Myocardin-related transcription factor B is required for normal mouse vascular development and smooth muscle gene expression


  • Ke Wei,

    1. Division of Cardiology, Department of Medicine, David Geffen School of Medicine, University of California at Los Angeles, Los Angeles, California
    2. Department of Molecular, Cellular, and Integrative Physiology, David Geffen School of Medicine, University of California at Los Angeles, Los Angeles, California
    Search for more papers by this author
  • Nam Che,

    1. Division of Cardiology, Department of Medicine, David Geffen School of Medicine, University of California at Los Angeles, Los Angeles, California
    Search for more papers by this author
  • Fabian Chen

    Corresponding author
    1. Division of Cardiology, Department of Medicine, David Geffen School of Medicine, University of California at Los Angeles, Los Angeles, California
    2. Division of Pediatric Cardiology, Department of Pediatrics, David Geffen School of Medicine, University of California at Los Angeles, Los Angeles, California
    3. Department of Molecular, Cellular, and Integrative Physiology, David Geffen School of Medicine, University of California at Los Angeles, Los Angeles, California
    • Division of Cardiology, Department of Medicine, David Geffen School of Medicine, University of California at Los Angeles, CA 90095-1760
    Search for more papers by this author


Smooth muscle gene expression is required for the proper development and function of multiple organ systems. Expression of smooth muscle genes is critical for contractile function and tissue architectural integrity. One critical transcription factor for smooth muscle gene expression is the Serum Response Factor (SRF). SRF is expressed ubiquitously, but tissue-specific transcriptional regulation is conferred by its binding to cofactors such as myocardin. Myocardin-related transcription factor B (MRTF-B) is a member of a family of genes (Myocardin(Myocd),Myocardin-related transcription factor A(MRTF-A),MRTF-B) that provides tissue-specificity and potentiate SRF-dependent transcription. Unlike myocardin, which is expressed specifically in smooth and cardiac muscle, MRTF-B is expressed in a wide variety of tissues. To examine the function of MRTF-B, we generated mice containing an insertional mutation of MRTF-B. MRTF-B homozygous mutants die in late gestation with vascular defects and liver hemorrhage. At E9.5, MRTF-B is expressed strongly in the septum transversum mesoderm critical for development of the vitelline system that produces the liver sinusoids and portal venous system. MRTF-B deficiency results in defective smooth muscle gene expression in the liver sinusoids, vitelline veins, and yolk sac, which contributes significantly to the lethal phenotype. These data support our hypothesis that MRTF-B has a unique role in regulating smooth muscle genes important for liver, yolk sac, and portal vascular development. Developmental Dynamics 236:416–425, 2007. © 2006 Wiley-Liss, Inc.


One critical transcription factor in vertebrate development is the Serum Response Factor (SRF), which binds a DNA sequence known as a CArG box, which has been shown to be particularly important in smooth and cardiac muscle development (Harvey and Rosenthal, 1999; Miano, 2003; Yoshida and Owens, 2005). However, SRF is expressed ubiquitously, and SRF-null mice die during gastrulation before heart and vascular development has begun (Arsenian et al., 1998). The ability of SRF to confer smooth and cardiac muscle specificity is due to its ability to bind tissue-specific cofactors such as myocardin (Wang and Olson, 2004; Pipes et al., 2006). Myocardin expression is restricted to heart and smooth muscle, and it also contains a potent transcriptional activation domain. Recently, it has been demonstrated that myocardin and ternary complex factors compete for a common site on SRF to control whether SMCs undergo differentiation or proliferation (Wang et al., 2004).

Myocardin knockout mice die by E10.5 from an absence of differentiated vascular SMCs (Li et al., 2003). Although a dominant-negative myocardin mutant affects cardiac development in Xenopus (Wang et al., 2001), heart formation was unaffected in the Myocardin-null mice presumably due to the presence of myocardin-related transcription factors MRTF-A and MRTF-B, which share significant homology with myocardin (Wang et al., 2002; Cen et al., 2003). Thus, there appears to be functional redundancy between Myocardin, MRTF-A, and MRTF-B. MRTF-A and MRTF-B also interact with SRF, and act as potent cotransactivators. Unlike Myocardin, MRTF-A and MRTF-B are expressed in a broad range of adult and embryonic tissues (Wang et al., 2002). MRTF-A (also called Mkl1/MAL/BSAC) and MRTF-B (also referred to as Mkl2/MAL16) can also promote SRF-dependent transcription in response to intracellular cytoskeletal changes by translocating to the nucleus in response to Rho GTPase and actin signaling (Miralles et al., 2003; Du et al., 2004; Kuwahara et al., 2005). Recently, it was shown that MRTF-A null mice are unable to nurse their young due to a defect in myoepithelial cells important for lactation (Li et al., 2006; Sun et al., 2006). The phenotype of the MRTF-A null mice is quite restricted despite the broad expression of MRTF-A in multiple tissues.

