Fgd1, the Cdc42 guanine nucleotide exchange factor responsible for ...

2 downloads 0 Views 985KB Size Report
FGD1, the gene responsible for the inherited disease faciogenital dysplasia, encodes a guanine nucleotide exchange factor (GEF) that specifically activates the.
© 2001 Oxford University Press

Human Molecular Genetics, 2001, Vol. 10, No. 5

485–495

Fgd1, the Cdc42 guanine nucleotide exchange factor responsible for faciogenital dysplasia, is localized to the subcortical actin cytoskeleton and Golgi membrane Lourdes Estrada1, Emmanuelle Caron3 and Jerome L. Gorski1,2,+ 1Department

of Human Genetics and 2Department of Pediatrics and Communicable Diseases, The University of Michigan Medical School, Ann Arbor, MI 48109, USA and 3MRC Laboratory for Molecular Cell Biology, University College of London, London, UK

Received 6 November 2000; Revised and Accepted 29 December 2000

FGD1, the gene responsible for the inherited disease faciogenital dysplasia, encodes a guanine nucleotide exchange factor (GEF) that specifically activates the p21 GTPase Cdc42. In order, FGD1 is composed of a proline-rich N-terminal region, adjacent GEF and pleckstrin homology (PH) domains, a FYVE-finger domain and a second C-terminal PH domain (PH2), structural motifs involved in signaling and subcellular localization. Fgd1, the mouse FGD1 ortholog, is expressed in regions of active bone formation within osteoblasts and in the osteoblast-like cell line MC3T3-E1, a finding consistent with its role in skeletal formation. Here, we use subcellular fractionation studies to show that endogenous Fgd1 protein is localized in the cytosolic and Golgi and plasma membrane fractions of mouse calvarial cells. Immunocytochemical studies performed with osteoblast-like MC3T3-E1 cells and other mammalian cell lines confirm the localization of Fgd1 and show that the proline-rich N-terminal region is necessary and sufficient for Fgd1 subcellular localization to the plasma membrane and Golgi complex. In contrast, the FYVE-finger and PH2 domains do not appear to direct the localization of Fgd1 or the activation of Cdc42. In addition, microinjection studies indicate that the N-terminal Fgd1 domain inhibits filopodia formation, suggesting that this region down-regulates GEF function. These results characterize the function of the Fgd1 domains for both protein localization and Cdc42 activation and indicate that the Fgd1 Cdc42GEF protein is involved in the regulation of Cdc42 activity at the subcortical actin cytoskeleton and Golgi complex. INTRODUCTION FGD1 is implicated as an important participant in skeletal formation because mutations in this gene result in the disease

faciogenital dysplasia (FGDY) or Aarskog syndrome (1,2), an X-linked skeletal dysplasia which adversely affects a characteristic set of skeletal structures (OMIM 305400). Most FGD1 mutations in FGDY patients are predicted null alleles; thus the X-linked recessive phenotype appears to be due to the absence of the gene product in affected males (2). The cardinal features of this disease include disproportionate acromelic short stature and specific craniofacial abnormalities; radiographic abnormalities include maxillary and mandibular hypoplasia, hypoplastic phalanges, retarded bone maturation and cervical vertebral anomalies (2). The FGD1 mouse ortholog, Fgd1, was cloned and mapped to a syntenic region of the mouse X chromosome (3); FGD1 and Fgd1 share the same structural organization and are 95% identical in sequence. During embryogenesis, Fgd1 expression is restricted to regions of incipient and active endochondral and intramembranous ossification including skeletal elements involved in the FGDY phenotype (4). Fgd1 protein is also expressed in mouse osteoblasts, human osteosarcoma cell lines and permanent osteoblast-like cell lines including MC3T3-E1 (4). Thus, the amassed data indicate that FGD1 signaling plays a critical role in skeletal formation. FGD1 encodes a guanine nucleotide exchange factor (GEF) for the p21 GTPase Cdc42, a member of the Rho (Ras homology) family of GTPase proteins (5,6). RhoGEFs activate Rho GTPases by catalyzing the exchange of bound GDP for GTP (7). The mammalian Rho protein family consists of at least 10 distinct proteins and three major subfamilies: Cdc42, Rac and Rho; each Rho protein has at least 50% homology with any of the others and 30% homology with Ras. Together, these Rho proteins regulate diverse cellular processes including the organization of the actin cytoskeleton, vesicular transport, cell cycle progression and the transcriptional regulation of gene expression (8). Rho GTPases play a critical role in the regulation of the actin cytoskeleton in a wide variety of eukaryotic cells. Rho regulates the assembly of contractile actin–myosin filaments (stress fibers) and focal adhesion complexes; Rac controls the assembly of actin filaments at the cell periphery to produce lamellipodia and membrane ruffles. In contrast, Cdc42 regulates the assembly of actin-rich surface protrusions called filopodia (9). By activating Cdc42, FGD1 stimulates

+To whom correspondence should be addressed at: Division of Pediatric Genetics, Room 3570 Medical Science Research Building II, Box 0688, University of Michigan Medical School, Ann Arbor, MI 48109 0688, USA. Tel: +1 734 647 2908; Fax: +1 734 763 9512; Email: [email protected]

486

Human Molecular Genetics, 2001, Vol. 10, No. 5

fibroblasts to form filopodia, cytoskeletal elements involved in cellular signaling, adhesion and migration (5). Through Cdc42, FGD1 also activates the c-Jun N-terminal kinase (JNK) signaling cascade (5,6), stimulates the passage of fibroblasts through the G1 phase of the cell cycle (10) and causes the tumorigenic transformation of NIH 3T3 fibroblasts (11). These accumulated data indicate that FGD1 is a specific Cdc42GEF and that FGDY is a developmental disorder of dysregulated FGD1/Cdc42 signaling. Rho protein signaling requires precise activation (9); this, in turn, requires the specific localization and regulation of RhoGEF activity. The human and mouse FGD1 protein is composed of (in order) a proline-rich N-terminal region, a GEF [or Dbl homology (DH)] domain, an adjacent pleckstrin homology (PH) domain, a FYVE-finger domain and a second C-terminal PH domain (PH2), structural motifs involved in signaling and subcellular localization. FGD1 is a founding member of a family of highly related FGD1-like genes which includes the mouse (and human ortholog) Fgd2 (FGD2) and Fgd3 (FGD3) genes (12,13) and the rat Frabin gene (14). Like FGD1, each of these genes contains a conserved core which spans at least 540 residues and includes the DH, PH, FYVE-finger and PH2 domains. Biochemical and microinjection studies show that, like FGD1, Fgd3 and frabin activate Cdc42 (13,14). However, each of these genes contains a different N-terminal domain. This suggests that, although FGD1-related proteins are likely to function as RhoGEFs in similar or related signaling pathways, each FGD1 family member may interact with a distinct set of protein-specific signaling and/or regulatory molecules. The speed and precision of intracellular signal transduction depends upon specific protein–protein interactions which regulate accurate protein recruitment and subcellular localization (15). The substrate for FGD1, Cdc42, has been localized to a variety of different subcellular regions (16–18); thus, Fgd1 could potentially act on several different substrate pools to affect different signaling pathways. In this study, we have used cellular fractionation and immunocytochemical methods to localize the Fgd1 protein in osteoblasts and other mammalian cell types. Our data indicate that, in addition to being found in the cytosol, a significant portion of endogenous Fgd1 is found in the Golgi and plasma membrane fractions of primary mouse calvarial (PMC) cells. Transient transfection studies performed with osteoblast-like MC3T3-E1 cells and other cell lines confirm the localization of Fgd1 in the subcortical actin cytoskeleton and Golgi apparatus. We find that the Fgd1 proline-rich N- terminal region contains the signal for its accurate subcellular targeting. In addition, microinjection studies indicate that the N-terminal Fgd1 domain inhibits filopodia formation, suggesting that this region down-regulates GEF function. Surprisingly, the absence of the FYVE-finger and PH2 domains, motifs common to signaling molecules involved in vesicular transport and protein localization, do not appear to alter the localization of Fgd1 or the activation of Cdc42. These results characterize the function of the Fgd1 domains for both protein localization and Cdc42 activation.

