Mammalian Actin-binding Protein-1/Hip-55 Interacts with FHL2 and Negatively Regulates Cell Invasion*

Mammalian actin-binding protein-1 (mAbp1) is an adaptor protein that binds actin and modulates scission during endocytosis. Recent studies suggest that mAbp1 impairs cell invasion; however, the mechanism for the inhibitory effects of mAbp1 remain unclear. We performed a yeast two-hybrid screen and identified the adaptor protein, FHL2, as a novel binding partner that interacts with the N-terminal actin depolymerizing factor homology domain (ADFH) domain of mAbp1. Here we report that depletion of mAbp1 or ectopic expression of the ADFH domain of mAbp1 increased Rho GTPase signaling and breast cancer cell invasion. Moreover, cell invasion induced by the ADFH domain of mAbp1 required the expression of FHL2. Taken together, our findings show that mAbp1 and FHL2 are novel binding partners that differentially regulate Rho GTPase signaling and MTLn3 breast cancer cell invasion.

Here we report a novel interaction between the N-terminal ADFH domain of mAbp1 and FHL2 that modulates cell invasion. In the yeast, but not the mammalian form, the N terminus of Abp1 binds Arp2/3 (2,28). We have now identified a novel binding partner that interacts with the N terminus of mAbp1. We found that depletion of mAbp1 or ectopic expression of the ADFH domain enhances cell invasion. Moreover the increased invasion observed with ectopic expression of the ADFH domain requires expression of FHL2. Our findings suggest that mAbp1 inhibits invasive migration of breast cancer cells at least in part through its interaction with FHL2 and the modulation of Rho GTPase signaling. These results identify a novel role for the interaction between mAbp1 and FHL2 in modulating MTLn3 breast cancer cell invasion.
Cell Transfections-Transient transfections of MTLn3 cells were performed with Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions. In brief, 1 ϫ 10 5 cells were cultured in a 6-well plate, and 2 g of DNA and 3 l of Lipofectamine 2000 reagent were used. Transfection mixtures incubated at room temperature in Opti-MEM media (Gibco for 30 min before adding to cells. Cells were incubated with the transfection mixture 4 h before washing 3ϫ with PBS. Cells then were cultured by standard methods as described under "Cell Culture" above. HEK 293 transient transfections were performed by the standard calcium phosphate precipitation method. All cells, with the exception of HEK 293 (cultured at 10% CO 2 ), were maintained at 5% CO 2 at 37°C.
DNA and siRNA Constructs-All constructs were generated by PCR amplification of product with cut sites in the primers as follows: full-length human mAbp1 and the ADFH, proline-rich region, or SH3 domains of mAbp1 were cloned into the pEGFP-C1 vector using the XhoI and BamHI cut sites. GFP-mAbp1-W415K was generated by site-directed mutagenesis of the wild-type construct. GST-mϩAbp1 was cloned into pGex-4T1 with BamHI and XhoI cut sites. RFP-FHL2 was cloned into tagRFP-C1 vector with HindIII and BamHI cut sites. Myc-FHL2 was cloned into pcDNA3.1 myc vector with XhoI and EcoRI cut sites and the myc tag on the N terminus of FHL2. His-FHL2 was cloned into pTrcHisA vector with XhoI and EcoRI cut sites.
Retroviral Transduction of Cells-Stable knockdown MTLn3 lines were generated with the pSUPER.retro RNAi system per the manufacturer's instructions (OligoEngine, Seattle, WA). Briefly, to generate retrovirus for infection of MTLn3 cells, we plated 3 ϫ 10 6 PhxA cells on a 10-cm dish. The PhxA cells were transfected with 12 g of DNA using the calcium phosphate precipitation procedure. 24-h virus was discarded, and 4 ml of new media were added. The 48-h and 72-h retrovirus was harvested, filtered through a 0.2-m syringe filter, and mixed with 1ϫ Polybrene before adding to a 6-cm dish of 1 ϫ 10 5 MTLn3 cells. Cells were incubated in retrovirus mixture for 6 h at 32°C. Retroviral media was removed, and cells were recovered in standard media for 24 h before selection. Cells containing retrovirus were selected with 1 g/ml puromycin (Sigma) for 5 days. Retroviral transduction of MDA-MB-231 cells was performed similarly with the exception that virus from 24, 48, 72, and 96 h was used for infection and virus-containing media remained on cells 24 h instead of 6. This longer infection was necessary for successful transduction of these cells.
