Interaction of the N- and C-terminal Autoregulatory Domains of FRL2 Does Not Inhibit FRL2 Activity*

Formin homology proteins are a highly conserved family of cytoskeletal remodeling proteins best known for their ability to induce the formation of long unbranched actin filaments. They accomplish this by nucleating the de novo polymerization of F-actin and also by acting as F-actin barbed end “leaky cappers” that allow filament elongation while antagonizing the function of capping proteins. More recently, it has been reported that the FH2 domains of FRL1 and mDia2 and the plant formin AFH1 are able to bind and bundle actin filaments via distinct mechanisms. We find that like FRL1, FRL2 and FRL3 are also able to bind and bundle actin filaments. In the case of FRL3, this activity is dependent upon a proximal DAD/WH2-like domain that is found C-terminal to the FH2 domain. In addition, we show that, like other Diaphanous-related formins, FRL3 activity is subject to autoregulation mediated by the interaction between its N-terminal DID and C-terminal DAD. In contrast, the DID and DAD of FRL2 also interact in vivo and in vitro but without inhibiting FRL2 activity. These data suggest that current models describing DID/DAD autoregulation via steric hindrance of FH2 activity must be revised. Finally, unlike other formins, we find that the FH2 and N-terminal dimerization domains of FRL2 and FRL3 are able to form hetero-oligomers.

Formin homology proteins (formins) are a highly conserved family of cytoskeletal regulatory proteins. Over 30 formins have been described to date, with more than 15 family members found in vertebrates (1). Formin activity is required in vivo for a diverse array of cellular functions, such as stress fiber formation, endosome motility, cell motility, cytokinetic ring formation, cell-cell junction assembly, filopodia formation, induction of cell polarity, and activation of the MAL/serum response factor signaling pathway (2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13). It is thought that at the core of all of these activities is the ability of formins to regulate actin cytoskeletal dynamics. This is achieved through the activity of two conserved formin homology domains, FH1 and FH2. FH1 consists of proline-rich repeats of varying sizes that can serve as ligands for Src homology 3 and WW domains as well as the small actin-binding protein profilin (14,15). FH2 domains form a dimer that induces the polymerization of long, unbranched filaments. It does this by nucleating de novo actin polymerization and by associating with the F-actin barbed end as a "leaky capper," allowing filament elongation while antagonizing the function of capping proteins (16). However, not all formins nucleate or elongate F-actin with equal potency. For example, mDia1 is an extremely efficient nucleator of polymerization, and its association with the barbed end of actin filaments accelerates elongation (16,17). In contrast, the isolated FH2 domain of FHOD1 does not induce actin polymerization in vivo (18). In addition, FH2 activity may also be influenced by the activity of adjacent domains. For example, in INF2, a WH2 G-actin binding motif antagonizes the nucleation activity of its associated FH2 domain (19). More recently, it has been recognized that the FH2 domains of FRL1, mDia2, and AFH1 are able to bind and bundle actin filaments using at least two distinct mechanisms (20,21).
