CAP1 (Cyclase-Associated Protein 1) mediates the cyclic AMP signals that activate Rap1 in stimulating matrix adhesion of colon cancer cells

Abstract

We previously reported that CAP1 (Cyclase-Associated Protein 1) regulates matrix adhesion in mammalian cells through FAK (Focal Adhesion Kinase). More recently, we discovered a phosphor-regulation mechanism for CAP1 through the Ser307/Ser309 tandem site that is of critical importance for all CAP1 functions. However, molecular mechanisms underlying the CAP1 function in adhesion and its regulation remain largely unknown. Here we report that Rap1 also facilitates the CAP1 function in adhesion, and more importantly, we identify a novel signaling pathway where CAP1 mediates the cAMP signals, through the cAMP effectors Epac (Exchange proteins directly activated by cAMP) and PKA (Protein Kinase A), to activate Rap1 in stimulating matrix adhesion in colon cancer cells. Knockdown of CAP1 led to opposite adhesion phenotypes in SW480 and HCT116 colon cancer cells, with reduced matrix adhesion and reduced FAK and Rap1 activities in SW480 cells while it stimulated matrix adhesion as well as FAK and Rap1 activities in HCT116 cells. Importantly, depletion of CAP1 abolished the stimulatory effects of the cAMP activators forskolin and isoproterenol, as well as that of Epac and PKA, on matrix adhesion in both cell types. Our results consistently support a required role for CAP1 in the cAMP activation of Rap1. Identification of the key role for CAP1 in linking the major second messenger cAMP to activation of Rap1 in stimulating adhesion, which may potentially also regulate proliferation in other cell types, not only vertically extends our knowledge on CAP biology, but also carries important translational potential for targeting CAP1 in cancer therapeutics.

1. Introduction

CAP (Cyclase-Associated Protein) was originally identified in the budding yeast S. cerevisiae where it regulates the actin cytoskeleton while also mediating Ras regulation of adenylyl cyclase (1–3). CAP is now best established as a conserved actin-regulating protein in all eukaryotes (1,3–10). Studies focusing mainly on yeast CAP and mammalian CAP1 have found that CAP promotes actin filament turnover by facilitating all the key rate-limiting steps in the cofilin-driven actin dynamics (11), mediated by all three of its structural domains (4,6,10). Mammals possess two CAP isoforms, CAP1 and CAP2 (12,13), and CAP1 is the ubiquitously expressed isoform believed to fulfill the protein functions in most tissue or cell types with CAP2 in a complementary role. Consistent to its function in actin filament turnover, silencing of CAP1 in mammalian cells leads to actin cytoskeletal phenotypes suggesting compromised rate of actin dynamics, especially enhanced stress fibers (14,15). Dynamic actin filament turnover provides the primary driving force for cell migration, which underlies cancer cell invasion and metastasis (16,17), and findings from silencing CAP1 appear to support a required role for CAP1 in the motility and invasiveness in some cell types (15,18–23).

We and others have unraveled additional cell functions for CAP1, including that in matrix adhesion (14,15,21,24,25), as well as in cell proliferation that is limited to certain cell types (19–22,26–28). We first identified the CAP1 function in matrix adhesion in HeLa cells (14), and our consequent findings support conservation of this function in other cells, including breast and pancreatic cancer cells (15,21). However, the CAP1 function in matrix adhesion is cell type-specific, and silencing of CAP1 either stimulates or compromises matrix adhesion (14,15,21). In HeLa and the metastatic breast cancer cells, CAP1 depletion led to enhanced adhesion and elevated activity of FAK (Focal Adhesion Kinase), a key adhesion regulator (29), while knockdown of CAP1 inhibited FAK activity and compromised matrix adhesion in pancreatic and non-metastatic breast cancer cells (14,15,21). Adhesion of cells to the extracellular matrix is critical in regulating a variety of physiological or pathological responses, including cell migration and cancer cell metastasis (30).

