Functional and Structural Insights into ASB2α, a Novel Regulator of Integrin-dependent Adhesion of Hematopoietic Cells*

By providing contacts between hematopoietic cells and the bone marrow microenvironment, integrins are implicated in cell adhesion and thereby in control of cell fate of normal and leukemia cells. The ASB2 gene, initially identified as a retinoic acid responsive gene and a target of the promyelocytic leukemia retinoic acid receptor α oncoprotein in acute promyelocytic leukemia cells, encodes two isoforms, a hematopoietic-type (ASB2α) and a muscle-type (ASB2β) that are involved in hematopoietic and myogenic differentiation, respectively. ASB2α is the specificity subunit of an E3 ubiquitin ligase complex that targets filamins to proteasomal degradation. To examine the relationship of the ASB2α structure to E3 ubiquitin ligase function, functional assays and molecular modeling were performed. We show that ASB2α, through filamin A degradation, enhances adhesion of hematopoietic cells to fibronectin, the main ligand of β1 integrins. Furthermore, we demonstrate that a short N-terminal region specific to ASB2α, together with ankyrin repeats 1 to 10, is necessary for association of ASB2α with filamin A. Importantly, the ASB2α N-terminal region comprises a 9-residue segment with predicted structural homology to the filamin-binding motifs of migfilin and β integrins. Together, these data provide new insights into the molecular mechanisms of ASB2α binding to filamin.

Characterization of hematopoietic stem cell and leukemia stem cell properties and understanding the molecular mechanisms that control the early steps of hematopoietic differentiation, which are deregulated in leukemia cells, are major challenges. These issues are relevant not only for the development of therapeutic approaches to target leukemia stem cells in vivo, but also for engraftment of normal hematopoietic stem cells following transplantation. Hematopoietic stem cells reside predominantly in a complex bone marrow microenvironment, the stem cell niche (1,2). Hematopoietic stem cell fate decisions are governed by the integrated effects of niche-independent intrinsic and niche-dependent extrinsic signals. Recent studies indicate that changes in the hematopoietic stem cell niche may have a role in hematopoietic malignancies (3)(4)(5). Available data show that during development and following transplantation, integrin adhesion molecules play a major role in anchoring stem cells in the hematopoietic niche (6 -8). Although adhesion of CD34 positive cells to fibronectin inhibits cell proliferation, adhesion of acute myeloid leukemia (AML) 5 cells to fibronectin stimulates proliferation (9). In addition, AML cells showed increased survival as a result of the interaction of ␤1 integrins (␣4␤1/VLA-4 and ␣5␤1/VLA-5) on leukemia cells with fibronectin leading to their reduced chemosensitivity (10,11). Accordingly, antibodies directed against VLA-4 prolong survival of mice in a bone marrow minimal residual disease model (10). Moreover, the FNIII14 peptide of fibronectin that impairs VLA-4-and VLA-5-mediated adhesion to fibronectin overcomes cell adhesion-mediated drug resistance (12). In this context, proteins controlling integrin-dependent adhesion of hematopoietic cells may represent novel therapeutic targets in AML.
