Filamin B deficiency in mice results in skeletal malformations and impaired microvascular development

Abstract

Mutations in filamin B (FLNB), a gene encoding a cytoplasmic actin-binding protein, have been found in human skeletal disorders, including boomerang dysplasia, spondylocarpotarsal syndrome, Larsen syndrome, and atelosteogenesis phenotypes I and III. To examine the role of FLNB in vivo, we generated mice with a targeted disruption of Flnb. Fewer than 3% of homozygous embryos reached term, indicating that Flnb is important in embryonic development. Heterozygous mutant mice were indistinguishable from their wild-type siblings. Flnb was ubiquitously expressed; strong expression was found in endothelial cells and chondrocytes. Flnb-deficient fibroblasts exhibited more disorganized formation of actin filaments and reduced ability to migrate compared with wild-type controls. Flnb-deficient embryos exhibited impaired development of the microvasculature and skeletal system. The few Flnb-deficient mice that were born were very small and had severe skeletal malformations, including scoliotic and kyphotic spines, lack of intervertebral discs, fusion of vertebral bodies, and reduced hyaline matrix in extremities, thorax, and vertebrae. These mice died or had to be euthanized before 4 weeks of age. Thus, the phenotypes of Flnb-deficient mice closely resemble those of human skeletal disorders with mutations in FLNB.

Filamins are large actin-binding proteins that stabilize the actin cytoskeleton, link the actin network with cellular membranes, and mediate interactions between actin and transmembrane receptors (1). In mouse, there are three filamin genes, Flna, Flnb, and Flnc. Flna and Flnb, which encode large proteins with 70% structural similarity, are ubiquitously expressed, whereas Flnc, encoding a much smaller protein, is expressed in heart and skeletal muscles. It has been proposed that filamins are important for fetal development by regulating the communication between extracellular signals and the cellular cytoskeleton to guide migration of cells into appropriate anatomical sites (1). In addition to actin, filamins are capable of binding to a wide range of molecules involved in cellular signaling and transcriptional regulation (2, 3).

FLNB was first isolated as a protein that interacts with the cytoplasmic tail of glycoprotein Ibα (4, 5). The 2,603-aa chain of FLNB contains an amino-terminal actin-binding domain and a backbone of 24 Ig-like rod domain repeats disrupted by two hinge regions (6). Mouse Flnb is located on chromosome 14 and consists of 47 exons. In mouse embryos, Flnb is expressed in vertebral bodies, and it has been suggested that Flnb may play a role in vertebral segmentation, joint formation, and endochondral ossification (7).

Mutations in the human FLNB gene have been found in several skeletal disorders, including spondylocarpotarsal syndrome, autosomal-dominant Larsen syndrome, atelosteogenesis I and III (7), and boomerang dysplasia (8). The skeletal disorders caused by mutations in FLNB are very similar to those caused by mutations in FLNA (9). Interestingly, FLNA and FLNB are coexpressed within chondrocytes (7) and physically interact with each other in neurons (10), suggesting that FLNA and FLNB may form both homodimers and heterodimers.

Future research on FLNB disorders would benefit from the availability of relevant animal models. To address this issue and to define the physiologic importance of FLNB, we produced Flnb-deficient mice and cell lines and analyzed the consequences of Flnb deficiency on cell migration, the actin cytoskeleton, and embryonic and postnatal development. We show that Flnb deficiency leads to impaired microvascular development and severe skeletal malformations that recapitulate human FLNB disorders.

Results

Production of Flnb^−/− Mice.

We generated Flnb^−/− mice with a mutant embryonic stem (ES) cell line containing an insertional mutation in intron 20 of Flnb (Fig. 1A), producing a fusion between the amino terminus of Flnb and β-geo, which is a fusion of β-galactosidase and neomycin phosphotransferase II. Genotyping of genomic DNA from mouse tails by PCR yielded a 2,004-bp fragment from the wild-type allele and a 2,504-bp band from the mutant allele (Fig. 1B). In livers from Flnb^−/− mice, wild-type Flnb transcripts and full-length Flnb protein were undetectable as judged by RT-PCR (Fig. 1C) and Western blots with an antibody that recognizes sequences in the carboxyl terminus (Fig. 1D). There was no up-regulation of Flna expression in the liver as a consequence of Flnb deficiency as judged by semiquantitative RT-PCR (Fig. 1C).

