Expression of bone morphogenetic protein receptor 1A (BMPR1A) is attenuated in the lung vessels of patients with pulmonary arterial hypertension, but the functional impact of this abnormality is unknown. We ablated Bmpr1a in cardiomyocytes and vascular smooth muscle cells(VSMCs) by breeding mice possessing a loxP allele of Bmpr1a(Bmpr1aflox) expressing R26R with SM22α-Cre mice. SM22α-Cre;R26R;Bmpr1aflox/flox mice died soon after embryonic day 11 (E11) with massive vascular and pericardial hemorrhage and impaired brain development. At E10.5, SM22α-Cre;R26R;Bmpr1aflox/flox embryos showed thinning of the myocardium associated with reduced cell proliferation. These embryos also had severe dilatation of the aorta and large vessels with impaired investment of SMCs that was also related to reduced proliferation. SM22α-Cre;R26R;Bmpr1aflox/flox mice showed collapsed telencephalon in association with impaired clearing of brain microvessels in areas where reduced apoptosis was observed. Transcript and protein levels of matrix metalloproteinase (MMP) 2 and 9 were reduced in E9.5 and E10.5 SM22α-Cre;R26R;Bmpr1aflox/floxembryos, respectively. Knock-down of BMPR1A by RNA interference in human pulmonary artery SMCs reduced MMP2 and MMP9 activity, attenuated serum-induced proliferation, and impaired PDGF-BB-directed migration. RNA interference of MMP2 or MMP9 recapitulated these abnormalities, supporting a functional interaction between BMP signaling and MMP expression. In human brain microvascular pericytes, knock-down of BMPR1A reduced MMP2 activity and knock-down of either BMPR1Aor MMP2 caused resistance to apoptosis. Thus, loss of Bmpr1a, by decreasing MMP2 and/or MMP9 activity, can account for vascular dilatation and persistence of brain microvessels, leading to the impaired organogenesis documented in the brain.
Bone morphogenetic protein receptors (BMPRs) are members of the transforming growth factor β superfamily of receptors(de Caestecker, 2004; Mehra and Wrana, 2002). Heteromeric complexes form between BMPR1 and BMPR2(Gilboa et al., 2000). Aberrant BMP signaling has been linked to pulmonary arterial hypertension(PAH). Various germline mutations in BMPR2 have been identified in familial and even sporadic forms of the disease(Deng et al., 2000; Lane et al., 2000; Thomson et al., 2001). Moreover, independent of a mutation, expression of BMPR2(Atkinson et al., 2002) and BMPR1A (Du et al.,2003) is reduced in lungs of PAH patients.
PAH is a potentially fatal disease(Abenhaim et al., 1996)characterized by both obliteration of proximal pulmonary arteries resulting from vascular smooth muscle cell (VSMC) proliferation and migration(Jeffery and Morrell, 2002),and loss of distal arteries associated with endothelial cell (EC)(Campbell et al., 2001) and pericyte apoptosis (Zhao et al.,2003). These pathological features account for the progressive increase in pulmonary vascular resistance culminating in right-side heart failure (Humbert et al., 2004; Rubin, 1997).
Mice homozygous null for Bmpr2(Beppu et al., 2000), Bmpr1a (Mishina et al.,1995), the ligand Bmp4(Winnier et al., 1995) and the effector Smad4 (Sirard et al.,1998) die early in embryonic life owing to a lack of mesodermal induction. In mice with Flk1-targeted deletion of Bmpr1a(Flk1-Cre;Bmpr1aflox/flox) (Flk1 is also known Kdr - Mouse Genome Informatics)(Park et al., 2006), lethality occurs between E10.5 and E11.5, in association with massive abdominal hemorrhage. These mice exhibit dilatation of large vessels owing to poor recruitment of VSMCs around the EC layer, but it is not clear whether the vascular phenotype is due to Bmpr1a-deficient ECs or SMCs(Park et al., 2006).
In this study, we determined whether VSMC deletion of Bmpr1a could cause abnormalities in vasculogenesis that might explain a propensity to PAH. We bred mice expressing floxed Bmpr1a and ROSA26 with SM22α-Cre mice [SM22α is also known as transgelin (Tagln) - Mouse Genome Informatics]. Progeny homozygous for deletion of Bmpr1a,SM22α-Cre;R26R;Bmpr1aflox/flox, died soon after E11 with massive vascular and pericardial hemorrhage. These mice had a thin ventricular wall and aneurysmal dilatation of large vessels associated with reduced myocyte proliferation related to decreased MMP9 and MMP2 activities. Defective brain development documented in the SM22α-Cre;R26R;Bmpr1aflox/flox mice was associated with impaired clearing of brain microvessels related to a resistance of pericytes to apoptosis and decreased levels of MMP2.
