Regulation of Cross-linked Actin Network (CLAN) Formation in Human Trabecular Meshwork (HTM) Cells by Convergence of Distinct β1 and β3 Integrin Pathways

Abstract

Purpose.: To determine the β1/β3 integrin-mediated pathways that regulate cross-linked actin network (CLAN) formation in human trabecular meshwork (HTM) cells. CLANs form in glaucomatous and steroid-treated TM cells, which may contribute to reducing outflow facility through the TM.

Methods.: Expression of CD47 (an αvβ3 integrin coreceptor/thrombospondin-1 receptor) and integrins αvβ3 and β1 was assessed by FACS. CLANs were induced by plating cells on fibronectin (a β1 integrin ligand) in the absence or presence of the β3 integrin-activating mAb AP-5 and were identified by phalloidin labeling. The role of Src kinases, PI-3 kinase (PI-3K), Rac1, and CD47 was determined by incubating cells with the inhibitors PP2 and EPA (Src kinases), LY294002 (PI-3K), or NSC23766 (Rac1). Tiam1 and Trio siRNAs and dominant-negative Tiam1 were used to determine which Rac1-specific guanine nucleotide exchange factor was involved. The role of CD47 was determined using the thrombospondin-1-derived agonist peptide 4N1K and the CD47 function blocking antibody B6H12.2.

Results.: HTM cells expressed CD47 and integrins αvβ3 and β1. β3 Integrin or CD47 activation significantly increased CLAN formation over β1 integrin-induced levels, whereas anti-CD47 mAb B6H12.2 inhibited this increase. PP2, NSC23766, and Trio siRNA decreased β3-induced CLAN formation by 72%, 45%, and 67%, respectively, whereas LY294002 and dominant negative Tiam1 had no effect. LY294002 decreased β1 integrin-mediated CLAN formation by 42%, and PP2 completely blocked it.

Conclusions.: Distinct β1 and αvβ3 integrin signaling pathways converge to enhance CLAN formation. β1-Mediated CLAN formation was PI-3K dependent, whereas β3-mediated CLAN formation was CD47 and Rac1/Trio dependent and might have been regulated by thrombospondin-1. Both integrin pathways were Src dependent.

It is well established that in both humans and animals, treatment with a glucocorticoid (GC) such as dexamethasone (DEX) can increase intraocular pressure (IOP) both in vivo and in cultured anterior segments. ^1–8 In some cases this can cause damage to the optic nerve and can result in a steroid-induced glaucoma (SIG). Studies in cultured anterior segments ³ and cultured trabecular meshwork (TM) cells ^9–11 treated with DEX have suggested that steroid treatment can lead to a rearrangement of the actin cytoskeleton into cross-linked actin networks (CLANs) that resemble geodesic domes or polygonal actin networks. ^12–14 CLANs have also been observed in cultured TM cells and in TM cells in isolated meshworks from glaucomatous donor eyes in the absence of any DEX treatment, ^5,15 which suggests these actin structures are involved in the pathogenesis of SIG and other forms of primary open angle glaucoma (POAG). ^3,9,11,16 CLANs have also been found in normal TM cells in isolated meshworks, albeit at a lower frequency than in glaucomatous TMs. ¹⁵  

The function of CLANs in the TM remains unclear at this time. CLANs can be found in both spreading ^12,17,18 and nonspreading cells ^9,19,20 and were originally thought to be precursors to actin stress fibers ¹² or reorganized sarcomeres. ²¹ It has been suggested that CLANs are specialized structures that participate in maintaining cellular tensegrity. ²² Recently, it has been suggested ³ that CLAN formation in TM cells may reduce the contractility of the tissue by increasing the rigidity of the cells and thus rendering them unable to change shape and “relax” under pressure. Alternatively, CLAN formation could be impacting other actin-mediated biological processes of the TM that are required for normal outflow facility such as attachment to the extracellular matrix (ECM), phagocytosis, and gene expression. ^16,23  

CLANs are made up of interconnected F-actin bundles (spokes) radiating outward from central vertices (or hubs). The vertices appear to be composed of molecular complexes (vertisomes) composed of α-actinin, syndecan-4, phosphatidylinositol 4,5-bisphosphate (PIP₂), and filamin in addition to actin. ¹⁷ Outside the vertisomes, filamin, myosin, and tropomyosin localize along the F-actin bundles. ^12–14,17 In TM cells, CLAN formation can be controlled by cooperative signaling between β1 and β3 integrins in the absence of steroid treatment ¹⁷ and by TGF-β2 (Hoare MJ. IOVS 2009;49:ARVO E-Abstract 4876). 

Integrins are transmembrane receptors that consist of a heterodimer of α and β subunits. They recognize ECM proteins by binding to the amino acid sequence Arg-Gly-Asp (RGD) or its homologues within a given protein. Signaling from integrins is dependent on the formation of supramolecular complexes with both integral or peripheral membrane proteins and cytoplasmic molecules. These complexes provide bidirectional signaling that allows integrins to transduce extracellular signals to the actin cytoskeleton and within the intracellular environment (outside-in signaling) and intracellular signals to the outside environment (inside-out signaling). Thus, the specific arrangement of molecules associated with integrins forms an important physical link between the extracellular and intracellular environment that regulates cell function and the organization of the actin cytoskeleton. ^24,25  