Collagen Contraction-Collagen contraction experiments were performed as follows. Desired volume of collagen was mixed 1:1 with 150 mM HEPES (Cellgro) in 2ϫ PBS for a stock concentration of 4.5 mg/ml collagen and incubated on ice. 1 ϫ 10 6 MDA-MB-231 cells were counted and resuspended in 656 l of media. 444 l of collagen stock was mixed with cells for a final collagen concentration of 2 mg/ml and immediately transferred to 1 well of a 6-well plate. The plate was rocked to spread collagen across the entire surface then cultured at 37°C for 1 h to polymerize. When polymerized, 2 ml of media were added to well, and collagen was loosened from the well with gentle rocking. Cells were cultured, and images were taken at 24 h and 48 h. Percent contraction was quantified on metamorph imaging software.
Immunoblotting and Immunoprecipitation-Cells were scraped into lysis buffer (20 mM HEPES, 50 mM KCl, 1 mM EDTA, 1% Nonidet P-40, 0.2 mM PMSF, 1 g/ml pepstatin, 2 g/ml aprotinin, and 1 g/ml leupeptin) on ice and clarified by centrifugation. Protein concentrations were determined with a bicinchoninic acid protein assay kit (Thermo Fisher Scientific, Waltham, MA) according to the manufacturer's instructions. Equal protein levels were denatured in SDS and loaded on mAbp1 Inhibits FHL2 Cell Invasion 4 -20% SDS-polyacrylamide gradient gels. Separated proteins were transferred to nitrocellulose, probed with the appropriate antibodies, and imaged with an Odyssey Infrared Imaging System (Li-Cor Biosciences, Lincoln, NE). For GFP immunoprecipitation experiments, transiently transfected HEK 293 cells were cultured to ϳ80% confluency on 10-cm plates, washed once with PBS, and lysed in lysis buffer. Lysates were clarified by centrifugation and incubated with 3 g of rabbit-anti-GFP antibody. Immune complexes were captured on Gammabind G-Sepharose beads (GE Healthcare), washed three times in lysis buffer, and analyzed by immunoblotting.
Rho Pulldown Assay-Rho GTPase pulldown assays were performed with Rhotekin-RBD Protein GST beads (Cytoskeleton, Inc.) according the manufacturer's instructions with some modifications. In brief, 3 ϫ 10 6 MTLn3 cells were cultured on 15-cm dishes, serum-starved for 24 h, then treated with 5 nM EGF for 5 min. Cells were immediately lysed on ice in Rho lysis buffer (50 mM Tris, pH 7.5, 10 mM MgCl 2 , 100 mM NaCl, 1% Triton X-100, and 1% PMSF, leupeptin, and aprotinin) and centrifuged at 14,000 rpm for 5 min for clarification. A sample of lysate was removed for Western blot loading control, and the remainder of lysate was placed in a tube containing 60 g of Rhotekin-RBD beads to capture active Rho. Tubes were incubated for 30 min at 4°C before centrifuging for 1 min at 1000 rpm and washed 3ϫ with wash buffer (25 mM Tris, pH 7.5, 30 mM MgCl 2 , 40 mM NaCl, and 1% PMSF, leupeptin, and aprotinin). Meanwhile, the positive and negative control lysates were incubated at room temperature for 15 min with 15 mM EDTA and 0.2 mM GTP␥S or 1 mM GDP, respectively. GTP␥S is a non-hydrolysable GTP analog so that active Rho is pulled down from cell lysates. GDP inactivates Rho and blocks its binding to rhotekin RBD beads. MgCl 2 (60 mM final concentration) was added to stop all reactions, and lysates were incubated on rhotekin beads and washed as described above. All samples were run on SDS-PAGE. The amount of Rho pulled down was calculated by densitometry analysis of Western blots. The ratio of active Rho pulled down to input Rho was calculated first. Then the ratio of active Rho pulled down per condition was normalized to control shRNA cells.
