ASB2α, an E3 Ubiquitin Ligase Specificity Subunit, Regulates Cell Spreading and Triggers Proteasomal Degradation of Filamins by Targeting the Filamin Calponin Homology 1 Domain*

Background: The E3 ubiquitin ligase ASB2α triggers proteasomal degradation of filamins and regulates cell spreading and differentiation. Results: ASB2α targets filamin's calponin homology 1 (CH1) domain, and mutations in CH1 render filamin ASB2α-resistant. Conclusion: ASB2α-resistant filamins protect cells from ASB2α-mediated inhibition of spreading. Significance: ASB2α exerts its effect on cell spreading via targeted degradation of filamins. Filamins are actin-binding and cross-linking proteins that organize the actin cytoskeleton and anchor transmembrane proteins to the cytoskeleton and scaffold signaling pathways. During hematopoietic cell differentiation, transient expression of ASB2α, the specificity subunit of an E3-ubiquitin ligase complex, triggers acute proteasomal degradation of filamins. This led to the proposal that ASB2α regulates hematopoietic cell differentiation by modulating cell adhesion, spreading, and actin remodeling through targeted degradation of filamins. Here, we show that the calponin homology domain 1 (CH1), within the filamin A (FLNa) actin-binding domain, is the minimal fragment sufficient for ASB2α-mediated degradation. Combining an in-depth flow cytometry analysis with mutagenesis of lysine residues within CH1, we find that arginine substitution at each of a cluster of three lysines (Lys-42, Lys-43, and Lys-135) renders FLNa resistant to ASB2α-mediated degradation without altering ASB2α binding. These lysines lie within previously predicted actin-binding sites, and the ASB2α-resistant filamin mutant is defective in targeting to F-actin-rich structures in cells. However, by swapping CH1 with that of α-actinin1, which is resistant to ASB2α-mediated degradation, we generated an ASB2α-resistant chimeric FLNa with normal subcellular localization. Notably, this chimera fully rescues the impaired cell spreading induced by ASB2α expression. Our data therefore reveal ubiquitin acceptor sites in FLNa and establish that ASB2α-mediated effects on cell spreading are due to loss of filamins.

Filamins are actin-binding and cross-linking proteins that organize the actin cytoskeleton and anchor transmembrane proteins to the cytoskeleton and scaffold signaling pathways. During hematopoietic cell differentiation, transient expression of ASB2␣, the specificity subunit of an E3-ubiquitin ligase complex, triggers acute proteasomal degradation of filamins. This led to the proposal that ASB2␣ regulates hematopoietic cell differentiation by modulating cell adhesion, spreading, and actin remodeling through targeted degradation of filamins. Here, we show that the calponin homology domain 1 (CH1), within the filamin A (FLNa) actin-binding domain, is the minimal fragment sufficient for ASB2␣-mediated degradation. Combining an in-depth flow cytometry analysis with mutagenesis of lysine residues within CH1, we find that arginine substitution at each of a cluster of three lysines (Lys-42, Lys-43, and Lys-135) renders FLNa resistant to ASB2␣-mediated degradation without altering ASB2␣ binding. These lysines lie within previously predicted actin-binding sites, and the ASB2␣-resistant filamin mutant is defective in targeting to F-actin-rich structures in cells. However, by swapping CH1 with that of ␣-actinin1, which is resistant to ASB2␣-mediated degradation, we generated an ASB2␣-resistant chimeric FLNa with normal subcellular localization. Notably, this chimera fully rescues the impaired cell spreading induced by ASB2␣ expression. Our data therefore reveal ubiquitin acceptor sites in FLNa and establish that ASB2␣-mediated effects on cell spreading are due to loss of filamins.
ASB2␣ 5 (named for ankyrin repeat-containing protein with a suppressor of cytokine signaling box-2) was initially identified as a retinoic acid-response gene in myeloid leukemia cells, where it was shown to induce growth inhibition and chromatin condensation, events critical to retinoic acid-induced differentiation (1,2). ASB2␣ is the specificity subunit of an E3 ubiquitin ligase complex (3,4) and contains an N-terminal region followed by 15 predicted ankyrin repeats and a C-terminal suppressor of cytokine signaling (SOCS) box (Fig. 1A). The SOCS box mediates interaction with a cullin family member (Cullin 5) and RING finger proteins (Rbx1/2) by interacting with elongin BC to form an E3 ubiquitin ligase complex (3,4). ASB2␣ was therefore proposed to regulate myeloid cell proliferation and differentiation by targeting a specific regulatory protein, or proteins, for degradation (1,3).