One potential component in this supramolecular signaling complex is CD47 (integrin-associated protein [IAP]). ^26–29 CD47 was initially identified as a 50-kDa protein associated with αvβ3 integrin signaling and was later shown to be a receptor for the carboxyl terminal domain of thrombospondin-1 (TSP1). ^26–29 It is an atypical member of both the immunoglobulin superfamily and the G-protein-coupled receptor (GPCR) family of membrane proteins. Although CD47 has only five transmembrane domains rather than the seven that are typical of GPCRs, it has been suggested that a complex formed by CD47 and an integrin heterodimer such as αvβ3 could function as a GPCR. ^26–29 It is unknown whether CD47 expression in HTM cells is altered in response to GC treatment; however, it has been shown that expression of its ligand TSP1 is increased in DEX- or TGF-β1-treated HTM cells. ³⁰ In addition to associating with αvβ3, CD47 can interact with integrins αIIbβ3, α4β1, α5β1, and α2β1; however, CD47 appears to differentially regulate β1 and β3 integrin signaling. ^28,31 A CD47 antibody that blocks β3 integrin signaling causes activation of β1 integrin signaling, suggesting differences in the physical interaction of this receptor with different integrin species. The CD47/integrin complex regulates such activities as platelet activation, cell motility, adhesion, migration, phagocytosis, cytokine synthesis, and integrin cross-talk. ^28,32 The molecular basis for how CD47 affects integrin signaling is unclear because it is not found in focal adhesions with integrins. However, αvβ3/CD47 signaling complexes can be found either in ³³ or out of lipid rafts. ³⁴ Data suggest that CD47 may modulate integrin signaling by activating G_αi-containing heterotrimeric GTPases and stimulating the phosphorylation of FAK (focal adhesion kinase) and the Src kinase Lyn. ^28,33,35  

In this study we used various pharmacologic agents, activating and dominant negative peptides, function-blocking antibodies, and siRNAs to examine the signaling components used by β1 and β3 integrins to promote CLAN formation in HTM cells. These data indicate that the cooperative β1 and β3 integrin signaling pathways that enhance CLAN formation in TM cells spread on a fibronectin are distinct. The β1 integrin-mediated pathway is dependent on Src kinases and phosphatidylinositol 3-kinase (PI-3K). The β3 integrin-mediated pathway is also dependent on Src kinases, along with Rac1, the Rac1 GEF Trio, and CD47. The studies also suggest CD47 activation by a TSP1-specific peptide may play a central role. 

Methods

Cell Culture

The A7–1, N27TM-1, and N27TM-2 strains of HTM cells were isolated from a 30-year-old and two 27-year-old donors, respectively, who had no known history of ocular disease, as previously described. ^36–38 Both N27TM cell strains were found to perform similarly to the A7–1 strain in our spreading assay, though the basal levels of CLAN formation were lower in these strains than in A7–1 HTM cells. The A7–1 strain was used in all experiments except where noted in the figure legends. Cells were cultured in low-glucose DMEM (Sigma, St. Louis, MO), 15% fetal bovine serum (Atlanta Biologicals, Atlanta, GA), 2 mM l-glutamine (Sigma), 1% amphotericin B (Mediatech, Herndon, VA), 0.05% gentamicin (Mediatech), and 1 ng/mL FGF-2 (PeproTech, Rocky Hill, NJ). ^36,37  

Fluorescence-Activated Cell Sorting Scan Analysis

FACS analysis was performed as described previously. ³⁹ Cells were incubated with anti-human CD47 (clone 2D3; eBioscience, San Diego, CA), anti-human αvβ3 (LM609; Millipore Corp., Temecula, CA), or anti-human β1 integrin (Hb1.1; Millipore Corp.) at 2 μg/mL prepared in Tris-buffered saline (20 mM Tris pH 7.6, 150 mM NaCl) with 1% bovine serum albumin (BSA) for 30 minutes on ice. After incubation with FITC-conjugated anti-mouse IgG (Pierce, Rockford, IL) diluted 1:200 in TBS containing 1% BSA, the cells were analyzed by FACScan caliber flow cytometer (Becton-Dickinson, Franklin Lakes, NJ). 

Spreading Assays

Spreading assays were performed using duplicate determinations, as described previously. ¹⁷ Before replating onto coverslips precoated with 20 nM fibronectin, suspended cells were preincubated for 30 to 60 minutes in the absence or presence of 0.25, 1, 5, or 20 μM of the Src family kinase (SFK) inhibitor PP2 (EMD Biosciences, Inc., San Diego, CA), 60 μM of the selective SFK inhibitor cis-5,8,11,14,17-eicosapentaenoic acid (EPA, Sigma), 20 μM of the PI-3K inhibitor LY294002 (EMD Bioscience, Inc.), 20 μM of the Rac1 inhibitor NSC23766 ⁴⁰ (kindly provided by Yi Zheng, Children's Hospital Research Foundation, Cincinnati, OH), or 100 nM of the dominant negative (DN) form of the Rac1 guanine nucleotide exchange factor (GEF) Tiam1. Cells pretreated with dimethyl sulfoxide (DMSO) were used as controls for PP2, EPA, and LY294002 treatments. Treated cells were then plated in the absence or presence of 8 μg/mL mAb AP-5, which activates β3 integrins. In some experiments, cells were plated in the presence of 100 or 200 μg/mL CD47 agonist peptide 4N1K (KRFYVVMWKK) derived from the carboxyl-terminal domain of TSP1, ²⁷ 200 μg/mL control peptide 4NGG (KRFYGGMWKK), or 20 μg/mL CD47 function blocking mAb B6H12.2 (Abcam, Inc., Cambridge, MA). All cells were allowed to spread for 3 hours and then were fixed with 4% p-formaldehyde plus 0.18% Triton X-100 for 30 minutes. Human plasma fibronectin was prepared as described. ⁴¹  