Cell Fractionation-Cell fractionation was performed as previously described (32). In brief, cells were scraped from confluent 10-cm dishes on ice into 1 ml of ice-cold PBS. Cells were briefly centrifuged, and supernatant was removed. Cells were then resuspended in 900 l of 0.1% Nonidet P-40 in PBS and titrated 3ϫ. 300 l of lysate was removed as the whole cell lysate. Remaining lysate was briefly centrifuged, and supernatant was removed as the cytosolic fraction. Nuclear pellet was resuspended in 180 l of SDS buffer. Samples were loaded on SDS page and analyzed on Western blot.
Immunofluorescence-Coverslips were acid-washed, ethanol-sterilized, coated with 2 mg/ml gelatin, and subsequently coated with 10 g/ml fibronectin overnight at 4°C. 3.0 ϫ 10 4 cells were plated in ␣-minimum Eagle's medium on coverslips for 6 h at 37°C and 5% CO 2 . Cells were fixed with 3.7% formaldehyde for 10 min, quenched with 0.15 M glycine for 10 min, permeabilized with 0.2% Triton X-100 for 10 min, washed with PBS 3 times, and blocked with 5% goat serum for 1 h or overnight at 4°C. Cells were incubated with primary antibody and/or rhodamine phalloi-din for minimum of 1 h, washed, then incubated with fluorophoreconjugated secondary antibodies for 1 h.
Cells were imaged either with a Zeiss Observer Z.1 inverted microscope or with a Nikon Eclipse TE300 inverted microscope. On the Zeiss Observer Z.1 inverted microscope, images were captured with a Coolsnap ES 2 camera or with an Evolve 512 Delta EM-CCD camera for confocal mode (Photometrics, Huntington Beach, CA) using a 63ϫ oil/NA 1.4 Plan Apo immersion objective (Zeiss Immersol 518F oil) and processed with Zen (blue edition) software (Carl Zeiss). Images from the Nikon Eclipse TE300 were captured with a Coolsnap ES 2 camera using a 60ϫ oil/NA 1.4 Plan Apo oil immersion objective and processed with Zen imaging software.
Invasion Assays-Cells were serum-starved for 24 h, and 2 ϫ 10 5 cells were seeded into top chamber of 8-m pore boyden chambers coated with Matrigel (BD Biosystems). Serum-starving medium for MTLn3 cells contained 1% FBS and antibiotics. Serum-starving medium for other cell types contained 0.2% fatty acid free BSA and no serum. Cells were allowed to invade for 16 or 48 h as indicated toward serum-containing media. Cells were removed from the top chamber, and cells remaining on the membrane were stained with a Hema-3 stain kit (Thermo Fisher Scientific). Cells were counted from 12 representative 20ϫ field of view images per condition from at least three independent experiments. All images were acquired with an Olympus 1X-70 inverted microscope (Olympus America, Melville, NY) using a 20ϫ/NA 0.4 objective. Images were processed with Metavue imaging software v6.2 (Universal Imaging, Downingtown, PA).
Statistical Analyses-Statistical analyses used were unpaired t test or one-way analysis of variance (ANOVA) with Tukey post tests. p Ͻ 0.05 was considered significant.

Results
mAbp1 Inhibits Invasion of Breast Cancer Cells-We previously reported that mAbp1 impairs the migration of Src-transformed NIH 3T3 cells (8). To investigate whether mAbp1 also negatively regulates invasion of breast cancer cells, we depleted mAbp1 in MTLn3 and MDA-MB-231 breast cancer cell lines using retroviral shRNA (Fig. 1, A, B, and H). Cell invasion through Matrigel-coated boyden chambers was enhanced 2-4-fold in mAbp1-deficient breast cancer cell lines ( Fig. 1, C, D, and I). This enhanced invasion was rescued to control levels with exogenous transient expression of GFP-mAbp1 in mAbp1-deficient MTLn3 cells (Fig. 1F). GFP-tagged mAbp1 localized to the cytoplasm and perinuclear regions (Fig. 1G) similar to endogenous mAbp1 as detected by immunofluorescence (Fig. 1B). Taken together, these findings identify an inhibitory role for mAbp1 during invasive migration of breast cancer cell lines.