MATERIALS AND METHODS
Experimental model: SM22α-Cre;R26R;Bmpr1aflox/flox mice
We crossed SM22α-Cre mice with mice homozygous for the floxed Bmpr1a gene (Mishina et al., 2002) and the Cre reporter gene ROSA26(R26R) (Soriano,1999) (Bmpr1aflox/flox;R26R+/+). F2 breeding was then realized by backcrossing F1 mice(SM22α-Cre;R26R+/-;Bmpr1aflox/+)with the Bmpr1aflox/flox;R26R+/+ mice to produce mice that were SM22α-Cre;R26R;Bmpr1aflox/flox(flox/flox). All studies were performed under a protocol approved by the Animal Care Committee at Stanford University in accordance with the guidelines of the American Physiological Society.
Preparation of embryos for histological analyses
Isolated E9.5-11 mouse embryos were fixed with formalin or 4%paraformaldehyde (PFA) in phosphate-buffered saline (PBS), embedded in paraffin and cut transversely (7 μm).
Histology and immunostaining
Paraffin sections of brains, hearts and dorsal aortae of formalin-fixed embryos were stained with Hematoxylin and Eosin (H&E) to assess the phenotype resulting from deletion of Bmpr1a. To assess apoptosis, we performed the TUNEL assay using the ApopTag Peroxidase In Situ Oligo Ligation Apoptosis Detection Kit (Chemicon International, Temecula, CA). Sections were counterstained with Methyl Green (Vector Labs, Burlingame, CA).
To assess alpha smooth muscle actin (αSM-actin) or the proliferating cell nuclear antigen (PCNA), formalin-fixed tissue sections were incubated with either mouse anti-αSM-actin antibody (1:200, Sigma-Aldrich, St Louis, MO) or with biotinylated mouse anti-PCNA antibody (1:100, Zymed, South San Francisco, CA). For αSM-actin staining, sections were then incubated with goat anti-mouse-biotinylated antibody (1:500, Jackson ImmunoResearch,West Grove, PA). For both αSM-actin and PCNA staining, sections were incubated with streptavidin-horseradish peroxidase (HRP)-conjugated antibody(1:500, Jackson ImmunoResearch). Brown immunoreactivity was observed by subjecting the sections to diaminobenzidine substrate (DAB; Vector Labs). Sections stained with antibodies to αSM-actin and PCNA were counterstained with Hematoxylin and Methyl Green, respectively.
To assess apoptosis in brain pericytes, TUNEL assay using the ApopTag Red In Situ Apoptosis Detection Kit (Chemicon) was followed by immunostaining for the pericyte marker NG2 (CSPG4 - Mouse Genome Informatics) (primary antibody,1:100, Chemicon) on formalin fixed-head sections.
Expression of MMP2 and MMP9 in aortic walls and heart was analyzed in tissue sections of PFA-fixed embryos incubated with either an anti-MMP2 (Ab-4)mouse mAb (75-7F7) or an anti-MMP9 (Ab-3) mouse mAb (56-2A4) (1:100,Calbiochem, EMD Biosciences, San Diego, CA) followed by Alexa Fluor 488 goat anti-mouse IgG (H+L, 1:200, Molecular Probes, Invitrogen, Carlsbad, CA).
To assess BMP10 signaling in embryo hearts, PFA-fixed tissue sections were incubated with a p57KIP2 (CDKN1C - Mouse Genome Informatics) primary antibody(clone 57P06, 1:100, Neomarkers, Fremont, CA) followed by a biotinylated rabbit anti-mouse secondary antibody (1:250, BMK-2202, MOM Kit, Vector Labs)and ABC Reagent (PK6100, ABC Elite Kit, Vector Labs). Sections were then subjected to DAB+ (DAKO, Carpinteria, CA) and counterstained with Hematoxylin.
Whole-mount lacZ staining
E8.5-10.5 PFA-fixed mouse embryos were stained with 0.7 mg/ml X-Gal for assessment under the microscope or were sectioned and counterstained with Nuclear Fast Red (Vector Labs).
Whole-mount PECAM staining
PFA-fixed E10.5 mouse embryos were incubated with PECAM antibody [1:100,rat anti-mouse CD31 (PECAM1), clone MEC13.3, BD Pharmingen, BD Biosciences,San Jose, CA] followed by HRP-conjugated goat anti-rat IgG (1:500, Jackson ImmunoResearch). Embryos were then subjected to DAB substrate (Vector Labs),cleared (benzyl alcohol/benzyl benzoate) for better visualization of the vascular tree and then assessed under the microscope or sectioned and counterstained with Methyl Green (Vector Labs) for histological analysis.