Our previous work identified ASB2 as a retinoic acid response gene and a target gene for the oncogenic promyelocytic leukemia retinoic acid receptor ␣ (PML-RAR␣) fusion protein in acute promyelocytic leukemia cells (13,14). Expression of PML-RAR␣ has been shown to induce the myeloid differentiation arrest observed in acute promyelocytic leukemia (15)(16)(17)(18). At the molecular level, PML-RAR␣ acts as a transcriptional repressor that interferes with gene expression programs normally leading to full myeloid differentiation. Recently, PML-RAR␣ was shown to be bound to the ASB2 promoter in acute promyelocytic leukemia cells in the absence of retinoic acid leading to hypoacetylation of histone H3 (19). Moreover, following retinoic acid treatment of acute promyelocytic leuke-mia cells, hyperacetylation and recruitment of RNA polymerase II to the ASB2 promoter were observed (19). Furthermore, ASB2 is also a target of another oncoprotein that acts as a transcriptional repressor, the AML1-ETO fusion protein, 6 indicating that ASB2 mis-expression is associated with AML. However, ASB2 is specifically expressed in normal immature hematopoietic cells (13,14) and so is likely to be relevant during early hematopoiesis. Importantly, Notch activation stimulated ASB2 expression (20). ASB2 encodes two isoforms, a hematopoietic-type (ASB2␣) and a muscle-type (ASB2␤) that are involved in hematopoietic and myogenic differentiation, respectively (21,22). ASB2 proteins belong to the family of ASB proteins that harbor a variable number of ankyrin repeats (ANK) followed by a suppressor of cytokine signaling box located at the C-terminal end of the protein (23). These proteins are the specificity subunits of E3 ubiquitin ligase complexes (21,22). Indeed, suppressor of cytokine signaling box-mediated interactions with the Elongin B-Elongin C (EloB-EloC) complex and the Cul5/Rbx2 module allow ASB2 proteins to assemble a multimeric E3 ubiquitin ligase complex, and so regulate the turnover of specific proteins involved in cell differentiation. We have recently shown that ASB2␣ ubiquitin ligase activity drives proteasome-mediated degradation of actinbinding proteins filamin A (FLNa), FLNb, and FLNc (24,25). In addition to their role as actin cross-linkers, FLNs bind many adaptor and transmembrane proteins (26 -28). In this way, FLNs can regulate cell shape and cell motility. We have demonstrated that ASB2␣-mediated degradation of FLNs can regulate integrin-mediated spreading of adherent cells and initiation of migration of both HT1080 and Jurkat cells (24,25,29). FLNs are composed of an N-terminal actin-binding domain followed by 24 immunoglobulin-like domains (IgFLN(1-24)) (30). The CD face of Ig-like repeats of FLNa (IgFLNa), the major nonmuscle isoform of FLNs, represents a common interface for FLN-ligand interaction (31)(32)(33). Interestingly, it was recently demonstrated that FLN ligands can associate with several IgFLNa domains belonging to the same subgroup (34). Among group A, which contains seven IgFLNa repeats, IgFLNa21 binds GPIb␣, ␤7 integrin, and migfilin with the highest affinity (31,32,34). Here, molecular modeling, site-directed mutagenesis, and cell biological studies were used to obtain structural and functional insights into the ASB2␣ E3 ubiquitin ligase complex.
Cell Adhesion to Fibronectin-Fibronectin (BD Biosciences) was immobilized overnight at 4°C in 96-well plates (50 g/ml) in PBS. Wells were then saturated with 5% BSA in PBS for 1 h at room temperature and washed three times with PBS. PLB985 cells stably transfected with zinc-inducible vectors encoding ASB2 proteins or the corresponding empty vector were cultured with or without ZnSO 4 for 16 h, loaded with 0.5 M calcein AM in HBSS without Ca 2ϩ and Mg 2ϩ (HBSS Ϫ ) containing 0.5% BSA, and washed once in HBSS Ϫ , 1 mM EDTA. Adhesion to fibronectin was assayed using 200,000 cells/well in HBSS Ϫ , 0.5% BSA wells in the absence or presence of 1 mM MnCl 2 to activate integrins for 10 min. Nonadherent cells were removed with 1 to 3 gentle washes with PBS containing 1 mM MnCl 2 (to maintain integrin activation) or PBS alone (to suppress integrin activation). Fluorescence intensity was quantified using a microplate fluorescence reader FLx-800 (Bio-TEK). The percentage of adherent cells was calculated as follows: (fluorescence intensity of adherent cells/fluorescence intensity of cells plated) ϫ 100. Each assay was performed in triplicate and at least four independent experiments were done. Statistical anal-yses were performed using Prism software. All p values were calculated using the Mann-Whitney t test.
Cell Spreading-Cell spreading assays were carried out as described (24).