Cell adhesion is interconnected to the actin cytoskeleton, as in the example reflected by the role of talin (31), and the two also collaborate in cell migration, which is essential for both normal biological processes such as embryonic development, wound healing and immune response, as well as pathological conditions such as cancer metastasis. Cell adhesion to the matrix provides the traction force essential for pulling the cell body forward in movement. On the other hand, repeated cycles of formation of new adhesions at the leading edge and disassembly of the old adhesions at the trailing edge is required for directional movement of the cell (32). Accordingly, cell adhesion that is too strong or too weak can both hinder cell migration. The role of CAP1 in cell migration is therefore highly complex, given its functions in both the actin cytoskeleton and cell adhesion, and even more so due to the context-dependent nature of the latter. While loss of the CAP1 capacity in promoting actin filament turnover negatively impacts the motility of the CAP1- knockdown cells, in HeLa and metastatic breast cancer cells depletion of CAP1 actually stimulated cell motility and invasiveness (14,21), in which the enhanced adhesion is believed to have overcome the negative effect from the reduced actin dynamics.

We previously identified a phosphor-regulation mechanism for CAP1, which represents the only known regulation mechanism for any CAP homologue. The discovery opened up an uncharted area towards understanding how the cell signaling system controls the relevant cell functions through CAP1 (24,25), whose dys-regulation also underlies human cancers. Multiple cell signals function in concert to facilitate transient phosphorylation at the S307/S309 tandem site, including CDK5 (Cyclin-Dependent Kinase 5) and GSK3 (Glycogen Synthase Kinase 3) as the kinases for the site and cAMP and PKC (Protein Kinase C) signals that induce its dephosphorylation. Transient phosphorylation of CAP1 facilitates alternative binding of CAP1 with cofilin and actin (24), which is critical for CAP1 to facilitate all the key steps in the cycle of cofilin-driven actin dynamics (6). Results from re-expression of the phosphor mutants in the CAP1-knockdown HeLa cells support the critical importance for the phosphor-regulation in the CAP1 function in the actin cytoskeleton (24,25).

Importantly, the phosphor-regulation is also critically important for CAP1 functions in cell adhesion (24,25), migration and invasiveness (15,21), as well as proliferation in cancer cells (21). Thus, CAP1 links the relevant cell signals to control the actin cytoskeleton, cell adhesion, and cell proliferation, which underlie the roles for CAP1 in both the invasiveness and proliferative transformation in cancer cells. However, while the CAP1 function in the actin cytoskeleton is universal and molecular mechanisms well elucidated, the CAP1 functions in adhesion and proliferation are highly cell type-specific, with poorly understood molecular mechanisms. Elucidating how the cell signaling system regulates the actin cytoskeleton and related functions such as adhesion represents an important area in life and biomedical sciences that impacts multiple disciplines, and it will also help us understand the interplay between proliferative transformation and metastatic progression, arguably the two most prominent hallmarks of cancer (33).

The recent finding from Altschuler’s group (34) that CAP1 binds Rap1 and is required for the membrane localization of Rap1 provided a stimulating hint leading to our reasoning that CAP1 may mediate signals from cyclic AMP (cAMP) to activate Rap1, which is another key regulator of cell adhesion. In addition to the role of cAMP in phosphor-regulating CAP1, cAMP also activates Rap1 in stimulating cell adhesion (35–37) while the activated Rap1 can also regulate cell proliferation (38). cAMP is a major second messenger that is activated by a wide variety of extracellular stimuli, including hormones, growth factors, and neurotransmitters that control relevant cell functions including adhesion and proliferation. Thus, CAP1, Rap1, and cAMP share functions in matrix adhesion and proliferation. Interestingly, dephosphorylated CAP1 predominantly localizes to the cell periphery (plasma membrane) (24), suggesting that CAP1 dephosphorylation induced by cAMP may indeed facilitate Rap1 localization to the cell membrane leading to its activation. Despite our extensive effort, we could not identify a phosphatase that executes CAP1 dephosphorylation downstream of cAMP, instead our results support that activated cAMP signaling prevents access of CAP1 to its kinase CDK5 (25). We further unraveled that cytoplasmic effectors of cAMP, Epac (Exchange proteins directly activated by cAMP) and PKA (Protein Kinase A), are both involved in relaying the cAMP signals that induce CAP1 dephosphorylation in HeLa and HEK293T cells (25). Interestingly, Epac is also a GEF (Guanine-Nucleotide-Exchange Factor) of Rap1 itself (39), which activates the G-protein by promoting GTP binding of Rap1 (36).