Primary cell cultures and RNA interference (RNAi)
Adult human pulmonary artery smooth muscle cells (HPASMCs) and human brain vascular pericytes (HBVPs) were cultured as previously described(El-Bizri et al., 2008). Cells were transiently transfected with control, human BMPR1A, MMP9 or MMP2 siRNA (Dharmacon, Lafayatte, CO) in Opti-MEM I (Gibco,Invitrogen) using Lipofectamine 2000 (Invitrogen). `Starvation media' (media supplemented with 0.1% FBS) were added 6 hours later for a total of 48 hours.
Quantitative real-time PCR (qRT-PCR)
Total RNA was extracted from whole E9.5 mouse embryos or from HPASMCs or HBVPs and then reverse transcribed as previously described(El-Bizri et al., 2008). Gene expression levels were quantified using preverified Assays-on-Demand TaqMan primer/probe sets (Applied Biosystems, Foster City, CA) and normalized to 18S RNA and B2M for murine and human samples, respectively, using the comparative delta-CT method.
Cell proliferation (MTT) and apoptosis (caspase 3 and 7 activity)
Forty-eight hours following transfection in 0.1% FBS, HPASMCs were exposed to 10% FBS for 72 hours, and cell growth was assessed by the MTT Cell Proliferation Assay (American Type Culture Collection, Manassas, VA) and by cell counts. Transfected HBVPs were kept under serum-free conditions for an additional 24 hours, after which apoptosis was assessed by measuring caspase 3 and 7 activity using the Caspase 3/Caspase 7 Luminescent Assay Kit(Caspase-Glo, Promega, Madison, WI), and proliferation was assessed by the MTT assay.
Cell migration assay (Boyden Chamber)
Migration was assessed using a modified Boyden Chamber (BD Falcon, BD Biosciences) as previously described(Leung et al., 2004). SiControl- and SiBMPR1A-transfected HPASMCs were stimulated to migrate for 6 hours in 0.1% FBS for baseline measurements and in response to 10% FBS or PDGF-BB (20 ng/ml) (R&D Systems, Minneapolis, MN) as chemoattractants in the lower compartments of the chambers.
Conditioned media collected from the upper compartments of the Boyden Chambers to evaluate production of MMPs in HPASMCs migrating in response to 0.1% FBS, 10% FBS or PDGF-BB (20 ng/ml) and from HBVPs after 48 hours of serum starvation as well as extracts of individual mouse embryos were used for gelatin zymography. The supernatants were subjected to electrophoresis in an 8% SDS-PAGE gel co-polymerized with gelatin (1 mg/ml, Sigma-Aldrich)(Cann et al., 2008). The gelatinolytic activities were detected as transparent bands against the background of Coomassie Brilliant Blue-stained gelatin and quantified using ImageJ.
Values for each determination are expressed as mean±s.e.m. For comparisons between two groups, statistical significance was determined using the unpaired two-tailed t-test. For comparisons of multiple groups,one-way analysis of variance (ANOVA) followed by Bonferroni's multiple-comparison test was carried out. The number of mouse embryos or samples used in each experiment is provided in the figure legends.
Embryonic lethality in transgenic mice with SM22α-targeted deletion of Bmpr1a
Mouse embryos were genotyped as described under Materials and methods and are illustrated in Fig. 1A. Ubiquitous expression of Bmpr1a is observed early in mouse development, starting at E6.5 (Mishina et al., 1995; Roelen et al.,1997). To assess the profile of Cre activity reflecting areas of Bmpr1a deletion, we performed whole-mount lacZ staining on SM22α-Cre;R26R;Bmpr1aflox/+ mouse embryos at E8.5-10.5 (Fig. 1B). Cre activity was evident in the heart from E8.5 by blue lacZ staining(data not shown), and in the heart and vasculature at E9.25(Fig. 1Ba). At E10.5, smaller intersomitic vascular branches and somitic myotomes showed positive lacZ staining (Fig. 1Bb).
SM22α-Cre;R26R;Bmpr1aflox/flox mice appeared normal until E9.5 (Fig. 1C, compare b with a). By E10.5, they were somewhat smaller than wild-type (WT) (Fig. 1C,compare d with c), and at E11 the mice showed massive perivascular and pericardial hemorrhage (Fig. 1C, compare f with e) and died soon after. Following the breeding strategy described in the Materials and methods, the expected frequency of the flox/flox genotype was 25%. This frequency was observed by genotyping embryos at different ages up to E11-11.5; the frequency was 10% at E12.5, and there were no fetuses with this genotype by E18.0 or in postnatal mice assessed after weaning (Fig. 1D).
