* The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Mammalian and fungalDiaphanous-related formin homology (DRF) proteins contain several regions of conserved sequence homology. These include an amino-terminal GTPasebinding domain (GBD) that interacts with activated Rho family members and formin homology domains that mediate targeting or interactions with signaling kinases and actin-binding proteins. DRFs also contain a conservedDia-autoregulatory domain (DAD) in their carboxyl termini that binds the GBD. The GBD is a bifunctional autoinhibitory domain that is regulated by activated Rho. Expression of the isolated DAD in cells causes actin fiber formation and stimulates serum response factor-regulated gene expression. Inhibitor experiments show that the effects of exogenous DAD expression are dependent upon cellular Dia proteins. Alanine substitution of DAD consensus residues that disrupt GBD binding also eliminate DAD biological activity. Thus, DAD expression activates nuclear signaling and actin remodeling by mimicking activated Rho and unlatching the autoinhibited state of the cellular complement of Dia proteins.
Diaphanous-related formin homology
GTPase binding domain
Wiskott-Aldrich syndrome protein
verprolin cofilin acidic
green fluorescent protein
serum response factor. CRIB, Cdc42-Rac interactive binding region
Formin homology (FH)1 proteins and Rho small GTPases modulate cytoskeletal remodeling during cytokinesis, polarized cell growth, and development (
). This study identifies a new homology domain unique to the Diaphanous-related FH family members termed the Dia-autoregulatory domain or DAD.
The DRFs bridge signaling and cell remodeling pathways by binding to several signaling kinases and scaffolding proteins via SH3 domain interactions with the proline-rich FH1 domain. These include the Src nonreceptor-tyrosine kinase family (
). Expression of ΔGBD-mDia1 and -mDia2 in fibroblasts activates a Rho-regulated signaling pathway that leads to the activation of theserum response factor (SRF) transcriptional regulator in the nucleus (
). It binds to the carboxyl-terminal verprolin,cofilin acidic (VCA) region located in the carboxyl terminus. Upon binding to Cdc42, the CRIB domain releases the carboxyl terminus, and the exposed VCA region of WASP functions as an effector by direct binding to the Arp2/3 actin-nucleating complex (
). The amino terminus of mDia1 has been shown to interact with the carboxyl terminus in a manner that is disrupted by activated Rho binding to the GBD. Here, a conserved carboxyl-terminal autoregulatory domain, termed DAD, that facilitates intramolecular binding is identified. When exogenously expressed in cells, DAD is a potent activator of the cellular complement of DRFs and causes the induction of both cytoskeletal remodeling and the SRF signaling.
Cell Culture, Microinjection, and Fluorescence Microscopy
NIH 3T3 cells grown on glass coverslips were maintained in Dulbecco's modified essential medium (Life Technologies, Inc.) containing 10% (v/v) fetal calf serum (Life Technologies, Inc.) until 24 h prior to microinjection when cells were changed to medium containing 0.1% (v/v) FCS. The NIH 3T3 SRE-FosHA reporter cell line HA13 was used in all SRE gene expression studies (
). Purified plasmid DNA expression vectors were microinjected at a concentration of 10 μg/ml each in a buffer of phosphate-buffered saline/dH20 (1:1), unless indicated otherwise. Empty vector (pEFm) was included in experiments to normalize injected DNA concentrations when necessary. For fluorescent detection experiments, cells were fixed 3 h after microinjection with 3.7% formaldehyde in phosphate-buffered saline and permeabilized with 0.3% Triton X-100 (Sigma) prior to staining. SRE-regulated FosHA staining was detected by indirect immunofluorescence using primary rabbit anti-HA antisera (Y-11, Santa Cruz Biotechnology) followed by aminomethyl coumarin acetate (AMCA)-coupled anti-rabbit (Jackson). Filamentous actin was monitored in cells by staining with TRITC-labeled phalloidin (Sigma). After staining, coverslips were mounted in gelvatol. Fluorescent images were captured with a digital camera (SPOT R100, Diagnostics) mounted on a Nikon E400 epifluorescence microscope using fixed exposure times with either × 40 or × 100 magnification (1.4 NA) where indicated. Images were saved as TIFF files and assembled into figures using Clarisdraw.