The Diaphanous-related formins contain three dimerization interfaces: DID/DAD interactions mediate Diaphanous-related formin autoinhibition, whereas the N-terminal dimerization domain and the C-terminal FH2 each are able to form homodimers. The presence of multiple dimerization domains has called into question previous models that suggest that Diaphanous-related formins in the autoinhibited conformation are monomeric (15,(22)(23)(24). We reported previously that the dimerization and FH2 domains of mDia1 and mDia2 do not form hetero-oligomeric complexes. We wished to determine the general applicability of this observation to other formins. To this end, we performed an initial structure-function analysis of the highly similar Diaphanous-related formins FRL2 and FRL3 to determine the basic organization of these proteins and test their ability to form hetero-oligomers. As part of this study, we identified two C-terminal WH2/DAD-like motifs in FRL2 and FRL3. The more distal motif is required for FRL3 autoregulation, whereas the more proximal sequence is required for efficient F-actin bundling by both proteins. Moreover, we found that although FRL3 is autoregulated, the highly similar FRL2 is not. Indeed, we find that full-length FRL2 is constitutively active, although its DID is able to bind to its DAD both in vivo and in vitro. Finally, we find that, unlike mDia1 and mDia2, both the FH2 and N-terminal dimerization domains FIGURE 1. FRL3 is an autoregulated formin. Epitope-tagged derivatives of FRL3 corresponding to the full-length protein, N terminus, FH1 ϩ FH2, and FH2 were expressed by transient transfection in NIH 3T3 cells (0.3 g of DNA). Protein expression was detected with 9E10 anti-Myc monoclonal (red); F-actin was visualized with fluorescein phalloidin (green). A, full-length FRL3 is distributed diffusely throughout the cytoplasm of transfected cells (100% of cells) and fails to induce stress fiber formation (94%, n ϭ 105). B, FRL3N distributes throughout the cytoplasm and shows apparent membrane targeting in NIH 3T3 cells (99%). FRL3N-expressing cells have reduced levels of F-actin (97%, n ϭ 113). C, expression of FRL3.F1F2 induces robust stress fiber accumulation (93%) and dense accumulation of F-actin at the cell periphery that tightly associates with the overexpressed protein (89%, n ϭ 102). D, FRL3.FH2 induces stress fibers (73%), dense patches of F-actin in lamellopodia-like structures at the periphery of the cell and granular patches of actin throughout the cytoplasm (86%, n ϭ 109). E, expression of FRL3.F1F2 and FH2 (1 g of DNA transfected) induces robust activation of an SRF reporter gene. Reporter gene activity was standardized to activation induced by expression of an SRF-VP16 control fusion protein. Error bars, S.E., n ϭ 3. F, schematic of FRL3 derivatives. of FRL2 and FRL3 proteins are able to form hetero-oligomers in vivo.

MATERIALS AND METHODS
Plasmids-The constructs pMLV-LacZ, p3D.A-Luc, and pEF-SRFVP16 were described previously (25). The constructs f.l.   (11) with FLAG and Myc epitope tags, respectively, as well as pET-30a for expression in bacteria.
Cell Culture-NIH3T3 cells were cultured in Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 10% fetal bovine serum (Wisent) and antibiotics (50 units/ml penicillin and 50 g/ml streptomycin). Raw 264.7 cells were cultured in Dulbecco's modified Eagle's medium (Wisent) supplemented as above. Cells were maintained at 37°C in a humidified atmosphere with 10% CO 2 for NIH3T3 cells and 5% CO 2 for Raw 264.7 cells.
Immunofluorescence-Immunocytochemistry was performed as previously described (26,27). Briefly, NIH3T3 cells were transfected with polyethyleneimine (1 mg/ml; 4:1 ratio with plasmid DNA) and maintained in 0.5% fetal bovine serum in Dulbecco's modified Eagle's medium overnight. Raw 264.7 cells were transfected using Lipofectamine 2000 according to the supplied protocol. Transfected cells were fixed the next day in 3.7% para-formaldehyde and permeabilized with 0.3% Triton X-100. DNA was stained with 4Ј,6-diamidino-2phenylindole; F-actin was detected using fluorescein-phalloidin (catalog number F432; Molecular Probes) at a dilution of 1:40; FLAGtagged proteins