RESULTS Fgd1 is localized to the subcortical actin cytoskeleton and Golgi complex The subcellular distribution of Fgd1 was determined by performing total cellular fractionation and immunoblotting equivalent amounts of protein from each of the fractions of homogenized PMC cells. Isolated PMC cells are a standard preparation for osteoblasts (19). We previously showed that PMC cells constitutively express Fgd1 in culture; in cell extracts derived from these cells, anti-Fgd1 antibody detected a protein migrating at 107 kDa (4), the predicted size of the Fgd1 protein (1). As shown in Figure 1A, 50 µg of each fraction showed that Fgd1 was not present in the nuclear fraction (P3K), but was present in both the high-speed cytosolic (S100K) and the postnuclear membrane pellet (P100K) fractions. These results indicated that significant amounts of endogenous Fgd1 protein were present in both the cytosol and membranes of PMC cells. To further define the subcellular distribution of Fgd1, the PMC cell P100K (Fig. 1A) was resolved by Percoll density gradient ultracentrifugation. As shown in Figure 1B, an analysis of Percoll fraction aliquots showed that GRP94, a component of the endoplasmic reticulum (ER), was recovered in a broad range of fractions (fractions 10–24; buoyant density 1.042–1.080 g/ml). In contrast, Fgd1 was recovered in a relatively narrow range of fractions (fractions 20–24; buoyant density ≤ 1.042 g/ml). Like Fgd1, syntaxin 6 and β1 integrin, resident proteins of the Golgi complex and plasma membranes, respectively, were also recovered from fractions 20–24 (Fig. 1B). These results indicated that, in addition to being present in the cytosol, endogenous Fgd1 was localized in the plasma membrane and/ or Golgi apparatus of cultured osteoblasts. To further examine the distribution of Fgd1, we generated a tagged Fgd1 protein with green fluorescent protein (GFP) fused to the N-terminus of Fgd1 (construct fl–Fgd1), expressed the fusion protein in a variety of cultured cells by transient transfection and observed its subcellular localization by confocal microscopy. In MC3T3-E1 cells, a permanent nontransformed osteoblast-like cell line derived from mouse calvarial cells (20), fl–Fgd1 protein was observed in the plasma membrane and the perinuclear region (Fig. 2A). In contrast, the GFP produced from the control vector was predominantly localized in the nucleus (Fig. 2B). A similar distribution of fl–Fgd1 protein was observed in transiently transfected COS-7 cells; like the MC3T3-E1 cells, upon transfection into COS-7 cells, the fl–Fgd1 protein was localized to the plasma membrane and perinuclear region (Fig. 2D). A similar distribution of tagged Fgd1 protein was observed in transiently transfected NIH 3T3 cells; like the MC3T3-E1 and COS-7 cells, upon transfection into NIH 3T3 cells, the fl–Fgd1 protein was localized to the plasma membrane and perinuclear region (data not shown). Variation in the level of fl–Fgd1 protein expression was observed among transfected cells; although over-expressing cells contained relatively greater amounts of the epitope-tagged protein within the cytoplasm, fl–Fgd1 protein was consistently observed at the plasma membrane and perinuclear region in transfected cells. A myc epitope-tagged expression construct of the Fgd1 protein (construct pTGE–myc–Fgd1) was used to exclude the

Human Molecular Genetics, 2001, Vol. 10, No. 5

487

Figure 1. Density gradient fractionation of PMC cell homogenates. (A) PMC cells were homogenized and fractionated as described in Materials and Methods. Fifty micrograms of each fraction was loaded for each sample and resolved by SDS–PAGE on a 7% gel, transferred to an Immobilon-P membrane and probed with anti-Fgd1 polyclonal antibody. Proteins were visualized using enhanced chemiluminescence. Anti-Fgd1 antibody detected a protein migrating at 107 kDa; P, pellet; S, supernatant. (B) Percoll gradient analysis of Fgd1 in PMC cell homogenates. The P100K fraction resulting from the centrifugation of PMC cell homogenate was further fractionated through a self-forming Percoll density gradient at 20 000 g for 4 h. Fractions (1 ml) were collected by gravity; each fraction was spun twice at 100 000 g for 30 min to remove Percoll. A 50 µl aliquot of every other fraction was subjected to SDS–PAGE analysis and transferred to Immobilon-P membranes. Membranes were incubated with anti-Fgd1, anti-GRP94, anti-β1 integrin and anti-syntaxin 6 antibodies; blots were incubated with a secondary HRP-conjugated antibody and detected using chemiluminescent reagents.

possibility that the GFP tag was altering the localization of the Fgd1 protein. Like the fl–Fgd1 protein, when expressed in MC3T3-E1 and COS-7 cells, the myc epitope-tagged Fgd1 protein was expressed in the plasma membrane and perinuclear region of the cells (Fig. 2C). Colocalization studies were performed to confirm the subcellular distribution of Fgd1. COS-7 cells were transfected with fl–Fgd1 (Fig. 2D), fixed and stained with TRITCconjugated phalloidin (Fig. 2E) to detect filamentous actin. As shown by the yellow pseudocolor of the merged images (Fig. 2F), the GFP-tagged Fgd1 protein colocalized with filamentous actin; these results localized Fgd1 to the subcortical actin cytoskeleton. Confocal microscopy was used to study the localization patterns of the fl–Fgd1 protein and a Golgi marker to determine whether the perinuclear fluorescence associated with fl–Fgd1 expression was due to the presence of Fgd1 in the Golgi apparatus. COS-7 cells transfected with fl–Fgd1 (Fig. 2G) were treated with a Golgi-specific vital stain (Fig. 2H) and observed alive. As shown by the yellow pseudocolor of the merged images shown in Figure 2I, the fl–Fgd1 protein colocalized with the Golgi-specific vital stain; these results suggested that fl–Fgd1 was distributed in the Golgi apparatus. However, since the GFP staining is present in an area larger than that delineated by the Golgi marker, it was also apparent that fl–Fgd1 was localized in the cytosol, a finding consistent with the results of the cellular fractionation studies. In order to confirm the Golgi localization of fl–Fgd1, brefeldin A (BFA) was used to disperse the cis and medial Golgi apparatus (21). BFA-treated COS-7 cells transfected with the fl–Fgd1 expression construct (Fig. 2J) were treated with a Golgi-specific vital stain (Fig. 2K) and observed alive. As shown by the yellow pseudocolor of the merged images shown in Figure 2L, in BFA-treated

cells, the fluorescent signal of the fl–Fgd1 protein corresponded exactly with that of the Golgi-specific vital stain, indicating that the localization of Fgd1 to the Golgi apparatus was authentic. In contrast, an ER-specific vital stain failed to colocalize with fl–Fgd1 (data not shown). Based on these studies, we concluded that, in addition to the cytoplasm, Fgd1 was distributed in both the subcortical actin cytoskeleton and the Golgi apparatus. This pattern of subcellular localization was consistent with the results obtained by density gradient centrifugation. Together, these data strongly suggested that the Fgd1 protein expressed in COS-7 and MC3T3-E1 cells had the same distribution as that constitutively expressed in PMC cells. These results also indicated that transiently transfected cells would provide useful models for studying the determinants of Fgd1 subcellular localization. The N-terminal region targets Fgd1 protein localization A collection of GFP and myc epitope-tagged Fgd1 fusion protein constructs were generated to identify the structural domains that are involved in targeting Fgd1 to the plasma membrane and Golgi complex. As shown in Figure 3A, these expression constructs were designed to express the following portions of the Fgd1 protein with a 5′ GFP tag: the proline-rich N-terminal region (Nter), the DH and PH domains (DH/PH); the FYVE-finger and C-terminal PH domains (FYVE/PH2); the N-terminal region and the DH and PH domains (Nter+DH/PH); and the DH, PH, FYVE-finger and C-terminal PH domains (DH/PH+FYVE/PH2). As shown in Figure 3Bi, in cells expressing the Nter construct, Nter protein was localized to the subcortical actin cytoskeleton and intensely localized to the perinuclear Golgi region. Similarly, when expressed in COS-7 cells, the Nter+DH/PH protein was localized to the subcortical