We have shown that ASB2␣ expression triggers rapid proteasomal degradation of filamins in various cell types (5)(6)(7). Furthermore, retinoic acid-induced expression of ASB2␣ correlates with filamin down-regulation in myeloid leukemia cells induced to differentiate (5), and knock down of endogenous ASB2␣ delays filamin degradation and differentiation (5). Loss of filamins may therefore play a key role in hematopoietic cell differentiation.
Filamins are homodimeric actin-binding and cross-linking proteins that operate in a diverse range of cellular functions, including cell motility, maintenance of cell shape, differentiation, transcriptional regulation, and mechano-transduction (8). The mammalian filamin family consists of three highly homologous isoforms, FLNa, FLNb, and FLNc, of which FLNa is the most abundant and widely distributed (9). Each filamin subunit is composed of an N-terminal actin-binding domain (ABD), followed by 24 Ig-like repeats (IgFLN1-24) (Fig. 1A). The Iglike repeats are interrupted by two hinge regions, H1 and H2, and mediate interaction with the majority of filamin-binding partners (10 -12). Filamins dimerize via IgFLN24 and form a flexible V-shaped structure that regulates organization of the actin cytoskeleton, anchors transmembrane proteins, and scaffolds a wide range of signaling molecules (9,11,13,14).
In addition to a possible role in hematopoietic cell differentiation, filamins have also been implicated in differentiation of other cell types as follows: filamin down-regulation correlates with myoblast differentiation (15) and FLNa knockdown in human mesenchymal stem cells significantly increases chondrocyte differentiation (16). In keeping with their important roles in cytoskeletal organization, signaling and differentiation filamins are essential to mammalian development (11,12). Mutations in the most widely expressed human filamin, FLNa, result in a neuronal migration disorder and cardiovascular defects as well as a broad range of other congenital malformations, affecting craniofacial structures, skeleton, viscera, and urogenital tract (17)(18)(19)(20). Mutations in FLNb lead to severe skeletal malformations and impaired differentiation of chondrocyte precursors (21,22), whereas mutations in FLNc result in myofibrillar myopathy, muscular dystrophy, and defective differentiation of muscle cells (23-25).
As described above, the known roles of filamins in cytoskeletal organization, signaling, and differentiation, coupled with the clear evidence that ASB2␣ targets filamins for degradation, suggest that ASB2␣-mediated filamin depletion is likely to be central to ASB2␣'s effects on cell adhesion, spreading, and differentiation. Moreover, a recent study using ASB2␣ conditional knock-out mice supports roles for ASB2␣ and FLNa in dendritic cell spreading and maturation (26). However, other ASB2␣ substrates have recently been reported, including Janus kinases (27,28) and mixed lineage leukemia (29). A definitive test of filamins' roles in ASB2␣ function requires a better understanding of ASB2␣-mediated degradation and identification of functional ASB2␣-resistant filamin mutants. In our previous studies, we provided insight into the mechanism by which ASB2␣ targets filamins for degradation, and we identified the ABD of filamin as sufficient and necessary for ASB2␣-mediated degradation (7). We now map the ASB2␣ target region to the calponin homology domain 1 (CH1) in the FLNa actin-binding domain and identify a specific cluster of lysine residues in CH1 that are required for FLNa proteasomal degradation. We use this information to generate point mutant or chimeric FLNa that is resistant to ASB2␣ allowing us to show that cell spreading defects associated with ASB2␣ expression can be fully reversed by restoring FLNa expression.
Immunofluorescence-Cells plated on fibronectin-coated (5 g/ml) coverslips were fixed in 4% paraformaldehyde in phosphate-buffered saline (PBS), pH 7.4, for 15 min and permeabilized for 30 min with PBS containing 0.2% bovine serum albumin (BSA), 50 mM NH 4 Cl, and 0.3% Triton X-100. The coverslips were then incubated with primary antibody or fluorophore-conjugated phalloidin for 1 h at room temperature, washed in PBS, and incubated with secondary antibody for 1 h at room temperature. Coverslips were mounted using the ProlongGold anti-fade mounting agent (Invitrogen). Images were acquired using Nikon Eclipse Ti with ϫ40 or ϫ100 objective using Micro-Manager open source software (32) and analyzed using ImageJ (National Institutes of Health, Bethesda, rsb.info.nih.gov).