were detected using mouse anti-FLAG antibodies (Sigma F3165) at a dilution of 1:500; and Myc-tagged proteins were detected using rabbit anti-c-Myc antibodies (catalog number sc-789; Santa Cruz Biotechnology, Inc., Santa Cruz, CA) at a dilution of 1:100. Secondary antibodies used were donkey Alexa594 anti-rabbit (Molecular Probes), donkey anti-rabbit Cy3 (Jackson ImmunoResearch) at a dilution of 1:100, and donkey anti-mouse Cy5 (Jackson ImmunoResearch) at a dilution of 1:100. All optical sections were captured with a ϫ63 oil immersion objective (numerical aperture ϭ 1.4) using the apotome on a Zeiss Axioimager.Z1. FIGURE 2. FRL2 is not autoregulated. Epitope-tagged derivatives of FRL2 corresponding to the full-length protein, N terminus, FH1 ϩ FH2, and FH2 were expressed by transient transfection in NIH 3T3 cells. Protein expression was detected with 9E10 anti-Myc monoclonal (red); F-actin was visualized with fluorescein phalloidin (green). A, full-length FRL2 accumulates at the cell periphery, strongly induces stress fiber formation, and associates with F-actin in transfected cells (97%, n ϭ 101). B, FRL2N is distributed diffusely throughout the cytoplasm (96%). FRL2N-expressing cells have reduced levels of F-actin (99%, n ϭ 103). C, expression of FRL2.F1F2 induces robust stress fiber accumulation (92%) and dense accumulation of F-actin at the cell periphery that tightly associates with the overexpressed protein (96%, n ϭ 105). D, FRL2.FH2 induces stress fibers, dense patches of F-actin in lamellopodia-like structures at the periphery of the cell, and granular patches of actin throughout the cytoplasm (98%, n ϭ 1-2). E, expression of full-length F1F2 and FH2 derivatives of FRL2 induces robust activation of an SRF reporter gene. Reporter gene activity was standardized to activation induced by expression of an SRF-VP16 control fusion protein. Error bars, S.E., n ϭ 3. F, schematic of FRL2 derivatives.
Immunoprecipitation-Co-immunoprecipitation of FLAGand Myc-tagged proteins expressed by transient transfection in NIH 3T3 cells were performed as previously described (26,27).
Reporter Gene Assay-The serum response factor (SRF) 2 reporter gene assays were performed as previously described (26,27). Activation was standardized to an SRF-VP16 internal control included in all experiments. For inhibition and dominant negative studies, activation in absence of the inhibitor is set to 100% to allow better comparison of the relative levels of inhibition.
Actin Bundling Assay-The in vitro actin bundling assays were performed as previously described (20, 26 -28). Briefly, FRL2 and FRL3 FH2 and N-terminal deletion derivatives were expressed in Escherichia coli as His tag fusion proteins and purified on Ni 2ϩ resin. F-actin was generated from pyrene-labeled rabbit muscle actin (Cytoskeleton Inc.) according to the supplier's instructions. For actin bundling experiments, FH2 derivatives (0, 0.15, 0.3, and 0.6 M) were incubated with 2 M F-actin for 10 min at room temperature. Samples were centrifuged at 16,000 ϫ g for 10 min at 4°C. The top 80% of supernatant was transferred to a new tube, precipitated with 5 volumes of acetone, and resuspended in 20 l of 1ϫ SDS gel loading buffer. The remaining supernatant was removed from the pellets, which were then resuspended in 25 l of 1ϫ SDS gel loading buffer. Equal volumes of supernatant and pellet samples were subjected to SDS-PAGE, and proteins were visualized by Coomassie staining. For the inhibition experiments, FRL2N was incubated with the FH2 derivatives for 10 min at room temperature prior to the addition of 2 M F-actin.
Pyrene Actin Polymerization Assay-The in vitro actin bundling assays were performed as previously described (26,27). Briefly, FRL2 FH2, FRL2ϩ3 FH2, and FRL2N were purified as above. FH2 (60 nM) derivatives were preincubated with the indicated concentrations of FRL2N or buffer for 10 min at room temperature and aliquoted into a black quartz 96-well plate. Polymerization was initiated by the addition of 5% pyrene-labeled actin for a final G-actin concentration of 2 M. Fluorescence (excitation ϭ 365 nm; emission ϭ 407 nm) was read every 15 s in a Spectromax M2 fluorimeter (Molecular Devices).