488

Human Molecular Genetics, 2001, Vol. 10, No. 5

actin cytoskeleton and the Golgi apparatus (Fig. 3Bii), a distribution of protein identical to that observed in cells

expressing fl–Fgd1 (Fig. 2). In marked contrast, when transiently expressed in cells, the DH/PH and the FYVE/PH2

Figure 2. Fgd1 is colocalized to the subcortical actin cytoskeleton and Golgi apparatus. MC3T3-E1 cells (A and B) were transfected with either an expression vector encoding a GFP–Fgd1 fusion construct (fl–Fgd1) (A) or an expression vector encoding GFP alone (B). COS-7 cells were transfected with a myc epitope-tagged expression construct (pTGE–myc–Fgd1) containing the nearly full-length Fgd1 cDNA (C). In (A) and (C), arrows indicate the localization of epitope-labeled Fgd1 protein at the plasma membrane. COS-7 cells were transfected with the GFP–Fgd1 fusion construct, fl–Fgd1 (D–L). (D), (G) and (J) show the localization of the fl–Fgd1 protein. (E) COS-7 cell stained with TRITC-conjugated phalloidin to show the distribution of filamentous actin. The yellow signal generated in the overlayed images of (D) and (E) shows essentially complete colocalization of fl–Fgd1 protein and filamentous actin at the subcortical actin cytoskeleton (F). (G–L) Transfected COS-7 cells were vitally stained with TR-BODIFY ceramide C6, a Texas red-tagged Golgi-specific vital dye, and examined alive. (H) COS-7 cells stained with TR-BODIFY ceramide C6 to show the Golgi apparatus. When the images of (G) and (H) are overlayed, the generated yellow signal shows significant colocalization of the Golgi apparatus and fl–Fgd1 protein (I). (J–L) Transfected COS-7 cells vitally stained with TR-BODIFY ceramide C6 and treated with 5 µg/ml brefeldin A (BFA). (K) COS-7 cells stained with TR-BODIFY ceramide C6 and treated with BFA to show a dispersed Golgi apparatus. When the images of (J) and (K)are overlayed, the generated yellow signal shows essentially complete colocalization of the dispersed Golgi apparatus and fl–Fgd1 protein (L). Twelve hours after transfection, cells were fixed and examined by confocal microscopy. Magnification, 400×.

Human Molecular Genetics, 2001, Vol. 10, No. 5

489

Figure 3. Fgd1 domains responsible for protein localization. (A) The top bar is a schematic diagram of FGD1 showing the domains that comprise this protein including the proline-rich N-terminal region (hatched bar), the Dbl-homology (DH or GEF) and adjacent pleckstrin homology (PH) domains (DH/PH; gray bar), the FYVE-finger domain (FYVE; open bar) and the second C-terminal PH domain (PH2; black bar). Numbers denote the residues present at the beginning and end of a domain. Schematic figures below the top illustration show the composition of GFP–Fgd1 fusion constructs used to perform the localization experiments. The GFP protein is fused to the N-terminus of each construct. The fl–Fgd1 construct (residues 18–960) expresses the nearly full-length Fgd1 protein. Nter (residues 18–385) contains the N-terminal region; DH/PH (residues 375–706) contains the DH and PH domains; and FYVE/PH2 (residues 712–925) contains the FYVE-finger and PH2 domains. The Nter+DH/PH construct (residues 18–706) contains the N-terminal proline-rich region, the DH and PH domains; the DH/PH+FYVE/PH2 construct (residues 359–922) contains the DH, PH, FYVE-finger and PH2 domains. The same regions of the Fgd1 protein were used to generate myc epitope-tagged Fgd1 constructs. (B) Subcellular localization of Fgd1 fusion proteins in transfected COS-7 cells. Panels show COS-7 cells transfected with the following constructs: Nter (i), Nter+DH/PH (ii), DH/PH (iii) and FYVE/PH (iv). Arrows indicate recombinant Fgd1 protein at the plasma membrane (i and ii). Magnification, 400×.

proteins localized to cell nuclei (Fig. 3Biii and iv), like the control GFP. The nuclear localization of the DH/PH and FYVE/PH2 proteins was also associated with dramatic morphological changes in transfected COS-7 cells; as shown in Figure 3B, these morphologic changes included marked vacuolar enlargement and the formation of polynuclear cells. Similar morphologic changes were previously observed in fibroblasts following Fgd1 transfection (11). To rule out any possible influence of the GFP tag in the localization of our fusion proteins, we transfected cells with myc-tagged Fgd1 constructs containing the Nter+DH/PH, DH/PH and DH/PH+ FYVE/PH2 fragments. When transiently expressed in cells, the myc-tagged constructs showed a pattern of subcellular localization (and morphologic changes) identical to that previously observed with the GFP-tagged proteins (data not shown). These studies indicated that the DH, PH, FYVE-finger and Cterminal PH domains did not contain signals sufficient for targeting recombinant Fgd1 protein to the plasma and/or Golgi membranes. These studies also showed that, when expressed in cells, the N-terminal region of Fgd1 showed the same subcellular localization as that demonstrated by recombinant fulllength and endogenous Fgd1 protein. Thus, these data indicated that the N-terminal region of Fgd1 contained a motif that was necessary and sufficient for targeting Fgd1 to the actin cytoskeleton and Golgi complex. Fgd1 domains regulate filopodia formation and protein localization Although the transient transfection studies identified the N-terminal Fgd1 fragment as the region necessary for accurate Fgd1 subcellular targeting, an analysis of COS-7 actin filaments

failed to identify consistent changes in the actin cytoskeleton attributable to the different Fgd1 expression constructs. Compared with other cells, serum-starved Swiss 3T3 cells have a markedly reduced amount of fibular F actin and are relatively more responsive actin cytoskeletal regulators; numerous studies have demonstrated the utility and validity of using these cells to molecularly dissect cytoskeletal signaling (8,9). We had previously used quiescent Swiss 3T3 cells to perform microinjection experiments to show that a myc epitope-tagged FGD1 DH/PH expression construct promoted the Cdc42-dependent formation of filopodia (5). These results were consistent with the demonstration that recombinant DH/PH protein selectively bound Cdc42 (but not Rac1 or RhoA) and acted as a GEF for Cdc42 (6). These studies showed that the activation of filopodia formation by FGD1 was dependent upon Cdc42 and an intact DH domain, and that the activation of Cdc42 by FGD1 was blocked by dominant Cdc42 inhibitors (5). Thus, to identify Fgd1 domains that influence filopodia formation and confirm the localization studies, Fgd1 expression constructs were used to perform nuclear microinjection experiments in quiescent Swiss 3T3 cells. Swiss 3T3 cells were microinjected with Fgd1 expression constructs, fixed and stained with TRITC-conjugated phalloidin to detect filamentous actin; the results of these studies are summarized in Table 1. Expressed in Swiss 3T3 cells, the fl–Fgd1 protein stimulated the formation of filopodia in ∼80% of cells microinjected. In marked contrast, filopodia were observed in only 14% of cells microinjected with empty vector. However, since the cells were scored for the presence of two or more filopodia, the difference between these two groups was more dramatic than the numbers convey.