Plasmids and GenBankTM/EBI Accession Numbers
mDia1, mDia2, and various domain expression constructs were made in pEFm (courtesy of R. Marais), pEFHA, pEFmEGFP, pT7-plink, pGEX-KG, and pGAD10 from polymerase chain reaction products using standard methods and confirmed by direct sequencing; complete details are available upon request. In vitro translation plasmids were made using pT7-plink (
). GenBankTM/EBI accession numbers for the gene products discussed are mDia1, U96963; mDia2, AF094519; DIA156, NP006720; Diaphanous, AAA67715; Bni1, P41832; and SepA, AAB63335. For most of the experiments, plasmids encoded the following amino acids for mDia2: GBD, 101–216; FH1, 521–630; FH2, 801–910; and DAD, 1031–1171 unless otherwise indicated.
In Vitro GST Pull-down and Two-hybrid Assays
Two-hybrid assays and in vitro translation/GST pull-down assays were conducted as described previously (
). In short, the indicated bait proteins were generated by subcloning the indicated cDNAs into pGBT9 Gal4 DNA binding domain plasmid; prey were Gal4 activation domain fusion proteins generated in either pGAD10 or pSE1107. HF7c (CLONTECH) reporter yeast strain was cotransformed with the indicated plasmids and selected on appropriate plates for bait and prey auxotrophic markers and then restreaked onto His-plate to select for Gal4-regulated His reporter expression. Levels of reporter activity were monitored by replicate streaking onto Trp/Leu/His-plates with increasing concentrations (0–64 mm) of 3-aminotriazole.
For in vitro translation and pull-down assays, plasmids were constructed using pT7-plink or pCAN (
), GBD, or FH1 and were in vitro translated using the TNT kit (Promega) using35S-labeled methionine. 4 μl of labeled in vitro translation product were incubated with 10 μg of GST, GST·DAD, or GST·SrcSH3, bound to glutathione beads as indicated. Beads were incubated at 4 °C with rocking for 2 h in 25 mm Tris pH 7.2, 100 mm NaCl, and 10 mm MgCl2. Binding reactions were warmed to 30 °C before addition of recombinant RhoA-V14 (
). Beads were collected by centrifugation and after washing beads three times in 5 bed volumes binding buffer, beads were suspended in SDS sample buffer, electrophoresed, and then stained with Coomassie Blue R250 and dried prior to autoradiography.
Identification of DAD
Amino acid sequence alignments of the growing FH protein superfamily have delineated the proline-rich FH1 and FH2 regions of homology (
). Comparative alignment of only the Dia-related subfamily yields a conserved domain in the carboxyl termini that is shown in Fig. 1. The consensus sequence (G/A)(V/A)MDXLLEXL(K/R/Q)X(G/A)(S/G/A)(A/P) was designated the Dia-autoregulatory domain or DAD. There is also a conserved region of basic residues several residues toward the carboxyl termini. The DAD domain was found in all the three mouse/human DRFs2 (
). To determine whether DAD could mediate this interaction in mDia2, the DAD domain was tested for binding to the GBD in both two hybrid and GST pull-down assays as shown in Fig. 2,A and B, respectively. In vitro both mDia2 and the isolated GBD bound specifically to the DAD domain (data not shown). The DAD·GBD or DAD interaction with mDia2 was tested for regulation by activated RhoA-V14. Both in vitro translated [35S]methionine-labeled GBD and mDia2 were incubated with increasing concentrations of purified recombinant RhoA-V14 protein.2 Activated Rho disrupted the association of both the isolated mDia2 GBD and mDia2 itself as shown in Fig. 2Bbut did not have any effect on FH1-SrcSH3 binding.