Fig. 1.

Insertional mutation in Flnb and production of Flnb^−/− mice. (A) Schematic illustration of the mutation in Flnb. Numbered boxes represent exons. The mutant allele yields an in-frame fusion transcript with exons 1–20 of Flnb and β-geo. Sites for forward and reverse oligonucleotide primers are indicated (arrows). SA, splice acceptor. (B) Genotyping of mouse tails by genomic PCR. (C) RT-PCR analysis of Flnb and Flna transcripts from Flnb^+/+, Flnb^+/−, and Flnb^−/− mouse liver. Amplification of actin mRNA is provided as internal control. (D) Western blot of total protein from liver of Flnb^+/+, Flnb^+/−, and Flnb^−/− mice, immunoblotted with an anti-Flnb antibody that recognizes near carboxyl terminus. Immunoblot with actin is included as loading control. Flnb^−/− embryos die after E11.5. (E) Genotyping of embryos and pups from Flnb^+/− parents of the F₂ generation compared with the expected Mendelian genetics. Embryos at different embryological days were screened for their genotype. Only 2.3% of embryos reached term (P < 0.0001, χ² test).

Reduced Survival of Flnb^−/− Embryos After Embryonic Day 11.5 (E11.5).

Genotyping of 173 offspring from intercrosses of Flnb^+/− mice identified 99 Flnb^+/+, 70 Flnb^+/−, and 4 Flnb^−/− mice (Fig. 1E), which suggests that Flnb deficiency is associated with high embryonic lethality. To determine when the Flnb^−/− embryos die, we genotyped embryos from heterozygous intercrosses at E8.5, E9.5, E11.5, E13.5, and E15.5. The expected ratio of 25% for the Flnb^−/− embryos declined after E11.5 (Fig. 1E).

Strong Expression of Flnb in Endothelial Cells and Chondrocytes.

Colocalization studies on serial sections stained with antibodies directed against Flnb and CD31, a commonly used endothelial cell surface maker, suggested that Flnb was expressed in the vascular endothelium in Flnb^+/+ embryos (Fig. 2 A and B). Lack of immunohistochemical staining of the same vessel for LYVE-1 indicated that it is a blood vessel, not a lymphatic vessel (Fig. 2C). We also detected expression of Flnb in adult endothelial cells of the aortic wall (Fig. 2D). In Flnb^−/− mice, a strong LacZ expression was detected in endothelial cells as judged by X-Gal staining of capillary structures between fibers in skeletal muscle (Fig. 2E), intimal layer of the aorta (Fig. 2F), and cardiac capillaries (Fig. 2G). Double-staining of heart sections with X-Gal and an anti-CD31 antibody indicated that the LacZ-expressing cells are of endothelial origin (Fig. 2H).

Fig. 2.

Flnb is expressed by vascular endothelial cells. Serial sections stained with antibodies against Flnb (A, brown) and CD31 (B, brown) indicate that Flnb-expressing cells are of endothelial cell origin in Flnb^+/+ embryos at E13.5. (C) No LYVE-1 expression in the same vessel. (D) Expression of Flnb by endothelial cells in the adult aortic intima (arrow). (E–G) LacZ expression (turquoise) was highly expressed by adult endothelial cells in different organs of Flnb^−/− mice, including the capillary structures between the skeletal muscle fibers (E), intimal layer of the aortic wall (F), and cardiac capillaries (G) in an Flnb^−/− 3-week-old mouse. (H) Double-staining with an anti-CD31 antibody (brown) and X-Gal (turquoise) in the same heart section indicated that LacZ-expressing cells are endothelial cells. Sections were slightly counterstained with Mayer's hematoxylin.

Flnb was ubiquitously expressed in Flnb^+/+ embryos, as judged by immunohistochemistry using anti-Flnb antibody (Fig. 3 A–C); no Flnb expression was detected in Flnb^−/− embryos (Fig. 3 D–F). Higher magnification of developing vertebral bodies of Flnb^+/+ embryos showed strong expression of Flnb in chondrocytes (Fig. 3 B and C). This expression pattern was confirmed by X-Gal staining of Flnb^−/− embryos (Fig. 3 G–I).

Fig. 3.