490

Human Molecular Genetics, 2001, Vol. 10, No. 5

Table 1. Stimulation of filopodia formation by recombinant Fgd1 proteins Fgd1 expression constructs

Percentage of cells showing filopodiaa

No. of independent experimentsb

fl–Fgd1

77.7 ± 5.5

2

Nter+DH/PH

23.7 ± 4.9

5

DH/PH

62.8 ± 3.5

2

DH/PH+FYVE/PH2

64.3 ± 2.5

2

Mock injection

14.4 ± 3.7

2

aCells with more than one actin-rich surface protrusion were scored as showing filopodia. Percentages are indicated as arithmetic mean ± SEM. bFor each independent experiment, at least 50 cells expressing the construct were examined; typically, >90% of microinjected cells express the GFP-tagged construct.

Expressing cells injected with fl–Fgd1 formed numerous filopodia; the filopodia of a typical fl–Fgd1-expressing cell are shown in Figure 4B. In contrast, positive control cells were generally limited to a total of two to three filopodia. The DH/PH and DH/PH+FYVE/PH2 constructs stimulated the formation of filopodia in ∼60% of cells microinjected. Like the response to fl–Fgd1, these constructs stimulated a profuse response and the formation of abundant filopodia (Fig. 4F and H). In marked contrast, the Nter+DH/PH construct failed to give a robust response and stimulated filopodia in relatively few cells (24%). Like the control cells, filopodia formation in Nter+DH/PHexpressing cells was limited to two to three filopodia; the actin cytoskeleton of a typical Nter+DH/PH-expressing cell is

shown in Figure 4D. These data showed that the Fgd1 DH/PH domains were effective in inducing the formation of filopodia, a finding consistent with earlier studies (5). These data also indicated that, in the absence of the FYVE-finger and PH2 domains, the Fgd1 N-terminal region inhibited filopodia formation, suggesting that this region down-regulated the activity of the DH domain. The microinjection experiments confirmed the protein localization results of the transient transfection studies. Expressed in Swiss 3T3 cells, fl–Fgd1 protein was again localized to the subcortical actin cytoskeleton and perinuclear Golgi region. In several cells, fl–Fgd1 protein also appeared to localize to the internal actin cytoskeleton (Fig. 4A); however, this was not uniformly observed. Consistent with the transfection studies, Nter+DH/PH protein was also localized to the subcortical actin cytoskeleton and the Golgi complex (Fig. 4C). In contrast, DH/PH and DH/PH+FYVE/PH2 proteins were distributed throughout the cytoplasm (Fig. 4E and G), a finding consistent with earlier microinjection studies (5). However, these proteins differed in their cytoplasmic distribution. In contrast to the relatively uniform distribution of the DH/PH+FYVE/PH2 protein, the DH/PH protein had a punctate intracellular distribution throughout the cytoplasm (Fig. 4E and G); the significance of this distribution is unclear. These results indicated that it was not essential to deliver the DH/PH domains to the plasma or Golgi membrane to induce filopodia. These results also indicated that localization of the Fgd1 DH/PH domain to the actin cytoskeleton was not, in and of itself, sufficient to stimulate filopodia formation. These data also confirmed the results that showed that the N-terminal region was necessary and

Figure 4. Fgd1 domains responsible for filopodia formation and protein localization. Quiescent, serum-starved Swiss 3T3 fibroblasts were microinjected with recombinant Fgd1 expression constructs. (A), (C), (E) and (G) show the localization of the expressed fusion protein; (B), (D), (F) and (H) show the actin cytoskeleton stained with TRITC-conjugated phalloidin. Cell nuclei were microinjected with the following expression constructs: fl–Fgd1 (A and B), Nter+DH/PH (C and D), DH/PH (E and F) and DH/PH+FYVE/PH2 (G and H). Arrows indicate recombinant Fgd1 protein at the subcortical actin cytoskeleton (C); arrowheads indicate filopodia (B, F and H). In marked contrast to cells injected with the other constructs containing the DH domain, cells injected with Nter+DH/PH did not form filopodia (D). Magnification, 400×.

Human Molecular Genetics, 2001, Vol. 10, No. 5

491

Figure 5. A model for Fgd1 localization and signaling. The proline-rich N-terminal domain is localized to the subcortical actin cytoskeleton. Proteins containing domains that bind to polyproline sequence tracts (SH3, WW, EVH1 and profilin) are shown within the actin cytoskeleton. A different set of proteins may be involved in targeting Fgd1 to the Golgi complex (data not shown). The Fgd1 DH/PH domain is shown activating Cdc42 through the catalytic exchange of GDP for GTP; Cdc42 activation leads to filopodia formation and the activation of the JNK signaling cascade. The FYVE-finger domain (FYVE) and C-terminal PH domain (PH2) are shown interacting with phosphorylated phosphatidylinositol (PtdIns) molecules in the plasma membrane; FYVE-finger and PH domains have been shown to bind to different PtdIns derivatives (see text). Our data suggest that the N-terminal region down-regulates the Fgd1 DH domain; the positive regulatory influences of the FYVE-finger and PH2 domains are speculative.

sufficient for Fgd1 targeting to the actin cytoskeleton and Golgi complex. DISCUSSION In this study, we demonstrate that Fgd1 has a limited subcellular distribution and that, in addition to being present in the cytosol, Fgd1 is specifically localized to the subcortical actin cytoskeleton and Golgi apparatus. Several lines of evidence support this localization. Density gradient fractionation and immunoblotting studies demonstrate this protein distribution in primary osteoblasts, cells that constitutively express Fgd1. This localization is further substantiated by showing that transiently expressed GFP and myc epitope-tagged Fgd1 fusion proteins demonstrate this same subcellular distribution in a variety of cell types including the osteoblast-like cell line MC3T3-E1, COS-7 cells and Swiss 3T3 fibroblasts. The localization of Fgd1 to the actin cytoskeleton and the Golgi complex is confirmed by showing that Fgd1 colocalizes with actin filaments within the subcortical actin cytoskeleton and with BFA-sensitive Golgi-specific vital dyes. Transient transfection and microinjection studies show that the proline-rich N-terminal region is necessary and sufficient for Fgd1 subcellular localization to the plasma membrane and Golgi complex. These same studies also show that the remainder of the Fgd1 protein is not required for accurate subcellular localization; specifically, the FYVE-finger and the C-terminal PH domains are not required. In addition, microinjection studies indicate that the N-terminal Fgd1 domain inhibits filopodia formation, suggesting that, in the absence of the FYVE-finger and PH2 domains, the N-terminal region down-regulates DH domain function. Although we cannot rule out the formal possibility

that, in this particular construct, the N-terminal region induces an improperly folded and inactive DH domain, the absence of a similar effect on the ‘full-length’ Fgd1 construct argues against this possibility. In addition, our results do not rule out the possibility that the N-terminal region may interact with the C-terminal Fgd1 domains to regulate GEF activity; the mechanism by which the N-terminal region inhibits filopodia formation awaits additional investigation. Together, these results strongly suggest that the Fgd1 Cdc42GEF protein is involved in the regulation of Cdc42 activity at the subcortical actin cytoskeleton and the Golgi complex and that the Fgd1 N-terminal domain is critical to the correct distribution of this signaling component. These results are summarized schematically in Figure 5. Intracellular signal transduction relies on specific protein– protein interactions to provide accurate protein recruitment and precise subcellular localization (15). Our studies show that non-cytosolic Fgd1 is specifically localized to the subcortical actin cytoskeleton and the Golgi apparatus; the fact that this distribution is observed in both constitutively and exogenously expressing cells strongly suggests that this localization is accurate and biologically relevant. The substrate for Fgd1, the Rho GTPase Cdc42, has been localized to several different subcellular regions. In addition to an inactive cytosolic form bound to the Rho–GDP dissociation inhibitor protein (16), membrane-bound forms of Cdc42 have been localized to both the plasma membrane (18) and a Golgi complex (17). Thus, our results suggest that, by being distributed to the plasma and Golgi membranes, Fgd1 is poised to activate Cdc42 signaling in either subcellular region. It is firmly established that Cdc42 plays a critical role in the regulation of the actin cytoskeleton in a wide variety of eukaryotic cells, and that in mammalian