Cardiac defect in SM22α-Cre;R26R;Bmpr1aflox/flox embryos:thinning of the myocardium associated with reduced proliferation
Cre activity was confined to atrial and ventricular myocytes, with no expression in the endocardium of E10.5 flox/flox embryos as assessed by whole-mount lacZ staining (Fig. 2A,B). Myocardial deletion by SM22α-Cre is consistent with transient expression of SM22α in the developing heart(Li et al., 1996; Umans et al., 2007).
To assess the sequelae of loss of Bmpr1a in cardiomyocytes on embryonic cardiac development, histological analysis of heart sections of viable E10.5-11 SM22α-Cre;R26R;Bmpr1aflox/flox and age-matched littermate control embryos (WT) was carried out to show the four chambers and outflow tract anatomy at multiple levels. We noted thinning of the ventricular wall in the flox/flox versus WT hearts(Fig. 2, compare D with C),quantified as a ∼35% reduction in the number of ventricular cells per heart section (P<0.05) (Fig. 2E). The cardiac phenotype was not due to enhanced apoptosis as only the occasional TUNEL-positive cell was seen(Fig. 2F,G), but rather was associated with attenuated cell proliferation. There was a reduction in the percentage of PCNA-positive cells over the total number of ventricular cells in heart sections of the flox/flox(Fig. 2I) versus WT(Fig. 2H) (P<0.05)at E9.5 that persisted at E10.5-11 (Fig. 2J). As Bmp10-deficient embryos show thinning of the myocardium associated with decreased cell proliferation and ectopic expression of p57KIP2 (Chen et al.,2004), we assessed the expression of p57KIP2 by immunofluorescence to address the possibility that a deletion of Bmpr1a in the heart might lead to a defect in BMP10 signaling. Our results showed no difference in p57KIP2 immunoreactivity between mutant(Fig. 2L) and WT hearts(Fig. 2K).
Vascular defect in SM22α-Cre;R26R;Bmpr1aflox/flox embryos:dilatation of large vessels associated with reduced proliferation of vascular smooth muscle cells
To characterize the vascular defect resulting in perivascular hemorrhage and lethality in flox/flox embryos, we performed whole-mount PECAM staining on embryos at E10.5. Gross morphological examination revealed massive dilatation of the large vessels appreciated in the dorsal aortae, mesenteric(Fig. 3Ab,d,f) and cranial vessels (not shown) of the mutants versus WT littermates(Fig. 3Aa,c,e). There were more ramifications or interconnections in the interlimb vessels of the flox/flox (Fig. 3Af)versus WT (Fig. 3Ae) embryos. H&E-stained transverse sections of E10.5 embryos showed dilated aortae(Fig. 3Ah) in flox/flox embryos relative to WT controls(Fig. 3Ag).
Whole-mount staining revealed poor investment of SM22α-Cre-expressing lacZ-positive cells in the dilated aortic wall of the SM22α-Cre;R26R;Bmpr1aflox/flox mutants(Fig. 3Aj) as compared with WT(Fig. 3Ai), where strong lacZ staining was evident. The lacZ-positive cells were identified as being of smooth muscle lineage by immunoperoxidase staining using an antibody for αSM-actin. There were also fewer surrounding mesenchymal cells expressing αSM-actin in SM22α-Cre;R26R;Bmpr1aflox/flox(Fig. 3Bb) versus WT(Fig. 3Ba) embryos. TUNEL staining on sections of aorta revealed only occasional positive mesenchymal cells (Fig. 3Bc,d, arrows). Instead, the decreased number of αSM-actin-positive perivascular cells was consistent with reduced proliferation as assessed by PCNA staining(Fig. 3B, compare f with e,arrows). Quantitative analysis revealed a ∼53% reduction in the percentage of PCNA-positive SMCs forming the vessel wall(Fig. 3Bg)(P<0.05). PECAM staining of sections did not reveal a difference in the number of ECs surrounding the dilated vessels, but the cells appeared`stretched' (data not shown).