RESULTS
Previous work from our group has suggested that, despite extensive sequence homology, neither the FH2 domains nor N-terminal dimerization motifs of the closely related formins mDia1 and mDia2 are able to form heterodimers (26,27). To test the generality of this observation, we wanted to 2 The abbreviation used is: SRF, serum response factor. assess the ability of the related proteins FRL2 (also known as FMNL3b) and FRL3 (also known as FMNL2b) to form hetero-oligomeric complexes. These proteins share extensive homology at the protein level across their entire sequence (74% identical, 87% similar). We first generated epitopetagged derivatives of FRL2 and FRL3 comprising the fulllength protein, the isolated N terminus, and C-terminal derivatives containing FH1/FH2 (F1F2) and FH2 domains. These proteins were expressed by transient transfection in NIH 3T3 cells, and their subcellular localization and effects on actin stress fiber formation were assessed by immunofluorescence ( Figs. 1 and 2).
As expected, expression of full-length FRL3 did not induce F-actin accumulation, consistent with the presence of well conserved autoregulatory N-terminal DID and C-terminal DAD domains (Fig. 1A). The full-length protein exhibited a diffuse cytoplasmic distribution similar to that seen with full-length mDia1 (27). Expression of the FRL3 N terminus (FRL3N) inhibited stress fiber formation (Fig. 1B). The FH1/FH2 derivative of FRL3 (FRL3.F1F2) induced robust accumulation of stress fibers and a dense band of F-actin at the periphery of the cell. The protein was mostly distributed diffusely in the cytoplasm with some concentrated at the periphery and apparently co-localized with F-actin (Fig. 1C). FRL3.FH2 also strongly induced stress fiber formation and F-actin accumulation in lamellopodial-like projections at the periphery of the cell; the epitopetagged protein showed apparent tight association with the F-actin in these lamellae (Fig. 1D). The immunofluorescence results were confirmed using an SRF reporter gene assay (Fig. 1E). This reporter gene is activated by the actin/ MAL/SRF pathway in response to depletion of the cellular pool of G-actin and serves as a reliable quantitative measure of the activity of factors that induce actin polymerization (11-13, 25, 26, 29, 30). As expected, expression of FRL3.F1F2 or FRL3.FH2 strongly activated the reporter gene, whereas expression of full-length FRL3 and FRL3N had no effect.
Expression of N-terminal (FRL2N), FH1/FH2 (FRL2.F1F2), and FH2 (FRL2.FH2) deletion derivatives of FRL2 yielded similar results as the corresponding derivatives of FRL3 with the notable exception that expression of fulllength FRL2 also resulted in the robust accumulation of stress fibers in transfected cells (Fig. 2, A-D). These results were confirmed in the SRF reporter gene assay (Fig. 2E). This suggests that, despite the presence of highly conserved DID and DAD regulatory motifs, FRL2 is not subject to autoregulation.
To investigate the basis for the apparent constitutive activity of full-length FRL2, we took advantage of the ability of the DID/DAD autoregulatory interaction to be recapitulated in trans using isolated N-and C-terminal deletion derivatives. Notably, in the case of mDia1 and mDia2, the isolated N terminus of mDia1 is able to inhibit the isolated C terminus of mDia2 and vice versa. Using a similar strategy, we wanted to determine if the failure in FRL2 autoinhibition lies with the DID or DAD of FRL2 (Fig. 3). FRL3.F1F2 was co-expressed with FRL2N or FRL3N in NIH 3T3 cells, and the effects on actin polymerization were assessed by immunofluorescence. In both cases, expression of either FRL2N or FRL3N was sufficient to completely inhibit the ability of FRL3.F1F2 to induce stress fiber formation (Fig. 3, A and B). Consistent with this being a direct effect, FRL3.F1F2 was extensively co-localized with FRL2N and FRL3N. FRL3.F1F2-induced activation of SRF was also completely inhibited by co-expression of either FRL2N or FRL3N (Fig. 3E). In contrast, expression of either FRL2N or FRL3N had no effect on FRL2.F1F2 activity as assessed by immunofluorescence and the SRF reporter gene assay (Fig. 3, C-E). Thus, FRL2 contains a functional DID motif that is capable of inhibiting FRL3 activity in trans but does not contain a functional DAD.