492

Human Molecular Genetics, 2001, Vol. 10, No. 5

cells Cdc42 regulates the assembly of actin-rich filopodia (9). It is also established that Cdc42 directly binds to the γ-subunit of the coatomer (γCOPI) complex (22) and regulates vesicular membrane transport and epithelial cell polarity (22,23). Thus, based on the localization of Fgd1 to the Golgi complex and subcortical actin cytoskeleton, it is reasonable to hypothesize that Fgd1 acts on Cdc42 to regulate vesicular transport and the actin cytoskeleton within the osteoblast. Our data indicate that the N-terminal region of Fgd1 is solely responsible for its targeting to the subcortical actin cytoskeleton and Golgi complex. The 5′ region of the Fgd1 protein is remarkably proline-rich and proline constitutes ∼22% of the first 250 amino acid residues; more importantly, the overwhelming majority of these proline residues are clustered to form short polyproline tracts that are dispersed throughout the N-terminal region (1,3). A variety of different structural domains have been shown to bind to short polyproline sequence tracts; these include the Src homology 3 (SH3) domain, the WW domain, the profilin protein and the EVH1 domain (Fig. 5) (24). The Fgd1 N-terminal region is predicted to encode two partially overlapping SH3-binding domains (1,3). Studies have shown that peptide ligands can bind to an SH3 domain in one of two orientations; either an N→C or C→N terminal orientation relative to the SH3 domain (25). The two predicted SH3–BD motifs (residues 170–187) oppose each other in opposite orientation; thus these sites potentially provide two distinct ligands for different SH3 domains. SH3 domains are typically part of cytoskeletal proteins and proteins involved in signal transduction such as Abl, actin-binding protein, cortactin, Grb2, PI3K regulatory subunit, PLCγ and spectrin (24). Thus, it is reasonable to hypothesize that the predicted SH3-binding domains may be involved in Fgd1 protein localization. The actin-binding protein profilin has also been shown to bind polyproline sequence tracts with high affinity. Based on the sequences of identified profilin-binding proteins, a profilin-binding consensus sequence, ZPPX, was proposed, with Z representing proline, glycine, alanine or a charged amino acid, and X representing a hydrophobic residue (26). Profilin-binding proteins were found to contain at least two copies of this motif in a short amino stretch of not more than 100 residues (26). The Fgd1 N-terminal region contains at least six potential profilin-binding motifs (3). Similarly, WW and EVH1 domains have been shown to selectively bind short polyproline sequences; however, no definitive ligand consensus sequences are identified for these domains (24). Proline-rich ligands have typically been observed in protein– protein interactions involving the recruitment or interchange of several proteins (24). Although we cannot exclude the possibility that additional signaling domains are held within the N-terminal region, our results suggest that the Fgd1 N-terminal region is recruited to the subcortical actin cytoskeleton and Golgi complex through protein–protein interactions involving the polyproline sequence tracts contained in this region (Fig. 5). Studies directed towards identifying Fgd1-interacting proteins are underway. The recognition that the Fgd1 N-terminal region is involved in protein targeting has significance for understanding other FGD1-related proteins. Although every FGD1-related family member contains a conserved core structure of four domains (in order: DH, PH, FYVE-finger and C-terminal PH domains), each of the genes (Fgd1, Fgd2, Fgd3 and Frabin) encodes a

different N-terminal domain (12,13). The Frabin N-terminal region encodes an actin-binding domain (ABD) (14). Studies show that this N-terminal ABD is necessary and sufficient for accurate Frabin localization to the subcortical actin cytoskeleton, and that none of the other Frabin domains are required for protein localization (27). Although Fgd1 and Frabin are targeted to different subcellular regions, these results are remarkably consistent in that the N-terminal regions of Fgd1 and Frabin are solely responsible for their localization. Based on these results and the observed conservation among family members, it seems reasonable to predict that N-terminal domains will be shown to specify the subcellular distribution of the other FGD1-related proteins. Since each of these proteins contains a different N-terminal domain, these results suggest that each FGD1-related protein may interact with a different set of recruiting and/or regulatory molecules. Despite these targeting similarities, some of our Fgd1 expression results differ from those obtained with Frabin. For example, our microinjection studies show that the Fgd1 DH/PH domains stimulate filopodia formation but the Nter+DH/PH construct does not. In contrast, the Frabin DH/PH construct does not stimulate filopodia and the Nter+DH/PH does (27). The basis of these observed differences in the ability to stimulate filopodia formation between the similar fragments of Fgd1 and Frabin is not known. However, these observations do suggest that the Nterminal region of the Frabin protein may be involved in the regulation of GEF activity, a finding consistent with our results. Although our data do not indicate the function of the Fgd1, FYVE-finger and PH2 domains, the highly conserved features of these domains strongly suggest that they play an important role in the biology of FGD1-related protein family members. PH signaling domains bind the charged headgroups of specific polyphosphoinositide molecules and phosphotyrosine residues (28). FYVE-finger domains are evolutionarily conserved cysteine-rich zinc-binding motifs. Most commonly these domains have been identified in proteins involved in membrane trafficking such as the yeast vacuolar sorting proteins, Vac1p and Vps27p, the yeast phophatidylinositol-4phoshphate-5-kinase, Fab1, and the mammalian ATPase, Hrs-2 (29). Recently, it was shown that the FYVE-finger domains selectively bind to phosphatidylinositol-3-phosphate, and that this binding is critical for protein targeting (29,30). Our data failed to identify a localization role for either the FYVE-finger domain or the PH2 domain, a finding consistent with the Frabin data (27). However, these experiments could not exclude the possibility that these domains might modulate Fgd1 activity through domain–ligand interactions (Fig. 5). Based on these results, a clearer image of the role of Fgd1/FGD1 in mammalian skeletal formation is beginning to emerge. Expressed in osteoblasts and localized to the subcortical actin cytoskeleton and Golgi complex, this Cdc42 activator is poised to influence osteoblast cytoskeletal organization, cell polarity and/or membrane transport, cellular processes critical to normal cellular function and differentiation. In Saccharomyces cerevisiae, Schizosaccharomyces pombe, Drosophila melanogaster and Caenorhabditis elegans, the loss of RhoGEF activity is functionally equivalent to the loss of the target Rho protein (31); thus, it is likely that a loss of FGD1 activity results in disrupted osteoblast Cdc42 signaling. Several lines of evidence indicate that Cdc42 signaling and actin cyto-