Immunofluorescence Microscopy and Quantification of CLANs and Cell Area

Fixed cells were labeled with Alexa Fluor 488-conjugated phalloidin and Hoechst 33342 (both from Invitrogen, Carlsbad, CA) to visualize F-actin and nuclei, respectively. DN-Tiam1 was localized in cells using a mouse monoclonal antibody to the DYKDDDDK tag on Tiam1 (Anti-Flag M2; Sigma) and the Alexa Fluor 546 goat anti-mouse IgG secondary antibody (Invitrogen). Fluorescence was observed with an epifluorescence microscope (Axioplan 2; Zeiss, Thornwood, NY) equipped with a digital camera (Axiocam HRm; Zeiss) and image acquisition software (Axiovision version 4.5; Zeiss). 

To quantify the number of CLAN-positive cells (CPCs), five to eight low-power (200×) fluorescence images from each treatment group were captured. The minimum requirement for an actin structure to be counted as a CLAN was the presence of three intensely fluorescent vertices connected by three actin spokes. Representative images of what we counted as CLANs are shown in Figure 1. The number of CPCs per image, along with the total number of cells, was counted to calculate the percentage of CPCs per image. Data were pooled from three experiments for each treatment and represent the mean percentage CPC ± SD. To determine any effects on cell spreading, the cell areas from six to eight of the images captured from the groups treated with or without PP2 were analyzed using Axiovision software (see Fig. 3C). All data were pooled from three experiments and represent the mean ± SD cell area. Statistical analysis comparing the different treatment groups for CLAN formation or for cell area was performed using ANOVA (see Figs. 3, 4, 6) or Student's t-test (see Fig. 5). Where pairs of treatment groups had to be compared (Figs. 3, 4), ANOVA analysis was used in conjunction with the Tukey HSD test. 

Figure 1.

 

Representative images of CLANs. (A–F) Various types of CLANs observed in all the treatment groups. (A) Minimal structure for a CLAN (inset, asterisk). (B) A small CLAN (inset, asterisk). Triangle: minimum requirement for CLAN designation. (C) Two CPCs side by side. The cell on the left contains a slightly larger CLAN than in (B), whereas the cell on the right contains a larger, more extensive CLAN. (D) A moderately sized CLAN. (E, F) Two examples of large extensive CLANs. There appeared to be a general tendency for higher order CLANs to be more frequent in cells treated with mAb AP-5. No attempt was made, however, to quantify any differences in CLAN size with respect to specific treatments. The N27TM-1 HTM strain was used to acquire these images. Scale bar, 20 μm (A).

siRNA-Mediated Silencing

siRNA against human Trio (ON-TARGETplus SMARTpool L-005,047–00-0005), human Tiam1 (ON-TARGETplus SMARTpool L-003,932–00-0005), and a nontargeting control (ON-TARGETplus siCONTROL Non-targeting Pool D-001,810–10-05) was obtained from Dharmacon (Lafayette, CO). Differentiated monolayers of N27TM-2 cells were transfected with 125 nM siRNA using a cationic lipid-based transfection reagent (Lipofectamine 2000; Invitrogen) according to the manufacturer's protocol. After 48 hours, a spreading assay was performed and CPCs were quantified, as described. 

RNA Extraction and RT-PCR

Total RNA was extracted from siRNA-transfected N27TM-2 cells using an RNA purification kit (RNeasy Plus Mini Kit; Qiagen, Valencia, CA), according to the manufacturer's instructions. The RNA was then reverse transcribed (RETROscript Kit; Applied Biosystems, Foster City, CA). Each reaction was performed using 2 μg RNA, 5 μM random decamers, 500 μM each dNTP, 10 U RNase inhibitor, and 100 U MMLV reverse transcriptase. Reactions were incubated at 44°C for 1 hour and then at 92°C for 10 minutes. After first-strand synthesis, the cDNA was amplified by PCR using 1 U DNA Taq polymerase (Platinum Taq; Invitrogen), 1.5 mM MgCl₂, 125 μM each dNTP, and 5 μM gene-specific primer pairs. Sequences used for the Trio and Tiam1 primers have been published previously. ¹⁸ Amplification conditions were 94°C for 1 minute, 30 cycles at 94°C for 30 seconds, 55°C for 30 seconds, and 72°C for 30 seconds, and a final extension at 72°C for 5 minutes. The PCR products were separated on a 2.5% agarose gel and stained with ethidium bromide. 

Cloning, Expression, and Purification of DN-Tiam1

N-terminal-truncated Tiam1 cDNA (C1199) was used as a template for polymerase chain reaction amplification of a DNA sequence encoding amino acids 393 to 854 of Tiam1. The Tiam1 cDNA was a gift of John G. Collard (The Netherlands Cancer Institute, Amsterdam, The Netherlands). The sense primer, 5′-AATAACCGGTATGAGTACCACCAACAG-3′, generated an AgeI restriction site (in boldface), and the antisense primer, 5′-AATAGCATGC TCACGCGTCTGACTTCTCAA-3′, introduced a SphI restriction site (in boldface) and a stop codon (underlined). The amplified DNA was ligated into the pGEX-PTD4-FLAG vector provided by Jennifer Faralli (University of Wisconsin, Madison, WI) and contains a PTD4 transduction sequence (YARAAARQARA) ⁴² and a FLAG tag (DYKDDDDK) upstream of the multiple cloning site. The vector was introduced into Escherichia coli BL21-competent cells (CodonPlus; Stratagene, La Jolla, CA) according to the manufacturer's instructions. 