Defective brain development of SM22α-Cre;R26R;Bmpr1aflox/flox embryos associated with impaired clearing of small vessels
Cre activity was seen in the forebrain of an E10.5 WT embryo by whole-mount lacZ staining (Fig. 4A). To characterize and better visualize any brain development abnormality, we examined heads of embryos incubated with ethidium bromide under UV light. Compared with WT (Fig. 4Ba,c), flox/flox mutant embryos(Fig. 4Bb,d) showed brain compression and collapse of telencephalic vesicles. These defects were apparent in H&E-stained transverse sections of the heads at multiple levels (Fig. 4Bf,h,j,l). To determine how loss of Bmpr1a in SM22α-expressing cells could impair brain development, we performed whole-mount PECAM staining on embryos at E9.5 and E10.5. We observed similar brain vessel distribution in the WT and flox/flox mutants at E9.5 (data not shown); however, at E10.5, we noted evidence of clearing of telancephalic vessels in the WT(Fig. 4Ca,c) but not the mutants (Fig. 4Cb,d).
Transverse sections of the brains stained for PECAM at the level of the nasal-mandibular processes showed histologic evidence of clearing of vessels in the WT heads (Fig. 4Ce),whereas flox/flox mutant heads(Fig. 4Cf) showed persistent vessels (brown). To determine whether the clearing of vessels is related to apoptosis, the TUNEL assay was performed on brain sections. TUNEL-positive cells were plentiful in the WT (Fig. 4Cg) but were almost absent from the flox/flox mutant heads (Fig. 4Ch). Quantitative analysis showed a ∼62% reduction in the percentage of TUNEL-positive cells over the total number of cells in the flox/flox group(Fig. 4Ck)(P<0.05).
Since pericytes express SM22α(Ding et al., 2004), we speculated that loss of Bmpr1a in these cells led to resistance to apoptosis and reduced clearing of brain microvessels. We therefore performed a fluorescent TUNEL assay followed by fluorescent immunostaining for NG2, a pericyte marker. Reduced apoptosis was associated with an increased number of pericytes in the mutant (Fig. 4Cj) versus WT (Fig. 4Ci) brains. Because co-localization of the TUNEL and NG2 staining was not observed in the WT brain (Fig. 4Ci), we could not confirm ongoing apoptosis of pericytes,suggesting that this occurred before E10.5. Persistence of brain microvessels was not due to enhanced cell growth, as PCNA immunoreactivity showed no difference between E10.5 WT and flox/flox mutants(Fig. 4Cl).
Reduced MMP9 and MMP2 in embryos with SM22α-targeted deletion of Bmpr1a
We investigated the level of expression of candidate genes dysregulated by loss of Bmpr1a that could account both for resistance to apoptosis in pericytes and repression of proliferation in VSMCs. For example, BMPs increase MMP activity and mRNA expression (Mishina et al., 2004; Palosaari et al., 2003) and MMPs can regulate cell survival(Jones et al., 1997) and induce proliferation of VSMCs (Zempo et al., 1994). Quantitative RT-PCR was applied to embryonic extracts to assess differential expression of MMPs and other extracellular matrix genes that could be modulated by loss of Bmpr1a and might account for these altered vascular cell phenotypes. Many of the genes that we assessed are modified in other embryonic mouse models of vascular dilatation(Oh et al., 2000). We also assessed the transcript levels of genes implicated in vasculo/angiogenesis -vascular endothelial growth factor (Vegf; also known as Vegfa) and the angiopoietins (Angpt1, Angpt2) - and of the phosphatase and tensin homolog gene (Pten), a known gene downstream of BMP signaling implicated in juvenile polyposis(He et al., 2004) that might also impact cell growth (Beck and Carethers, 2007). RNA was extracted from E9.5 embryos, preceding the appearance of the phenotype in SM22α-Cre;R26R;Bmpr1aflox/flox mice. We found a significant decrease in the mRNA expression of Mmp9(P<0.05) and Mmp2 (P<0.05), and trends toward reduced expression of tenascin C (Tnc), fibronectin, connective tissue growth factor (Ctgf) and urokinase plasminogen activator(uPA; Plau - Mouse Genome Informatics) were observed(Fig. 5A). No differences in tissue plasminogen activator (tPA; Plat)(Fig. 5A), Angpt1, Angpt2,Vegf (data not shown) and Pten(Fig. 5A) mRNA levels were noted between the WT and mutants.
This decrease in Mmp2 and Mmp9 transcripts was associated with a decrease, although not statistically significant, in the pro (40%) and active (30%) forms of MMP2 in SM22α-Cre;R26R;Bmpr1aflox/flox mutants versus WT, as assessed by gelatin zymography on mouse embryos (data not shown).
To determine whether the decrease in mRNA levels of Mmp9 and Mmp2 in whole E9.5 mutant mouse embryos is translated into reduced protein expression at a later age, we performed immunostaining of MMP9 and MMP2 at E10.5. We found abundant MMP9 and, to a greater extent, MMP2, in the aortic walls of WT embryos (Fig. 5Ba,c) and only weak immunoreactivity in the mutants(Fig. 5Bb,d). However, a low and diffuse immunostaining was noted in the heart and brains of WT and mutants(data not shown).