A model based on these observations is shown in Fig. 2Cafter Watanabe et al. (
) In cells with low levels of activated Rho·GTP, Dia protein assumes an inactive state with the carboxyl-terminal DAD directly interacting with the amino-terminal GBD. Small GTPase activation and GBD binding induces release of the DAD and then effectors are recruited through the FH1 and FH2 domains. The carboxyl-terminal DAD, like the VCA domain of WASP, could be a bifunctional regulator. In this event, DAD would not only associate with the GBD but also recruit effectors such as the ARP2/3 complex in a manner analogous to the WASP-VCA domain (
). If DAD lacks effector activity, it is possible that expression of this isolated domain in cells might disrupt the intramolecular GBD-DAD association. In this case, DAD expression might activate endogenous DRFs in cells. These two possibilities were then examined.
DAD Expression Activates Actin Remodeling and SRF
The mDia2 DAD domain was fused to green fluorescent protein (EGFP) in a mammalian expression plasmid (pEFmEGFP-DAD (
)) and was microinjected into NIH 3T3 cells previously maintained for 24 h in 0.1% serum, which diminishes the appearance of actin fibers. 3 h after injection, the effects on actin reorganization and activation of SRF were assayed. Actin polymerization was observed by fixing and staining cells with fluorescent TRITC-phalloidin, and SRF-regulated gene expression was monitored by indirect immunofluorescence by staining HA13 cells for the induction of a stably transformed SRE-controlled c-fos reporter gene that contains an HA-epitope tag (
). The effects of EGFP·DAD were compared with similar EGFP fusion proteins containing other mDia2 domains, including the GBD, FH1, and FH2 sequences, in addition to the previously described activated ΔGBD variants (
). Whereas none of the other homology domains had an effect on actin or SRF activity, EGFP·DAD expression strongly induced the formation of actin filaments in cells as shown in Fig. 3A.
The localization of the fusion proteins differed greatly. The EGFP·FH1 fusion was predominantly nuclear with some diffuse cytoplasmic localization. However, upon stimulation with LPA, the EGFP·FH1 fusion began to decorate actin stress fibers as shown in the example in Fig. 3B. The inset shows a region from the EGFP·FH1-expressing cell in the top row; merged images clearly show overlapping EGFP·FH1/stress fibers which appearyellow. EGFP·FH2 localization was consistently diffuse throughout the cell, whereas EGFP·DAD was excluded from the nucleus. DAD did appear to concentrate at the ends of a subset of actin fibers in cells expressing higher levels of DAD but not to the extent seen with EGFP·FH1 (data not shown).
EGFP·DAD strongly induced SRF as summarized in Fig. 3C, where bars represent the number of FosHA-positive GFP fusion-expressing cells. This result is consistent with previous results showing a role for the DRFs in an SRF-signaling pathway (
). The model (Fig. 2C) predicts that overexpression of the GBD would squelch the effects of DAD domain expression through direct binding. To test this, plasmids encoding both GBD (pEFm-GBD) and DAD (pEFmEGFP-DAD) were microinjected into cells, and actin and SRF activity were monitored as before. As shown in Fig. 4A, GBD expression blocked DAD-induced actin remodeling. Coinjection of either FH1 or FH2 domains were without effect. Similar results were obtained when SRF activity was assayed (Fig. 4B). These results showed that the GBD expression inhibited the DAD domain intrans and was, in effect, squelching the DAD domain.
The GBD could also block SRF and actin remodeling if DAD was activating Rho. Expression of other Rho GTPase binding domains have been shown to block Rho signaling (
). DAD activation of SRF via Rho was tested despite the presumption that the DRFs were downstream effectors of Rho signaling. DAD was coexpressed with either the amino terminus of PKN (PKN.N) or C3 transferase, which also inhibits Rho signaling but not the activated DRFs (
). Neither C3 or PKN.N inhibited DAD activation of SRF (Fig. 4B) though each blocked serum activation of SRF as previously reported (data not shown). Because of this specific blockade of DAD, it was concluded that DAD was triggering signals downstream of the Rho GTPases to activate SRF. These experiments also show that DAD is biologically active and is capable of specifically binding the GBD in cells. The molecular nature of DAD activity in cells was then examined.