Expression of the FLAG tagged TAT-DN-Tiam1 (DN-Tiam1) fusion protein was induced with 0.4 mM isopropyl β-D-thiogalactopyranoside at 30°C, for 3 hours. After induction, the bacteria were resuspended in buffer consisting of 20 mM Tris-HCl (pH 7.4), 150 mM NaCl, 4 mM KCl, and 0.1 mM dithiothreitol and then was lysed with 0.5 mg/mL lysozyme, and 25 μg/mL DNaseI for 20 minutes. This was followed by sonication at 20% for 5 minutes in pulse mode. The lysate was incubated with 1% Triton X-100 for 20 minutes, and the insoluble material was sedimented by centrifugation at 10,000g for 30 minutes at 4°C. The DN-Tiam1 was isolated from the clarified lysate by affinity chromatography on glutathione Sepharose 4B, followed by on-column cleavage by thrombin and incubation with p-aminobenzamidine Sepharose 6B, as described previously. ⁴³  

The biological activity of the recombinant DN-Tiam1 was confirmed by examining its effect on cell morphology and cell contacts in monolayers of immortalized human TM-1 cells. ⁴⁴ DN-Tiam1 caused extensive cell rounding of TM-1 cells when they were transduced with DN-Tiam1 (not shown). This activity is consistent with the role that Tiam1 plays in maintaining cell-cell adhesions. ⁴⁵  

Results

Participation of CD47 in β3 Integrin-Mediated CLAN Formation

Previous studies demonstrated that the basal level of CLANs formed by plating cells onto a β1 integrin binding substrate is enhanced by activation of β3 integrin using the activating β3 mAb AP-5. ¹⁷ Since CD47 is a known coreceptor for β3 integrin, FACs analysis was performed to determine whether CD47 is present on the surfaces of TM cells and thus could also be involved. Figure 2 shows that HTM cells express CD47 in addition to αvβ3 and β1 integrins. To determine whether CD47 participated in CLAN formation, HTM cells were plated in the presence or absence of the CD47 agonist peptide 4NIK, ²⁷ which contains a peptide sequence from the cell-binding domain of TSP1. The peptide 4NGG was used as a control. Figure 3 shows that both 100 and 200 μg/mL 4N1K caused an approximately twofold increase (P < 0.0001) in CLAN-positive cells (CPCs) relative to untreated cells or cells treated with the 4NGG control peptide, whereas the β3 integrin-activating mAb AP-5 caused a 3.5-fold increase in CPCs relative to untreated cells (P < 0.0001). The increase in CPCs in the presence of mAb AP-5 was significantly greater than that seen in the presence of either 4N1K concentration (P < 0.01). In contrast, treatment with the control peptide 4NGG showed only 7% to 8% of the cells as CLAN positive, which is comparable to the level of CLAN formation previously observed in untreated cells. ¹⁷ Treatment of HTM cells with either 4N1K or 4NGG did not have any obvious effect on cell spreading compared with untreated cells or cells treated with mAb AP-5 (not shown). 

Figure 2.

 

FACS analysis of HTM cells for expression of CD47, αvβ3, and β1 integrins. HTM cells were labeled with either nonspecific mouse IgG or antibodies against (A) CD47, (B) αvβ3 integrin, or (C) β1 integrins before FACS analysis. Primary antibodies were detected with FITC-conjugated anti-mouse IgG.

Figure 3.

 

CD47 regulates CLAN formation through αvβ3 integrin. HTM cells were plated onto fibronectin-coated coverslips in the absence (NT) or presence of soluble β3 integrin activating mAb AP-5, 4NGG, or 4N1K; other treatment groups also included soluble CD47 blocking antibody B6H12.2 together with mAb AP-5 or the peptides. Cells were fixed and labeled with phalloidin. The percentage of CPCs is shown as the mean ± SD; n ranged from 3139 to 2439 for each group. The percentage of CPCs in β3 mAb- or 4N1K-treated (100 and 200 μg/mL) cells was significantly greater than in untreated cells or cells treated with the 4NGG control peptide, respectively (P < 0.0001). The percentage of CPCs in mAb AP-5-treated cells was significantly greater than either concentration of peptide 4N1K by itself (P < 0.01). The percentage of CPCs in cells treated with CD47 blocking antibody B6H12.2 plus mAb AP-5 was significantly lower than in cells treated with mAb AP-5 only (P < 0.0001). The percentage of CPCs in cells treated with mAb B6H12.2 plus 100 μg/mL 4N1K was significantly lower than in cells treated with 100 μg/mL 4N1K only (P < 0.01). The percentage of CPCs in cells treated with mAb B6H12.2 plus 200 μg/mL 4N1K was significantly greater than in untreated cells, cells treated with the control 4NGG peptide, or cells treated with mAb B6H12.2 plus 100 μg/mL 4N1K (P < 0.01). There was no statistical difference between untreated cells, 4NGG-, B6H12.2 plus AP-5-, and B6H12.2 plus 100 μg/mL 4N1K-treated cells.