We report here our findings that knockdown of CAP1 in SW480 and HCT116 colon cancer cells caused opposite matrix adhesion phenotypes that are similar to the case in breast cancer cells (21). We further reveal that elevated activities of both FAK and Rap1 underlie the enhanced matrix adhesion in the CAP1-knockdown HCT116 cells, while both FAK and Rap1 had reduced activities in the CAP1-knockdown SW480 cells that show reduced matrix adhesion. More importantly, we identify a novel signaling pathway, namely cAMP/Epac-PKA/CAP1/Rap1, through which CAP1 mediates the cAMP signals to activate Rap1 in stimulating matrix adhesion in both cell types. Depletion of CAP1 abolished the stimulatory effects of cAMP activators on matrix adhesion, including that of the physiological cAMP activator isoproterenol. Similarly, CAP1 was also indispensable for activators of the cAMP effectors Epac and PKA to stimulate the matrix adhesion in these cells. In addition, CAP1 knockdown caused actin-related phenotypes, including development of multi-nucleated SW480 cells that suggests a disrupted cytokinesis in the knockdown cells that is reported for the first time; however, our results do not support a significant role for CAP1 in the proliferation of colon cancer cells.

2. Materials and methods

2.1. Cell lines, culture media and reagents for manipulating cell signals

Colon cancer cell lines Caco-2 (Catalog# HTB-37), HCT116 (Catalog# CCL-247), SW480 (Catalog# CCL-228) and SW620 (Catalog# CCL-227), and FHC (Catalog# CRL-1831) that serves as the control cell line were all purchased from the ATCC (American Type Culture Collection). FHC cells were cultured in RPMI-1640 medium supplemented with 20% FBS (Fetal Bovine Serum). HCT116, SW480, and SW620 cells were cultured in RPMI-1640 supplemented with 10% FBS, and Caco-2 cells were cultured in DMEM (Dulbecco’s Modified Eagle’s Medium) supplemented with 10% FBS. Cell cultures were maintained at 37°C and supplemented with 5% CO₂. PDGF (Platelet-Derived Growth Factor) was purchased from Fisher Scientific (Waltham, MA). Forskolin (Catalog# F-9929) was from LC Laboratories (Woburn, MA). The biologically inactive forskolin analogue (1,9-Dideoxyforskolin from Coleus forskohlii) and the PKG inhibitor (Catalog# KT5823) were from Sigma-Aldrich (St. Louis, MO). Isoproterenol (Catalog# I6504), the general cAMP analogue 8-Br-cAMP, the specific PKA activator 6-Bnz-cAMP (Catalog# B4560), the specific Epac activator 8PCT-2Me-cAMP (Catalog# C8988), and the PKC activator PMA (Phorbol-12-Myristate-13-Acetate) were all from MilliporeSigma (Burlington, MA). The PKA inhibitor PKI 14–22 amide (myristoylated) and the Epac inhibitor ESI 09 were from Tocris Bioscience (Minneapolis, MN).