Invasive migration is dependent on Rho GTPase activity and subsequent cell contractility. To determine how mAbp1 impairs cell invasion, we used MDA-MD-231 cells that grow well in three-dimensional collagen gels to examine how mAbp1 expression modulates cell contraction. We found that in accordance with increased invasion, depletion of mAbp1 increased the contraction of collagen gels, indicating that mAbp1-deficient cells have enhanced cell contractility and force application in three-dimensional collagen (Fig. 1, J, and  K). Likewise, mAbp1-deficient cells had higher RhoA activity (Fig. 2, A and B). Cell forces are applied to the extracellular matrix via adhesion sites known as focal adhesions. We found that mAbp1-deficient cells had larger focal adhesions and were more spread on two-dimensional surfaces (Fig. 2, C-F). Taken together, these results suggest that mAbp1 negatively regulates the invasion of breast cancer cells, at least in part through the modulation of Rho GTPase signaling. The Inhibitory Effects of mAbp1 on Cell Invasion Requires the C-terminal SH3 Domain-Mammalian Abp1 is an adaptor protein that binds to the actin cytoskeleton through its N-terminal ADFH domain and to dynamin (2,33,34), WIP (7), and other proteins through its C-terminal SH3 domain (Fig. 3A). To determine whether the SH3 domain is necessary for the inhibitory effects of mAbp1 on invasion, we expressed either fulllength mAbp1, the ADFH domain alone, or an SH3 domain mutant (mAbp1-W415K) in mAbp1-deficient MTLn3 cells and measured invasion through Matrigel (Fig. 3, A-D). mAbp1depleted cells had increased invasion, and expression of WT GFP-mAbp1 rescued this effect. Interestingly, expression of the GFP-mAbp1-W415K or the GFP-ADFH domain alone was not sufficient for rescue (Fig. 3, B-D). A caveat is that the expression levels of the ADFH domain were low in the rescue lines as shown by immunostaining (Fig. 3B). Regardless, these results    ). B, control or mAbp1 shRNA MTLn3 cells were transiently transfected with GFP constructs, and lysates were probed for mAbp1, GFP, and actin as a loading control. The Western blot was cut to immunostain GFP and mAbp1 separately; however, GFP staining of all constructs can be seen in Fig. 4D. The arrow denotes endogenous mAbp1 expression in control and mAbp1 shRNA cells. C, transiently transfected MTLn3 cells were serum-starved for 24 h, seeded in Matrigel coated boyden chambers, and allowed to invade toward serum-containing media for 16 h. Representative images of invaded cells at 20ϫ magnification. Scale bar ϭ 100 m. D, quantification of invasion of transiently transfected control or mAbp1-deficient cells. Average number of cells invaded from 12 images taken at 20ϫ magnification. n ϭ 4. **, p ϭ 0.0067; ***, p ϭ 0.0010; **, p ϭ 0.0080; ***, p ϭ 0.0003. E, MTLn3 cells overexpressing GFP, GFP-mAbp1, or GFP-mAbp1-W415K were serum-starved then treated with 5 nM EGF for 5 min. Active Rho was pulled down with Rhotekin binding beads and probed for RhoA and actin as a loading control. Overexpression of mAbp1, but not the SH3 mutant mAbp1-W415K, reduced RhoA activity in mAbp-deficient cells. F, quantification of active RhoA pulled down by densitometry analysis. The amount of active RhoA was normalized to control; n ϭ 3. **, p ϭ 0.0056. ns, not significant. Error bars represent S.E.

mAbp1 Inhibits FHL2 Cell Invasion
suggest that a functional SH3 domain is necessary for the inhibitory effects of mAbp1 on invasion. We next asked if a functional SH3 domain was required for mAbp1 inhibition of RhoA activity. We overexpressed GFP, GFP-mAbp1, or GFP-mAbp1-W415K in MTLn3 cells and pulled down active Rho from lysates (Fig. 3, E and F). We found that RhoA activity was decreased with mAbp1 overexpression but not with mAbp1-W415K, suggesting that the SH3 domain interactions are required for the regulation of Rho activity by mAbp1.