Flow Cytometry Assays of ASB2␣ Activity-Flow cytometry assays were described previously (7). Briefly, for the serial transfection, CHO cells were transfected with the indicated GFPtagged constructs, and 24 h after transfection, cells were re-transfected with either dsRed-ASB2␣ or dsRed-ASB2␣⌬S. 48 h after the second transfection, cells were detached and washed in PBS, and GFP fluorescence intensity of dsRed-expressing cells was quantified using an LSRII instrument (BD Biosciences). Flow cytometry data analysis was carried out using FlowJo analysis software. Percent of protein remaining was defined as F␣/F␣⌬S ϫ 100, where F␣ is the GFP geometric mean fluorescence intensity of dsRed-ASB2␣-expressing cells, and F␣⌬S is the GFP geometric mean fluorescence intensity of dsRed-ASB2␣⌬S-expressing cells.
Binding Assays-GST fusion proteins were produced in Escherichia coli BL21 Gold (Stratagene, La Jolla, CA) and purified on glutathione-Sepharose 4 Fast Flow medium (GE Healthcare) according to manufacturer's instructions. CHO cells were transiently transfected with GFP-ASB2␣, harvested 24 h later, and lysed. Cell lysates were incubated overnight with GST, GST-FLNaABD, or GST-FLNaABD K42R/ K43R/K135R bound to glutathione-Sepharose beads, washed, and resuspended in SDS sample buffer. Bound proteins were fractionated by SDS-PAGE and analyzed by Western blotting using anti-GFP antibody.
Cell Spreading Assay-HeLa cells were plated on fibronectincoated (5 g/ml) coverslips, and 3 h after plating, the cells were fixed in 4% paraformaldehyde in phosphate-buffered saline (PBS), pH 7.4. Cell areas were measured by manually rendering the cell contour in phase contrast.

ASB2␣ Targets CH1 of FLNa for Degradation-Using
Western blotting, immunofluorescence, and flow cytometry assays, we previously showed that ASB2␣ expression triggers polyubiquitination and proteasomal degradation of filamins (5)(6)(7)31). In keeping with our earlier studies (7), flow cytometric degradation assays show that the FLNa ABD is both necessary and sufficient for ASB2␣-mediated degradation as the isolated ABD is efficiently degraded following ASB2␣ expression, although FLNa lacking the ABD is resistant to degradation (Fig. 1B). The ABDs of the three filamin isoforms show high sequence and structural similarity (30), and we previously demonstrated that the ABDs of FLNa, FLNb, and FLNc are efficiently targeted for degradation by ASB2␣ (7). Consistent with our flow cytometric data, Western blotting shows that co-expressing FLNaABD-GFP with ASB2␣, but not the inactive mutant ASB2⌬S (which lacks the SOCS box and so is unable to engage the rest of the E3 ubiquitin ligase complex), leads to loss of the FLNaABD (Fig. 1C). Furthermore, addition of the proteasome inhibitor, MG132, protects FLNaABD from ASB2␣-mediated degradation. Thus, as we reported previously (7), ASB2␣ triggers proteasomal degradation of FLNa by targeting the ABD.
The ABD of filamin is composed of two calponin homology domains (CH1 and CH2) (30,33,34) separated by a linker region and contains three predicted actin-binding sites (ABS1, ABS2, and ABS3), two in CH1 and one in CH2 that together are thought to support binding to F-actin (Figs. 1A and 2A) (35). To narrow down the region within the FLNaABD that is targeted for polyubiquitination and proteasomal degradation by ASB2␣, we generated GFP-tagged constructs corresponding to CH1 and CH2 ( Fig. 2A). When expressed in cells, the CH1 of FLNa exhibited strong targeting to actin stress fibers, although CH2 failed to accumulate on stress fibers (Fig. 2B). Western blotting established that FLNaABD-GFP, FLNaCH1-GFP, and FLNaCH2-GFP were all of the expected sizes ( Fig. 2C) indicating that the cytoplasmic localization of FLNaCH2-GFP is not due to a GFP degradation product (Fig. 2C). Our data therefore suggest that CH1 is sufficient for F-actin binding. This is consistent with published studies of CH domains in other ABD-containing pro-teins (35,(37)(38)(39)(40)(41). Furthermore, as was reported for the isolated ␣-actinin CH1 domain (35), the CH1 of FLNa exhibited aberrant clumping on stress fibers (Fig. 2B), suggesting that CH2 normally regulates a CH1-mediated interaction of FLNaABD with actin filaments.
Having validated the expression and targeting of the GFPtagged constructs, we tested the ability of the individual CH domains to be targeted for degradation by ASB2␣ using the flow cytometry-based assay described previously (7). Similar to FLNaABD-GFP, levels of FLNaCH1-GFP were substantially decreased in ASB2␣-expressing cells (Fig. 2D). However, the level of FLNaCH2-GFP was unaffected, as was the level of FLNa⌬ABD (an FLNa mutant lacking the ABD), which we previously showed to be resistant to ASB2␣-mediated degradation (Figs. 1A and 2D). These data point to FLNaCH1 as the minimal region of FLNa sufficient for ASB2␣-mediated degradation and suggest that FLNaCH1 is also sufficient for interaction with ASB2␣.