Loss of BMPR1A attenuates proliferation and directed migration of vascular smooth muscle cells and induces pericyte resistance to apoptosis via reduced MMP9 and MMP2
Subsequent studies were carried out using cultured human pulmonary artery smooth muscle cells (HPASMCs) and human brain (micro)vascular pericytes(HBVPs) to determine (1) whether reducing levels of BMPR1A by RNAi would result in suppression of MMP9 and/or MMP2 activities and (2) whether reducing BMPR1A, MMP9 and/or MMP2, represses proliferation of VSMCs and induces resistance to apoptosis in pericytes. We reduced the mRNA level of BMPR1A by 66% by transfecting HPASMCs with siRNA (SiBMPR1A),and observed, by gelatin zymography, that SiBMPR1A-transfected cells had decreased levels of the pro and active forms of MMP9 (P<0.05 for both) and of MMP2 (P<0.01 and P<0.05, respectively)versus SiControl-transfected cells (Fig. 6A).
Consistent with our hypothesis and our findings in the mouse embryo, we showed that RNAi-mediated reduction in mRNA of BMPR1A (by 66%), MMP9 (to undetectable levels) or MMP2 (by >80%), resulted in a 35-40% reduction in HPASMC proliferation in response to 10% FBS as assessed by the MTT assay (P<0.001, Fig. 6B) and cell counts (data not shown). Since MMP9 and MMP2 levels increase in migrating SMCs(Bendeck et al., 2002; Franco et al., 2006; Kuzuya et al., 2003; Mason et al., 1999), we determined whether the chemotactic migratory behavior of SMCs was impaired by loss of BMPR1A, in association with reduced MMP9 and/or MMP2. A deficiency in SMC migration could also account for the lack of SMC investment of the aneurysmally dilated vessels in flox/flox embryos. We serum starved HPASMCs in 0.1% FBS for 48 hours and then assessed their response to a 6-hour treatment with PDGF-BB (20 ng/ml) using a modified Boyden Chamber assay. The MMP9 and MMP2 activities in SiControl HPASMCs, as assessed by gelatin zymography, were repressed in SiBMPR1A-treated cells(P<0.01 for MMP9 and P<0.001 for proMMP2 and MMP2)(Fig. 7A). Although basal levels of migration were increased in SiBMPR1A-transfected HPASMCs(P<0.001), these cells did not significantly migrate in response to PDGF-BB when compared with SiControl HPASMCs (P<0.05)(Fig. 7B).
We then investigated whether pericytes with loss of BMPR1A would be resistant to apoptosis owing to a reduction in MMP2 activity, as MMP2 activity is proapoptogenic in pericytes of diabetic patients(Yang et al., 2007). Using RNAi under conditions of serum starvation (0.1% FBS) for 48 hours, we showed a 53% reduction in BMPR1A transcript levels in HBVPs. To induce apoptosis, cells were serum deprived for an additional 24 hours, after which a reduction in both pro and active forms of MMP2 was demonstrated by gelatin zymography (P<0.05) (Fig. 8A). We then used RNAi to reduce mRNA levels of MMP2 in HBVPs and confirmed the decrease in pro and active forms of MMP2 by gelatin zymography (P<0.001) (Fig. 8B). A control experiment showing the sensitivity of gel zymography to detect gelatinase activity in a dose-dependent manner is provided in Fig. S1 (see supplementary material). Transfecting HBVPs with SiBMPR1A, or with SiMMP2, induced resistance to apoptosis when compared with SiControl HBVPs (P<0.001), as assessed by caspase 3 and 7 activities (Fig. 8C), without affecting cell proliferation as assessed by the MTT assay (data not shown). Therefore, lack of MMP2 in pericytes could account for the resistance to apoptosis seen in highly vascularized areas of flox/flox mutant brains.
Bmpr1a in cardiac development
The cardiac phenotype in E10.5 SM22α-Cre;R26R;Bmpr1aflox/flox mouse embryos was characterized by thinning of the ventricular wall and was attributed to reduced cell proliferation evident at E9.5. Bmp10-null mice also develop hearts with hypoplastic walls owing to reduced proliferation of cardiac myocytes at E9.0-9.5 (Chen et al., 2004). However, our results showed that BMP10 signaling in E10.5 SM22α-Cre;R26R;Bmpr1aflox/flox heart sections was not affected by Bmpr1a deletion.