Expression of MRTF-B is also found in multiple tissues including smooth muscle. The overlapping expression of Myocardin, MRTF-A, and MRTF-B suggests that these transcription cofactors may have some functional redundancy. Similar to the MRTF-A mutant mouse, there may be specific circumstances where MRTF-B has critical roles in smooth muscle gene regulation. To further investigate the function of MRTF-B, we analyzed a mouse mutant for MRTF-B generated by insertional mutagenesis. Other gene-trap models of MRTF-B displayed perinatal lethality (Skarnes et al., 1992; Li et al., 2005). Germ-line knockout of MRTF-B resulted in earlier lethality at E13.5 that is characterized by cardiac and branchial arch defects (Oh et al., 2005). MRTF-B was shown to be important for smooth muscle differentiation of the neural crest–derived branchial arch arteries leading to a spectrum of aortic arch development defects in addition to thin myocardium and ventricular septal defects. In our analysis, we found these cardiac and outflow tract defects to be incompletely penetrant. However, here we make the unique finding that mice homozygous for our MRTF-B mutant gene die in late gestation with defects in liver and portal vascular development.


Generation and Initial Characterization of MRTF-BΔMutant Mice

To investigate the function of the SRF cofactor, MRTF-B, we generated a mouse mutant for MRTF-B using a gene-trap mutant ES cell line from Baygenomics (ES cell line RRJ478). The insertion is located in intron 10 of MRTF-B gene and results in the production of a fusion protein encoded by the first 10 exons of MRTF-B fused in-frame to a β-galactosidase–neomycin phosphotransferase IIgeo) reporter gene. The mutant MRTF-B lacks the transcriptional activation domain, but contains the amino-terminal domain (amino acids 1 to 731), basic region, glutamine-rich region, and SAP domains conserved between the myocardin family of transcription factors (Fig. 1A). We refer to the mutant MRTF-B allele as MRTF-BΔ.

Figure 1.

Generation and characterization of MRTF-BΔ mutant mice. A:MRTF-B gene structure and gene-trap strategy. The 15 exons and conserved domains are shown above. The splice acceptor (SA)-neomycin phosphotransferase II (Neo)–β-galactosidase (β-gal) cassette is inserted between exons 10 and 11. The splice acceptor is flanked by two loxP sites (arrowheads). B: Southern analysis of BglII-digested genomic DNA hybridized with an exon 10 probe (underlined). Positions of wild-type (WT) and mutant (mut) bands are shown. C: Northern analysis of E15.5 liver RNA with a MRTF-B cDNA probe. Northern analysis repeated with a β-actin cDNA probe shown below.

Disruption of the MRTF-B gene was demonstrated by Southern blot analysis (Fig. 1B). MRTF-B RNA expression is significantly decreased in E15.5 liver from MRTF-BΔ homozygous mutant mice as demonstrated by Northern analysis (Fig. 1C). The Northern blot utilized a cDNA probe containing the entire coding region, and did not demonstrate a second RNA transcript corresponding to the mutant MRTF-BΔ RNA transcript. However, the mutant MRTF-BΔ transcript was detected by RT-PCR (data not shown).

To determine if the MRTF-B fusion lacking the transcriptional activation domain was functional or had a dominant-negative effect, we performed transient transfection experiments with the smooth muscle22α (SM22α) and α-cardiac actin promoters fused to a luciferase reporter gene. Consistent with previously reported studies (Wang et al., 2002; Li et al., 2005), full-length MRTF-B was able to activate both promoters. The truncated molecule with the βgeo fusion molecule by itself had no transcriptional activity (Fig. 2A). When cotransfected in equal amounts with wild type MRTF-B, MRTF-BΔ had no significant effect on the SM22α promoter, and a modest additive effect on the α-cardiac actin promoter. We performed a dose-response transfection analysis of MRTF-BΔ to further determine if the mutant MRTF-BΔ molecule had any transcriptional activity in the presence of the wild type MRTF-B molecule. Escalating expression of wild type MRTF-B resulted in increased transcriptional activation of the SM22α promoter (Fig. 2B). Escalating expression of MRTF-BΔ had only a minimal additive effect on the transcriptional activation of the SM22α promoter with a baseline amount of wild type MRTF-B expression. Thus, expression of MRTF-BΔ in our mutant mouse is unlikely to have significant transcriptional activity that could affect the phenotype.

Figure 2.