DAD Expression Unlatches the GBD·DAD Autoregulatory Mechanism
If DAD was an effector domain analogous to the WASP-VCA domain (
), GBD squelching of DAD activity would be predicted. The GBD would also be expected to block DAD if the alternative occurred; DAD inhibits a negative intramolecular DAD-GBD association. In this case, DAD domain expression would be activating endogenous Dia proteins by unlatching the GBD·DAD autoregulatory mechanism. This was tested by expressing DAD and simultaneously inhibiting endogenous mDia1 by coinjection of affinity-purified anti-mDia1 with the EGFP·DAD expression vector (
). NIH 3T3 cells express mDia1 but not mDia2; anti-mDia1 antibody recognizes mDia1 amino acids 66–77 (YGDDPTAQSLQD). Anti-mDia1 effectively blocked DAD activity toward actin remodeling (data not shown) and SRF (Fig. 5,A and B). Anti-mDia1 did not inhibit ΔGBD-mDia2 or ΔGBD-mDia1 SRF induction because the deletion constructs lack the peptide sequence necessary for recognition by anti-mDia1 (
). Both reagents effectively blocked DAD-induced stress fiber formation and SRF (Fig. 5B).
To correlate DAD-DRF binding with DAD biological activity, a series of alanine substitutions were introduced into the DAD consensus region and were tested for binding to mDia2 by two-hybrid analysis. After determining the effects on DAD-GBD association, the resultant DAD domain mutants were then expressed in cells. The results of the binding experiments performed by two-hybrid analysis are summarized in Fig.6A. Mutations that disrupted the mDia2-DAD interaction are indicated by the filled arrowheads; mutations that had no effect on mDia2 interaction are shown by the open arrowheads. M1041A, L1044A, L1048A (data not shown) (amino acid positions are from the mDia2 peptide sequence; conserved residues are indicated by black boxes, similar residues by grey boxes) substitutions of conserved DAD residues disrupted binding and were inactive toward the activation of both SRF (data not shown) and actin fiber formation as shown in Fig.6B. The L1044A localization was also significantly altered. Instead of the diffuse membrane localization seen with wild-type DAD, L1044A had a punctate localization in the membrane and appeared at the cell edge. The L1044A substitution allowed the fusion protein to appear in the nucleus and also caused the formation of lamellipod-like extensions. Alanine substitutions of the nonconserved residues S1043A and Q1049A had no effect on binding, and both were able to induce fiber formation and SRF (data not shown). Interestingly, the E1046A substitution of the conserved glutamate residue was also inactive, suggesting that DAD·mDia2 binding is largely mediated through hydrophobic interactions. The interaction is also dependent upon the stretch of basic residues adjacent to the core DAD sequence. The introduction of a stop codon at position 1050 weakened but did not eliminate the interaction as determined by two-hybrid assay and also reduced the number and density of apparent fibers in cells (data not shown). The laboratory is in the process of examining the importance of these residues and the variable distance from the DAD core in its biological activity. Taken together, the results of these DAD mutational experiments show that complete DAD activity is dependent upon its ability to bind to a full-length DRF.
The current study defines a regulatory domain that accounts for Rho GTPase activation of the Diaphanous-related FH protein subfamily. The DAD·GBD autoregulatory mechanism is analogous to the interaction of the WASP GTPase binding and the VCA domains (
); the DRFs are regulated by intramolecular GBD·DAD binding where Rho-GTP activates the DRFs by disrupting GBD-DAD interaction. DAD is highly conserved, and its identification further explains the nature of several prior observations where truncation of Bni1, mDia1, and mDia2 GTPase binding domains resulted in their activation (
). Here, it is shown that ectopic DAD expression mimics Rho·GTP binding by interfering with normal autoregulation of cellular DRFs. Based on this model, it may be possible that overexpression of the GTPase binding domain would also unlatch the autoinhibited Dia proteins. This is not the case as DAD clearly activates both actin fiber production and SRF, whereas the GBD was inactive and inhibits DAD in coexpression experiments. These results suggest the GBD is truly a bifunctional autoinhibitory domain in addition to the GTPase binding domain. Even if GBD binding to the carboxyl terminus of the endogenous Dia protein in transunlatched intramolecular autoinhibition, it would likely interfere with endogenous Dia function by masking exposed effector domains contained in the carboxyl terminus or by inducing the inactive state of the Dia proteins through direct binding. This may be reflected in results obtained by Nakano et al. (