Immunofluorescence Microscopy-Immunofluorescence analyses were performed essentially as described (24). To better visualize ASB2␣ and FLNa localization on stress fibers 8 h after transfection of HeLa cells, immunofluorescence analyses were performed in cytoskeleton buffer containing 10 mM MES, pH 6.9, 5 mM glucose, 150 mM NaCl, 5 mM EGTA, 5 mM MgCl 2 and the cells were pretreated with the same buffer containing 0.1% Triton X-100 for 30 s before paraformaldehyde fixation. Secondary antibodies used were Cy3-coupled goat anti-mouse (Jackson Laboratories). F-actin was visualized with Alexa 633phalloidin (Fisher) diluted 1:500. Slides were viewed with a Zeiss Axio Imager M2 using a ϫ63/1.3 oil DIC Plan Apochromat objective (Zeiss). Images were acquired and processed using AxioVision software and AxioCam MRm camera (Zeiss). Pearson correlation coefficients of the ASB2 protein co-localization with FLNa are mean Ϯ S.E. from 5 random regions from 5 ASB2␣-expressing cells.
2 g of anti-FLNa antibodies or control IgG and 30 l of protein A-Sepharose suspension (GE Healthcare) were added to 880 g of whole cell extracts of demembranated cells. After 3 h in ice, beads were washed twice in XT buffer (50 mM Pipes, 50 mM NaCl, 150 mM sucrose, 40 mM Na 4 P 2 O 7 ⅐10H 2 O, 0.05% Triton X-100) supplemented with 1 mM Na 3 VO 4 , 50 mM NaF, 25 mM ␤-glycerophosphate, 2 mM sodium pyrophosphate, and 1% protease inhibitor mixture. After 5 min of boiling in Laemmli buffer, samples were resolved by SDS-PAGE and analyzed by immunoblotting.
In Vitro Binding Assays-7 l of 1PNA serum or 15 l of its preimmune counterpart and 30 l of protein A-Sepharose suspension was added to 500 g of whole cell extracts of demembranated cells. After 3 h in ice, beads were washed twice in XT buffer supplemented with 1 mM Na 3 VO 4 , 50 mM NaF, 25 mM ␤-glycerophosphate, 2 mM sodium pyrophosphate, and 1% protease inhibitor mixture. The immobilized GFP-ASB2␣ANK(1-10) protein was further incubated with 10 g of Escherichia coli extracts expressing GST or GST-IgFLNa21 in whole cell extract buffer for 3 h in ice. Beads were washed twice as described above. After 5 min of boiling in Laemmli buffer, samples were resolved by SDS-PAGE and analyzed by immunoblotting. Alternatively, pulldown assays were performed using 5 g of purified GST-IgFLNa21, GST-IgFLNa21AA/DK, or GST-Ig-FLNb21 bound to glutathione-agarose beads (Macherey-Nagel) and 0.4 g of purified ASB2␣ANK(1-10)-His 6 . After two washes in dialysis buffer, bound proteins were fractionated by SDS-PAGE and analyzed by protein staining and Western blotting.
Molecular Modeling-Modeling was performed using the Accelrys modules Homology, Discover, Docking, and Biopolymer, run within InsightII (2005 version) on a Silicon Graphics Fuel work station.
Docking of the ASB2␣ N-terminal Peptide into the IgFLNa21 Domain-The FLNa-binding motif within the 20-residue ASB2␣ N-terminal peptide was identified and modeled based on structural homology to the integrin ␤2 (PDB code 2JF1_T), integrin ␤7 (PDB code 2BRQ_C), and migfilin (PDB code 2W0P_C) motifs, using the Align 123 and Homology programs. The residues located at the N-and C-terminals of the ASB2␣ ␤-strand motif were initially built in an extended conformation. Energy was minimized using the Discover consistent valence force field, a forcing constant of 100 kcal/mol, and the steepest descent and conjugate gradients protocols. The common ␤-strand served to preposition the ASB2␣ N-terminal peptide into the IgFLNa21 CD groove in the same way as migfilin. The resulting preliminary structure was then submitted to the Affinity program, an automatic flexible docking refinement procedure applied to predefined residues from the binding interface. The final interaction energy was Ϫ90 and Ϫ42 kcal for the van der Waals and electrostatic components, respectively. The criteria used for identifying hydrogen bonds were a donor-acceptor distance Յ3 Å and a minimum donor-protonacceptor angle of 120°. The criterion used for identifying an interaction between hydrophobic groups was a distance Յ5 Å.