Transcriptional activation of the smooth muscle 22α (SM22α) and α-cardiac actin promoters fused to a luciferase reporter gene. A: MRTF-B can activate the SM22α and α-cardiac actin promoters. MRTF-BΔ–βgeo fusion molecule alone lacks transcriptional activity. When co-transfected with full-length MRTF-B, MRTF-BΔ–βgeo fusion molecule alone has no significant effect on the SM22α promoter, and modestly increases activation of the α-cardiac actin promoter. B: Increasing concentrations of MRTF-B and MRTF-BΔ–βgeo were cotransfected with a baseline amount of MRTF-B (0.2 μg) to assess activation of a SM22α–promoter–Luciferase construct. MRTF-B results in a dose-dependent increase in transcriptional activation. MRTF-BΔ–βgeo results in a modest increase in activation of the SM22α –promoter in the presence of full-length MRTF-B. All transfection data utilized HeLa cells. Transfections performed in triplicate with 0.1-μg promoter reporter plasmid. Cotransfected pCMV-Renilla (0.25 μg) used to control for transfection efficiency.

Analysis of MRTF-B Gene Expression With X-gal

MRTF-B expression was examined with X-gal staining for β-galactosidase activity. βgeo is fused with the truncated MRTF-B, and should be expressed wherever MRTF-B is present. Previous in situ hybridization studies have shown that MRTF-B expression is widely distributed throughout the embryo, especially in late gestation (Wang et al., 2002). Our expression analysis of the MRTF-BΔ β-galactosidase activity demonstrated that at E10.0, MRTF-B is expressed in the heart, dorsal aorta, and liver primordium (Fig. 3A). Histological analysis at E9.5 demonstrated expression of MRTF-B in the ventricular myocardium, and the septum transversum (Fig. 3B) that originates from the lateral plate mesoderm. At E9.5, MRTF-B is not expressed significantly in the hepatoblasts, which unlike the septum transversum, express Hnf1α (which was identified by immunohistochemistry) (Fig. 3C). During liver development, the hepatic diverticulum invades the septum transversum, which secretes multiple signals required for hepatogenesis (Gualdi et al., 1996; Rossi et al., 2001). The septum transversum gives rise to phagocytic Kupffer cells, hematopoietic cells, blood vessels, and connective tissue that comprise the liver. At E14.5, βgeo is expressed strongly in the liver compared to a littermate wild-type control (Fig. 3D,E). Histological analysis of the E14.5 liver demonstrated that MRTF-B expression is prominent at the liver periphery consistent with its expression in the septum transversum-derived mesothelium (Fig. 3F). As shown in the avian embryo, the mesothelium that surrounds the liver gives rise to mesenchymal cells that invade the liver to form the endothelial and stellate cell populations of the developing sinusoids (Perez-Pomares et al., 2004). At E14.5, MRTF-B expression is also notable in the developing central vessels and sinusoids of the liver parenchyma (Fig. 3G).

Figure 3.

MRTF-BΔ is expressed in multiple embryonic tissues including liver. X-Gal staining of (A) E10.0, (B,C) E9.5 heterozygous MRTF-B+/Δ embryos; (D,F,G) liver from heterozygous E14.5 MRTF-B+/Δ embryos (+/−), (E) liver from wild type E14.5 littermate (+/+) (D). A: In an E10.0 embryo, note blue staining in liver primordium (red arrow), heart (white arrow), dorsal aorta (arrowhead). B,C: Adjacent sagittal histological sections of E9.5 heterozygous MRTF-B+/Δ embryo. X-Gal staining for MRTF-B expression is present in the heart (ht) and septum transversum (arrow). C: Hepatoblasts from the liver stain brown by immunohistochemistry with antibody for Hnf1α (arrowhead). These hepatoblasts do not have significant X-Gal staining as shown in the adjacent section (B) (arrowhead). D: Note blue staining in liver at E14.5 (posterior view). E: No blue staining in wild-type littermate liver (posterior view). F: E14.5 liver has X-Gal staining at the liver periphery consisting of septum transversum-derived mesothelium. Longitudinal section of left lateral liver lobe. G: E14.5 liver has X-Gal staining in central veins (arrowhead) and surrounding sinusoids in liver parenchyma. B,C: Hematoxylin/eosin counterstaining. F,G: Nuclear Fast Red counterstaining.

Phenotype of MRTF-BΔ/ΔMutant Embryos

Heterozygous MRTF-B+/Δ mice were fertile and indistinguishable from their wild-type littermates. Live MRTF-BΔ/Δhomozygous embryos were obtained in near-mendelian ratios from heterozygote intercrosses up to E15.5 (Table 1). The MRTF-BΔ/Δ embryos died primarily between E15.5 and E17.5. A small number of MRTF-BΔ/Δ homozygous mice survived weaning and appeared grossly normal. This incomplete penetrance may be due to a genetic background effect, the result of residual wild type MRTF-B expression with the MRTF-BΔ/Δ mutation, or due to functional redundancy provided by myocardin and MRTF-A.