The N-terminal ADFH Domain of mAbp1 Interacts with FHL2-To determine how mAbp1 inhibits invasive migration, we performed a yeast two-hybrid screen with full-length human mAbp1 and identified several novel binding partners ( Table 1). One protein of particular interest was FHL2, as it has been implicated in breast cancer progression. To confirm the interaction between mAbp1 and FHL2, we performed in vitro GST pulldown assays (Fig. 4A). Beads coated with GST alone did not pull down His-tagged FHL2, while beads coated with GST-mAbp1 interacted with His-FHL2. To confirm the interaction in mammalian cells, we performed a co-immunoprecipitation experiment by pulling down endogenous FHL2 and probing for mAbp1 (Fig. 4B). In addition, we expressed GFP or GFP-mAbp1 with myc-FHL2 in HEK 293 cells and immunoprecipitated GFP; myc-FHL2 interacted with GFP-mAbp1 but not GFP alone (Fig. 4C).
To determine which mAbp1 domain interacts with FHL2, we expressed either the GFP-tagged ADFH domain, proline-rich region, or the SH3 domain alone, along with His-FHL2 and performed co-immunoprecipitation experiments (Fig. 4D). Surprisingly, we found that FHL2 interacted with the ADFH domain but not the SH3 or proline-rich region regions of mAbp1. Moreover, the interaction between FHL2 and the ADFH domain was more robust than FHL2 interaction with full-length mAbp1. This result suggests that conformation of full-length mAbp1 may modulate FHL2 binding. Although GFP-ADFH domain expression was low relative to full-length GFP-mAbp1 (as shown by immunofluorescence), immunoprecipitation of GFP enriched GFP-ADFH to comparable levels (Fig. 4D). Taken together, these findings suggest that FHL2 interacts directly with the ADFH domain of mAbp1. Interestingly, the ADFH domain of mAbp1 is not conserved in cortactin, thus providing a key distinction between the binding partners of mAbp1 and cortactin. It is an intriguing idea that FHL2 interactions may play a role in the differential effects of mAbp1 and cortactin on cell invasion.
FHL2 Is Necessary for Invasive Migration of Breast Cancer Cells-FHL2 expression is up-regulated in breast cancer (20,21); however, the role of FHL2 during breast cancer cell invasion is not known. To determine how FHL2 affects invasive migration of breast cancer cells, we depleted FHL2 in MTLn3 cells using retroviral shRNA (Fig. 5, A and B) and found that even with a partial knockdown, invasive migration was significantly impaired (Fig. 5, C and D). To confirm the role of FHL2 during invasive migration, FHL2-deficient cell invasion was rescued with transient exogenous expression of RFP-FHL2 (Fig. 5, E-G). Consistent with previous reports, exogenous RFP-FHL2 localized at focal adhesions and along stress fibers in MTLn3 cells (Fig. 5H).
To determine if FHL2 modulates focal adhesions, we imaged focal adhesions and stress fibers in control and FHL2-deficient cells. Similar to exogenous RFP-FHL2 expression, endogenous immunofluorescence of FHL2 was localized to focal adhesions FIGURE 4. mAbp1 interacts with FHL2 through its ADFH domain. A, an in vitro GST pulldown assay was performed with purified His-tagged FHL2 and GST beads or GST-tagged mAbp1 beads. His-tagged FHL2 interacted with GST-mAbp1 but not GST beads alone. His-tagged FHL2 was probed using anti-His antibody. The presence of GST and GST-mAbp1 (arrow) was analyzed by staining the blot with Amido Black stain (lower panel). B, co-immunoprecipitation (IP) of endogenous FHL2 and mAbp1 from MTLn3 cell lysate. FHL2 was immunoprecipitated with anti-FHL2 but not with IgG alone, and co-immunoprecipitated mAbp1 was probed by Western blot. C, co-immunoprecipitation of mAbp1 and FHL2 overexpression in mammalian cells. GFP-mAbp1 and myc-FHL2 were transiently overexpressed in HEK-293 cells. Cells were lysed and incubated with GFP antibody overnight, then harvested with gamma-G-Sepharose beads. myc-FHL2 did not pull down with GFP alone but did interact with GFP-mAbp1. The blot was probed for GFP and myc. The arrow indicates IgG band. D, the ADFH domain of mAbp1 interacts with FHL2. GFP constructs containing full-length mAbp1, the ADFH domain, the proline-rich region, or the SH3 domain alone were transiently expressed with myc-FHL2 in HEK 293 cells. GFP antibody was used to pull down GFP fusion proteins, and immunoblotting was performed to detect co-immunoprecipitated myc-FHL2. The arrow indicates IgG band. All experiments were performed in triplicate. and stress fibers (Fig. 6A). We found that FHL2-deficient MTLn3 cells had reduced focal adhesion area, although the total numbers of focal adhesions were not significantly different (Fig. 6, B and C). Accordingly, we also found that FHL2-deficient cells had reduced RhoA activity by pulldown assay in the presence of EGF stimulation (Fig. 6, D and E). These findings are consistent with the reported roles for FHL2 in heart muscle and contractility of the sarcomere (35), wound healing (36), and the invasion of glioblastoma cells (25,26). These findings sug-gest that FHL2 is necessary for efficient invasion of MTLn3 breast cancer cells, likely in part through its effects on Rho GTPase activity.