Further examination of the FRL2 and FRL3 C-terminal sequences revealed the presence of two WH2/DAD-like motifs in both proteins (Fig. 4A). To identify which of these motifs is required for the inhibitory DID interaction, we generated a C-terminal deletion derivative of FRL3.F1F2 that removed the more distal DAD motif (FRL3.F1F2.1045). The ability of FRL3N to inhibit the activity of this C-terminal truncation mutant was assessed by immunofluorescence. FRL3N was unable to inhibit FRL3.F1F2.1045-induced stress fiber formation (Fig.  4, B and C). Similar results were obtained with these derivatives in the SRF reporter gene assay (Fig.  4E). Deletion of the distal DAD motif of FRL3 had only a moderate effect on activity as assessed by immunofluorescence or in the SRF reporter gene assay. Based on these results, we refer to the more distal motif as DAD and the more proximal motif as WH2 (see below).
Having mapped the DAD domain in FRL3, we then wanted to determine if it is the DAD domain and sequences C-terminal to it in FRL2 that are insensitive to DID inhibition or if it is a general property of the FH2 domain of FRL2. We constructed chimeric FRL2 derivatives containing the FRL3 DAD and sequences C-ter-   1 and 2 with lanes 3 and 4 and lanes 5-8 with lanes 9 and 10) 1 and 2 with lanes 3 and 4 and lanes 5-8 with lanes 9 and 10). NOVEMBER 28, 2008 • VOLUME 283 • NUMBER 48 minal to it (FRL2.F1F2.2ϩ3 and FRL2.FH2.2ϩ3) and tested the ability of FRL2N to inhibit their activity. As expected, FRL2N was unable to inhibit FRL2.F1F2 or FH2 activity (Fig.  5, A and D); nor did it inhibit the activity of FRL2 derivatives that lack the DAD domain (FRL2.F1F2.986 or FH2.986) (Fig.  5, B and D). However, unlike the other FRL2 derivatives, the chimeric protein was inhibited by co-expression of either FRL2N or FRL3N (Fig. 5, C and D). Similar results were obtained with chimeric versions of the full-length FRL2 and FRL3 proteins (supplemental Fig. 1) Thus, it is the FRL2 DAD or sequences C-terminal to it that fail to support DIDdependent inhibition of FH2 activity.

F-actin Binding by FRL2 and FRL3
The inability of FRL2 C-terminal derivatives to be inhibited by FRL2N or FRL3N implies two possible models; either the FRL2 DID and DAD domains are unable to form a stable complex, or they form a complex that is not competent to repress activation of the adjacent FH2 domain. To distinguish between these two models, we used co-immunoprecipitation assays to determine if the DID and DAD domains of FRL2 are able to interact in vivo. In these assays, we tested the ability of FLAG-tagged FRL2N or FRL3N to co-immunoprecipitate FRL2 and FRL3 C-terminal derivatives either containing or lacking the DAD motif. As expected, we found that FRL2N and FRL3N were able to efficiently co-immunoprecipitate FRL3.FH2 in a DAD-dependent manner (Fig. 6, A  and B). Unexpectedly, we found that FRL2N and FRL3N were also able to co-immunoprecipitate FRL2.F1F2 with a similar efficiency as with FRL3.FH2 or FRL2.F1F2.2ϩ3 (Fig.  6, A and B). The interaction between FRL2N and FRL2.F1F2 was also DAD-dependent (Fig. 6, A and B). Thus, the DID and DAD domains of FRL2 are able to interact in vivo, but in the case of FRL2.F1F2 or FRL2.FH2, this interaction is not inhibitory.