Table 1. Genotype Distribution From MRTF-B+/Δ Intercrossesa
Genotype:+/++/−−/− (%)Total 
  • a

    Embryonic lethality of homozygous MRTF-B embryos occurs after E15.5. Offspring resulting from heterozygous MRTF-B intercrosses were genotyped as indicated. Number in parentheses indicates percentage of offspring at that age. Chi square analysis was performed and no significant (NS) differences were found from mendelian ratios prior to E15.5, and significant differences were found after E15.5 (P value indicated).

E11.5363 (25)12NS
E12.510168 (24)34NS
E13.5245514 (15)93NS
E14.5478136 (22)164NS
E15.55911535 (17)209P < 0.05
E16.58816439 (13)291P < 0.001
E17.519455 (8.5)59P < 0.01
Newborn1061766 (2.1)288P < 0.001
3 weeks1081877 (2.3)302P < 0.001

We expected that MRTF-BΔ/Δ mutant mice would succumb to heart failure and/or vascular defects given the known role of myocardin, MRTF-A, and MRTF-B as potent transcriptional activators of multiple cardiac and smooth muscle genes (Wang et al., 2001, 2002; Li et al., 2003; Selvaraj and Prywes, 2003; Oh et al., 2005). However, we identified no gross evidence of heart failure such as peripheral edema or pericardial effusion in any of the MRTF-BΔ/Δ homozygous mutant embryos. Unlike the previously reported myocardin knockout that lacked smooth muscle, our MRTF-BΔ/Δ knockout had smooth muscle (Li et al., 2003). We sequentially examined 15 MRTF-BΔ/Δ embryos at E15.5–E16.5 with careful attention paid to the heart and great vessels. Five mutant embryos had a ventricular septal defect (Fig. 4A,B), two had thin myocardium, and one had truncus arteriosus (Fig. 4C,D). Aortic arch and ventricular septal defects in isolation should not result in fetal demise until after birth when separate systemic and pulmonary circulations exist, and the newborn begins to breathe. The incomplete penetrance of cardiac and aortic arch defects, the expression of MRTF-B in multiple tissues, coupled with the significant prenatal lethality, made us consider it likely that the embryonic lethality in our MRTF-BΔ/Δ mice was not entirely explained by ventricular septal or cardiac outflow tract defects.

Figure 4.

Cardiovascular abnormalities in MRTF-BΔ/Δ mutant mice. A: Immunohistochemistry of sagittal sections of the heart performed with antibody for smooth muscle α-actin (stained brown). This E16.5 MRTF-BΔ/Δ embryo has a ventricular septal defect (*) and thin myocardium (arrowhead) that is not present in the littermate wild-type control (B). C: Hematoxylin/eosin stained section of MRTF-BΔ/Δ E15.5 embryo shows truncus arteriosus (TA) with a single great vessel arising from the heart, compared to the normal arrangement of the great arteries of a littermate wild-type embryo (D), which has the almost perpendicular pulmonary artery (PA) adjacent to the ascending aorta (Ao).

We did find that 40% (8/20) of unresorbed E15.5–E16.5 MRTF-BΔ/Δ embryos had aneurysmal dilation of the vitelline veins that return blood and nutrients from the yolk sac to the embryo, and that give rise to the future portal system (Fig. 5A–D). Given the known importance of MRTF-B in smooth muscle differentiation, we used immunohistochemistry to assess smooth muscle α-actin (Acta2) expression. We found decreased smooth muscle α-actin expression (Fig 5E,F) in the vitelline vessels, which is likely responsible for the aneurysmal dilation of the vitelline vein. The yolk sacs were pale and also had decreased smooth muscle α-actin expression in addition to reduced cellularity compared to wild type, which likely also affects its function (Fig 5G,H).

Figure 5.

MRTF-BΔ/Δ mutant mice have defective vitelline vein and yolk sac development. A,B:MRTF-BΔ/Δ homozygous mutant (−/−) E15.5 embryo (A) with aneurysmal dilation of the vitelline vein (white arrowhead), which is not observed in littermate wild-type control embryo (+/+) (B). Note the MRTF-BΔ/Δ embryo yolk sac is pale, and the yolk sac vessels are poorly perfused compared to the littermate wild-type embryo. C,D: Histologic section of E15.5 MRTF-BΔ/Δ embryo with aneurysmal dilation of the vitelline vein (black arrowhead) (C) compared with normal littermate wild-type control with no aneurysmal dilation (D). E,F: Immunohistochemistry of histological sections of the vitelline vein (arrowhead) and artery (arrow) of an E15.5 MRTF-BΔ/Δ embryo using an antibody for smooth muscle α-actin (stained brown). The smooth muscle α-actin expression in the vitelline vein and artery is decreased in the mutant (E) compared to littermate wild-type control (F). G,H: Immunohistochemistry of sections of yolk sacs with antibody for smooth muscle α-actin (stained brown) of E15.5 MRTF-BΔ/Δ (G) and wild-type embryo (H), showing decreased smooth muscle α-actin expression in yolk sac of MRTF-BΔ/Δ embryos in addition to reduced cellularity compared to wild type.