The ADFH Domain of mAbp1 Increases Invasive Migration, and This Effect Requires FHL2-To determine whether mAbp1 affects invasion through its interaction with FHL2, we overexpressed WT mAbp1, ADFH domain alone, and mAbp1-W415K and assessed invasion through Matrigel in control and FHL2-deficient cells (Fig. 7). As expected, overexpression of

mAbp1 Inhibits FHL2 Cell Invasion
full-length GFP-mAbp1 in wild-type cells impaired invasive migration. By contrast, ectopic expression of the ADFH domain alone dramatically increased invasive migration of MTLn3 cells (Fig. 7, B and C), suggesting that the ADFH domain may block the inhibitory effects of endogenous mAbp1 on invasion. In support of this idea, we found that overexpression of the SH3 domain mutant (GFP-mAbp1-W415K) also enhanced invasive migration. Our results confirm the hypothesis that mAbp1 binds FHL2 through the ADFH domain and inhibits FHL2 invasive activity via its SH3 domain or its interactions. Although the enhanced invasion of full-length mAbp1-W415K-expressing cells was not as dramatic as expression of the ADFH domain alone, this difference may be due to reduced levels of FHL2 interaction with full-length mAbp1 in comparison with the ADFH domain alone (Fig. 4D).
To determine how the interaction between mAbp1 and FHL2 modulates invasive cell migration, we determined if expression of FHL2 was necessary for the invasive migration induced by expression of the ADFH domain of mAbp1. We found that FHL2 expression was required for the induction of invasive migration with ectopic expression of the ADFH domain or W415K-mAbp1 mutant (Fig. 7, B and C). However, a potential caveat is that invasive migration may be dependent on FHL2 expression independent of mAbp1 effects. Taken together, these findings support the idea that mAbp1 sequesters FHL2 and impairs FHL2-Rho signaling to affect cell invasion. Furthermore, these results suggest that the inhibitory effect of mAbp1 on invasion requires the SH3 domain and its specific binding partners.

Discussion
Here we report a novel interaction between mAbp1 and the multifunctional adaptor protein FHL2. We show that mAbp1 impairs MTLn3 cell invasion and Rho GTPase signaling, while FHL2 increases MTLn3 invasion and RhoA activity. We show that mAbp1 interacts with FHL2 through its N-terminal ADFH domain. Our data suggest that mAbp1 regulates cell invasion at least in part through its effects on FHL2 and Rho GTPase activation. In support of this idea, expression of the ADFH domain of mAbp1, lacking the inhibitory SH3 domain, is sufficient to induce invasive cell migration, and this effect requires the expression of FHL2.
FHL2 is an adaptor protein that has known functions at focal adhesions and also may function as a transcriptional cofactor in the nucleus. The interaction between mAbp1 and FHL2 is particularly interesting as FHL2 is up-regulated during invasive migration of metastatic osteosarcoma cells, colon cancer, and glioblastomas (25,26,38) and has been implicated in migration during wound closure (20,21,25,36). In colon cancer, FHL2 promotes epithelial to mesenchymal transition by suppressing E-cadherin and promoting vimentin and MMP-9 expression (25,39), while in vitro, FHL2 has been shown to alter transactivation of key breast cancer related genes (13,17,18). Although FHL2 expression can alter the invasive capacity of cancer cells, it remains unclear whether FHL2 mediates its effects on invasion through focal adhesions or its transcription cofactor activity. Our findings suggest that FHL2 and mAbp1 regulate cell invasion through effects on focal adhesions and the activation of Rho GTPase signaling.