ASB2␣-induced FLNa Degradation Regulates Adhesion of
Hematopoietic Cells to Fibronectin-We previously found that ASB2␣ expression prevented cell spreading and inhibited initiation of migration, and that these effects were recapitulated by knocking down FLNa and FLNb (24,29). To assess the role of ASB2␣-induced FLN degradation in hematopoietic cell adhesion, myeloblastic PLB985 cells stably transfected with zinc inducible vectors encoding ASB2␣, E3 ubiquitin ligase defective mutant ASB2␣LA, or the corresponding empty vector control were cultured with or without ZnSO 4 , labeled with calcein AM, and allowed to adhere on fibronectin-coated wells in the absence or presence of Mn 2ϩ to activate integrins. Loss of FLNa was observed only in cells expressing ASB2␣ (Fig. 1A). As expected, when adhesion was performed in the absence of Mn 2ϩ , only a low level of adhesion of cells was observed (supplemental Fig. S1A). However, cell adhesion was greatly enhanced with Mn 2ϩ (Fig. 1B). Intriguingly, upon integrin activation by cultivating the cell in the presence of Mn 2ϩ , percentages of adherent cells expressing ASB2␣ after the first and second washing steps in PBS alone were significantly increased (Fig. 1B). However, no significant differences were observed following washes in PBS containing Mn 2ϩ (supplemental Fig.  S1B). To extend and further validate the finding that ASB2␣ controls integrin-mediated adhesion of hematopoietic cells to fibronectin, we established PLB985 cells lacking FLNa expression by transfecting cells with a vector encoding an shRNA against FLNa (Fig. 1C). As was observed for ASB2␣-expressing cells, adhesion of PLB985 FLNaKD cells was significantly higher than adhesion of PLB985 LucKD control cells after washing steps in PBS alone (Fig. 1D). Taken together, our results suggest that integrin-dependent adhesion of hematopoietic cells is sustained in the absence of FLNa and establish a role for ASB2␣ in the regulation of hematopoietic cell adhesion through FLNa degradation.
The N-terminal Region and ANK 1 to 10 of ASB2␣ Are Necessary to Target FLNa-As previously observed in transfected HeLa cells (24), (i) ASB2␣ was transiently colocalized with F-actin (Fig. 3A); (ii) ASB2␣ is diffuse throughout the cytoplasm when FLNa is degraded (Fig. 3B); and (iii) E3 ubiquitin ligase defective mutants of ASB2␣ do not degrade FLNa and accumulate on stress fibers (Fig. 3, A and B). These results suggest that colocalization of ASB2␣ with F-actin may be the result of ASB2␣ association with FLNa. We therefore examined the sub-cellular localization of the ASB2⌬N N-terminal deletion mutant in transfected HeLa cells. ASB2⌬N did not accumulate on stress fibers (Fig. 3, A and B). Accordingly, deletion of the N-terminal region in ASB2␣ abrogated its ability to induce degradation of endogenous FLNa in transfected HeLa cells (Fig. 3,  B and C). Collectively, our data indicated that ASB2␣ residues 1-20 encompass a major determinant for ASB2␣ colocalization with FLNa and subsequent polyubiquitylation and degradation by the proteasome. We next questioned whether ASB2␣ residues 1-20 were sufficient for colocalization with FLNa onto stress fibers. This domain was therefore fused to GFP and expressed in HeLa cells. The resulting GFP-ASB2␣ (residues 1-20) protein was diffuse throughout the cytoplasm and the nucleus (Fig. 4A) indicating that the N-terminal region specific to ASB2␣ is not sufficient for ASB2␣ recruitment to stress fibers. To further delineate the ASB2␣ domain required for the targeting of FLNa, several C-terminal deletion mutants of ASB2␣ were constructed and their subcellular localization assessed in HeLa cells. In contrast to GFP-ASB2␣ANK(1-3), GFP-ASB2␣ANK(1-5), GFP-ASB2␣ANK(1-7), and GFP-ASB2␣ANK(1-9), GFP-ASB2␣ANK(1-12) accumulates onto stress fibers (Fig. 4A). As expected, deletion of amino acids 1-20 (GFP-ASB2⌬NANK(1-12)) abrogated stress fiber localization of this deletion mutant (Fig. 4A). To better visualize ASB2␣ and FLNa localization on stress fibers, cells were permeabilized with Triton X-100 before fixation. This treatment removes the cell membrane and membrane-associated proteins but leaves behind the cytoskeleton and cytoskeletally associated proteins. Eight hours post-transfection, GFP-ASB2␣ANK(1-12) and GFP-ASB2␣ANK(1-10) colocalized with FLNa and F-actin in stress fibers (Fig. 4B). These results indicate that ANK 1 to 10, together with the N-terminal region specific to ASB2␣, is required for colocalization of ASB2␣ with FLNa on stress fibers.