Identification of Lysine Residues Required for ASB2␣-mediated FLNaABD Degradation-We previously showed that ASB2␣ targets filamins for polyubiquitination and proteasomal degradation (5,6). Furthermore, we demonstrated that the ABD of filamin is necessary and sufficient for ASB2␣-mediated degradation (7). Our finding that CH1 of FLNa is the minimal fragment targeted for degradation by ASB2␣ implies that CH1 harbors lysine residues targeted for ubiquitination by ASB2␣. The ABD of FLNa contains 20 lysine residues ( Fig. 2A), and 11 reside within CH1, whereas two are within the linker region ( Fig. 2A). Thus, we sought to identify lysine residue(s) important for ASB2␣-mediated degradation and subsequently to generate an ASB2␣-resistant FLNaABD by mutagenesis of the target lysine residues.
First, we performed a series of mutagenesis reactions to generate a mutant FLNaABD in which all 13 lysine residues within CH1 and the linker region ( Fig. 2A) were substituted with charge-conserving arginine (FLNaABD Lys/Arg). When transiently expressed in cells, this mutant FLNaABD exhibited impaired association with F-actin at stress fibers and was mainly cytoplasmic (Fig. 3A). Western blotting revealed GFP-tagged protein of the expected size (Fig. 3B), excluding the possibility that GFP cleavage might explain the cytoplasmic localization of the FLNaABD Lys/Arg mutant. Although FLNaABD Lys/Arg mutant exhibits impaired association with F-actin at stress fibers (Fig. 3A), it is resistant to ASB2␣-mediated degradation ( Fig. 3C and Table  1). Thus, despite the presence of seven lysine residues in the CH2 and 20 lysines in the GFP tag, mutation of all 13 lysines in the CH1 plus linker region renders FLNaABD resistant to ASB2␣-induced degradation indicating that CH1 lysine residues are required for ASB2␣-mediated degradation.
To determine which of the 13 lysine residues, or which combination of lysine residues, are required for FLNaABD degradation, various permutations of the FLNaABD Lys/Arg mutants were tested in the flow cytometric assay (Table 1). An in-depth analysis revealed that no single or double lysine to arginine substitution tested inhibited FLNaABD degradation (Table 1). However, all constructs containing a triple lysine to arginine substitution at positions 42, 43, and 135 were resistant to ASB2␣-mediated degradation (Table 1; Fig. 4C). Notably, all three of these lysines must be mutated to render FLNaABD resistant to degradation as the presence of either lysine 135 or of lysines 42 and 43 is sufficient to support ASB2␣-triggered degradation ( Fig. 4C) even in ABD mutants containing a total of nine other lysine to arginine mutations (Table 1). This suggests that lysines 42, 43, and 135 are likely to be acceptor sites for polyubiquitination and subsequent proteasomal degradation and that polyubiquitination of any of these residues will support degradation.
ASB2␣-mediated FLNaABD Degradation and F-actin Targeting Are Separable-In addition to characterizing their susceptibility to ASB2␣-mediated degradation, we analyzed the lysine to arginine mutant FLNaABD constructs for their association with F-actin as assessed by stress fiber targeting. This revealed four lysine residues (at positions 42, 43, 127, and 135) whose mutation to arginine resulted in impaired stress fiber targeting ( Table 1). Each of these lysine residues resides within the predicted actin-binding sites on CH1 ( Fig. 2A), and it is likely that their mutation interferes with or disrupts F-actin binding. Thus, residues required for ASB2␣-mediated degradation and for F-actin binding overlap, raising the question of whether the processes are linked.