Given the late gestation lethality, the aneurysmal dilation of the vitelline vessels, and pale yolks sacs, we then focused on the liver that receives blood from the yolk sac via the vitelline vessels. In late gestation, the liver undergoes dramatic growth and development. The livers of all of our unresorbed MRTF-BΔ/Δ embryos had multiple red lesions consistent with hemorrhage (Fig. 6A,B). Histological examination further demonstrated hemorrhage within the MRTF-BΔ/Δ liver parenchyma (Fig. 6C–F). We examined multiple smooth muscle markers in the embryonic livers by immunohistochemistry, and found that expression of smooth muscle α-actin and SM22α were markedly decreased in the large central veins and the sinusoids of the liver parenchyma (Fig. 7A–D). Expression of smooth muscle myosin heavy chain was not obviously affected compared with smooth muscle α-actin and SM22α (Fig. 7E,F). Thus, MRTF-B deficiency resulted in decreased expression of smooth muscle genes, including smooth muscle α-actin and SM22α, which affects development of the related vitelline veins, yolk sac, and liver vasculature.

Figure 6.

MRTF-BΔ/Δ homozygous mutant embryos have liver defects. A: Posterior view of liver of E15.5 MRTF-BΔ/Δ embryo with hemorrhage (arrows), compared to normal liver from wild type littermate (B). Abnormal extension of liver parenchyma found in severely affected liver (arrowhead). C: Hematoxylin/eosin-stained section of MRTF-BΔ/Δ liver at E15.5 with hemorrhage (arrow), compared to normal liver from wild type littermate (D). E,F: higher magnification of the sections of C,D showing hemorrhage (arrow) in MRTF-BΔ/Δ liver (E).

Figure 7.

Smooth muscle gene expression is decreased in MRTF-BΔ/Δ liver. Immunohistochemistry of liver histological sections using antibodies for smooth muscle α-actin (A,B), SM22α (C,D), and SM-MHC (E,F) comparing E15.5 MRTF-BΔ/Δ (A,C,E) and wild type littermate embryos (B,D,F). A,B: Smooth muscle α-actin expression is decreased in the central veins (arrowheads) and in the liver parenchyma surrounding sinusoids (arrows) in MRTF-BΔ/Δ liver (A) compared to the littermate wild type (B). C,D: SM22α expression is decreased in the liver vasculature (arrowheads) in the MRTF-BΔ/Δ (C) compared to the littermate wild type (D). Note comparable staining for SM22α in the intestinal smooth muscle (asterisks) (C,D). E,F: SM-MHC expression is similar in a large hepatic vein comparing MRTF-BΔ/Δ (E) and wild type littermate (F). Note the lack of detectable SM-MHC expression in the liver parenchyma in both MRTF-BΔ/Δ (E) and wild type (F).

Tissue-Specific Inactivation of the MRTF-BΔ Mutation

Complicating the analysis of MRTF-B is its expression in a broad variety of tissues. In addition, MRTF-B and its related family member are expressed in multiple types of smooth muscle. The smooth muscle cell term is used here to refer to any connective tissue cell that wraps around an endothelial tube (Mahoney and Schwartz, 2005). This includes smooth muscle cells found in the wall of arteries and veins, and smooth muscle-like cells found in the liver sinusoidal, mammary gland myoepithelial, and kidney glomerular mesangial cells. The smooth muscle of the ascending aorta arch has a neural crest origin, and MRTF-B has previously been shown by others to be important in neural crest and aortic arch development (Li et al., 2005). The embryonic origin of smooth muscle-like cells expressing smooth muscle genes in the yolk sac and liver sinusoids is not clear, and has not been attributed to neural crest.

An advantage of the gene-trap MRTF-BΔ mouse model is the ability to modify the mutant allele using Cre/loxP technology. In cells expressing Cre recombinase, there is excision of the loxP flanked splice acceptor present in the gene-trap vector. Given the previously described abnormalities detected in the MRTF-BΔ/Δ mutant embryos, we sought to further dissect the tissue-specific roles of MRTF-B in neural crest (Li et al., 2005). We crossed mice to generate MRTF-BΔ/Δ mice with the Wnt1-Cre (Chai et al., 2000) transgene. The Wnt1-Cre transgene allows us to examine the importance of MRTF-B in neural crest. This should affect neural crest–derived smooth muscle differentiation in the ascending aortic arch that forms from the branchial arch vessels. X-Gal staining of newborn MRTF-B+/ΔWnt1-Cre+ hearts showed a mild decrease in blue staining on the ascending aorta and pulmonary artery, while no significant differences were observed in dorsal aorta, atria and ventricles (Fig. 8A,B). We genotyped 145 newborn mice from a MRTF-B+/Δ/Wnt1-Cre+ X MRTF-B+/Δ /Wnt1-Cre+ intercross (Table 2). Thirteen out of 113 (11.5%) (25% expected if no lethality) MRTF-BΔ/Δ/Wnt1-Cre+ were observed. This is in comparison to 2 of 32 (6.3%) (25% expected if no lethality) MRTF-BΔ/Δ/Wnt1-Cre newborns. The observed number of MRTF-BΔ/Δ/Wnt1-Cre+ live newborn mice is indicative of partial lethality as determined by chi-square analysis (P < 0.01). In comparison to the initial MRTF-B+/Δ heterozygote intercross (Table 1, 6/288, 2.1% MRTF-BΔ/Δ observed at the newborn stage), a higher fraction of MRTF-BΔ/Δ/Wnt1-Cre+ mice survive (Table 2, 13/113, 11.5%). Thus, despite continued embryonic lethality with neural crest Wnt1-Cre inactivation of the MRTF-BΔ mutation, these results do suggest improved survival of the MRTF-BΔ/Δ mice with Wnt1-Cre.