To further confirm that CD47 plays a role in β3 integrin-mediated CLAN formation, HTM cells were incubated with the CD47 function blocking mAb B6H12.2. ²⁷ As shown in Figure 3, this mAb not only blocked 100 μg/mL 4N1K-induced CLAN formation as expected (P < 0.01), it completely blocked AP-5 induced CLAN formation (P < 0.01). The effect of mAb B6H12.2 could be overcome by increasing the concentration of the 4N1K peptide to 200 μg/mL. These data suggest that modulation of β3 integrin activity by activating CD47 with the TSP1 peptide promotes CLAN formation in HTM cells. 

Roles of SRC Family Kinases and PI-3K in CLAN Formation

To determine whether β1 and β3 integrins differentially regulated CLAN formation, we examined the role of several signaling molecules known to be involved in integrin signaling. The first molecules examined were members of the Src family kinases (SFK). SFK members are important regulators of integrin signaling and are among the first signaling molecules activated on integrin engagement. ⁴⁶ To assess the potential role that SFK members played in regulating CLAN formation, we used the inhibitor PP2, which inhibits all SFKs. ^47–50 As shown in Figures 4A and 4C, cell spreading in the presence of 5 or 20 μM PP2 was strikingly impaired compared with untreated cells or cells treated with DMSO alone. Both 5 μM and 20 μM PP2 decreased the average area of cell spreading by 58% and 71% (P < 0.0001), respectively, and, not surprisingly, CLAN formation was completely absent (Fig. 4B) even in cells that exhibited partial spreading in the presence of PP2. 

Figure 4.

 

β1- and β3-mediated CLAN formation are dependent on a Src family kinase, whereas only β1-mediated CLAN formation is PI-3K dependent. Cells were plated onto fibronectin-coated coverslips, fixed, and labeled with phalloidin. (A) Representative photomicrographs of A7–1 HTM cells treated with or without 5 or 20 μM PP2 in the presence or absence of mAb AP-5. Scale bar, 50 μm. (B) CLAN formation in A7–1 HTM cells spread in the absence (NT) or presence of 0.1% DMSO, 20 μM PI-3K inhibitor LY294002, or the SFK inhibitor PP2 with or without soluble mAb AP-5. β1-Mediated CLAN formation was significantly inhibited by LY294002 or PP2 (5 and 20 μM) (P < 0.01) compared with controls. β3-Mediated CLAN formation (mAb AP-5 only) was significantly greater than untreated cells (P < 0.0001). Although β3-mediated CLAN formation was statistically lower than in AP-5 only-treated cells when LY294002 was added (P < 0.01), there was no statistical difference between the vehicle-treated cells (β3 mAb + 0.1% DMSO) and cells treated with β3 mAb + LY294002. (C) A7–1 cells spread in the absence or presence of 0.1% DMSO or PP2 (5 or 20 μM) with or without mAb AP-5. Spreading in the presence of 5 or 20 μM PP2 was statistically less than in control cells (P < 0.0001) when β3 mAb AP-5 was absent. Cell spreading in the presence of AP-5 and either 5 or 20 μM PP2 was statistically greater than in cells treated with PP2 alone (P < 0.0001). (D) CLAN formation in spread N27TM-1 cells in the absence or presence of 0.1% DMSO, 20 μM PP2, or 60 μM EPA with or without mAb AP-5. β1-Mediated CLAN formation (no AP-5) in the presence of EPA was not statistically different from vehicle-treated cells, whereas β3-mediated CLAN formation in the presence of EPA was statistically less than in vehicle-treated cells (P < 0.01). The percentage of CPCs is shown as the mean ± SD; n ranged from 2899 to 1146 for each group.

To determine whether β1 integrin-mediated CLAN formation could be separated from cell spreading, experiments were repeated using lower concentrations of PP2 (0.25 μM and 1 μM PP2). These concentrations of PP2 effectively blocked CLAN formation by greater than 10-fold (P < 0.01) compared with control cells. In untreated cells or 0.1% DMSO-treated cells, the percentages of CLAN-positive cells were 2.95% ± 1.2% CPC and 2.6% ± 1.2% CPC, respectively. In contrast, the percentages of CLAN-positive cells in cultures treated with 0.25 μM or 1 μM PP2 were 0.3% ± 0.6% CPC and 0% CPC. Despite the fact that CLAN formation was nearly completely blocked by these lower concentrations of PP2, cell spreading was only partially affected. Compared with the mean area of spread cells in the 0.1% DMSO control, cell spreading was decreased by 48% in 1 μM PP2-treated cultures (1690 ± 1432 μm² vs. 874 ± 1025 μm²; P < 0.0001) and by 14% in 0.25 μM PP2-treated cultures (1690 ± 1432 μm² vs. 1450 ± 1249 μm²; P < 0.0001). Thus, β1-mediated CLAN formation could be decoupled from cell spreading on fibronectin, suggesting that SFK signaling directly regulates β1 integrin-mediated CLAN formation. 

When β3 integrin signaling was activated with the AP-5 antibody in HTM cells pretreated with 5 μM or 20 μM PP2, cell spreading significantly recovered. As shown in Figures 4A and 4C, the average cell area increased by 66.5% and 58%, respectively, in cells pretreated with 5 μM (P < 0.0001) or 20 μM (P < 0.0001) PP2 in the presence of soluble mAb AP-5 compared with cells treated with PP2 alone. However, despite the recovery of cell spreading, 5 μM PP2 reduced CLAN formation by 72% (P < 0.0001) relative to cells treated with DMSO only (Fig. 4B), and 20 μm PP2 essentially blocked all CLAN formation. Thus, SFK signaling also plays a direct role in β3 integrin-mediated CLAN formation. 