As summarized in Table 1, all ASB2␣-resistant mutants are defective in stress fiber targeting. However, failure to target to stress fibers is not sufficient to impart ASB2␣ resistance as several FLNaABD mutants, which exhibit cytoplasmic localization, are targeted for degradation by ASB2␣ (Table 1), illustrating that FLNaABD localization to stress fibers is dispensable for ASB2␣-mediated degradation. As shown in Fig. 4A, FLNaABD mutant constructs K33R/K42R/K43R, K135R, and K42R/  K43R/K135R all exhibit impaired targeting to stress fibers despite expression of GFP-tagged protein of the expected sizes as assessed by Western blotting (Fig. 4B). However, although FLNaABD K33R/K42R/K43R and FLNaABD K135R are targeted for degradation by ASB2␣ (Fig. 4C), FLNaABD K42R/K43R/ K135R is resistant to ASB2␣-mediated degradation (Fig. 4C), confirming that any one of the three lysine residues (Lys-42, Lys-43, or Lys-135) is sufficient to sustain ASB2␣-mediated degradation but showing that all three residues are required for F-actin association. Notably, lysine residues 42, 43, and 135 are in close proximity and are exposed at the surface and readily accessible from the outside as shown in the ribbon diagram of FLNaABD (Fig. 5A). Our identification of putative polyubiquitination sites within the actinbinding sites of FLNaCH1, which are likely to be buried when FLNa is bound to F-actin, suggests that FLNaABD bound to F-actin may be protected from degradation and that ubiquitination may occur only upon dissociation from F-actin.
FLNaABD Lys-42/43/135 Retains ASB2␣ Binding Activity-ASB2␣-triggered proteasomal degradation of FLNa requires that the ASB2␣ E3 ubiquitin ligase complex associates with FLNa and catalyzes the addition of polyubiquitin chains onto lysine residues on FLNa. Our mutagenesis data establish that arginine substitution of lysine 42, 43, and 135 in FLNa ABD renders it resistant to ASB2␣-mediated degradation, suggesting that these CH1 lysine residues may either serve as ubiquiti-nation sites and/or possibly be involved in mediating interaction with ASB2␣ as well as F-actin. To test whether mutation of these lysine residues disrupts the ability of FLNaABD to interact with ASB2␣, we performed pulldown assays. Both GST-FLNaABD and GST-FLNaABD K42R/K43R/K135R bound to ASB2␣, although only background binding was seen to the GST beads (Fig. 5B). Our results therefore indicate that lysine to arginine mutation on residues 42, 43, and 135 does not alter ASB2␣ interaction with FLNaABD, instead suggesting that these lysines serve as polyubiquitination sites during ASB2␣mediated proteasomal degradation. This effect is highly specific as we demonstrate that FLNaABD K42R/K43R/K135R is not targeted for degradation (Fig. 4C) even though all other lysine residues except 42, 43, and 135 are available and surface-exposed (Fig. 5A). To assess polyubiquitination, FLNaABD wild-type and FLNaABD K42R/K43R/K135R were coexpressed with ASB2␣ in the presence of MG132 (to prevent proteasomal degradation of the polyubiquitinated protein) and cell lysates were subjected to Western blot analysis. Indeed, a smear of higher molecular species, indicative of polyubiquitination, was seen for FLNaABD-GFP but not for FLNaABD K42R/K43R/ K135R -GFP proteins (Fig. 5C). We further confirmed that, as expected, FLNaCH1-GFP also exhibited a smear of higher molecular weight species when co-expressed with ASB2␣ (Fig. 5D).

Full-length FLNa K42R/K43R/K135R Mutant Is Resistant to
ASB2␣-mediated Degradation-The preceding results demonstrate that FLNaABD K42R/K43R/K135R is resistant to ASB2␣-mediated degradation and is still capable of interacting with ASB2␣, and the results suggest that lysine residues 42, 43, and 135 are required for ubiquitination and degradation of FLNa. Furthermore, our localization assays suggest that lysine residues 42, 43, and 135 are also important for F-actin association as mutagenesis of these lysine residues impairs FLNaABD targeting to stress fibers. We next asked whether mutagenesis of these specific lysine residues would render full-length FLNa resistant to ASB2␣-mediated degradation and affect FLNa targeting to stress fibers. As shown in Fig. 6A, full-length FLNa K42R/K43R/K135R also exhibits impaired targeting to stress fibers, and unlike FLNaABD K42R/K43R/K135R (Fig. 4A), some weak targeting is observed. We suggest that this is likely to be due to dimerization of the mutant FLNa with endogenous CHO cell FLNa. Analysis of lysates from transfected cells revealed bands of the expected size for FLNa K42R/K43R/K135R (Fig. 6B). Importantly, similar to FLNaABD K42R/K43R/K135R , full-length FLNa K42R/K43R/K135R is resistant to ASB2␣-mediated degradation (Fig. 6C). These results validate our earlier findings and suggest that ASB2␣ targeted degradation of full-length FLNa can be mediated by three different lysine residues.