Flnb is expressed by vertebral chondrocytes. Flnb^+/+ (A–C) and Flnb^−/− (D–F) mouse embryos at E13.5 were immunohistochemically stained by using anti-Flnb antibody and visualized by using DAB chromagen (brown), and the nucleus of the cells was slightly counterstained with Mayer's hematoxylin (blue). (A) Ubiquitous expression of Flnb in the Flnb^+/+ embryo. Note the presence of developing vertebral discs in the Flnb^+/+ embryo (B, black arrow), which is lacking in the Flnb^−/− embryo. (C) Expression of Flnb in the developing vertebral discs and bodies (white arrows). (D–F) Lack of Flnb expression in Flnb^−/− embryos. (G) Ubiquitous expression of LacZ in Flnb^−/− embryos at E15.5. (H and I) Strong expression of LacZ in the developing Flnb^−/− vertebral bodies.

Flnb Deficiency Impairs Microvascular Development.

Immunofluorescence staining of whole Flnb^+/+ and Flnb^−/− embryos at E11.5 with anti-CD31 antibody showed that the microvascular network in the central nervous system and perivertebral areas was poorly developed in Flnb^−/− embryos (Fig. 4 A and B). High resolution of these areas indicated impaired vascular network compared with Flnb^+/+ embryos (Fig. 4 C and D). We also observed impaired development of larger blood vessels, such as the middle cerebral artery, by histological analyses of the Flnb^−/− embryos at E13.5 (Fig. 4 E–H). In this poorly developed arterial tree, X-Gal staining colocalized with endothelial cells (Fig. 4H Inset). In addition to the central nervous system, the perivertebral areas of the Flnb^−/− embryos at E13.5 exhibited a disorganized and poorly developed vascular network (Fig. 4 I and J).

Fig. 4.

Impaired microvasculature patterning in Flnb^−/− embryos. (A–D) Whole-mount immunofluorescence staining of an Flnb^+/+ (A) and Flnb^−/− (B) embryo with an anti-CD31 antibody at E11.5 (red). The vascular network around the vertebral column and in the central nervous system is indicated (white arrows) to demonstrate an impaired vascularization in the Flnb^−/− embryo at E11.5. (C and D) An area of these embryos is enlarged to compare the vascular patterning. Note the proper sprouting of microvasculature in the Flnb^+/+ embryo (C, small white arrows), which is severely impaired in the Flnb^−/− embryo. (E and F) Histological pictures of the central nervous system stained with hematoxylin/eosin of Flnb^+/+ (E) and Flnb^−/− (F) mouse embryos at E11.5. (G and H) Middle cerebral artery between midbrain (mb), diencephalon (de), and trigeminal ganglion (tg) is enlarged. In this area of the Flnb^−/− embryo, note that the middle cerebral artery is rudimentary (as also indicated in A, right arrow), and endothelial cells covering small vascular structures express LacZ (H Inset). (I and J) Embryological sections stained with an anti-CD31 antibody demonstrate impaired and disorganized vascular network around the vertebral column at E13.5.

Flnb^−/− Mice Develop Severe Skeletal Malformations.

Flnb^+/− mice were born at the expected Mendelian frequency and exhibited normal growth rate and appeared healthy and fertile. However, <3% of Flnb^−/− embryos reached term. The Flnb^−/− mice died or had to be euthanized before the age of 4 weeks because of severely reduced body weight, limited movement, and a rigid posture. The Flnb^−/− pups exhibited skeletal malformations particularly in the vertebral column, including kyphotic changes in the thoracic part and scoliotic changes in the abdominal part (Fig. 5A). Moreover, the Flnb^−/− mice had a striking reduction in hyaline cartilage in the ribs (Fig. 5A), metacarpal bones, phalanges (Fig. 5B) and tarsal bones. The body weight of Flnb^−/− pups was reduced 2- to 3-fold compared with Flnb^+/− and Flnb^+/+ siblings at 4 weeks (Fig. 5C). Histological examinations revealed that intervertebral discs, including the nucleus pulposus and surrounding cartilage, were absent in the Flnb^−/− mice (Fig. 5 D–G), which resulted in fusion of the vertebral bodies, likely explaining the rigid posture and limited movement. The absence of intervertebral discs was evident already in E13.5 embryos (Fig. 3E).

Fig. 5.

