments the incidence and severity of the phenotypes in Atxn1L-null mice, suggesting that Atxn1 partially compensates for the loss of Atxn1L function. Interestingly, our data consistently show that Atxn1L is more critical than Atxn1 in the manifestation of all the phenotypes. This phenomenon seems to be closely associated with the dominant role of Atxn1L in stabilization of Cic and raises the question of how this occurs. One simple explanation is that the absolute number of ATXN1L molecules might be much greater than that of ATXN1 such that ATXN1L-CIC complexes predominate in most cells. Another explanation is that CIC may prefer ATXN1L, rather than ATXN1, as a binding partner. In this case, it would be important to do an in-depth comparative structural analysis between ATXN1 and ATXN1L, especially for the AXH domain that is responsible for the interaction of ATXN1 with CIC (Lam et al., 2006). Overall, the data in this study show that Atxn1 and Atxn1L are functionally redundant and that Atxn1L plays a pivotal role in development.
Our data showed that loss of Atxn1L causes the lung alveolarization defect associated with overexpression of Mmp genes and reduction in elastic fibril formation within alveolar walls. Many genetic studies of mice that are defective in regulation of ECM remodeling have pointed to the pivotal role of the ECM remodeling in lung alveolarization. Transgenic overexpression of either MMP1 or MMP9 in lungs leads to progressive adult-onset (at around 4-6 months after birth) emphysema, whereas mice lacking Timp3 spontaneously develop enlarged air spaces in the lung at 2 weeks after birth, with disease severity progressing with age (Foronjy et al., 2003; Foronjy et al., 2008; Leco et al., 2001). These studies not only suggest that a tight regulation of Mmp activity is critical for normal lung development, but also indicate that hyperactivation of multiple Mmp enzymes causes an earlier onset of lung alveolarization defects rather than the overexpression of a single MMP gene. We found that Atxn1L-null mice have increased levels for Mmp genes (Mmp8, Mmp9, Mmp12 and Mmp13) in the lung at P6 and develop the alveolarization defect leading to enlarged air spaces at late alveolar stages (P17-P23). Comparing our findings with the previous reports, increased levels of Mmp8, Mmp9, Mmp12 and Mmp13 might cooperatively contribute to the onset of lung alveolarization defects in Atxn1L−/− mice. It is interesting that the Atxn1 and Atxn1L double mutant mice have other phenotypes, which are also reminiscent of defects in ECM formation. Hydrocephalus is caused by obstruction in circulation of cerebrospinal fluid (CSF) leading to accumulation of CSF and expansion of ventricles in brain. The ECM composition in the meninges, where the arachnoid villi are located, is regarded as an important environment for absorption of CSF into arachnoid villi (Zhao et al., 2010). Several studies using genetically engineered mouse models have indicated that alterations in Mmp9 levels are associated with development of hydrocephalus (Zechel et al., 2002; Muñoz et al., 2006; Oshima et al., 1996). Therefore, our finding that Mmp9 levels are up-regulated in the meninges from Atxn1L−/− mice may explain the onset of hydrocephalus in Atxn1L−/− mice. Abdominal wall hernias are considered a disease of the ECM because disturbances in collagen metabolism are strongly associated with these developmental defects in humans (Antoniou et al., 2009). Several clinical studies have shown that MMP1, MMP2, MMP9 and MMP13 are overexpressed in patients with abdominal wall hernias (Antoniou et al., 2009). In this regard, the omphalocoele in Atxn1−/−; Atxn1L−/− double mutant embryos might result from the role of Atxn1 protein family in ECM formation. Taken together, our findings suggest that the ATXN1 protein family plays a critical role in ECM remodeling during development.
In this study, we found that Cic levels are decreased whereas Pea3 group genes are overexpressed in Atxn1−/−; Atxn1L−/− double mutant animals. We also verified that expression of Pea3 group genes is regulated by Atxn1/Atxn1L-Cic transcriptional repressor complexes in lung cells. Given that more than one purine rich PEA3 element (A/CGGAA/T) exists in promoter regions of most Mmp genes, overexpression of a subset of Mmp genes in Atxn1L−/− mice may be mediated by derepression of Pea3 group genes. Indeed, we showed that derepression of Etv4 due to knock-down of Cic mediates the overexpression of Mmp9 in the alveolar macrophage cells. It has also been shown that Etv4 regulates Mmp13 expression in hepatic stellate cells (Díaz-Sanjuán et al., 2009). Altogether, our study demonstrates that ATXN1/ATXN1L-CIC complexes directly regulate the expression of PEA3 group genes, thereby mediating the regulation of a subset of MMP genes.