FLNa K42R/K43R/K135R Does Not Rescue Impaired Spreading of Filamin Knockdown Cells-We previously showed that ASB2␣ expression induces proteasome-mediated degradation of all three filamin isoforms and impairs cell spreading (5, 6). As cell knockdown for both FLNa and FLNb is also impaired in spreading (5,6), it was proposed that ASB2␣ regulates hematopoietic cell spreading by targeting filamins for degradation (5,26,31). To test whether filamin degradation is the only mechanism by which ASB2␣ impacts cell spreading, we first sought to generate ASB2␣-resistant filamin constructs for use in reconstitution studies. As described above, we have now identified an ASB2␣-resistant full-length FLNa mutant (FLNa K42R/K43R/K135R). However, this mutant is defective in stress fiber targeting suggesting that its F-actin binding activity is impaired. To determine whether the mutant FLNa is functional, we first tested its ability to rescue spreading of HT1080 double FLNa and FLNb knockdown (FLNabKD) cells. We previously showed that knockdown of FLNa and FLNb impairs cell spreading and that expressing an shRNA-resistant full-length FLNa rescues the spreading phenotype (Fig. 6D) (6). However, whereas re-expression of knockdown-resistant full-length FLNa rescues the spreading defect of FLNabKD cells, the FLNa K42R/K43R/K135R mutant does not (Fig. 6D), suggesting that F-actin binding is important for filamin's role in cell spreading.
FLNa Chimera Containing CH1 of ␣-Actinin1 Is Resistant to ASB2␣-mediated Degradation-We have identified putative ubiquitination sites on FLNa and generated an ASB2␣-resistant FLNa mutant. However, as this mutant is also defective in stress fiber localization, presumably due to impaired F-actin binding, the resulting protein is nonfunctional. We therefore sought an alternative approach to generate ASB2␣-resistant filamin with a normal subcellular localization. We previously demonstrated that ASB2␣ specifically triggers degradation of the ABD of filamins but not of ␣-actinin1 (7), despite their related structure and similar sequence (Fig. 7A) (30). Thus, we sought to generate another ASB2␣-resistant filamin by swapping CH1 of FLNaABD with that of ␣-actinin1. Two ABD-GFP chimeras were gen- erated as follows: one contains CH1 of FLNa followed by linker region of FLNa plus CH2 of ␣-actinin1 (CH1f-CH2␣), and the other contains CH1 of ␣-actinin1 also followed by linker region of FLNa plus CH2 of FLNa (CH1␣-CH2f) (Fig.  7B). When expressed in CHO cells, both chimeras produce GFP-tagged proteins of the expected sizes as analyzed by Western blotting (Fig. 7D). When their subcellular localization was analyzed, both chimeras show strong targeting to F-actin at stress fibers (Fig. 7B) illustrating that they are functional. These results are consistent with previous studies that showed that similar filamin and ␣-actinin1 CH domain chimeras interact with F-actin in vitro and in vivo (35).
We next tested ASBS2 ␣-mediated degradation of the two chimeras. We employed the previously described flow cytometry-based degradation assay (7) to show that CH1f-CH2␣ but not CH1␣-CH2f is targeted for degradation by ASB2␣ (Fig. 7E). These results extend our earlier findings and confirm that CH1 is the minimal FLNa fragment sufficient for ASB2␣-mediated degradation and that, despite their structural similarity, ASB2␣-mediated degradation is specific for CH1 of FLNa over CH1 of ␣-actinin.
As the chimera comprising CH1␣-CH2f is resistant to ASB2␣mediated degradation (Fig. 7E), we sought to generate a full-length FLNa resistant to ASB2␣-mediated degradation and capable of binding F-actin by swapping the CH1 of full-length FLNa with that of ␣-actinin1 (CH1␣-FLNa). This construct expressed at the expected size as illustrated by Western blotting (Fig. 8B). Importantly, although the chimera exhibited some aberrant clumping, it associated with F-actin at stress fibers (Fig. 8A) and is resistant to ASB2␣-mediated degradation (Fig. 8C). Furthermore, as shown in Fig. 8D, the ASB2␣-resistant full-length FLNa chimera (CH1␣-FLNa) rescues the spreading phenotype of FLNabKD HT1080 cells, suggesting that it is functional, whereas FLNa⌬ABD does not rescue the spreading defect (Fig. 8D). We have therefore generated an ASB2␣-resistant FLNa that targets to stress fibers and supports cell spreading. with either dsRed-ASB2␣ or dsRed-ASB2␣⌬S, and FLNa degradation was assessed as described in Fig. 1 (see "Experimental Procedures" for details). Bar chart depicts mean percentage of GFP-tagged protein remaining Ϯ S.E. Data are from at least six independent experiments. D, HT1080 wild-type (WT) and FLNa and FLNb double knockdown cells (FLNabKD) or FLNabKD cells transiently expressing either wild-type FLNa GFP or FLNa K42R/K43R/K135R GFP were plated on fibronectin-coated coverslips. 3 h after plating, cells were fixed, and areas were measured by manually rendering the cell contour in phase contrast and normalized to the size of untransfected WT cells. Bar chart depicts mean relative cell area Ϯ S.E. for each condition from three independent experiments (Ͼ20 cells per condition). p values were calculated using the t test.