Skeletal abnormalities in Flnb^−/− mice. (A) In an Flnb^−/− mouse at lateral (Left) and supine (Right) positions, kyphotic and scoliotic malformations of the vertebral column are indicated (arrow) compared with an Flnb^+/+ mouse. Staining is with alcian blue (hyaline cartilage) and alizarin red (bone). (B) Reduced hyaline cartilage tissue (blue) is indicated (arrow) in metacarpal bones and phalanges of the Flnb^−/− mouse. (C) Body weight of this Flnb^−/− mouse was almost three times less than its siblings (n = 9). (D–G) Intervertebral discs are lacking in the kyphotic and scoliotic areas of the Flnb^−/− spine as indicated (G, arrow). (F) Hyaline cartilage (blue) around the Flnb^+/+ intervertebral disc as indicated (white arrows) is lacking in the Flnb^−/− vertebral bone. (H–K) Flnb^−/− extremities are shorter and weaker. (H) Flnb^−/− tibiae are significantly shorter than Flnb^+/− and Flnb^+/+ tibiae (*P < 0.05, unpaired t test). (I) Dual x-ray absorptiometry measurements indicate a significant reduction in the areal bone mineral density. Changes in the cortical thickness (J) and in the trabecular volumetrical bone mineral density (K) are measured by pQCT, and the data are given as mean ± SD values (∗, P < 0.05, unpaired t test).

Analyses of the bones of 4-week-old mice revealed that tibiae from Flnb^−/− mice were significantly shorter compared with tibiae from Flnb^+/− and Flnb^+/+ mice (Fig. 5H). The mid-diaphyseal area of tibiae from Flnb^−/− mice had significantly reduced bone mineral density compared with Flnb^+/+ controls, as quantified by dual x-ray absorptiometry (Fig. 5I). There was also a reduced cortical thickness (Fig. 5J) but no change in the trabecular volume (Fig. 5K) in tibiae of Flnb^−/− mice compared with Flnb^+/+ controls, as measured by peripheral quantitative computerized tomography (pQCT).

Expression of Flna and Flnc in Flnb^−/− Cells.

In Flnb^−/− fibroblasts, mRNA expression of either Flna or Flnc was not changed [supporting information (SI) Fig. 7]. However, Flna protein expression was reduced in Flnb^−/− fibroblasts (P < 0.05). Flna mRNA and protein were expressed by wild-type vascular endothelial cells, whereas there was no detectable Flnc mRNA or protein expression in these cells (SI Fig. 8). In Flnb^−/− endothelial cells, no alteration in the Flna mRNA expression was detected. The expression of Flna protein in chondrocytes of vertebral bodies was reduced in Flnb^−/− embryos compared with Flnb^+/+ chondrocytes (SI Fig. 9). Flnc protein expression was not detected in either Flnb^+/+ or Flnb^−/− chondrocytes (SI Fig. 9).

Reduced Migration of Flnb^−/− Embryonic Fibroblasts.

FLNB binds to the actin cytoskeleton. We tested whether the absence of Flnb would have an impact on cellular migration. The migration of primary fibroblasts from E13.5 Flnb^−/− embryos was reduced by 75% compared with fibroblasts from Flnb^+/+ embryos, as judged by a Boyden chamber assay (P < 0.05; Fig. 6A). In a wound-healing assay with cells grown on plastic plates, migration of Flnb^−/− fibroblasts was reduced by 20% (P < 0.05; Fig. 6B). There was no difference in the migration between Flnb^+/− and Flnb^+/+ fibroblasts.

Fig. 6.

Flnb^−/− fibroblasts migrate less than Flnb^+/− and Flnb^+/+ controls. (A) Fibroblasts are assayed in a Boyden chamber for migration 4 h after loading into cell culture dishes. Data are given as mean ± SD values (∗, P < 0.05, unpaired t test). Representative images of cells migrated through the holes in the membranes are given. (B) Decreased number of migrated Flnb^−/− fibroblasts into scraped areas in a wound-healing assay. Data are given as mean ± SD values (∗, P < 0.05, unpaired t test). Representative images of cells in the scraped areas (edges of wounds are shown by dashed red lines) captured at 0 and 12 h. (C) Up-regulated expression of RhoA protein in three different Flnb^−/− fibroblast lines compared with different Flnb^+/+ and Flnb^+/− fibroblasts. Data are given as mean ± SD values (∗, P < 0.05, unpaired t test). (D) Comparison of GTP-bound RhoA levels between Flnb^+/+ and Flnb^−/− fibroblasts. Data are expressed as the ratio of total amounts of GTP-bound RhoA in the presence of serum to starvation and normalized to Flnb^+/+ level. (E) Formation of actin filaments in Flnb^+/+ and Flnb^−/− fibroblasts induced by serum as detected by fluorescent staining by using phalloidin-TRITC.