Figure 8.

Wnt1-Cre inactivation of the MRTF-BΔ mutation. Note decreased X-Gal blue staining in the ascending aorta (asc) compared to the descending aorta (da, black arrow) and heart in the MRTF-B+/Δ/Wnt1-Cre+ (B) compared to MRTF-B+/Δ/Wnt1-Cre (A) newborn (between white arrows). pa, pulmonary artery; ra, right atrium; la, left atrium.

Table 2. Improved Newborn Survival of MRTF-BΔ/Δ Mico With Wnt1-Crea
Wnt1-Cre+MRTF-B+/Δ × Wnt1-Cre+MRTF-B+/Δ
MRTF-BΔ/Δ (%)Expected without lethalityMRTF-BΔ/Δ (%)Expected without lethality
  • a

    Newborns from Wnt1-Cre+MRTF-B+/Δ × Wnt1-Cre+MRTF-B+/Δ intercrosses were genotyped. The numbers and percentage of live newborn MRTF-BΔ/Δ are shown. The numbers and percentage of live newborn MRTF-BΔ/Δ mice expected without lethality are shown for comparison.

11313 (11.5%)28 (25%)322 (6.3%)8 (25%)

Gene Expression Analysis in MRTF-BΔ/Δ Mutant Embryos

In view of the abnormalities found in our MRTF-BΔ/Δ mutant embryos, we examined SM22α, smooth muscle α-actin, and smooth muscle myosin heavy chain mRNA levels from isolated liver and yolk sac using real-time quantitative RT-PCR of multiple wild-type and MRTF-BΔ/Δ embryos. E15.5 embryos were examined, given the significant lethality beginning at that embryonic stage. As expected, MRTF-B expression was decreased in the MRTF-BΔ/Δ embryo liver (31% of wild type) and yolk sacs (21% of wild type). Wild type MRTF-B expression was assessed using PCR primers specific for the 3′ end of MRTF-B distal to the gene trap insertion site. MRTF-A expression was not significantly affected by MRTF-B deficiency in either embryo livers or yolk sacs. Expression of smooth muscle genes such as SM22α and smooth muscle α-actin (Acta2) mRNA were decreased in MRTF-BΔ/Δ mutant compared to wild-type littermate E15.5 livers and yolk sacs, while SM-MHC expression was not significantly affected (Fig. 9). This RNA expression data is consistent with protein expression examined by immunohistochemistry (Fig. 7). Based on these results, MRTF-B deficiency affects liver and yolk sac development with alterations in smooth muscle gene expression, which is important for the mutant phenotype and also affects embryogenesis.

Figure 9.

Smooth muscle gene expression is decreased in the liver and yolk sac in the MRTF-BΔ/Δ E15.5 embryo as assessed by real-time PCR. The relative RNA expression levels of MRTF-BΔ/Δ mutant and littermate wild-type embryos are shown. MRTF-B expression is decreased, and MRTF-A is not significantly affected in the MRTF-BΔ/Δ embryo. SM22α and SM α-actin expression is significantly decreased in both liver and yolk sac. SM-MHC expression is not significantly changed. mRNA levels of each gene are normalized to Gapdh RNA levels. Each bar represents the mean ± SE (n = 4 to 6). *P < 0.05, significant difference between mutant and wild type.


MRTF-B encodes a SRF cofactor necessary for embryonic development of multiple tissues including the liver, yolk sac, and cardiovascular system. We have focused our attention on the role of MRTF-B in the vascular development of the liver and yolk sac, and demonstrate the importance of MRTF-B in the development of these structures necessary for embryonic development and viability. We demonstrate that in early midgestation (E9.5), MRTF-B is expressed in the septum transversum, a plate of mesoderm that lies below the pericardial cavity and extends to the stalk of the yolk sac. The septum transversum gives rise to the liver mesothelium important for formation of the liver sinusoids (Perez-Pomares et al., 2004).