DISCUSSION
ASB2␣ is thought to regulate hematopoietic cell differentiation by modulating cell spreading, actin remodeling, and cell adhesion (5,31), and the importance of ASB2␣ is supported by a recent knock-out mouse study showing that ASB2␣ is a critical regulator of spreading and migration in immature dendritic cells (26). We have established that ASB2␣ targets filamins for proteasomal degradation (5,6,31), have shown that the N-terminal region of ASB2␣ along with the first 10 ankyrin repeats supports ASB2␣ binding to filamins (31), and have revealed that the ABD of filamin mediates interaction with ASB2␣ and is sufficient for ASB2␣-mediated degradation (7). Here, we  NOVEMBER 1, 2013 • VOLUME 288 • NUMBER 44 extend those studies to show that ASB2␣ targets CH1 of FLNa for proteasomal degradation, and we identify a cluster of three lysine residues on the surface of CH1 that is required for ASB2␣-mediated proteasomal degradation of FLNa. We find that these lysines are also involved in FLNa binding to F-actin and that this is required to rescue spreading defects in filamin knockdown cells. We build on this information to show that an ASB2␣-resistant chimeric FLNa protein, in which the FLNa CH1 is replaced by the ␣-actinin CH1, retains localization to actin stress fibers and fully restores spreading in ASB2␣-ex-pressing cells, demonstrating that the spreading defect induced by ASB2␣ is due to the acute loss of filamins.

ASB2␣ Regulates Cell Spreading through Filamin Degradation
Our finding that CH1 of FLNa is the minimal fragment targeted for degradation by ASB2␣ implies that CH1 harbors both the ASB2␣-binding and ubiquitination sites. We previously showed that ASB2␣ targets filamins for polyubiquitination and proteasomal degradation (5-7) and narrowed potential ubiquitination sites from 156 lysines within full-length FLNa to 20 within the ABD (7) and here to 11 within CH1. Thus, we sought to generate an ASB2␣-resistant FLNaABD by mutagenesis of the lysine residues within CH1 and used an in-depth flow cytometry analysis to identify candidate sites of ubiquitination. We find that mutation of all CH1 lysine residues to arginine within the context of FLNaABD renders FLNaABD resistant to ASB2␣-mediated degradation. Interestingly, mutation of only three lysine residues (Lys-42, Lys-43, and Lys-135) is sufficient to make FLNaABD, or full-length FLNa, resistant to ASB2␣mediated degradation even though all other lysine residues remain available for ubiquitination.
The specificity of ASB2␣ is further illustrated by our observation that despite the structural similarity of the FLNa CH1 and CH2 domains (Protein Data Bank code 3HOP chain A: root mean square deviation of 1.35 Å over 75 carbon ␣ atoms and 23% sequence identity), only FLNa CH1 is targeted for ASB2␣mediated degradation. Likewise, the FLNa CH1 shares even greater structural and sequence similarity for the CH1 domain of ␣-actinin1 (Protein Data Bank code 2EYI residues 26 -140: root mean square deviation of 0.97 over 104 carbon ␣ atoms and 45% sequence identity), and yet ␣-actinin is not targeted by ASB2␣. The structural basis for ASB2␣ specificity is not yet known, but presumably requires both an ASB2␣-binding site and an appropriately positioned acceptor lysine. Notably, FLNa lysine residues 42, 43, and 135 are in close proximity, exposed at the surface, and readily accessible from the outside. Furthermore, our data suggest that the presence of any one of these three lysine residues is sufficient to sustain ASB2␣-mediated degradation. Importantly, we show that FLNaABD K42R/K43R/K135R is capable of interacting with ASB2␣ suggesting that mutation of these lysine residues has not interfered with ASB2␣ binding and that lysine residues 42, 43, and 135 in fact serve as polyubiquitination acceptor sites. This conclusion is supported by Western blotting showing that in the presence of proteasome inhibitors wildtype FLNaCH1 and FLNaABD, but not the FLNaABD K42R/ K43R/K135R triple mutant, exhibit higher molecular weight smears consistent with polyubiquitination.