) who have shown that expression of a portion the mDia1 amino terminus interfered with the integrity of the actin fiber network in MDCK cells. It is critical, however, to determine whether these types of interfering proteins block by binding to functional domains of endogenous Dia proteins or simply squelch activated Rho GTPases by binding to a cryptic GTPase binding domain found within these Dia regions. Unfortunately, there is currently little information regarding the specific structural requirements of the DRF GTPase binding domains. This interaction is likely complex, and Rho effector loop mutations suggest that the interaction of Rho with mDia2 is unique (
The carboxyl terminus of the DRFs, including regions between DAD and the FH2 domain, may still serve in a signaling or actin remodeling capacity like the WASP carboxyl terminus. DAD itself may also contain intrinsic effector activity. The L1044A-substituted mDia2 DAD, for example, was unable to bind the mDia2 GBD and therefore unlatch the autoinhibited cellular Dia proteins. Expression of this mutated DAD was still able to effect the actin cytoskeleton by causing the disruption of pre-existing actin fibers and the formation of lamella-like extensions. This suggests that either this mutant directly targets cellular components that participate in actin remodeling or it interferes with endogenous DRFs. Both possibilities are being explored as the complete induction of stress fibers by DAD will likely be shown to be dependent upon multiple activities being recruited to activated Rho- or DAD-bound Dia proteins. The current results also raise the possibility that expression of dual function autoregulatory domains from proteins like WASP may have multiple effects in cells; those that are dependent upon the recruitment of cellular factors to the domain of interest, but also those caused by the relief of self-inhibition.
Once the DRFs are activated by Rho binding, what is the mechanism of signal transduction to effectors? The current model suggests that once intramolecular inhibition because of GBD-DAD association is relieved by binding to activated Rho, the DRFs then recruit downstream effectors that mediate signals via the proline-rich FH1, the FH2, or other uncharacterized Dia domains. Whereas the integrity of the FH1 domain is required for the ability of ΔGBD-versions of both mDia1 and mDia2 to signal to actin and DRF (
). Src-dependent targeting may therefore be a permissive event. This is consistant with the conclusion of this study; ectopic DAD expression triggers endogenous Dia protein in a manner similar to activated Rho. In this situation, activated cellular Dia proteins are then directed to the membrane by factors in a FH1 domain-dependent manner. v-Src has been shown to be directed to focal adhesions in a Rho- and actin-dependent manner (
) that block activated Dia function, Src and Dia targeting may be mutually dependent events that are controlled by the rate-limiting Rho-activation step.
The DAD peptide represents a useful biological tool to study Rho signaling and cytoskeletal regulation pathways. Because DAD effects are largely dependent upon the endogenous complement of cellular DRFs, the results generated from further experiments using DAD will likely be more useful in the dissection of Dia-dependent signaling events. For example, in comparison to the activated ΔGBD-mDia variants, DAD expression induces the formation of stable microtubules whose orientation better reflects those observed after growth factor treatment.
) on DAD activity in cells and on its ability to interact with the GBD. The autosomal dominantDFNA1 mutation in the human Dia1 gene occurs at a donor splice site at the carboxyl terminus. The mutation causes a frameshift 15 amino acids away from the core DAD sequence, which introduces an anomalous 19 amino acids onto the protein. Thus far in our laboratory, the effects of truncations near this region of DAD have been equivocal, but this may suggest that it destabilizes the autoinhibitory GBD-DAD association. Expression of truncated versions of mDia1 by Watanabe et al. (
) appears to cause disruption of the actin cytoskeleton in some cells and resembles the results seen with DAD mutants shown here. However, the effects of the DFNA1 mutation on Dia1 protein function will not be clear until these Dia1 mutants are expressed in cells at in vivo levels.
The author thanks Gregg Gundersen, Andrew Thorburn, Michael Weinreich, and Nick Duesbery for discussions and/or comments on the manuscript and Martin Broome and Sara Courtneidge (Sugen, San Francisco, CA) for providing the anti-Src reagents.