Periostin/Filamin-A: A Candidate Central Regulatory Axis for Valve Fibrogenesis and Matrix Compaction

Figures

• 

  FIGURE 1. Protein periostin (PN) promotes phosphorylation of filamin A (FLNA) in mitral valve (MV) sections and in valve interstitial cell (VIC) ex vivo cultures. (A) Labels: The elliptic boxes (a1, a2), indicate the aortic valves. The boxes on a1 and a2 indicate the regions of the mitral valve that are shown at higher magnification on the corresponding b1 and b2. LV, left ventricle; LA, left atria; Myo, myocardium; Ao, Aorta; PT, pulmonary trunk. Green color represents immunostaining for myosin heavy chain (MF20). Blue color represents the nuclear staining Hoechst 33342. Contiguous sections from comparable regions from the mitral valve cushions of the same litter of e17.5 WT (b1, b2) or PN null mice (c1, c2) were used for immunostaining with FLNA and phospho-filamin A [p-FLNA (S2152)]. The p-FLNA (S2152) antibody used gave a clear, strong signal in WT mitral valve cushions at e17.5, which correlated closely with the expression of total filamin A (FLNA). In contrast, PN null valves showed significantly decreased p-FLNA. (B) Upper panel: Cell lysates from VICs from e17.5 days were isolated from wild type (WT), and from PN null mice that were transiently transfected with lentivirus encoded…

• 

  FIGURE 2. Activations of Cdc42 and Rac1 are associated with Pak1 activation and actin cytoskeletal assembly in VIC ex vivo cultures. (A) Specific interactions of active Rac1-GTP or Cdc42-GTP with GST-fusion protein of GTPase-binding domain (PBD-GST) of human PAK1 wild type (WT, full length) were measured. VIC lysates were incubated with 100 μM GTPγS or GDP and clarified before precipitation on PBD-GST beads. Supernatants (S) were separated from bead pellets (P) by centrifugation and analyzed by SDS-PAGE using the indicated antibodies. (B) Upper panel: Transduced VICs with the indicated shRNAs were serum starved for 16 h and incubated without or with 1 μg/ml PN at 37°C for 1 h. Cdc42-GTP was pulled-down on GTP-loaded PBD-GST beads. The relative amounts of the activation of small GTPases were quantified by normalizing the amount of bound Cdc42 to total Cdc42 in the lysates. Lower panel: Bars represent densitometric ratios of active Cdc42/total Cdc42 in the lysates. (C) Upper panel: Transduced VICs with the indicated shRNAs were serum starved for 16 h and incubated without or with 1 μg/ml of PN at 37°C for 1 h. Rac1-GTPs were pulled-down on GTP-loaded PBD-GST beads. The relative amounts of…

• 

  FIGURE 3. Activations of Cdc42 and Rac1 are associated with PN, which induced both the autophosphorylation and the kinase activity of Pak1 in the VIC ex vivo cultures. (A) Serum-deprived cells were cultured in the presence of either DMSO or Src inhibitor PP2 (5 μM). Pull-down assays for active Cdc42 were done using GTP-loaded PAK-1-binding domain (PBD)-GST beads. (B) Cells were either uninfected (control), or infected with lentivirus-expressing vector, or Fak (Y397F) and analyzed for active Cdc42 as in Figure 2B. (C) VICs were either transfected with DN Pak1 or DN Cdc42 constructs for 48 h. Cells were treated with 1 μg/ml PN for 1 or 5 min at 37°C, and a nonradioactive kinase assay was done using the traditional in-gel phosphorylation assay. For details of this experiment see the “Materials and Methods” section. Pak1 kinase activity was evaluated by phosphorylation of substrate MBP (pMBP). The amount of immunoprecipitated PAK was controlled by anti-PAK1 immunoblotting. Isotype control IgG was used as experiment control. (D) Autophosphorylation of PAK (pPAK) was measured after 1- and 5-min treatments with PN [substrate 1 μg of myelin basic protein (MBP)] and was visualized as a 65-kDa…