Cell motility and invasion are dependent on the tight regulation of Rho-ROCK signaling and cell contractility (40 -43).
Inhibition of Rho or ROCK impairs breast cancer invasion (44) and metastasis in vivo (45). Our work suggests that the actinbinding protein mAbp1 and FHL2 play opposing roles in modulating RhoA activity and cell invasion. We found that mAbp1 inhibits cellular contraction of collagen and inhibits RhoA activity while FHL2 facilitates RhoA activation. Not surprisingly, the activity of RhoA correlated with the invasive potential of mAbp1-and FHL2-depleted breast cancer cell lines. Future work will be needed to determine whether FHL2 and its interaction with mAbp1 regulate actomyosin contraction and invasive migration in three-dimensional collagen gels or within tissues in vivo.
Our findings suggest that the SH3 domain of mAbp1 is necessary for the inhibitory effects of mAbp1 on cell invasion. Moreover, our results suggest that mAbp1 may inhibit invasion by sequestering FHL2 and that this inhibitory effect requires the SH3 domain of mAbp1. Interestingly, we found that FHL2 interacted more strongly with the ADFH domain of mAbp1 compared to full-length mAbp1, suggesting that the conformation of full-length mAbp1 may regulate the interaction between mAbp1 and FHL2. Overexpression of the ADFH domain in cells with endogenous mAbp1 resulted in enhanced invasion FIGURE 7. FHL2 modulates invasion induced by the ADFH domain of mAbp1. A, MTLn3 lysates from control or FHL2 shRNA cells overexpressing GFP, GFP-mAbp1, GFP-ADFH, or GFP-mAbp1-W415K and probed for GFP, endogenous mAbp1, FHL2, and actin by immunoblotting. B, cells were serum-starved 24 h, seeded on Matrigel invasion chambers, and allowed to invade toward serum-containing media for 16 h. Representative images of invaded cells. Scale bar ϭ 100 m. C, quantification of cell invasion. Shown is the percentage of cells invaded relative to control from 12 images taken at 20ϫ magnification. Data are from independent experiments. **, p ϭ 0.0067; **, p ϭ 0.0080; ***, p Ͻ 0.0030. mAbp1 Inhibits FHL2 Cell Invasion dependent on FHL2 expression, suggesting that the ADFH domain was sufficient to overcome the inhibitory effects of fulllength mAbp1 on cell invasion. An increased affinity of the ADFH domain for FHL2 may be sufficient to block endogenous mAbp1 binding to FHL2 thereby modulating FHL2 function and Rho GTPase signaling.
These results suggest that the mAbp1-FHL2 interaction may be regulated by conformational changes in mAbp1 potentially through phosphorylation, additional protein interactions, SH3 domain modifications, or protease activity. Recently, putative phosphorylation sites have been identified by mass spectroscopy of His-tagged mAbp1 in HEK 293 cells (46). Serine 99 and threonine 114 are both located in the ADFH domain and are attractive targets that may alter FHL2 binding. mAbp1 serine 269 is also an interesting potential regulatory target since phosphorylation of mAbp1 serine 269 by the maternal embryonic leucine zipper kinase (MELK) promotes tumor growth and invasion of MCF-7 breast cancer cells. Our work is the first to suggest that endogenous mAbp1 functions as an inhibitor of cancer cell invasion. It will be interesting to determine whether phosphorylation of serine 269 affects the interaction between mAbp1 and FHL2 and cell invasion (47). Furthermore, it is intriguing that mAbp1 serine 269 is close to the calpain-2 cleavage sites of mAbp1 (7,30). Calpain-2 activity is required for several cancer lines including breast cancer (29,30,48,49). Calpain-2 cleavage of mAbp1 would effectively separate the N-terminal ADFH domain from the regulatory C-terminal SH3 domain and thus could alter cell invasion, possibly through ADFH interactions with FHL2. These possible modes of regulation of the mAbp1-FHL2 interaction remain to be tested and would provide further insight into potential targets that modulate the invasive migration of cancer cells.