Identification of an IgFLNa21-binding Motif Encompassing ASB2␣ Residues 8 -16-We next questioned whether ASB2␣ could target FLNa for degradation through an interaction between its N-terminal region and FLNa (Fig. 5A). A prototypic structural FLNa-binding motif has recently been characterized through a number of x-ray and NMR-based studies of complexes between the IgFLNa21 and peptides derived from either the tails of integrin ␤2 or ␤7, or the N-terminal region of migfilin (31,32,34,36). This 8 -10-residue long ␤-strand motif forms an anti-parallel ␤-sheet with the IgFLNa21 C ␤-strand and anchors into the CD groove through hydrophobic contacts, especially with Ile-2283 and Phe-2285 residues from the D ␤-strand (Fig. 5B,C). Alignment of the ASB2␣ N-terminal peptide with filamin-binding peptides derived from either the tails of integrin ␤2 or ␤7, or the N-terminal region of migfilin pre- dicts the presence of a 9-residue hydrophobic ␤-strand in the ASB2␣ N-terminal peptide (residues 8 -16) (Fig. 5A), structurally homologous to those in migfilin and ␤ integrins. Interestingly, the most stringent element of sequence conservation detected is Ser-11, which was shown to play a critical role in anchoring the motif to the D strand via hydrogen bonding. These results suggest that this newly identified structural ASB2␣ motif might dock into the cavity delineated by the IgFLNa21 CD hairpin similarly to the migfilin and integrin ␤ FLNa-binding motifs. Indeed, the molecular modeling of the complex between the ASB2␣ N-terminal peptide and the IgFLNa21 domain, based on the x-ray structure of the migfilin motif-IgFLNa21 complex, revealed that the ASB2␣ motif could fit into the hydrophobic IgFLNa21 CD groove without steric clashes (Fig. 5D). Binding is first mediated by hydrogen bonding between the ASB2␣ and IgFLNa21 C ␤-strand backbones and by a side chain/main chain hydrogen-bond between ASB2␣ Ser-11 and IgFLNa21 Ala-2281. The interaction is then anchored by a dense network of stacking interactions between ASB2␣ Phe-13 and IgFLNa21 Ile-2273 (C strand) and Ile-2283 and Phe-2285 (D strand). The complex is further stabilized by interactions involving residues specific to ASB2␣, such as Leu-12 and His-14, both stacked with IgFLNa21 Ala-2272.
Tyr-9, Ser-11, and Phe-13 of ASB2␣ Are Required to Target FLNa to Degradation-Because we previously showed that colocalization of ASB2␣ with F-actin may be the result of ASB2 association with FLNs, we examined subcellular localization of ASB2␣ proteins mutated within the putative IgFLNa-binding site in transfected HeLa cells. In contrast to GFP-ASB2␣, GFP-ASB2␣Y9F, GFP-ASB2␣S11D, and GFP-ASB2␣F13E did not co-localize with FLNa to stress fibers (Fig. 6A). To better quantify these observations, we measured the co-localization correlation coefficient between ASB2␣ and FLNa staining. The Pearson coefficients for GFP-ASB2␣ expressing cells was 0.539 Ϯ 0.117, for GFP-ASB2␣Y9F expressing cells, 0.006 Ϯ 0.085; for GFP-ASB2␣S11D expressing cells, Ϫ0.003 Ϯ 0.107; and for GFP-ASB2␣F13E expressing cells, Ϫ0.050 Ϯ 0.129. To understand the structural determinants of these observations, ASB2␣S11Dand F13E-mutated peptides were modeled and docked into the IgFLNa21 hydrophobic CD groove as described for the wild-type ASB2␣ N-terminal peptide (supplemental Fig.  S2, A and B). In both cases, there was hydrogen pairing between the ASB2␣-mutated motifs (residues 8 -16) and the C ␤-strand backbones, but the respective complexes presented different degrees of stability relatively to the wild-type complex, as estimated by an energy score. ASB2␣F13E mutation was the most  deleterious, inducing a marked decrease in the van der Waals energy absolute value, well in line with the loss of the critical hydrophobic contacts between ASB2␣ Phe-13 and IgFLNa21 Ile-2273, Ile-2283, and Phe-2285. Surprisingly, the ASB2␣S11D mutation caused a slight increase in the van der Waals energy absolute value, partially compensating for the decrease in the electrostatic energy absolute value, in correlation with a subtle reorganization and stabilization of the hydrophobic cluster. In addition, there was still a possibility of a side chain/main chain hydrogen bond between Asp-11 and IgFLNa21 Ala-2281.