Having shown that the DID/DAD interaction mediates heterodimerization between the N and C termini of FRL2 and FRL3, we used similar co-immunoprecipitation assays to assess the ability of the isolated N termini, FH2, and fulllength proteins to form heterodimers. In contrast to our previous findings with mDia1 and mDia2, we found that we were able to efficiently co-immunoprecipitate hetero-oligomeric complexes of both the N-terminal dimerization domains and FH2 domains of FRL2 with FRL3 (Fig. 7, A and  B). Consistent with these results, we were also able to efficiently co-immunoprecipitate hetero-oligomeric complexes of full-length FRL2 with full-length FRL3 (Fig. 7C).
Both FRL2 and FRL3 contain two C-terminal WH2/DADlike motifs (Fig. 4A). Previous reports have implicated similar motifs in modifying FH2 activity (19). Therefore, we generated additional FH2 C-terminal truncation mutants with each of these motifs deleted in turn and assayed the effect on FH2 activity in vivo. Expression of FRL3.FH2 induces the formation of thick actin stress fibers and punctate actin aggregates and the accumulation of F-actin in lamellopodialike projections (Fig. 8A). Deletion of the distal DAD motif (FRL3.FH2.1045) had little effect on FH2 activity in this assay (Fig. 8B). Further deletion of both distal and proximal WH2/DAD-like motifs (FRL3.FH2.1023) caused an apparent loss of the punctate F-actin aggregates and reduction in peripheral F-actin accumulation (Fig. 8C). We also assessed the function of the FRL3.FH2 deletion derivatives using the SRF reporter gene assay, where a progressive loss of activity was observed as each WH2/DAD-like motif was deleted. The effect of removing these motifs from FRL2.FH2 was less pronounced, as assessed by immunofluorescence, and had no  1 and 2 with lanes 3 and 4 and lanes 5 and 6 with lanes 7 and  8). FLAG-tagged FRL2N efficiently co-immunoprecipitated FRL2N. FLAG beads alone served as a negative control (lanes 1 and 2 and lanes 5 and 6). B, FRL2.FH2 and FRL3.FH2 form hetero-oligomers in vivo. FLAG-tagged FRL3.FH2 or FRL2.FH2 was co-expressed in NIH 3T3 cells with Myc-tagged FRL2.FH2 and FRL3.FH2. FLAG-tagged FRL3.FH2 efficiently co-immunoprecipitated FRL2.FH2 and FRL3.FH2 (compare lanes 1 and 2 with lanes 3  and 4 and lanes 9 and 10). FLAG-tagged FRL2.FH2 also efficiently co-immunoprecipitated FRL2.FH2 and FRL3.FH2 (compare lanes 5 and 6 with  lanes 7 and 8 and lanes 11 and 12). FLAG beads alone served as a negative control (lanes 9 -12). Vertical lines separate different exposures of the same immunoblot (lanes 1, 2, 5, 6, and 9 -12 versus lanes 3, 4, 7, and 8). C, full-length FRL2 and FRL3 form hetero-oligomers in vivo. FLAG-tagged FRL3 or FRL2 was co-expressed in NIH 3T3 cells with Myc-tagged FRL2 and FRL3. FLAG-tagged FRL3 efficiently co-immunoprecipitated FRL3 and FRL2 (compare lanes 1 and 2 with lanes 3 and 4 and lanes 9 and 10). FLAGtagged FRL2 also efficiently co-immunoprecipitated FRL2 and FRL3 (compare lanes 5 and 6 with lanes 7 and 8 and lanes 11 and 12). FLAG beads alone served as a negative control (lanes 9 -12). effect on FRL2.FH2 activity in the SRF reporter gene assay (Fig. 9).