Western blot analyses of total cellular extracts showed that the expression of RhoA was increased >3-fold in the Flnb^−/− cells compared with the Flnb^+/+ (Fig. 6C); however, the level of active RhoA induced by serum was reduced >20% (Fig. 6D). Stress fibers induced by serum showed more disorganized formation of filamentous actin in Flnb^−/− than in Flnb^+/+ cells (Fig. 6E). The level of total Rac protein or GTP-bound Rac was not changed in Flnb^−/− cells (SI Fig. 10). The expression of phosphorylated focal adhesion kinase (Tyr-397), Cdc42, and phosphorylated Akt (Ser-473) was not different in Flnb^−/− fibroblasts (SI Fig. 11). Immunoblotting confirmed the lack of Flnb protein in these Flnb^−/− cells.

Discussion

In this work, we generated Flnb knockout mice. Disruption of both alleles of Flnb was associated with embryonic death beginning at E11.5, indicating a pivotal role of Flnb in embryonic development. We demonstrated that vascular endothelial cells expressed high levels of Flnb and defined that the deficiency of Flnb led to an impaired development of microvasculature. In addition to endothelial cells, chondrocytes of the skeletal system strongly expressed Flnb, and the lack of Flnb caused severe skeletal malformations, particularly in the vertebral column. Finally, we demonstrated impaired migration of Flnb-deficient fibroblasts, indicating that Flnb may be important for the proper migration of cells into appropriate anatomical locations.

In our work, loss of Flnb expression resulted in embryonic death. Interestingly, a recent study of a chemically induced mutation in Flna showed embryonic lethality in male mice (the Flna gene is located on the X chromosome); the mutant embryos exhibited cardiac and skeletal malformations (11). Carrier females exhibited mild skeletal malformations, including sternum and palate defects. In another study of Flna deficiency in mice, Flna has been reported to be required for cell–cell contact in vascular development and cardiac morphogenesis (12). Similar to Flna deficiency in mice, we observed a defective microvascular development in Flnb^−/− embryos. Flna and Flnb are structurally related and ubiquitously expressed actin-binding proteins. Both overlapping and distinct expression patterns have been demonstrated (7). We argued that the absence of Flnb might result in a compensatory increase in the expression of Flna. On the contrary, the expression level of Flna protein was reduced in Flnb^−/− fibroblasts. The expression of Flna mRNA was not changed in endothelial cells. Expression of Flnc was not detected in either Flnb^+/+ or Flnb^−/− endothelial cells and chondrocytes. In our study, loss of Flnb expression resulted in severe skeletal malformations, both in the vertebral column and in ribs and phalanges. Although it is difficult to assess quantitatively the decreased level of Flna in Flnb^−/− chondrocytes, the results in mouse models of Flna and Flnb deficiency suggest a central role of filamins in skeletal development. We believe that the phenotypic abnormalities that appeared in homozygous Flnb-deficient mice are a combination of both chondrocyte/bone development dysfunction and generalized defective microvasculature that impairs delivery of nutrition to the developing skeletal system.

Our colocalization studies combining CD31 immunohistochemistry with Flnb immunohistochemistry or LacZ staining and RT-PCR analysis of Flnb transcript in embryonic endothelial cells demonstrated that Flnb is expressed in endothelial cells in agreement with earlier studies detecting FLNB in human endothelial cells (4) and normal brain vessels (13). However, the functional importance of FLNB in the vessel wall has not been studied. Our analysis of microvascular structures in the central system and perivertebral areas indicated disorganized microvascular network in the absence of Flnb, which may be the result of impaired migration of Flnb^−/− endothelial cells into the developing vessel wall during angiogenesis. Future studies are necessary to determine whether Flnb^−/− endothelial cells have a reduced capacity for migration. Impaired microvasculature in the perivertebral areas may also in turn affect recruitment of growth factors such as VEGF, which are capable of inducing skeletogenesis (14). An impaired recruitment of such factors could additionally worsen vertebral development in Flnb^−/− mice.