Remarkably, it was impossible to obtain a low energy conformation for the ASB2␣Y9F-IgFLNa21 complex, probably due to the observed perturbation in the initial orientations of the mutated motif side chains. These results further suggest that ASB2␣ residues 8 -16 comprise a determinant for ASB2␣ binding to FLNa. Accordingly, mutation of Tyr-9, Ser-11, or Phe-13 in ASB2␣ abrogated its ability to induce degradation of endogenous FLNa in transfected HeLa cells (Fig. 6B).
Although the ASB2␣Y9F protein has intrinsic E3 ubiquitin ligase activity (Fig. 6C), it does not induce degradation of endogenous FLNa in PLB985/MT-ASB2␣Y9F cells induced to express ASB2␣Y9F with zinc (Fig. 6D). We then assessed whether mutation of one of these key residues, Tyr-9, disrupted the ability of ASB2␣ to stabilize adhesion of hematopoietic cells. Indeed, in contrast to cells expressing wild-type ASB2␣, adhesion of ASB2␣Y9F-expressing cells was not sustained following integrin activation by Mn 2ϩ and washes with PBS alone (Fig. 6E). Furthermore, expression of ASB2␣Y9F but also of ASB2␣S11D and ASB2␣F13E did not affect the spreading of transfected NIH3T3 cells on fibronectin-coated slides (Fig. 6F) as previously observed for ASB2␣ E3 ligase defective mutants (24). Collectively, our data indicated that ASB2␣ residues 8 -16 encompass a major determinant for FLNa binding, subsequent polyubiquitylation and degradation by the proteasome, and regulate cell spreading and cell adhesion.

DISCUSSION
Physical contact between leukemia cells and the bone marrow microenvironment provides a refuge for minimal residual disease. Of importance, interaction of ␤1 integrins with fibronectin is involved in acquired chemoresistance of AML cells (10,11,37). In this context, deciphering the molecular mechanisms controlling integrin-dependent adhesion of normal hematopoietic and leukemia cells may ultimately lead to new treatment strategies that specifically target leukemia cells. Although players that activate integrins have been described, few players that inhibit integrins have been identified so far (38,39). Among them, FLNa has been proposed to compete with talin for binding to the cytoplasmic tail of integrin ␤ subunits (32). We recently demonstrated that ASB2␣ regulates FLNa functions via proteasomal degradation of FLNa (24), suggesting that ASB2 may contribute to integrin activation. We therefore assessed whether ASB2␣, through FLNa degradation, plays a role in the regulation of integrin-dependent functions in hematopoietic cells. We found that expression of ASB2␣ significantly sustained integrin-dependent adhesion of hematopoietic cells to fibronectin. Of importance, FLNa knockdown recapitulated ASB2␣ effects on hematopoietic cell adhesion and concomitant FLNb knockdown was not required. This is in contrast to the spreading or migration defects previously observed in adherent cells following ASB2␣ expression or FLNa and FLNb double knockdown (24,29). However, the levels of FLNb and FLNc are low in PLB985 cells (25) and so may be insufficient to compensate for the loss of FLNa in these cells. Our results are in agreement with previous work in HT1080 fibrosarcoma cells and Jurkat T lymphoblasts showing that loss of FLNs increased the percentage of non-motile cells plated on fibronectin (29).