Human Molecular Genetics, 2001, Vol. 10, No. 5

skeletal regulation play a critical role in the molecular and cellular biology of bone tissue. Transgenic mice (32–35) and in vitro biochemical studies (36,37) show that extracellular matrix molecules (ECM) and growth factors regulate osteoblast growth and function. Rho GTPases play a critical role in mediating signals from growth factors and the ECM to modify cellular proliferation, cytoskeletal organization and differentiation (8,9). In particular, Cdc42 signaling plays a crucial role in regulating integrin-dependent cellular adhesion and cytoskeletal organization (38). In embryonic chick osteoblasts, an intact actin cytoskeleton is required for the regulated expression of osteopontin in response to mechanical strain (39). Studies show that the actin cytoskeleton and the Rho signaling cascade are involved in the detection and response to fluid shear induced mechanical signaling in MC3T3-E1 cells (40). Additional studies have demonstrated that the actin cytoskeleton is critical for determining and maintaining osteocyte shape (41). These observations suggest that FGD1/Cdc42 signaling may play a role in the organization of the osteoblast actin cytoskeleton or in sensing mechanical stress. Although we cannot rule out the formal possibility that the FGD1 may be involved in additional signaling pathways, the accumulated data strongly indicate that FGDY is a disorder of the FGD1/ Cdc42 signaling in the developing skeleton. MATERIALS AND METHODS Primary cells, cell lines and antibodies PMC cells were isolated and cultured as previously described (4). The non-transformed osteoblast-like mouse cell line MC3T3-E1 (20) was kindly provided by Dr Franceschi (University of Michigan, Ann Arbor, MI). Permanent mouse embryonic NIH 3T3 fibroblasts and monkey kidney COS-7 cells were purchased from the American Tissue Culture Collection (ATCC). MC3T3-E1 cells were cultured in α-modified minimum essential medium (α-MEM); all other cell lines were cultured in Dulbecco’s modified Eagle′s Medium (DMEM). All media were supplemented with 10% fetal calf serum (FCS). Polyclonal anti-Fgd1 was produced by immunizing rabbits with a multiple antigen peptide (MAP) containing the synthetic peptide (Fgd1 residues 871–892) covalently linked to a polylysine core peptide as previously described (4). Rabbit polyclonal anti-GRP94 and mouse anti-syntaxin 6 were kindly provided by Dr Green (Saint Louis University, St Louis, MO) and Dr Stenmark (Norwegian Radium Hospital, Oslo, Norway), respectively. Anti-integrin β1 was obtained commercially (Transduction Laboratories). Monoclonal anti-myc antibody (9E10) and secondary anti-mouse IgG–FITC antibodies were purchased from Santa Cruz Biotechnology. For chemiluminescence, immunoblots were incubated with a secondary horseradish peroxidase (HRP)-conjugated antibody purchased from Pierce. Subcellular fractionation PMC cells were isolated as previously reported (4). Calvaria of 1- to 4-day-old mice were dissected, isolated from periosteum and subjected to sequential digestions of 20, 40, 90 and 60 min in 2 mg/ml collagenase A (Boehringer Mannheim) with 0.25% trypsin (Gibco BRL). Cells from the third and fourth digest

493

were washed, counted and plated in α-MEM supplemented with 10% FCS, 50 µg/ml ascorbic acid and 10 mM Na β-glycerol phosphate at a density of 55 000 cells/cm2. The third and fourth PMC cell fractions were previously shown to be the most competent in forming mineralized bone tissue as assayed by cultured bone nodule formation (4). PMC cells were cultured in a humidified incubator in 5% CO2 at 37°C for 7–14 days prior to homogenization. PMC cells (∼5 × 107 cells) were scraped with a rubber policeman into homogenization buffer (HB) consisting of 0.25 M sucrose with 1 mM EDTA pH 6.8. Cells were washed by centrifugation and resuspended in HB (∼5 ml cell pellet). Cells were homogenized manually with a Dounce glass homogenizer until at least 50% cell breakage occurred. The resulting homogenate was then centrifuged at 4°C for 10 min at 3000 g. The pellet (P3K) was washed with HB and centrifuged at 100 000 g for 90 min; the resulting pellet was resuspended in HB and centrifuged again at 100 000 g for 90 min. The resulting high-speed pellet (P100K; 0.5 ml) was fractionated through a self-forming Percoll density gradient essentially as described (42). This gradient was made by layering 22 ml of Percoll suspension (13.2 ml of 9:1 Percoll plus 8.8 ml of dissolved pellet in HB) onto 60% w/v aqueous sucrose, pH 6.8. For gradient calibration, a duplicate Percoll gradient (buoyant densities ranging from 1.15 to 1.042 g/ml) was formed using density marker beads (Amersham Pharmacia Biotech). After centrifugation with a Beckman model SW55Ti rotor at 20 000 g for 4 h at 4°C, 1 ml fractions were collected by gravity. Prior to western blot analysis, each fraction was spun twice at 100 000 g for 30 min to remove the Percoll. A 50 µl aliquot of every other fraction was subjected to SDS– PAGE analysis and transferred to Immobilon-P membranes as described (4). Membranes were incubated in anti-Fgd1 (1:7500), anti-GRP94 (1:10 000), anti-integrin B1 (1:5000) and anti-syntaxin 6 (1:5000) overnight at 4°C. Immunoblots were incubated with a secondary HRP-conjugated antibody (1:10 000–25 000) and detected using chemiluminescent reagents (ECL). Expression constructs and cell transfection The multiple cloning site of the GFP eukaryotic expression vector pEGFP–C1 (Clontech) was modified to facilitate the generation of GFP–Fgd1 expression constructs. To generate the vector pGFP1, pEGFP–C1 DNA was digested with EcoRI and ligated to annealed oligomers: 5′-AATTGCAGCGCTGGCTCAGCCCGTACGACCGGTGGGCGCGCCGAATTCG-3′ and 5′-GATCCGAATTCGGCGCGCCCACCGGTCGTACGGGCTGAGCCAGCGCTGC-3′. To generate vector pGFP2, pGFP1 DNA was digested with EcoRI–BamHI and ligated to annealed oligomers: 5′-AATTGCCTGCAGTCGACCGGATCCCGATCGGGTACCGCGGGCGAATCCC-3′ and 5′-GATCGGGATTCGCCCGCGGTACCCGATCGGGATCCGGTCGACTGCAGGC-3′. The Fgd1 BlpI–EcoRI cDNA restriction fragment [residues 18–960; (3)] was cloned in-frame into pGFP1 vector to generate the expression construct fl–Fgd1. The Fgd1 BamHI–EcoRI restriction fragment (residues 375– 706) of the previously reported construct pRK5 (5) was cloned into pGFP2 in-frame to generate the expression construct DH–PH. The FYVE–PH2 expression construct (residues 712–925) was generated by cloning the BamHI–EcoRIdigested PCR product derived from Fgd1 cDNA into pGFP2