To narrow down the possible SFK members that participated in cell spreading and subsequent CLAN formation in TM cells, the studies were repeated with the selective SFK inhibitor EPA, which inhibits the activity of Fyn and Lck but not Src. ^51,52 As shown in Figure 4D, 60 μM EPA, a concentration shown to be highly effective in other cell types, ^51,52 had little or no effect on spreading (not shown) or on β1-mediated CLAN formation (Fig. 4D). β3-Mediated CLAN formation was only slightly affected by 60 μM EPA, which decreased CLAN formation by 15% relative to the 0.1% DMSO control (P < 0.01). 

The next molecule examined was PI-3K, which is known to be downstream of Src in integrin signaling pathways and is also a key regulator of the actin cytoskeleton. ^53,54 Unlike the Src inhibitor PP2, treatment of cells with the PI-3K inhibitor LY294002 did not appear to dramatically alter overall cell spreading either in the absence or presence of mAb AP-5 compared with untreated cells or cells treated with DMSO only (not shown). LY294002, however, did affect CLAN formation. In the absence of mAb AP-5, basal levels of CLAN formation mediated by β1 integrins were decreased by 42% in cells treated with LY294002 (Fig. 4B) compared with cells treated with DMSO only (P < 0.01) and 52.5% compared with untreated cells (P < 0.01). This suggests that β1 integrin-mediated CLAN formation is dependent on PI-3K activity. In contrast, β3 integrin-mediated CLAN formation was not significantly altered after treatment with LY294002. The 20% decrease observed in these cells may be largely attributable to the DMSO solvent used to dissolve the inhibitor because DMSO alone decreased β3-mediated CLAN formation by 13% and there was no difference between the decrease observed in the presence of DMSO alone compared with that in the presence of LY294002. Together these data suggest that β1 integrin-mediated CLAN formation is dependent on Src/PI-3K signaling, whereas β3 integrin-mediated CLAN formation involves a Src-dependent, PI-3K-independent pathway. 

Involvement of Rac1/Trio Signaling in CLAN Formation

We then examined the role that Rac1 might play in regulating CLAN formation in HTM cells. Rac1 is a small GTPase known to regulate the polymerization of branched actin filaments during migration and lamellipodia formation during cell spreading. ⁵⁵ For this study we used the Rac1-specific inhibitor NSC23766, which blocks a subset of Rac1 signaling mediated by the GEFs Tiam1 and Trio. ⁴⁰ NSC23766 (20 μM), like LY294002, also differentially affected β1 and β3 integrin-mediated CLAN formation (Fig. 5). NSC23766 inhibited β3 integrin-mediated CLAN formation by nearly 45% (P = 0.0001), demonstrating a role for Rac1 signaling mediated by either Tiam1 or Trio in CLAN formation. In contrast, CLAN formation mediated by β1 integrins was not affected by NSC23766 (P = 0.23), indicating that basal CLAN formation is not regulated by Rac1/Tiam1 or Rac1/Trio signaling. As observed with cells spread in the presence of LY294002, NSC23766 did not visibly alter overall cell spreading compared with untreated cells spread under the same conditions (not shown). 

Figure 5.

 

β3-Mediated CLAN formation is dependent on Rac1/Tiam1 or Rac1/Trio signaling, but β1-mediated CLAN formation is not. HTM cells were plated onto fibronectin-coated coverslips in the absence (NT) or presence of 20 μM Rac1 inhibitor NSC23766, with or without soluble β3 integrin activating mAb AP-5. Cells were fixed and labeled with phalloidin. The percentage of CPCs is shown as the mean ± SD; n ranged from 2328 to 1964 for each group. The mean percentage of CPC in the presence of AP-5 and NSC23766 was significantly lower than in the presence of AP-5 alone (P = 0.0001).

Because the Rac1-specific inhibitor NSC23766 blocks both Tiam1 and Trio, we performed additional experiments to determine which Rac1 GEF was involved in regulating β3 integrin-mediated CLAN formation. In the first set of experiments, quiescent HTM cells were transfected with siRNA targeting either Tiam1 or Trio. As expected, cultures transfected with Trio or Tiam1 siRNA did not show a change in β1-mediated CLAN formation relative to control cells (Fig. 6A). In the presence of mAb AP-5, however, cultures transfected with Trio, but not Tiam1, siRNA demonstrated a 67% decrease in CLAN formation relative to control cells (P < 0.001). Tiam1-transfected cultures did not show any significant change in β3 integrin-mediated CLAN formation. 

Figure 6.