A notable effect of deleting the two WH2/DAD-like motifs of FRL3 was a decrease in FH2-induced punctate F-actin staining (Fig. 8C), which was also diminished in cells expressing the corresponding FRL2 construct (FRL2.FH2.965) (Fig. 9C). The related protein FRL1 has been shown to bind and bundle F-actin (20,28), and actin bundling by the actin-binding protein espin is dependent on its WH2 domain (31). The proximal WH2/DAD-like motif shares extensive homology with other WH2 domains (Fig.  10A). Therefore, we tested the ability of FRL2 and FRL3 FH2  deletion derivatives to bundle actin filaments in vitro. We found that purified FRL2.FH2 and FRL3.FH2 proteins are able to bundle actin filaments in vitro (Fig. 10, B and D). In the case of FRL3, removal of the distal DAD reduces actin bundling (Fig. 10C), and further truncation to remove the proximal WH2/DAD-like motif completely eliminates actin bundling by this protein (Fig. 10D). As we observed in vivo, the effects of deletion of the WH2/DAD-like motifs on FRL2.FH2 activity were less pronounced. In this case, deletion of the distal DAD had little effect on F-actin bundling (Fig. 10E), although further deletion of the more proximal motif reduced, but did not eliminate, F-actin bundling (Fig.  10F). The chimeric construct FRL2.FH2.2ϩ3 was also able to efficiently bundle F-actin in this assay. Thus, the proximal WH2/DAD-like motif is playing a role in F-actin bundling rather than autoregulation, and therefore we designate the proximal motif as WH2.
We used the F-actin bundling assay to investigate directly the inhibitory interaction between the N-and C-terminal regulatory domains of FRL2 and FRL3. FRL2N protein on its own did not induce F-actin bundling in vitro and did not pellet (Fig.  11A). Incubation of FRL2N with FRL3.FH2 or FRL2.FH2.2ϩ3 greatly inhibited the ability of either FH2 domain to bundle F-actin (Fig. 11, C and D). In contrast, incubation with FRL2N had essentially no effect on F-actin bundling by FRL2.FH2. However, although there was no inhibition of FRL2.FH2-induced actin bundling, FRL2N was still recruited to the actin pellet by FRL2.FH2 (Fig. 11B).
These results were confirmed using the pyrene-actin polymerization assay (Fig. 12). In this assay, both FRL2.FH2 and FRL2. FH2.2ϩ3 are able to induce actin polymerization. A 3-fold molar excess of FRL2N has no effect on FRL2 FH2-induced actin polymerization ( Fig. 12A) but is able to completely inhibit FRL2.FH2.2ϩ3 activity (Fig. 12B). Thus, as was the case in vivo, FRL2.FH2 binds directly to FR2N but in a noninhibitory complex.

DISCUSSION
We report here an initial structure-function analysis of the Diaphanous-related formins FRL2 and FRL3. We find that the C-terminal regions of both FRL2 and FRL3 contain two WH2/ DAD-like motifs. The more proximal motif we have designated as "WH2" based on homology and because of the role it plays in F-actin bundling. The more distal motif we have designated "DAD," since it is required for autoregulation of FRL3. We also show that these proteins are able to form hetero-oligomers through the dimerization of either their FH2 or N-terminal domains.