494

Human Molecular Genetics, 2001, Vol. 10, No. 5

(primer pair 5′-GGTCGTGGGATCCCCAAGCGGGCACCTACGCCC-3′ and 5′-TCACGATGAATTCTACGTGTCCCCTCGGCCCGC-3′). The Nter+DH/PH expression construct (residues 18–706) was generated by cloning the BlpI–EcoRIdigested PCR product derived from Fgd1 cDNA into pGFP1 (primer pair 5′-ACCGGTAGCGTCACCGGATCCATGCATGGCCACCGAGTC-3′ and 5′-ACCGGTGGGCGCGCCGAATTCCATCATCCCTGTTTGTTGAGT-3′). The DH/PH+ FYVE/PH2 expression construct (residues 359–922) was generated by cloning the BlpI–EcoRI-digested PCR product derived from Fgd1 cDNA into pGFP1 (primer pair 5′-ACCGGTAGCGCTGGCTCAGCCGAGAGACAGGAGTCTGTGGAG-3′ and 5′-ACCGGTGGGCGCGCCGAATTCCGTGTCCCCTCGGCCCGC-3′). The Nter expression construct (residues 18–385) was generated by cloning a BlpI–EcoRI-digested Fgd1 cDNA restriction fragment into pGFP1. The mycepitope mammalian expression vector pTGE was previously described (4). Selected restriction fragments were cloned into the pTGE vector in frame to generate myc epitope-tagged expression constructs containing the nearly full length Fgd1 cDNA [pTGE-myc-Fgd1; residues 18–960; (4)], the N-terminal, DH and PH domains (pTGE-myc-Nter+DH/PH; residues 18–706) and the DH, PH, FYVE and PH2 domains (pTGEmyc-DH/PH+FYVE/PH2; residues 359–922). The myc epitope-tagged DH/PH (residues 375–706) expression construct (pRK5) was reported previously (5). All constructs were verified by DNA sequencing. Fusion protein expression was confirmed by western blot analysis. For transfections, cells were plated into 6-well trays containing one coverslip per well. The following day, cells were transfected with 2–5 µg of DNA per well using lipofectaminePlus according to the recommendations of the supplier (Life Technologies). Cells were prepared for analysis 12 h after transfection. To study Golgi membrane dispersal, 3 h after transfection cells were treated with 5 µg/ml BFA (Molecular Probes) for 8 h. Immunofluorescence and microscopy Cells were fixed and permeabilized essentially as described (43). Briefly, cells were fixed in 4% paraformaldehyde/1% glutaraldehyde/PBS for 20 min, permeabilized in 0.2% Triton X-100/PBS for 10 min, and blocked with sodium borohydride (0.5 mg/ml) in PBS for 5 min. After blocking, cells were incubated with primary antibodies diluted in PBS for 1 h and secondary antibody for 45 min. To detect actin filaments, cells were stained with 0.1 µg/ml TRITC-conjugated phalloidin (Sigma) for 45 min. The Golgi apparatus was detected by staining live cells with the Golgi-specific vital dye TR-BODIPY ceramide C6, and the ER was stained with ER-tracker BlueWhite DPX vital dye following the manufacturer’s recommendations (Molecular Probes). Cells transfected with GFP fusion constructs were fixed 12 h after transfection, rinsed and mounted by inverting coverslips onto 5 µl of Fluoromount G (Electron Microscopy Sciences). Cells were examined with an MRC-600 BioRad confocal microscope and Comos software (BioRad), or a Zeiss Axiophot 2 epifluorescent microscope equipped with an Axiocam CCD camera and Axiovision digital imaging software (Zeiss).

Microinjection Serum-starved Swiss 3T3 were prepared as described (44). Eukaryotic Fgd1 expression vectors (0.1 mg/ml) were microinjected into the nucleus of at least 30 cells, in a temperatureand CO2-controlled chamber as previously described (45). Following injection, cells were returned to the incubator for 2–3 h. Cells were fixed in 4% paraformaldehyde/PBS for 20 min, blocked with 50 mM ammonium chloride for 10 min and permeabilized with 0.02% Triton X-100 for 5 min. Fixed cells were incubated with 0.1 µg/ml TRITC-conjugated phalloidin in PBS for 45 min. Injected cells were detected with AlexaFluor 350 streptavidin bound to biotin dextran (Molecular Probes) which was co-injected with each of the expression vectors. Coverslips were mounted by inverting them onto 5 µl Mowiol mountant (Calbiochem). Cells were examined by fluorescent microscopy as described above. ACKNOWLEDGEMENTS We thank Drs B.W. Donohoe and C.A. Edwards for technical advice and assistance.We are grateful to Dr Franceshi for the MC3T3-E1 cells, Dr Green for the rabbit polyclonal antiGRP94 antibody and Dr Stenmark for the mouse anti-syntaxin 6 antibody. We are grateful to Dr Hall and Dr Long for helpful advice and discussions. This work is supported, in part, by National Institutes of Health Grants 5T32HD07505-02 (L.E.) and HD34446 (J.L.G.). REFERENCES 1. Pasteris, N.G., Cadle, A., Logie, L.J., Porteous, M.E.M., Schwartz, C.E., Stevenson, R.E., Glover, T.W., Wilroy, R.S. and Gorski, J.L. (1994) Isolation and characterization of the faciogenital dysplasia (Aarskog-Scott syndrome) gene: a putative rho/rac guanine nucleotide exchange factor. Cell, 79, 669–678. 2. Gorski, J.L. (2001) Aarskog-Scott Syndrome. In Scriver, C.R., Beaudet, A.L., Sly, W.S. and Valle, D. (eds), The Metabolic and Molecular Basis of Inherited Disease, 8th edn. McGraw-Hill, New York, pp. 6153–6165. 3. Pasteris, N.G., de Gouyon, B., Cadle, A.B., Campbell, K., Herman, G.E. and Gorski, J.L. (1995) Cloning and regional localization of the mouse faciogenital dysplasia (Fgd1) gene. Mamm. Genome, 6, 658–661. 4. Gorski, J.L., Estrada, L., Hu, C. and Liu, Z. (2000) Skeletal-specific expression of Fgd1 during bone formation and skeletal defects in Faciogenital Dysplasia (FGDY; Aarskog syndrome). Dev. Dyn., 218, 573–586. 5. Olson, M.F., Pasteris, N.G., Gorski, J.L. and Hall, A. (1996) Faciogenital Dysplasia Protein (FGD1) and Vav, two related proteins required for normal embryonic development are upstream regulators of Rho GTPases. Curr. Biol., 6, 1628–1633. 6. Zheng, Y., Fischer, D.J., Tigyi, G., Pasteris, N.G., Gorski, J.L. and Xu, Y. (1996) The faciogenital dysplasia gene product FGD1 functions as a Cdc42Hs-specific guanine-nucleotide exchange factor. J. Biol. Chem., 271, 33169–33172. 7. Cerione, R.A. and Zheng, Y. (1996) The Dbl family of oncogenes. Curr. Opin. Cell Biol., 8, 216–222. 8. Van Aelst, L. and D’Souza-Schorey, C. (1997) Rho GTPases and signaling networks. Genes Dev., 11, 2295–2322. 9. Hall, A. (1998) Rho GTPases and the actin cytoskeleton. Science, 279, 509–514. 10. Nagata, K., Lamarche, N., Gorski, J.L. and Hall, A. (1998) Activation of G1 progression, JNK MAP kinase and actin filament assembly by the exchange factor FGD1. J. Biol. Chem., 273, 15453–15457. 11. Whitehead, I.P., Abe, K., Gorski, J.L. and Der, C.J. (1998) Cdc42 and FGD1 cause distinct signaling and transforming activities. Mol. Cell. Biol., 18, 4689–4697. 12. Pasteris, N.G. and Gorski, J.L. (1999) Isolation, characterization, and mapping of the mouse and human Fgd2 genes, Faciogenital Dysplasia (FGD1; Aarskog syndrome) gene homologues. Genomics, 60, 57–66.