 

The Rac1 GEF Trio, not Tiam1, regulates β3-mediated CLAN formation. N27TM-2 HTM cells were plated onto fibronectin-coated coverslips in the presence or absence of β3 integrin activating mAb AP-5. Cells were fixed, labeled with phalloidin, and stained for the FLAG-tag (B). (A) Quantification of CLANs in cells transfected with 125 nM Trio, Tiam1, or nontargeting control siRNA. The percentage of CPCs is shown as the mean ± SD; n ranged from 1273 to 1794 for each group. The mean percentage of CPCs in β3 antibody + Trio siRNA-treated cells is significantly lower than in β3 antibody alone-treated cells (P < 0.001). (B) Photomicrograph showing CLAN formation in N27TM-1 cells transduced with DN-TAT-Tiam1 and treated with soluble mAb AP-5. Scale bar, 20 μm. (C) Quantification of CLANS in N27TM-1 cells that spread in the absence or presence of DN-TAT-Tiam1. CLAN formation was also determined in cells that were Tiam1 positive, as determined by the presence of FLAG-tag staining. The percentage of CPCs is shown as the mean ± SD; n ranged from 2899 to 1146 for each group. There was no significant difference between untreated cells and those transduced with DN-TAT-Tiam1 that were positive for the FLAG-tag. (D) 2.5% agarose gel stained with ethidium bromide showing endogenous expression of Trio and the housekeeping gene Rig/S15 in N27TM-2 cells. Tiam1 expression was not detected. PCR primers for Trio, Tiam1, and Rig/S15 amplify 122-, 97-, and 361-bp fragments, respectively.

To confirm that Tiam1 is not involved in CLAN formation, HTM cells were transduced with a FLAG-tagged TAT-DN-Tiam1 fusion protein (DN-Tiam1). ⁵⁶ The TAT sequence was added to allow for the rapid transduction of recombinant DN-Tiam1 into cells, whereas the FLAG tag provided a means by which to detect the Tiam1 construct in the cells (Fig. 6B). ⁴² Using fluorescence microscopy, the transduction rate of DN-Tiam1 into the HTM cultures was found to be 50% to 54% (data not shown). As shown in Figure 6B, overall cell spreading in cultures transduced with DN-Tiam1 did not differ relative to untransduced cultures. Cultures transduced with DN-Tiam1 also did not show a significant change in either β1 integrin- or β3 integrin-mediated CLAN formation compared with control cells (Fig. 6C). 

RT-PCR analysis of Tiam1 and Trio RNA levels also indicated that Trio, rather than Tiam1, was the GEF involved in CLAN formation. The expression of Trio, but not Tiam1, RNA could be detected, indicating that HTM cells express Trio but not Tiam1 (Fig. 6D). The absence of Tiam1 in these cells was also confirmed by Western blot analysis (data not shown). 

Discussion

In this study, we found that distinct signaling pathways activated by β1 and β3 integrins converge and cooperate to promote CLAN formation (Fig. 7). The basal level of CLAN formation in HTM cells mediated by β1 integrins involved an SFK/PI-3K-dependent signaling pathway. Although it is likely that the SFK is Src, we have not yet ruled out that two other SFK members, Yes, and Lyn, ^57,58 are involved given that all SFK members can be inhibited by PP2 ^47–50 and EPA only inhibits Fyn and Lck. ^51,52 This basal level of CLAN formation induced via β1 integrin could be further enhanced by coactivating a β3 integrin signaling pathway that used distinct and separate signaling components. β3 integrins enhanced CLAN formation via a Src dependent, PI-3K-independent pathway that also involved Rac1-Trio. Furthermore this β3 integrin pathway could be activated by the G-protein coupled receptor CD47 via the TSP1-derived peptide 4N1K. These data suggest β1- and β3-mediated signaling pathways converge downstream of PI-3K and Rac1-Trio to promote CLAN formation (Fig. 7). 

Figure 7.

 

Schematic diagram showing that β1- and β3-mediated CLAN formation is the result of distinct signaling pathways. Both pathways are dependent on an SFK, possibly Src. The β1 integrin-activated pathway involves PI-3K, whereas the αvβ3-activated pathway involves CD47 and Rac1 signaling mediated by the GEF Trio. CD47 activation by TSP1 containing the carboxyl-terminal sequence found in peptide 4N1K (RFYVVMWK) may be an important regulator of CLAN formation. Convergence of the two pathways could potentially occur at two points. The β1 integrin pathway might also signal through Rac1 but via activation of a Rac1 GEF distinct from Trio. Alternatively, PI-3K could bypass Rac1 completely, and the two pathways would converge at a different downstream effector to promote CLAN formation.

CD47-induced CLAN formation was most likely mediated through interactions with αvβ3 integrins rather than αIIbβ3 or α2β1 integrins. ^59,60 The expression of αIIbβ3 integrins is restricted to megakaryocyte-derived cells, and previous studies showed that direct activation of α2β1 integrin did not induce significant CLAN formation. ¹⁷ Thus, the contribution of CD47-mediated activation of αIIbβ3 and/or α2β1 integrins to TM cell CLAN formation was minimal. Furthermore, the CD47 function blocking antibody B6H12.2 blocked CLAN formation induced by the β3 integrin activating antibody mAb AP-5, further suggesting an interaction between a β3 integrin and CD47. 

The involvement of CD47 in β3-induced CLAN formation raises some very interesting questions about how this might possibly relate to the induction of CLAN formation by DEX in cultured anterior segments and cells. ^3,9–11 CD47 is a receptor for TSP1, the expression of which is upregulated by both TGF-β1 and DEX treatment in human and mouse TM cells, and increased TSP1 deposition has been found in the ECM of the meshwork in cases of POAG and SIG. ³⁰ Additionally, Liu et al. ²³ showed that DEX caused a decrease in TSP1 released into culture medium, consistent with the idea that DEX treatment results in an increase in matrix- or cell-associated TSP1, similar to what has been demonstrated for other ECM components. ^44,61–63 Increased TSP1 within the ECM could lead to increased CD47 activation, thereby contributing to αvβ3 integrin-mediated CLAN formation in DEX-treated HTM cells. The suppression of TSP1 production and the subsequent decreased CLAN formation in HTM cells in response to the actin-disrupting drug LAT-A would further support the notion that TSP1 and CLAN formation are connected. ²³  