It is interesting that in the mouse in which Bmpr1a was deleted following activation of the cardiac myocyte-specific promoter alpha myosin heavy chain (αMHC-Cre;Bmpr1aflox/flox)(Gaussin et al., 2002), the ventricular thinning that took place at a later time point (E11.5-12.5) was attributed to enhanced apoptosis. Ventricular thinning was also seen at E11.5-12.5 in mice lacking Bmpr1a in cardiac progenitors(Islet1-Cre;Bmpr1a nulls) (Yang et al., 2006), or at E11.5 with cardiac-specific ablation of Smad4 (Song et al.,2007), and both were associated with attenuated proliferation and enhanced apoptosis of the ventricular septal myocytes. This suggests that differences in the timing of promoter activation and Bmpr1a deletion in cardiomyocytes might dictate whether the thinning of the ventricular wall will be the result of apoptosis and/or reduced proliferation.
Bmpr1a and vasculogenesis
We cannot exclude the possibility that the myocardial thinning is secondary to a hemodynamic abnormality caused by the vascular phenotype observed in the SM22α-Cre;R26R;Bmpr1aflox/flox embryos, and characterized by aneurysmal dilatation of the dorsal aorta and other large vessels. Dilatation of the aorta was observed in embryos with Flk1-targeted deletion of Bmpr1a(Flk1-Cre;Bmpr1aflox/flox)(Park et al., 2006) and in embryos null for Alk1 (Acvrl1 - Mouse Genome Informatics)(Oh et al., 2000) or Smad5 (Yang et al.,1999). In those models, the dilatation was attributed to a paracrine effect of Bmpr1a-deficient ECs repressing the recruitment of VSMCs or pericytes, as observed in mice lacking PDGF-BB or PDFG-Rβ(Hellstrom et al., 1999; Lindahl et al., 1997). Other possibilities suggested include poor transdifferentiation of ECs into SMCs, or a defect in SMC growth affecting vessel maturation and integrity(Park et al., 2006). The third explanation fits best with the further delineation of the phenotype of SM22α-Cre;R26R; Bmpr1aflox/flox embryos that we carried out.
We were able to assess the impact of Bmpr1a deletion in reducing VSMC proliferation in the tissue as well as in cultured cells, in which we also observed impaired PDGF-BB-directed VSMC migration. No defect in the EC layer was noted in the SM22α-Cre;R26R;Bmpr1aflox/flox embryos by whole-mount PECAM immunostaining (data not shown) that might explain the vascular defect through a non-cell-autonomous contribution. In addition, we did not observe upregulation of angiogenic factors, such as of Angpt1and Angpt2 as described in Alk1-null embryos(Oh et al., 2000), or of Vegf as observed in both Flk1-Cre;Bmpr1aflox/flox(Park et al., 2006) and Alk1 nulls (Oh et al.,2000). It follows that there was no concomitant angiogenic defect in SM22α-Cre;R26R;Bmpr1aflox/flox mutant embryos, such as the impaired yolk sac vascular remodeling seen in the Flk1-Cre;Bmpr1aflox/flox(Park et al., 2006) and the Alk1-null (Oh et al.,2000) embryos. Since Flk1 is a mesodermal marker and Alk1 is mostly expressed in ECs, the angiogenic defect in the yolk sac is likely to be due to the loss of Bmpr1a in ECs, a feature we reproduced by ablating Bmpr1a using Tie2-Cre (our unpublished observations) (Tie2 is also known as Tek). In contrast to other mice models of aneurysmal vascular dilatation, the SM22α-Cre;R26R;Bmpr1aflox/flox mice did not exhibit an increase in expression of proteases such as uPA and tPA (Oh et al., 2000; Park et al., 2006). In contrast to the Smad5-null embryos with dilated aorta, the SM22α-Cre;R26R; Bmpr1aflox/flox embryos did not show apoptosis in VSMCs or in neighboring mesenchymal cells(Yang et al., 1999).
When we assessed gene expression of extracellular matrix proteins and proteinases previously implicated in VSMC proliferation and migration, a consistent reduction in the expression of Mmp9 and Mmp2,genes downstream of BMP signaling in other cell types(Mishina et al., 2004; Palosaari et al., 2003), was observed. A direct association between reduced BMPR1A and impaired production of MMP9 and MMP2 was then demonstrated in cultured human VSMCs, in which knock-down of BMPR1A by RNAi attenuated MMP9 and MMP2 activities. The role of both MMP9 and MMP2 in VSMC proliferation and migration is well documented (Bendeck et al.,2002; Franco et al.,2006; Kuzuya et al.,2003; Mason et al.,1999). Expression of Pten, downstream of Bmpr1aand implicated in juvenile polyposis and affecting cell growth, was not modified in the mutants.