• 

  FIGURE 4. PN/integrin β1-regulated Cdc42 activation is associated with stimulation of both the autophosphorylation and the kinase activity of Pak1 and actin cytoskeletal assembly in the VIC ex vivo cultures isolated from ED 17.5 mice mitral valves. (A) Upper panel: Cell lysates were harvested from growth-arrested VIC cultures that were either treated with or without 1 μg/ml of PN for 1 h (in serum-starved condition), or first treated with 5 μg/ml of both α5 and β1 integrin blocking antibodies for 2 h and then treated with 1 μg/ml of PN for 1 h. Cell lysates were prepared, and Cdc42-GTP was pulled-down from the lysates on GTP-loaded PBD-GST beads. The relative amounts of the activation of Cdc42 were quantified by normalizing the amount of bound Cdc42 to that of total Cdc42 in the lysates. Lower panel: Bars represent densitometric ratios of active Cdc42/total Cdc42 in the lysates. (B) Upper panel: Pak1 activation was tested by immunoprecipitation with anti-Pak and Western blotting with anti-p-Pak1 [Thr (T) 423]. The total amounts of p-Pak1 in the Pak1 IPs were determined by normalizing the amounts to total Pak1 by densitometry. Lower panel: Bars represent densitometric ratios of phospho…

• 

  FIGURE 5. The effects of PN/integrin β1 on association of Pak1 with FLNA, a substrate for Pak1 kinase activity in VIC lysates. (A,B) VICs were transfected with lentivirus encoding Myc-FLNA for 48 h and then serum starved for 24 h. They were then either untreated or treated with 1 μg/ml of PN for 1 h. (A) Cell lysates were immunoprecipitated without (lane 1, control IgG) or with an antibody against Myc (for FLNA) and immunoblotted with antibodies against Pak1, Myc, or FLNA. (B) VICs were pretreated with cytoskeleton disrupter 2 μM cytochalasin D (CytoD) for 30 min and then treated with 1 μg/ml PN for 1 h (lane 4) and compared with VICs treated with PN alone (lane 3). Cell lysates were immunoprecipitated with an antibody against Myc (for FLNA) and immunoblotted with antibodies against Pak1, Myc, or for FLNA in WCL. Lane 1 is for control IgG. (C) Upper Panel: VICs were cotransfected with the FLNA S2152A mutant, or DN Pak1 containing Pak1 autoinhibitory domain (residues 83–149), or DN Cdc42, or 5 μg/ml of α5 and β1 integrin blocking antibodies for 2 h. After 48 h, the cells were serum starved, or treated with or without 1 μg/ml of PN as in the experiment in Figure 2G. Pak1 was…

• 

  FIGURE 6. FLNA is a substrate of Pak1, and the effects of FLNA mutations [FLNA (P637Q) and FLNA (G288R)] on PN induced FLNA phosphorylation, on activation of Rac1, and on cell migration are shown. (A) VICs were transfected with FLNA shRNA. The stable clones were isolated and designated as FLNA-deficient VICs (VICs Δ FLNA) WT VICs and FLNA Δ VICs were transfected with vector or with wild-type FLNA lentiviral constructs for 72 h before treatment. These transfected VICs were then embedded in three-dimensional collagen matrices to analyze collagen lattice compaction (see the “Materials and Methods” section for details). (B) Transwell migration assays are shown for WT VICs and VICs Δ FLNA pre-transfected with the indicated lentiviral constructs. Percentages of cells seeded in the upper chamber of Transwell filters that migrated overnight into the lower chamber are shown. Numbers of migrated cells were counted on six fields from three experiments. Detailed descriptions are in the “Materials and Methods” section. (C) Western blot analyses are shown for FLNA and β-actin (internal control) proteins in the lysates from WT VICs and VICs Δ FLNA pretransfected with the indicated lentiviral constructs.…

• 

  FIGURE 7. Effects of FLNA mutations [FLNA (P637Q) and FLNA (G288R)] on PN-induced cell differentiation measured by collagen gel compaction. (A) WT VICs were first transfected with the indicated constructs for 72 h before treatment with PN (1 μg/ml for 1 h). These transfected VICs and untransfected PN null VICs were then embedded in three-dimensional collagen matrices. (B) Matrix contractilities from the experiments in (A) were measured as % area of gel contracted and plotted in panel (B). Results in panel (A,B) are representative of ≥3 separate experiments. Error bars represent SD. *p < 0.01 was considered significant.

• 

  FIGURE 8. Model for the role of PN/α5β1 integrin-mediated Fak-Rac1-Cdc42-Pak1-FLNA signaling in VIC survival and remodeling functions. The figure summarizes our findings in a model that proposes how periostin and FLNA function as a central regulatory axis for valve morphogenesis. Overall, we conclude that a PN/α5β1 integrin interaction with FLNA engenders a unique signaling pathway that is transduced through the activation of Fak1/S6K signaling to Cdc42-Pak1-dependent activation of FLNA (Figures 1–7). This results in changes in adhesion and cytoskeletal organization that are linked to morphogenetic remodeling and maturation of valve primordia.