The role of ubiquitin-mediated proteasomal degradation in the control of hematopoiesis has recently been highlighted by the fact that c-Myc stability is controlled by the SCF Fbw7 E3 ubiquitin ligase in hematopoietic stem cells (40). In fact, ubiquitin-mediated degradation is one of the major pathways for controlled proteolysis in eukaryotes. In this pathway, E3 ubiquitin ligases that determine the specificity of protein substrates represent a class of potential drug targets for pharmaceutical intervention. Although proteasome inhibition has proved to be of therapeutic utility, the strategy of modulating the activity of E3 ubiquitin ligases is more specific. In this regard, characterization of the various E3 ubiquitin ligases and their respective substrates and understanding the signals that regulate specific ubiquitin ligation events should contribute to the development of new therapies that target the ubiquitin system. Our data provide evidence that the N-terminal region specific to the hematopoietic isoform of ASB2 plays roles in the targeting of FLNa to proteasomal degradation. Indeed, deletion of this domain or mutation of Tyr-9, Ser-11, or Phe-13 abolished the recruitment of ASB2␣ to actin stress fibers and completely abrogated the ability of ASB2␣ to induce FLNa degradation. Of interest, mutation of Tyr-9 of ASB2␣ abolished ASB2␣ effects on adherent cell spreading and adhesion of hematopoietic cells. The most striking feature of this N-terminal region is that it encompasses residues (8 to 16) that share structural homology with the binding domains of several IgFLNa21 ligands (31)(32)(33)(34).
However, our results indicate that, in addition to this region, ankyrin repeats 1 to 10 of ASB2␣ are necessary for in vivo colocalization with FLNa. Interestingly, we further showed that the ASB2␣ANK(1-10) protein, which contains the N-terminal region, can bind directly to the IgFLNa21 domain. As expected, mutation of Ala-2272 and Ala-2274 in strand C of IgFLNa21 strongly inhibited ASB2␣ANK(1-10) binding as previously observed for ␤7 integrin or migfilin (31,32,34), suggesting that ASB2␣, integrin ␤7, and migfilin may bind to a similar site on IgFLNa21. However, because the whole group A of IgFLNa repeats can bind a set of ligands including ␤ integrins and migfilin (34), we do not exclude that other regions of FLNa may also contribute to the ASB2␣-FLNa interaction.
Our results reinforce the view that the CD face of IgFLNa is a common binding interface for FLN partners and suggest that ASB2␣ may compete with FLN partners such as ␤ integrins and migfilin for FLNa binding. We therefore cannot exclude the possibility that ASB2␣ affects integrin-dependent functions through the dissociation of FLNa partners from FLNa. In this context, it should be mentioned that as observed in FLN-depleted cells following ASB2␣ expression or FLN knockdown, cells depleted in migfilin exhibit less cell spreading (41). Moreover, it is worth noting that binding of migfilin to FLNa may promote integrin activation by dissociating FLNa from integrins (42). Furthermore, displacement of FLNa from integrin tails or from migfilin by ASB2␣ may allow FLNa polyubiquitylation and subsequently, acute proteasomal degradation of all FLNa molecules. We have previously demonstrated that ASB2␣ induces degradation of all three filamins (29), whereas ASB2␤ induces degradation of FLNb but not FLNa (21,22). Our findings that the N-terminal region specific to the ASB2␣ isoform is required for FLNa degradation, and that ASB2␣-ANK(1-10) preferentially binds IgFLNa21, may help to explain the specificity of ASB2 proteins toward FLNa and FLNb. It will nonetheless be important to identify residues involved in the molecular interactions between ASB2 proteins and FLNa and FLNb to further our understanding of the specificity of ASB2 proteins toward FLNs.
In conclusion, our structural and cell biology studies have revealed a region of ASB2␣ that is involved in the recruitment of its substrate, FLNa. By inducing FLNa degradation by the proteasome, ASB2␣ may regulate integrin-dependent functions and thus hematopoietic stem cell fate within the niche.