The pathogenesis of many human diseases, such as rheumatoid arthritis, asthma, artherosclerosis, hypertension and fibrosis, is strongly associated with the misregulation of ECM remodeling due to alterations in MMP levels. Moreover, in cancer, altered proteolysis leads to unregulated tumor growth, inflammation, tissue invasion and metastasis (Kessenbrock et al., 2010). Indeed, it is well known that increased expression of MMPs is associated with cancer cell invasion and migration (Roy et al., 2009). Additionally, overexpression of PEA3 group genes have been found in many different human tumors including breast cancer, nonsmall cell lung carcinoma and leukemia (Kurpios et al., 2003). Given the critical roles of MMPs and PEA3 group proteins in tumorigenesis and cancer metastasis, our data suggest that the ATXN1/ATXN1L-CIC complex might potentially affect tumorigenesis or cancer metastasis. Intriguingly, one frameshift and three missense mutations in the CIC gene have been identified in patients with breast cancer and lung cancer cells (Sjöblom et al., 2006; Kan et al., 2010). Recently, mutations in CIC gene were found in tumor tissue from patients with oligodendrogliomas (OD), the second most common malignant brain tumor in adults (Bettegowda et al., 2011). Moreover, analysis on the expression profiles of ATXN1, ATXN1L and CIC in various cancer samples using the Oncomine database analysis tool (http://www.oncomine.org/) reveals that the levels of all three genes are significantly down-regulated in breast and brain cancer patients (data not shown), suggesting that loss of ATXN1/ATXN1L-CIC functions might contribute to tumorigenesis of particular types of cancer. In sum, this study reveals that ATXN1 protein family, together with CIC, is critical for development and for regulating key steps in transcriptional control of ECM remodeling. Moreover, these findings raise the interesting possibility about a potential role of these proteins in the pathogenesis of many human diseases associated with defects in ECM remodeling.
Generation of Atxn1L knock-out mice was carried out using a targeting construct previously used to generate a duplication allele (Bowman et al., 2007). To generate a replacement vector, the targeting construct was linearized using SfiI. Identification of targeted cells and generation of chimeras and Atxn1L+/− mice were done as previously described (Bowman et al., 2007).
To test viability of mutant mice at P21 (Table 1), we mated mice using the following combinations: Atxn1+/−; Atxn1L+/− and Atxn1+/−; Atxn1L+/−, Atxn1+/−; Atxn1L+/− and Atxn1−/−; Atxn1L+/− or Atxn1+/−; Atxn1L+/− and Atxn1+/−; Atxn1L−/−. To investigate the omphalocoele at E18.5-E19 (Table 1), we obtained the embryos from crosses between Atxn1+/−; Atxn1L+/− and Atxn1+/−; Atxn1L+/−, Atxn1+/−; Atxn1L+/− and Atxn1−/−; Atxn1L+/− or Atxn1+/−; Atxn1L+/− and Atxn1+/−; Atxn1L−/−.
Brain and lung tissues were dissected and fixed with 10% formalin (Sigma) at 4°C overnight. The lungs from mice older than 3 weeks were pressure perfused at 20cm H2O with 10% formalin at 4°C overnight. Tissues were dehydrated gradually by 70% to 100% EtOH, equilibrated with chloroform overnight, embeded in paraffin blocks, and sectioned at 6 μm thickness. Sections mounted onto slides were stained with H&E for histological analysis. Lungs from more than twenty 2.5-3.5 week old Atxn1L−/− and four 2-3 week old Cic-L−/− mice were analysed with H&E staining. For Verhoeff staining, ACCUSTAIN Elastic Stain kit (Sigma) was used according to manufacturer’s instruction. The step for staining in Van Gieson solution was skipped.
Four 10x images were captured from H&E-stained lung tissue sections from E17.5 double heterozygous and double null embryos. Each image was converted into 8 bit binary image with ImageJ software. The empty space area was determined by calculating the number of white pixels using ImageJ software and then divided by the number of pixels for whole image area to get the percentage of empty space area in the lung sections.
The lung tissues embeded in paraffin blocks were sectioned at 6 μm thickness and subjected to the immunohistochemistry using rabbit polyclonal antibodies for CCSP (Abcam, 1:2000) and SFTPC (Millipore, 1:5000). The VECTASTATIN Elite ABC kit (Vector Lab) was used for the immunohistochemistry according to the manufacturer’s instruction.