We demonstrated that Flnb was expressed in growth plate chondrocytes in mouse vertebral bodies in agreement with earlier results (7). The skeletal defects in Flnb-deficient mice are strikingly similar to those of the skeletal genetic disorders in the human. Localized mutations in FLNB, thought to confer gain of function, have been found in disorders that involve skeletal dysplasia accompanied by a spectrum of additional developmental abnormalities (7). We observed that Flnb^−/− mice have vertebral fusions and abnormalities and decreased hyaline cartilage in the vertebral, carpal, and tarsal bones similar to the clinical malformations related to vertebral segmentation, joint formation, and skeletogenesis in the syndromes of spondylocarpotarsal syndrome, atelosteogenesis I and III, Larsen syndrome (7), and boomerang dysplasia (8). Scoliotic and kyphotic abnormalities of the vertebral column in Flnb^−/− mice are similar to those seen in human boomerang dysplasia (8). Although we did not thoroughly study joints, we observed that Flnb^−/− mice have severely limited movement, which may indicate the presence of joint pathologies similar to human joint dislocations. Moreover, we did not observe the craniofacial abnormalities seen in most human cases. Further congenital abnormalities in humans include vascular pathologies, such as rarefaction of the retinal vessels (15), and we noticed that Flnb^−/− mice have impaired microvasculature defects and poorly developed larger vascular structures in the central nervous system.

Flnb^−/− fibroblasts in our work had an abnormal morphology and reduced ability for cellular migration. In contrast, Flna^−/− mouse fibroblasts have normal morphology and migratory properties (11, 12). However, the M2 human melanoma cell line, a well studied FLNA-deficient cell line, displays abnormal cell morphology and cell migration defects (16). This discrepancy in migration ability between FLNA-deficient fibroblasts and melanoma cells may be the result of a cell-specific effect of the interaction of FLNA with signaling molecules. In addition, Flnb^−/− cells exhibit more disorganized actin filaments that are formed during cellular migration. To explain the impaired migration of Flnb^−/− fibroblasts, we studied some of the molecules that have been earlier reported in the context of filamins. Of these molecules, we only observed increased expression of RhoA in Flnb^−/− fibroblasts. This increase may be a compensatory mechanism because of the decrease level of GTP-bound activated RhoA in Flnb^−/− cells. The activation of Rho family of small Ras-related GTPases and their downstream signaling molecules results in the contraction of actin fibers by recruiting filamin into the filopodial cytoskeleton (17). Cdc42 (3) and focal adhesion kinase (18) are also essential for filopodial formation, and Akt regulates migration (19), but expression of these molecules was not altered in Flnb^−/− fibroblasts. Detailed studies are necessary to determine the exact molecular mechanism behind the impaired migration in Flnb deficiency.

The demonstration of Flnb expression in vascular endothelial cells together with the skeletal malformations in Flnb^−/− mice indicates a pivotal role for Flnb in vasculature and vertebral morphogenesis. The range of abnormalities seen in homozygous embryos and cells is consistent with defects in cell migration. The phenotypic spectrum associated with mutation of Flnb highlights the importance of a functional actin cytoskeleton for many normal morphogenetic processes. Understanding the specific mechanisms by which Flnb deficiency leads to defects in microvascular abnormalities and skeletal morphogenesis will be facilitated by further investigations in Flnb-deficient mice.

Materials and Methods

Additional procedures are described in SI Materials and Methods.

Production of Flnb-Deficient Mice.

A mouse ES cell line (BCB085, strain 129/Ola) with an insertional mutation in Flnb was created in a gene-trapping program, BayGenomics (www.baygenomics.ucsf.edu). The gene-trapping vector pGT1lxf was designed to create an in-frame fusion between the 5′ exons of the trapped gene and a reporter, β-geo. The insertional mutation occurred in intron 20, which encodes the Ig-like domain repeat 16 just after hinge domain 1. Thus, the gene-trapped locus was predicted to yield a fusion transcript containing exons 1–20 of Flnb and β-geo. The ES cells were injected into C57BL/6 blastocysts to create chimeric mice. Five male high-percentage chimeric mice were obtained; all transmitted the mutation in the germ line. Chimeric mice were bred with wild-type C57BL/6 mice to generate heterozygous (+/−) Flnb-deficient mice. Thus, the mice used in this work had a mixed genetic background composed of ≈50% 129/Ola and ≈50% C57BL/6. The mice were weaned at 21 days of age, housed in a barrier facility with a 12-h light/dark cycle, and fed chow containing 4.5% fat (Ralston Purina, St. Louis, MO). All animal experiments were approved by the local animal ethical committee.