Human Molecular Genetics, 2001, Vol. 10, No. 5

13. Pasteris, N.G., Nagata, K., Hall, A. and Gorski, J.L. (2000) Isolation, characterization, and mapping of the mouse Fgd3 gene, a new Faciogenital Dysplasia (FGD1; Aarskog syndrome) gene homologue. Gene, 242, 237–247. 14. Obaishi, H., Nakanishi, H., Mandai, K., Satoh, K., Satoh, A., Takahashi, K., Miyahara, M., Nishioka, H., Takaishi, K. and Takai, Y. (1998) Frabin, a novel FGD1-related actin filament-binding protein capable of changing cell shape and activating c-Jun N-terminal kinase. J. Biol. Chem., 273, 18697–18700. 15. Pawson, T. and Nash, P. (2000) Protein-protein interactions define specificity in signal transduction. Genes Dev., 14, 1027–1047. 16. Ueda, T., Kikuchi, A., Ohga, N., Yamamoto, J. and Takai, Y. (1990) Purification and characterization from bovine brain cytosol of a novel regulatory protein inhibiting the dissociation of GDP and the subsequent binding of GTP to RhoB p20, a Ras p21-like GTP-binding protein. J. Biol. Chem., 265, 9373–9380. 17. Erickson, J.W., Zhang, C., Kahn, R.A., Evans, T. and Cerione, R.A. (1996) Mammalian Cdc42 is a brefeldin A-sensitive component of the Golgi apparatus. J. Biol. Chem., 271, 26850–26854. 18. Backlund Jr, P.S. (1993) Carboxyl methylation of the low molecular weight GTP-binding protein G25K. Biochem. Biophys. Res. Commun., 196, 534–542. 19. Majeska, R.J. (1996) Culture of osteoblastic cells. In Bilezikian, J.P., Raisz, L.G. and Rodan, G.A. (eds), Principles of Bone Biology. Academic Press, San Diego, CA, pp. 1229–1237. 20. Sudo, H., Kodama, H., Amagai, Y., Yamamoto, S. and Kasai, S. (1983) In vitro differentiation and calcification in a new clonal osteogenic cell line derived from newborn mouse calvaria. J. Cell Biol., 96, 191–198. 21. Lippincott-Schwartz, J., Yuan, L.C., Bonifacino, J.S. and Klausner, R.D. (1989) Rapid redistribution of Golgi proteins into the ER in cells treated with brefeldin A: evidence for membrane cycling from Golgi to ER. Cell, 56, 801–813. 22. Wu, W.J., Erickson, J.W., Lin, R. and Cerione, R.A. (2000) The γ-subunit of the coatomer complex binds Cdc42 to mediate transformation. Nature, 405, 800–804. 23. Korschewski, R., Hall, A. and Mellman, I. (1999) Cdc42 controls secretory and endocytic transport to the basolateral plasma membrane of MDCK cells. Nature Cell. Biol., 1, 8–13. 24. Kay, B.K., Williamson, M.P. and Sudol, M. (2000) The importance of being proline: the interaction of proline-rich motifs in signaling proteins with their cognate domains. FASEB J., 14, 231–241. 25. Cohen, G.B., Ren, R. and Baltimore, D. (1995) Modular binding domains in signal transduction proteins. Cell, 80, 237–248. 26. Witke, W., Podtelejnikov, A.V., Di Nardo, A., Sutherland, J.D., Gurniak, C.B., Dotti, C. and Mann, M. (1998) In mouse brain profilin I and profilin II associate with regulators of the endocytic pathway and actin assembly. EMBO J., 17, 967–976. 27. Umikawa, M., Obaishi, H., Nakanishi, H., Satoh-Horikawa, K., Takahashi, K., Hotta, I., Matsuura, Y. and Takai, Y. (1999) Association of Frabin with the actin cytoskeleton is essential for microspike formation through activation of Cdc42 small G protein. J. Biol. Chem., 274, 25197–25200.

495

28. Lemmon, M.A., Ferguson, K.M. and Schlessinger, J. (1996) PH domains: diverse sequences with a common fold recruit signaling molecules to the cell surface. Cell, 85, 621–624. 29. Gaullier, J.-M., Simonsen, A., D’Arrigo, A., Bremnes, B. and Stenmark, H. (1998) FYVE fingers bind PtdIns(3)P. Nature, 394, 432–433. 30. Stenmark, H. and Aasland, R. (1999) FYVE-finger proteins—effectors of an inositol lipid. J. Cell. Science, 112, 4175–4183. 31. Boguski, M.S. and McCormick, F. (1993) Proteins regulating Ras and its relatives. Nature, 366, 643–653. 32. Ducy, P., Desbois, C., Boyce, B., Pinero, G., Story, B., Dunstan, C., Smith, E., Bonadio, J., Goldstein, S., Gundberg, C., Bradley, A. and Karsenty, G. (1996) Increased bone formation in osteocalcin-deficient mice. Nature, 382, 448–452. 34. Xu, T., Bianco, P., Fisher, L.W., Longenecker, G., Smith, E., Goldstein, S., Bonadio, J., Boskey, A., Heegaard, A.M., Sommer, B. et al. (1998) Targeted disruption of the biglycan gene leads to an osteoporosis-like phenotype in mice. Nature Genet., 20, 78–82. 35. Delany, A.M., Amling, M., Priemel, M., Howe, C., Baron, R. and Canalis, E. (2000) Osteopenia and decreased bone formation in osteonectindeficient mice. J. Clin. Invest., 105, 915–923. 36. Xiao, G., Wang, D., Benson, M.D., Karsenty, G. and Franceschi, R.T. (1998) Role of the α2-integrin in osteoblast-specific gene expression and activation of the Osf2 transcription factor. J. Biol. Chem., 273, 32988–32994. 37. Xiao, G., Jiang, D., Thomas, P., Benson, M.D., Guan, K., Karsenty, G. and Franceschi, R.T. (2000) MAPK pathways activate and phosphorylate the osteoblast-specific transcription factor, Cbfa1. J. Biol. Chem., 275, 4453–4459. 38. Price, L.S., Leng, J., Schwartz, M.A. and Bokoch, G.M. (1998) Activation of Rac and Cdc42 by integrins mediates cell spreading. Mol. Biol. Cell., 9, 1863–1871. 39. Toma, C.D., Ashkar, S., Gray, M.L., Schaffer, J.L. and Gerstenfeld, L.C. (1997) Signal transduction of mechanical stimuli is dependent on microfilament integrity: identification of osteopontin as a mechanically induced gene in osteoblasts. J. Bone Miner. Res., 12, 1626–1636. 40. Pavalko, F.M., Chen, N.X., Turner, C.H., Burr, D.B., Atkinson, S., Hsieh, Y.-F., Qui, J. and Duncan, R.L. (1998) Fluid shear-induced mechanical signaling in MC3T3-E1 osteoblasts requires cytoskeletal-integrin interactions. Am. J. Physiol., 275, C1591–C1601. 41. Tanaka-Kamioka, K., Kamioka, H., Ris, H. and Lim, S.-S. (1998) Osteocyte shape is dependent on actin filaments and osteocyte processes are unique actin-rich projections. J. Bone Miner. Res., 13, 1555–1568. 42. Walling, H.W., Chan, P.T., Omura, T.H., Barmina, O.Y. and Fiacco, G.J. (1998) Regulation of the collagenase-3 receptor and its role in intracellular ligand processing in rat osteoblastic cells. J. Cell. Phys., 177, 563–574. 43. Nobes, C.D. and Hall, A. (1995) Rho, rac and cdc42 GTPases: regulators of actin structures, cell adhesion and motility. Biochem. Soc. Trans, 23, 456–459. 44. Lamarche, N., Tapon, N., Stowers, L., Burbelo, P.D., Aspenstrom, P., Bridges, T., Chant, J. and Hall, A. (1996) Rac and Cdc42 induce actin polymerization and G1 cell cycle progression independently of p65PAK and the JNK/SAPK MAP kinase cascade. Cell, 87, 519–529. 45. Caron, E. and Hall, A. (1998) Identification of two distinct mechanisms of phagocytosis controlled by different Rho GTPases. Science, 282, 1717–1721.

496

Human Molecular Genetics, 2001, Vol. 10, No. 5