F-actin Bundling-We found that both FRL2 and FRL3 are able to bind and to bundle F-actin in vitro. This activity is apparently dependent upon a WH2-like motif found immediately C-terminal to FH2 and proximal to the regulatory DAD motif. It has been noted previously that WH2 and DAD domains are quite similar at the sequence level. Indeed, the INF2 DAD is also a G-actin-binding WH2 domain that antagonizes the activity of the INF2 FH2 domain (19). The proximal WH2/DAD sequence in both FRL2 and FRL3  NOVEMBER 28, 2008 • VOLUME 283 • NUMBER 48 shares extensive homology with the WH2 motif found in INF2, N-WASP, and other proteins (Fig. 10A). Functionally, this sequence also exhibits more WH2-like behavior in that it is absolutely required for F-actin bundling by FRL3 and is required for efficient F-actin bundling by FRL2. This result is reminiscent of the ability of espin to bind and bundle F-actin in a WH2-dependent fashion (31). In this case, F-actin bundling is dependent upon G-actin binding by WH2 and the presence of two additional F-actin binding sites in espin (31). By analogy, our results would suggest a model where F-actin bundling by FRL3 is dependent on barbed end binding by the FH2 domain and an additional actin binding surface on WH2. Despite the overall homology between FRL2 and FRL3, the WH2 independence of F-actin bundling by FRL2 suggests that the two proteins are bundling actin filaments by distinct mechanisms. We favor a model where F-actin bundling by FRL2 is likely to be similar to FRL1 and dependent on dissociation of the FH2 dimer (20,28), whereas the WH2-dependent F-actin bundling by FRL3 is more similar to F-actin bundling by espins (31). Further experimentation is required to determine if the WH2 motif found in FRL2 and FRL3 is itself sufficient to confer F-actin binding when incorporated into a heterologous protein. A more complete analysis of how the C-terminal WH2 and DAD motifs of these proteins affect FH2-induced actin polymerization will be reported elsewhere. 3 Autoregulation-Our results show that full-length FRL2 is constitutively active when expressed by transient transfection in NIH 3T3 cells. This could result either from an inability of the DID and DAD domains to interact or a failure of this interaction to inhibit FH2 activity. Both in vivo and in vitro we were able to demonstrate that the DID of FRL2 is able to bind to DAD, but this association does not inhibit FH2 activity (Figs. 4, 6, 11B, and 12A). The inability of this interaction to inhibit FH2 activity involves a deficit within the C-terminal regulatory domain despite the presence of two DAD-like motifs in this region. Numerous lines of evidence suggest that this result is not an artifact of our system; nor is it an inherent property of the FH2 domain of FRL2. First, the N terminus of FRL2 is able to inhibit the activity of the FRL3 C terminus, demonstrating that FRL2 contains a functional DID (Fig. 3). Second, the C-terminal derivatives of FRL2 that contain the DAD of FRL3 are inhibited by either FRL2N or FRL3N in trans (Figs. 4, 6, 11B, and 12A). Third, a chimeric full-length FRL2 protein containing the DAD of FRL3 is autoinhibited (supplemental Fig. 1, E and G). Therefore, FRL2 FH2 activity can be inhibited through a DID/DAD interaction. Fourth, replacement of the DAD of FRL3 with that from FRL2 partially relieves FRL3 autoinhibition, although not to the same extent as deletion of the FRL3 DAD (supplemental Fig. 1, B, C, and F). The observation that, in the context of full-length FRL3, the DAD of FRL2 is able to support some level of autoinhibition suggests that the constitutive activity of FRL2 in vivo is not due to the absence of some unknown regulatory factor in NIH 3T3 cells. Instead, we favor the idea that the constitutive activity of FRL2 results from a combination of a suboptimal DAD sequence and an FH2 domain that is intrinsically more refractive to DID/DAD inhibition (see below). FHOD1, like FRL2 and FRL3, also possesses two C-terminal DAD-like motifs, with the more distal DAD mediating FHOD1 autoregulation (32). For FHOD1 and for mDia2, efficient DID/DAD dimerization and autoinhibition is dependent upon a cluster of basic residues C-terminal to DAD (32,33). Examination of the FRL2 and FRL3 primary sequences reveals that the number of basic residues varies significantly between the various alternative splice forms of each of these proteins (34). The FRL2b isoform used in this study has the least (3 arginines), whereas the FRL3b isoform used here has the most (9 arginines) (Fig. 4A). However, we found that the C terminus of FRL2 bound strongly to FRL2N in a DAD-dependent manner both in vivo and in vitro (Figs. 6 and 11). Therefore, it seems that in this case, the basic residues are not required so much for the DID/DAD interaction as for allowing this interaction to inhibit FH2 activity.