Genotyping and RT-PCR.

Genomic DNA (10–20 μg) from tail biopsies or yolk sacs from embryos was genotyped by PCR. A 2,004-bp wild-type Flnb fragment was amplified with forward primer 5′-GCC TCA AAG AGC TAC TGT CCA CGA-3′ located on exon 20 and reverse primer 5′-GGG TCA GAA TCA CGC AGG TTA CTT-3′ located on exon 21. A 2,504-bp mutant Flnb fragment was amplified with reverse primer 5′-GAC AGT ATC GGC CTC AGG AAG ATC G-3′ located in the β-geo insert and the wild-type Flnb forward primer.

RNA was extracted by using RNeasy mini kits (Qiagen, Valencia, CA) according to the manufacturer's instructions. RNA samples were treated with DNase (Invitrogen, Carlsbad, CA) followed by the generation of cDNA using SuperScript II (Applied Biosystems, Foster City, CA). The forward and reverse primers for Flnb were also used to study expression of Flnb mRNA by RT-PCR. Wild-type Flna transcript was amplified with forward primer 5′-AAG CCC TCT GCA GTT CTA TGT TGA T-3′ and reverse primer 5′-GCA AAC GTT TCA GCA GAC AGG GTT-3′. Additional primers are shown in SI Materials and Methods.

Western Blotting.

Proteins in total cellular extracts of murine tissues and cells were separated by SDS/PAGE and blotted onto a polyvinylidene difluoride membrane. Blocking was performed in TBS with 5% nonfat milk, followed by incubation with anti-Flnb recognizing near carboxyl terminus of Flnb (Calbiochem, San Diego, CA), anti-actin (Sigma, St. Louis, MO), and anti-RhoA (Santa Cruz Biotechnology, Santa Cruz, CA) antibodies in TBS with 5% nonfat milk. After several washes with TBS containing 0.1% Tween 20, the filters were incubated with anti-mouse or anti-rabbit IgG-horseradish peroxidase conjugate (Amersham Biosciences, Piscataway, NJ) in TBS with 5% nonfat milk. After several washes, proteins were visualized by using enhanced chemiluminescence (Amersham Biosciences) according to the manufacturer's recommendations. Densitometric readings of bands were carried out by using ImageQuant software (Bio-Rad, Hercules, CA). For additional antibodies used in immunoblotting and immunohistochemistry, see SI Materials and Methods.

Immunohistochemistry.

Immunohistochemical staining was performed as described earlier (20). Formaldehyde-fixed and paraffin-embedded sections were stained for Flnb (Calbiochem), CD31 (BD Biosciences, San Jose, CA), and LYVE-1 (21). The avidin/biotin detection system was used (ABComplex/horseradish peroxidase; Dako, Glostrup, Denmark). The reaction was visualized by the substrate 3-amino-9-ethylcarbazole or 3,3′-diaminobenzidine. Sections were counterstained with hematoxylin and mounted with cellulose triacetate.

X-Gal Staining.

Frozen tissue pieces were sectioned and fixed by using 0.25% glutaraldehyde in PBS for 5–10 min. The procedure for X-Gal staining was similar to that described previously (22). Briefly, sections were incubated at 37°C in PBS supplemented with 1 mg/ml X-Gal (Sigma)/5 mM potassium ferricyanide/5 mM potassium ferrocyanide/2 mM MgCl₂. Sections were rinsed with PBS and counterstained with Mayer's hematoxylin.

Whole-Mount Staining of Embryos.

Whole embryos were stained by using CD31 antibody as described previously (23). Images were captured by using an automated laser confocal fluorescent microscope, a custom-modified Ultraview LCI system (PerkinElmer, Waltham, MA). Eighty individual images of each embryo were automatically captured in a consecutive fashion at ×10 magnification at excitation 568 nm, by using QuantCapture 4.0, a program developed by Openlab Automator. The full-view embryo image was created from the 80 images to a perfectly aligned, seamless image montage by using virtual microscopy software (Quantum Image Lab) from Aragon Systems (Philadelphia, PA). These images covering the field of interest were processed for exact pixel alignment and flat fielding by using virtual microscopy. The final composite was assembled in Adobe Photoshop, version 8.0 (Adobe, San Jose, CA).