The point at which these two pathways converged to enhance CLAN formation could not be determined from these studies. Clearly, these signaling pathways must converge because activation of β3 integrin signaling alone is not sufficient to induce CLAN formation in TM cells. ¹⁷ One point may be Src because PP2 blocked both β1- and β3-mediated CLAN formation. However, Src is generally thought of as an early upstream signaling event before Rac1, so this would not explain the differential inhibition of β1- and β3-mediated CLAN formation by LY94002 and NSC23766, respectively. The most likely explanation is that the pathways converge at Rac1 or downstream of Rac1 activation at one of the steps involved in the formation of a branched actin network given that Rac1 is known to regulate the formation of these structures. ⁵⁵ However, only αvβ3 integrin-mediated CLAN formation was blocked by the Rac1 inhibitor NSC23766, which is puzzling because β1-mediated CLAN formation is regulated by PI-3K, and there are numerous studies that show Rac1 activation is dependent on PI-3K. ^64,65 The NSC23766 inhibitor that was used, however, specifically blocks Rac1 signaling mediated by the GEFs Tiam1 or Trio. Hence, it is possible that β1-mediated CLAN formation involves activation of Rac1 by another Rac1 GEF, such as Vav1. ⁶⁶ The fact that gene-silencing experiments indicated that β3-mediated CLAN formation involved Trio rather than Tiam1 supports the idea that CLAN formation in HTM cells is regulated by specific GEFs. 

It is not surprising that Trio was found to be the active GEF in HTM cultures. Trio is highly expressed in neuronal tissues, ⁶⁷ and HTM cells have a neural crest origin, ⁶⁸ whereas Tiam1 expression appears to be more widespread. ⁶⁹ It is interesting, though, that the β3 integrin signaling was linked to Trio-Rac1. Integrin signaling has previously been associated with Tiam1, ^70,71 whereas Trio has been associated with heterotrimeric G proteins. ⁷² To our knowledge, this is the first time that any integrin signaling has been linked to Trio-Rac1. However, the fact that CD47 is considered an atypical GPCR that can activate the Gα_i subfamily of heterotrimeric G proteins ³⁵ and is also a co-receptor for β3 integrin, supports the idea that the β3 integrin/CD47 complex activates Trio-Rac1 signaling. 

Additional data suggest that CLAN formation may involve Gα_i signaling. A recent study in Schwann cells ⁷³ showed that CLANs (geodesic actin networks) were formed in response to lysophosphatidic acid (LPA) or sphingosine 1-phosphate (S1P). Both LPA and S1P signal via GPCRs and activate the Gα_i signaling pathway. ^74,75 Interestingly, LPA- and S1P-induced CLAN formation in Schwann cells was also found to be dependent on Rac1, rather than RhoA, activation. Clearly, future studies will be needed to explore possible links among CD47, Rac1, and Trio in CLAN formation. 

Although the function of CLAN formation in HTM cells is still unknown, CLANs do not appear to be necessary for cell spreading because the assembly of actin filaments into a CLAN appears to involve a distinct process that is not involved in cell spreading. Under conditions when β1-mediated CLAN formation was inhibited by the Src kinase inhibitor PP2, cell spreading was only partially inhibited. In addition, activation of β3-mediated CLAN formation in the presence of PP2 induced a partial recovery of cell spreading but not CLAN formation. Thus, we were able to decouple cell spreading from CLAN formation and show that CLAN formation is not needed for cell spreading. We can only speculate as to why CLAN formation, but not cell spreading, was dependent on Src. However, Src-independent cell spreading has previously been observed in osteoclasts, ⁷⁶ and it is possible that this can occur in TM cells. 

Interestingly, it took a higher concentration of PP2 to inhibit β3 integrin-mediated CLAN formation than β1-mediated CLAN formation. It is possible that αvβ3-dependent activation of Src is a later event in the formation of CLANs than in β1-mediated CLAN formation and that activation of αvβ3 could lead to the additional recruitment of activated Src to the cell membrane. It has been shown in platelets ⁷⁷ and osteoclasts ⁷⁸ that a pool of inactive Src is bound to β3 integrins in the absence of any ligand binding. This Src pool becomes activated after β3 integrins are activated. Thus, the recruitment of this pool of Src on activation of αvβ3 via mAb AP-5 could have been sufficient to partially overcome the effects of PP2 at relatively low concentrations and could have led to a partial recovery of CLAN formation. 

In summary, these studies suggest that increases in CLAN formation might be attributed to a significant upregulation in one or the other integrin signaling pathways. Clearly, additional studies examining the physiological role of these signaling molecules in CLAN formation are warranted and should be useful in furthering our understanding of the organization of the actin cytoskeleton in the TM and its role in outflow facility and in the pathophysiology of glaucoma. 

Footnotes

 Supported by National Eye Institute Grants EY017006, EY012515 (DMP), EY02698 (PLK), EY16995 (NS), EY018274 (MKS), and Core Grant P30 EY016665 (Department of Ophthalmology and Visual Sciences); a Shaffer Grant from the Glaucoma Research Foundation (MSF); and a Career Development Award from the Research to Prevent Blindness Foundation (NS).

Footnotes

 Disclosure: M.S. Filla, P; M.K. Schwinn, None; N. Sheibani, None; P.L. Kaufman, None; D.M. Peters, P

Footnotes

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.