Measurement of mean linear intercept (MLI) was conducted as described previously with some modification (Robbesom et al., 2003). Sixteen 40x images were randomly captured from H&E-stained lung tissue sections from 2.5-3 week old WT, Atxn1L−/−, and Atxn1+/−; Atxn1L−/− mice and analyzed for the MLI. Total number of counted intercepts was about 900 per each genotype.
Bronchoalveolar lavage (BAL) fluid collection and quantitation of airway alveolar macrophage cells were carried out as previously described (Kheradmand et al., 2002). BAL cells were collected by serially instilling and withdrawing 0.7 ml aliquots of PBS from the tracheal cannula. After cells were washed and enumerated, aliquots of 105 cells were centrifuged onto glass slides, stained using modified Giemsa, and used to determine the absolute numbers of alveolar macrophage cells.
Measurement of alveolar wall elastin content was conducted as described previously with some modification (Foronjy et al., 2008). Five 40x images were captured from Verhoeff-stained lung tissue sections from 7-9 month old WT and Atxn1L−/− mice (n=4 in each genotype). Each 40x image was magnified 4 times to make the stained elastic fibrils easily discernable, and then analyzed for both the whole alveolar wall area and elastic fibrils-occupying area within alveolar walls using ImageJ software. The average elastin percent area for each genotype was calculated as described previously (Foronjy et al., 2008).
Hundred ng of lung total RNA from either 6 day-old WT, Atxn1−/− or Atxn1L−/− mice (3 pairs of WT and Atxn1−/− mice and 4 pairs of WT and Atxn1L−/− mice) was used for microarray analysis using Affymetrix Mouse Gene 1.0 ST Array. The data were analyzed using Limma package in the Bioconductor R (http://www.bioconductor.org/). The signal intensities were obtained and corrected for background by using Robust Multi-array Analysis (RMA) algorithm (Irizarry et al., 2003). Probes having low signal intensity (< 90) were excluded from the analysis to reduce the inconsistency associated with low-intensity signals. P values were calculated with a modified t-test in conjunction with empirical Bayes method (Smyth, 2004). Probes with fold-change > 1.2 and P values < 0.05 were considered to be differentially expressed. The GEO accession number of microarray data is GSE29551.
Tissues were dissected from mice at either E18.5, P6 or P20 and frozen quickly in liquid nitrogen. In case of lung tissues, the right four lobules were dissected and chopped into several small pieces in cold phosphate buffered saline (PBS) and large blood vessels were removed under dissecting microscope before frozen in liquid nitrogen. The tissues were sonicated in RIPA buffer (150mM NaCl, 50mM Tris-HCl pH 8.0, 2mM EDTA pH 8.0, 1% Triton X-100, 0.5% Deoxychloric acid, 0.1% SDS) with protease and phosphatase inhibitors and incubated on ice for 15 min. After centrifugation at 13,200 rpm for 15 min, the supernatant was taken and 20 μg of that was used for sample preparation. The protein samples were loaded on NuPage 4-20% Bis-Tris gel (Invitrogen) and transferred to Trans-Blot nitrocellulose membrane (Bio-Rad). Primary antibodies for Cic, Atxn1 (11750), and Atxn1L were used as described previously (Bowman et al., 2007). Rabbit polyclonal anti-MMP9 (Abcam) and anti-ETV4 (ProteinTech Group) antibodies were used at 1:1000 and 1:500 dilutions, respectively.
Twenty μg of protein extract from P6 lung or meninge tissues was loaded on Novex 10% Zymogram (Gelatin) gel (Invitrogen). After electrophoresis, developing and staining the gel were carried out according to the manufacturer’s manual (Invitrogen).
MH-S and A549 cells were cultured in RPMI 1640 media (Invitrogen) supplemented with 10% FBS. They were plated the day before transfection at about 50% confluence in 6cm dishes. The following day, the indicated amount of siRNAs for negative control, Cic, Etv4 (mouse) and ETV4 (human) was transfected using DharmaFECT I transfection reagent (Dharmacon) according to the manufacturer’s protocol. Three to four days later, cells were harvested. Protein samples for western blot analysis were prepared by lysis of the cells in cold RIPA buffer for 15 min followed by centrifugation at 1,3000 rpm for 15 min. Total RNA was extracted using Trizol (Invitrogen) and subjected to QRT-PCR analysis. The siRNAs are commercially available (Cic siRNA from Dharmacon, mouse Etv4 stealth siRNAs from Invitrogen and human ETV4 siRNAs from Ambion). The target sequences of siCic are 5′-GGUGCAACAAGGACCGAAA-3′. The target sequences of siEtv4-1 and siEtv4-2 are 5′-CCUUCUGCAGCAAAUCUCCCGGAAA-3′ and 5′-AGAAGCUCAGGUACCGGACAGUGAU-3′, respectively. The target sequences of siETV4-1 and siETV4-2 are 5′-GGACUUCGCCUACGACUCA-3′ and 5′-CCUGAUUUCCAUUCAGAAA-3′, respectively.