The major known function of FLNa CH domains is as the primary actin-binding site. We therefore tested whether the lysine to arginine mutations impacted the ability of the ABD to localize to actin stress fibers. Mutating all CH1 lysines very strongly impaired association with F-actin suggesting that lysine residues within CH1 are important for interaction with F-actin. Specifically, we identified lysine residues 42, 43, and 135 as important for association with F-actin and also found that mutation of Lys-127 partially impaired ABD localization (Table 1). Consistent with this, all four of these lysine residues reside within the two predicted actin-binding sites in CH1 (30,33,34). That CH1 residues required for degradation are also needed for stress fiber localization suggests that filamin bound to F-actin may be resistant to ASB2␣-mediated degradation as lysine residues 42, 43, 135 may be unavailable for ubiquitination. We speculate that FLN must dissociate from F-actin in order for the target lysine residues to be exposed for ubiquitination and subsequently targeted for proteasomal degradation (Fig. 9).
Our data on FLNa ABD localization are consistent with studies of other ABD-containing proteins (35,(37)(38)(39)(40)(41). We find that FLNa CH1 but not CH2 displays considerable association with F-actin at stress fibers. Furthermore, expression of CH1 alone also results in aberrant clumping in cells suggesting that the CH2 domain is essential for proper interaction of FLNaABD with actin filaments. Studies on CH domains of ␣-actinin1 suggest that the CH2 regulates interaction and dynamics of the ABD with actin filaments and that in the absence of CH2, ABS1 within CH1 may become more accessible and contribute to increased filament association (35).
We previously showed that ASB2␣ specifically targets the ABD of FLNa, FLNb, and FLNc for degradation and that the ABD of ␣-actinin1 is resistant to ASB2␣-mediated degradation (7). Therefore, having identified CH1 as the minimal fragment of FLNa sufficient for ASB2␣-mediated degradation, we complemented our mutagenesis studies with a chimeric protein approach and generated an ASB2␣-resistant FLNa by swapping CH1 of FLNa with that of ␣-actinin1. Consistent with previous studies, both the chimeric ABD and chimeric FLNa targeted to stress fibers and exhibited clumping in cells, presumably due to increased F-actin binding (35). Nonetheless, the chimeras were resistant to ASB2␣-mediated degradation, and the chimeric FLNa is functional and rescues spreading of filamin knockdown cells and ASB2␣-expressing cells.
The exact mechanism by which ASB2␣-mediated filamin degradation affects differentiation remains to be determined. Until recently, retinoic acid was the only pathway known to trigger ASB2␣ expression, and filamins were the only known ASB2␣ substrate. However, more recent studies found that ASB2␣ expression can be induced by Notch signaling (27) and identified mixed lineage leukemia (29), Jak2 (27), and Jak3 (28) as potential substrates. However, Jak2 and Jak3 are not regulated by ASB2␣ during hematopoietic differentiation (42). Our data establish that filamin degradation accounts for the spreading defects seen following ASB2␣ expression, and we pos- FIGURE 9. Schematic summary of proposed model. The ABD of FLN interacts with F-actin and ASB2␣. ASB2␣ targets lysine residues within FLN CH1 domain, and mutations in CH1 render FLN ASB2␣-resistant. FLN bound to F-actin is protected from ASB2␣-mediated degradation and regulates cell spreading. Upon dissociation from F-actin, lysine residues within the CH1 are exposed and available for polyubiquitination leading to subsequent proteasomal degradation. NOVEMBER 1, 2013 • VOLUME 288 • NUMBER 44 tulate that ASB2␣ contributes to hematopoietic differentiation, through filamin degradation.

ASB2␣ Regulates Cell Spreading through Filamin Degradation
In keeping with this, filamins have been implicated in differentiation of various cell types (5,15,16,(43)(44)(45) and are downregulated during differentiation of hematopoietic cells (5), muscle cells (15), and chondrocytes (16,45) as well as maturation of dendritic cells (26). One possible mechanism involves the role of filamins in cytoskeletal re-organization and cell shape, which are integral aspects of differentiation (46,47). ASB2␣ may therefore control differentiation by modulating cell spreading and actin remodeling through targeting of filamins for degradation.
In summary, proteasomal degradation of filamins provides a mechanism to explain ASB2␣-mediated effects on hematopoietic cell differentiation. Our studies provide detailed insight into ASB2␣-mediated filamin degradation, identify lysine residues required for targeted degradation, and establish that loss of filamin explains the effect of ASB2␣ on cell spreading. The ASB2␣-resistant filamins described here will provide essential tools for future functional studies to elucidate if, and how, filamins mediate ASB2␣'s effects on cell differentiation.