Our observations linking reduced MMP9 and MMP2 to aneurysmal dilatation might seem at odds with clinical studies in human tissue in which increased MMP2 and especially MMP9 are observed in abdominal aortic aneurysm(Goodall et al., 2001; Thompson et al., 1995). Moreover, reduction of MMP9 activity by Doxycycline protects against experimentally induced aortic aneurysm(Kaito et al., 2003), as does local expression of TIMP1, an inhibitor of MMP9 activity(Allaire et al., 1998; McMillan et al., 1995). In addition, mice that are null for Mmp9 are resistant to elastase-induced aortic aneurysms (Pyo et al., 2000). It therefore appears that during vascular development,a reduction in both MMP9 and MMP2 in SMCs is required to produce aneurysmal dilatation, as a result of reduced proliferation and perhaps migration of SMCs. It is interesting that the Mmp2/9 double nulls(Lambert et al., 2003) do not recapitulate our phenotype. This could reflect compensatory induction of other MMPs in response to a global, rather than a tissue-specific, deletion. Alternatively, the mixed background of the flox/flox mutants compared with the C57BL/6J background of the Mmp2/9 double nulls might account for the difference in the phenotype.
Bmpr1a expression in pericytes mediates vessel regression during brain development
SM22α-Cre;R26R;Bmpr1aflox/flox mutants showed severe brain asymmetry and collapse of telencephalic vesicles. A vascular defect produced by impaired BMPR1A signaling that has not previously been described might explain these abnormalities.
Regression of vessels is crucial in triggering mesenchymal condensation culminating in chondrogenesis and skeletogenesis(Yin and Pacifici, 2001). As MMP2 activity is linked to retinal pericyte apoptosis in diabetic retinopathy(Yang et al., 2007), we reasoned that suppression of MMP2 resulting from lack of BMPR1A signaling might make pericytes resistant to apoptosis, preventing EC apoptosis and microvessel clearing, and subsequently leading to defective brain development. Indeed, we showed that lack of BMPR1A or MMP2 by RNAi renders pericytes in culture resistant to apoptosis.
The same phenomenon might explain the enhanced ramification of the interlimb vessels seen in the flox/flox mutants, suggesting that vascular deletion of Bmpr1a might impair organogenesis of other tissues not investigated here. In the rat aortic model of angiogenesis, MMP9 and MMP2 expression and activity not only increased during the angiogenic growth phase of microvessels, but also remained elevated and were necessary for microvessel regression (Zhu et al.,2000). Consistent with this, maximal MMP2 activity is observed in the late corpus luteum concomitant with vessel regression(Duncan et al., 1998). The deletion of Bmpr1a in brain cells(Hebert et al., 2002) did not recapitulate the phenotype, further indicating the importance of the vasculature in this cell-autonomous mechanism.
The discrepancy between the phenotypes resulting from loss of BMPR1A in VSMCs and pericytes might be related to the fact that they have different basement membranes (Meyrick and Reid, 1979) and hence could exhibit different effects resulting from reduced MMPs (Fig. 9).
It is worth mentioning that our findings did not recapitulate any aspects of juvenile polyposis (JP), a condition associated with mutations in BMPR1A. However, the site of pathology in JP, the villus, forms after E15.5 (Batts et al., 2006) and SM22α-Cre;R26R;Bmpr1aflox/flox embryos die several days earlier.
Our study is the first to show that both MMP9 and MMP2 are developmentally regulated by expression of Bmpr1a and that attenuation in their levels could reduce the proliferation of SMCs leading to aneurysmal dilatation of large vessels. These observations could also explain the reduced cell proliferation that leads to thinning of the ventricular wall. Our findings linking repression of Bmpr1a-mediated MMP2 activity to reduced apoptosis of pericytes, point to a feature not only of developmental importance in clearing of microvessels, but potentially to a mechanism that might help in preserving or regenerating microvessels in disease. In our recent studies (El-Bizri et al.,2008), in which patchy deletion of Bmpr1a was induced in VSMCs, mice were actually protected against both the excessive muscularization and loss of distal vessels associated with chronic hypoxia-induced PAH.
This research is supported by the Intramural Research Program of the NIH,NIEHS to Y.M., and by NIH Grant R01 HL074186 to M.R. N.E. is supported by a fellowship from the American Heart Association (AHA)/Pulmonary Hypertension Association and M.R. by the Dunlevie Professorship. C.-P.C. is supported by funds from the National Heart Lung and Blood Institute (HL085345), AHA,Children Heart Foundation, March of Dimes Foundation and Baxter Foundation. K.S. is supported by an AHA postdoctoral fellowship.