MRTF-B deficiency in the septum transversum, portal, and hepatic vasculature contributes to liver hemorrhage and embryonic lethality by affecting smooth muscle gene expression. As shown in humans during late gestation, hepatic blood flow dramatically increases as liver expansion occurs (Bellotti et al., 2000). This increase in hepatic blood flow coupled with abnormal liver development can result in liver hemorrhage and fetal demise. Yolk sac abnormalities due to defective smooth muscle gene expression can also result in embryonic demise.

Similar to previous reports, we also detected cardiac and aortic arch defects characterized by ventricular septal defects and aortic arch malformations attributed to MRTF-B deficiency (Li et al., 2005; Oh et al., 2005). The MRTF-B homozygous mutant made by homologous recombination resulted in complete embryonic lethality that occurs abruptly at E14.5 (Oh et al., 2005). This time point for embryonic lethality is compatible with our data demonstrating lethality occurring in late gestation between E15.5 and the newborn stage. The difference in phenotypes may be due to differences in the mutant molecule being generated or due to residual wild type MRTF-B expression that can be found in the gene trap mutant. Some wild type MRTF-B RNA transcripts may be made due to bypassing of the gene trap splice acceptor. This is demonstrated in our real-time PCR analysis of MRTF-B RNA expression of liver and yolk sacs that demonstrated that MRTF-B expression is still present, but at significantly reduced levels (Fig. 9).

Our results are compatible with MRTF-B playing an important in role in neural crest–derived smooth muscle differentiation. However, the partial rescue of the embryonic lethality we observed using the Wnt1-Cre transgene is in contrast to the much higher rescue described by others (Li et al., 2005). Differences in the penetrance of the cardiovascular phenotype may be due to genetic background effects resulting from using a 129/Ola × C57BL/6 mixed genetic background. Genetic background effects may also be responsible for the increased survival of the MRTF-BΔ/Δ/Wnt1-Cre newborns compared to the survival of MRTF-BΔ/Δ newborns from the initial MRTF-B+/Δ heterozygote intercross. The absence of complete rescue of lethality suggests that Wnt1-Cre inactivation of the mutation is insufficient in neural crest, or that MRTF-B has important roles in non-neural crest tissues as our data suggests.

Conditional inactivation of the gene trap mutation has similar obstacles to that of a conditional knockout experiment, and much is dependent on the function of the Cre recombinase. Tissue-specific inactivation of the MRTF-BΔ gene-trap splice acceptor using a tissue-specific promoter driving Cre demonstrates the tissue-specific expression of MRTF-B and the potential for tissue-specific inactivation of the mutation. Several reasons may explain why the transgene used may not completely rescue the MRTF-BΔ/Δ embryonic lethal phenotype. The lack of rescue may be the result of technical problems from using the Cre recombinase approach and the resultant recombined MRTF-BΔ/Δ allele. The Cre-recombined MRTF-BΔ gene still contains the gene-trap vector, but with the splice acceptor removed. The presence of residual gene-trap vector DNA may still affect wild-type MRTF-B expression. In addition, the ability of the Wnt1-Cre recombinase transgene to deactivate the splice acceptor may occur too late or not efficiently enough to rescue the phenotype. Most likely, MRTF-B may also be required for multiple tissues, and inactivation of the mutation in only one tissue type may be insufficient to rescue the embryonic requirement for MRTF-B.

Smooth muscle gene expression was affected in the MRTF-BΔ/Δ mouse embryo, and it may be that MRTF-B has a unique role in specific vascular beds compared to myocardin and MRTF-A. It is possible that MRTF-B deficiency results in a decreased number of cells that express smooth muscle genes in the liver, yolk sac, and vitelline vessels. This would also result in the decreased smooth muscle gene expression that we demonstrate. Certainly, smooth muscle and neural crest-related cardiovascular abnormalities resulting from MRTF-B deficiency can contribute to the embryonic lethal phenotype. However, defects in aortic arch patterning may also be a consequence of abnormal cardiovascular hemodynamics resulting from vitelline vein and liver congestion. For example, obstruction of the vitelline vein with a microclip can induce cardiac malformations in the chick embryo (Stekelenburg-de Vos et al., 2003). However, given the broad expression of MRTF-B and the results of our MRTF-BΔ/Δ mutant mice analysis, it is likely that MRTF-B has important roles in multiple tissues.

Recently, it was demonstrated that MRTF-A is required for proper differentiation of mammary myoepithelial cells (Li et al., 2006; Sun et al., 2006). Homozygous MRTF-A mutant mothers are unable to nurse their offspring due to a defect in differentiation of mammary myoepithelial cell, a smooth muscle-like cell type required for milk ejection. Similarly, our findings reveal a unique role for MRTF-B in another group of specialized smooth muscle-like cells, the perisinusoidal cells in the liver. There was a reduction in smooth muscle gene expression in the liver, yolk sac, and vitelline vessels. Smooth muscle α-actin and SM22α expression was decreased more than smooth muscle myosin heavy chain. These results are not surprising given the heterogeneity of tissue types present in these structures, and differences in transcriptional regulation of the many smooth muscle genes.