Two to three μg of total RNA from lung tissues, MH-S or A549 cells was used for cDNA synthesis using M-MLV-RT (Invitrogen). Quantitative RT-PCR was carried out using commercially available or homemade primers and probes for studied genes. For Etv1, 5′-GACCAGCAAGTGCCTTACGT-3′ (Forward), 5′-GACATTTGTTGGTTTCTCGGTACA-3′ (Reverse), and 5′-TCACCAACAGTCAGCGTGGGAGAAA-3′ (Probe) were used. For Etv4, 5′-CACTCCCCTACCACCATGGA-3′ (Forward), 5′-GGACTTGATGGCGATTTGTC-3′ (Reverse), and 5′-AGCAGTGCCTTTACTCCAGTGCCTATGA-3′ (Probe) were used. For Etv5, 5′-GATTGACAGAAAGAGGAAGTTTGTG-3′ (Forward), 5′-GCAGCTGGCTAAGATCCTGAA-3′ (Reverse), and 5′-CAGATCTGGCTCACGATTCTGAAGAGTTG-3′ (Probe) were used. Taqman Gene Expression Assays (Applied biosystem) were used for detection of Mmp2 (Mm00439506_m1), Mmp3 (Mm00440295_m1), Mmp8 (Mm00772335_m1), Mmp9 (Mm00442991_m1), Mmp11 (Mm00485048_m1), Mmp12 (Mm00500554_m1), Mmp13 (Mm01168713_m1), Mmp14 (Mm01318966_m1), Timp1 (Mm00441818_m1), Timp2 (Mm00441825_m1) and Timp3 (Mm00441826_m1). For Gapdh, TaqMan Rodent GAPDH control reagent (Applied Biosystem) was used. The SYBR Green was used for detection of PCR amplification of human ETV4, MMP9 and GAPDH genes. The primer sequences for ETV4 are 5′-GCAGTTTGTTCCTGATTTCCA-3′ (Forward) and 5′-ACTCTGGGGCTCCTTCTTG-3′ (Reverse). The primer sequences for MMP9 are 5′-TCTTCCCTGGAGACCTGAGA-3′ (Forward) and 5′-GAGTGTAACCATAGCGGTACAGG-3′ (Reverse). The primer sequences for GAPDH are 5′-AGCCACATCGCTCAGACAC-3′ (Forward) and 5′-GCCCAATACGACCAAATCC-3′ (Reverse).
Chromatin immunoprecipitation (ChIP) was carried out as previously described (Chahrour et al., 2008) with some modifications. In this experiment, P6 lung tissues and rabbit polyclonal anti-Cic antibody (Abcam) were used. To detect association of Cic with Etv1, Etv4 and Etv5 promoters, the following primers were used for PCR. For Etv1 promoter, 5′-GAGGAACTTGGGCCTTTGTTATG-3′ (Forward) and 5′-TGCTTTCGAGCAGGGGGC-3′ (Reverse) were used for PCR. For Etv4 promoter, 5′-CGTGGAGAAGCTGCCGGGTC-3′ (Forward) and 5′-CTCCCCACCGGTTCCCTTTG-3′ (Reverse) were used for PCR. For Etv5 promoter, 5′-GGTGCAGGCCGAGGCCAGGG-3′ (Forward) and 5′-CATTGACCAATCAGCACCGG-3′ (Reverse) were used for PCR.
For statistical analysis, all experiments were carried out more than three times independently. The Western blots were quantified using ImageJ software package. Statistical significance between control values and experimental values was determined using Student’s t test (two-tailed, two-sample unequal variance). Statistical significance is represented with asterisks.
We thank Dr. Moghaddam for helping us with inflated fixation of lung tissues and Jounghwa Won for the morphometric analysis. We are grateful to members of the Zoghbi laboratory for helpful discussions and comments on the manuscript. The morphological studies were performed in the Molecular Morphology Core Laboratory of the NIDDK-sponsored Texas Medical Center Digestive Disease Center, DK 56338. This research was supported by NIH grants NS27699 and HD24064 to H.Y.Z. H.Y.Z. is an investigator with the Howard Hughes Medical Institute.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.