Staining of Bone and Cartilage.

After euthanasia of the mice, skin was peeled away, and all tissues were eviscerated. Animals were fixed in 95% (vol/vol) ethanol for 5 days and stained with 0.3% alcian blue and 0.2% alizarin red dissolved in glacial acetic acid and 70% (vol/vol) ethanol. Remaining soft tissues were cleared in 1% KOH for 5 days. Skeletal tissues were then incubated in increasing concentrations of glycerol and finally stored in 100% of glycerol.

Dual X-Ray Absorptiometry.

Measurement of areal bone mineral density of the mid-diaphyseal area of tibia ex vivo was performed with the Norland pDEXA SABRE (Norland, Fort Atkinson, WI) and the SABRE RESEARCH (London, U.K.) software (version 3.6) as described (24). Each group consisted of at least three tibiae, and the data are presented as mean ± SD values.

pQCT.

CT was performed with the PQCT XCT RESEARCH M (version 4.5B; Norland), operating at a resolution of 70 μm as described previously (24). Trabecular volumetrical bone mineral density was determined ex vivo, with a metaphyseal pQCT scan of the proximal tibia. The scan was positioned in the metaphysis at a distance from the distal growth plate corresponding to 4% of the total length of the tibiae, and the trabecular bone region was defined as the inner 45% of the total cross-sectional area. Cortical bone parameters were determined ex vivo with a mid-diaphyseal pQCT scan of the tibiae. Each group consisted of at least three tibiae, and the data are presented as mean ± SD values.

Embryonic Fibroblasts and Cell Migration Assays.

Primary fibroblasts were isolated from Flnb^+/+, Flnb^+/−, and Flnb^−/− embryos at E13.5. Migration assays were performed on cells in passages 2–4. Cell migration was quantified with a modified Boyden chamber assay (25) where the ability of cells to migrate through a micropore nitrocellulose filter (8-μm-thick, 8-μm-diameter pores) was measured. Approximately 3 × 10⁴ cells were seeded in the upper chamber in medium containing 2% serum; the chamber was lowered into a well containing 10% serum. After 4 h of incubation at 37°C, migrated cells were fixed to the filter in methanol and stained with Giemsa. All cells that had migrated through the filter were counted by using a light microscope. Quadruplicates of each sample are presented as mean ± SD values.

For wound-healing assay, 2 × 10⁶ cells were seeded in 60-mm plates and allowed to reach confluence. Wounds were created by scraping with a sterile P1000 pipette tip. After washing with culture medium, cells were cultured under standard conditions for up to 12 h, and images were captured by using a Zeiss Axiovert 100 inverted microscope (Carl Zeiss, Jena, Germany). Migrated cells were counted blindly in at least nine different areas, and data are presented as mean ± SD values.

To visualize actin filaments, cells were seeded in LabTek II slide chambers (Nalge Nunc International, Rochester, NY), fixed, rinsed, and incubated with phalloidin coupled with TRITC (Sigma). Finally, sections were mounted with Vectashield mounting medium with DAPI (Vectastain; Vector Laboratories, Burlingame, CA). Images were captured with a digital camera (Carl Zeiss) mounted on an AxioImager microscope (Carl Zeiss).

Measurement of GTP-Bound Form of RhoA and Rac.

Mouse embryonic Flnb^+/+ and Flnb^−/− fibroblasts at a confluence of 50% were washed once in serum-free medium, incubated for 24 h in medium containing 0.5% of FBS, and then incubated for 20 h in a serum-free medium to starve the cells that were treated with either 10% FBS (Invitrogen) for 5 min or 50 ng/ml platelet-derived growth factor-BB (ImmunoKontact, Abingdon, Oxon, U.K.) for 10 min to activate RhoA or total Rac, respectively. Protein lysates were collected, and activated GTP-bound RhoA or Rac was analyzed with a G-LISA activation assay biochemistry kit (Cytoskeleton, Denver, CO) according to the manufacturer's instructions.

Abbreviations

E

embryonic day

Flnb

filamin B

pQCT

peripheral quantitative computerized tomography.