As is the case for cardiac and aortic arch development, liver and yolk sac development involves the interaction of several tissue types. During early liver development (E9.5), cords of hepatoblasts delaminate from the liver bud and invade the mesenchyme of the septum transversum (Zhao and Duncan, 2005). The septum transversum contributes to the development of the mesoepithelium and sinusoids of the liver. Perisinusoidal cells in the space of Disse express smooth muscle genes such as smooth muscle α-actin. These perisinusoidal cells may use smooth muscle proteins to regulate sinusoidal blood flow (McCuskey, 2000; Rockey, 2001). These perisinusoidal cells also play a role in extracellular modeling in response to liver injury, and are implicated in the pathogenesis of many human liver fibrotic diseases. Smooth muscle gene expression in the adult liver is increased in the liver fibrotic response to injury and disease (Bataller and Brenner, 2005). Determining the function of MRTF-B in the vascular development of the liver, yolk sac, and portal system will provide insights into the potential role of MRTF-B in adult liver and cardiovascular disease.


Production of MRTF-BΔ Mice and Embryos

MRTF-BΔ/Δ mutant mice were generated from the BayGenomics embryonic stem (ES) cell line RRJ478 (Stryke et al., 2003; Skarnes et al., 2004). Clone RRJ478 was injected into C57BL/6J blastocysts, and the mice were maintained on a 129/Ola × C57BL/6J genetic background. Embryos were obtained from timed pregnant mice with noon the day of the vaginal plug as 0.5 days post-coitum. Genotype was determined by Southern blot analysis and PCR with specific primers for the wild type and mutant alleles. Wnt1-Cre mice (C57BL/6J) were kindly provided by Emmanuele Delot and Karen Lyons (UCLA) and originally obtained from Henry Sucov (USC) (Chai et al., 2000). UCLA's Animal Care Committee approved all animal protocols. All primer sequences are available upon request.

Immunohistochemistry, Antibodies, and Histology

Embryos and tissues were taken from various stages and fixed in 4% paraformaldehyde for 24 h, followed by dehydration for 24 h in increasing concentrations of ethanol and embedding in paraffin. Mice and tissues were not perfused with fixative. We cut 5-μm sagittal or transverse sections. Some were stained with hematoxylin and eosin. Immunohistochemical analysis utilized antibodies against smooth muscle α-actin (Sigma), Hnf1α (BD Transduction laboratories), SM22α (Santa Cruz), and smooth muscle myosin heavy chain (Sigma). For X-Gal staining, embryos and tissue were fixed for 15–45 min in a solution containing 1% formaldehyde, 0.2% glutaraldehyde, and 0.02% NP-40 in phosphate-buffered saline (PBS; pH 7.4). Fixed embryos were washed twice in PBS, and subjected to staining for 4–16 h in a solution containing 1 mg/ml X-Gal, 5 mM K3Fe(CN)6, 5 mM K4Fe(CN)6, and 2 mM MgCl2 in PBS. Hematoxylin and eosin staining and Nuclear Fast Red staining were used as counterstains.

RNA Analysis

RNA was prepared with Trizol (Life Technologies, Bethesda, MD). RNA (25 μg) from each embryo liver was used for Northern blots. RT-PCR was performed on RNA from liver tissue or yolk sac from E15.5 embryos. Quantitative RT-PCR comparison of gene expression was determined with the Biorad I-cycler. Gapdh was used as RNA input controls. Primer sequences are listed below.









  • Specific for nucleotides 2,518 to 2,742 (amino acids 839–914)



  • Specific for nucleotides 2,527 to 2,775 (amino acids 842–925)




Hela cells were transfected with Effectene (Qiagen, Chatsworth, CA). Luciferase reporter gene constructs (0.1 μg) included α-cardiac actinLuc (Chen and Schwartz, 1996) and SM22α–Luc (Du et al., 2004) constructs. The SM22α–Luc construct contains 441 bp of the mouse SM22α promoter fused to a luciferase reporter construct (pGL2-Basic, Promega, Madison, WI). The α-cardiac actinLuc construct contains −315 to +15 of the cardiac actin promoter relative to the start of transcription that is adjacent to a luciferase reporter gene. MRTF-B and MRTF-BΔ expression vectors (0.2 μg) were used. Renilla luciferase (Promega) was used to normalize for transfection efficiencies, and pcDNA3 (Invitrogen, La Jolla, CA) was used to equalize the amount of DNA in each transfection. Luciferase assays were performed using Luciferase Assay kit as directed (Promega).


We thank D. Lang, J. Chen, and W.R. MacLellan for support and review of this manuscript.