2.2. Western blotting and antibodies

Western blotting was performed following standard immunoblotting techniques and signals were detected through ECL (enhanced chemiluminescence). The antibodies against CAP1 (Catalog# sc-376286) used in Western blotting and its beads-conjugated version for immunoprecipitation (Catalog# sc-376286 AC), GAPDH (Glyceraldehyde 3-phosphate dehydrogenase), Cyclin D1, and cofilin were from Santa Cruz Biotechnology (Dallas, TX). The antibodies against phosphor-cofilin Ser3, phosphor-FAK Tyr379, FAK, ERK 1/2, phosphor-ERK 1/2 (Thr202/Tyr204), and Rap1A/Rap1B were from Cell Signaling Technology Inc. (Danvers, MA). The CAP2 antibody was from Invitrogen Life Technologies (Waltham, MA). The phosphor-specific antibody against the Ser307/Ser309 tandem site on mouse CAP1 (or the equivalent Ser308/Ser310 site on human CAP1) was developed in our laboratory and previously described (24). Horseradish peroxidase-conjugated goat anti-rabbit and goat anti-mouse secondary antibodies were from Jackson ImmunoResearch Laboratories Inc. (West Grove, PA).

2.3. Cell culture and treatments

For testing PDGF and serum in inducing CAP1 dephosphorylation, cells were cultured on 6-well plates overnight to approximately 80%−90% confluent, serum-starved for 22 hrs, then treated with PDGF, or the culture media containing 10% serum for the indicated time durations. For treatment of cells used in adhesion assays, the following procedures were followed. For activating cAMP with forskolin, cells were treated with 10 μM forskolin dissolved in DMSO for 120 minutes. For activating cAMP with isoproterenol, cells were treated with 5 μM (SW480) or 10 μM (HCT116) isoproterenol for 30 minutes or 60 min, respectively. For specific activation of Epac, cells were treated with 100 μM 8PCT-2Me-cAMP for 60 minutes, unless indicated otherwise. For specific activation of PKA, cells were treated with 100 μM 6-Bnz-cAMP for 60 minutes. For activating PKC in cell adhesion assays, cells were treated with 0.4 μM PMA for 120 minutes.

2.4. Generation of SW480 and HCT116 clones that harbor stable knockdown of CAP1 or Rap1, and CAP1 re-expression SW480 cells.

For shRNA silencing of CAP1, HCT116 and SW480 cells at approximately 50% confluency were transfected with the vector-based shRNA constructs S2 (targeting nucleotides 519–537) and S3 (targeting nucleotides 1074–1092) of human CAP1 or the empty vector as we did previously (14,15,21,40). The stable clones were established following selection with Neomycin (500 μg/ml) for two weeks. Colonies were isolated, amplified, and clones with efficient CAP1 knockdown were identified through Western blotting. The CAP1-knockdown stable cells were maintained in the presence of a selection pressure by supplementing the culture medium with 250 μg/ml Neomycin. For stable silencing of Rap1, Rap1 shRNA (Catalog# sc-36384-SH) and the scramble shRNA (Catalog# sc-108060) were purchased from Santa Cruz Biotechnology (Dallas, TX). HCT116 and SW480 cells were transfected using the FuGENE^® HD DNA Transfection Reagent (Dublin, Ohio). The stable clones were selected with puromycin (3 μg/ml) for two weeks. Colonies were isolated, amplified, and clones with efficient Rap1 knockdown were identified through Western blotting. The established Rap1-knockdown stable clones were maintained in the presence of a selection pressure by supplementing the culture medium with 1.5 μg/ml puromycin. For generating stable cells that re-express WTCAP1 or the phosphor mutant, SW480 CAP1-knockdown clone S3#2 was used, following procedures we used preciously as described (15,21,24,25). Stable re-expression clones were established through selection with Zeocin (80 μg/ml) for two weeks. The Re-expression clones were maintained in the presence of the selection pressure of both Neomycin (250 μg/ml) and Zeocin (50 μg/ml).

2.5. Rap1 activation assays

Rap1 activation assays were performed using the Rap1 Activation assay kit (Catalog# STA-406–1) from Cell Biolabs, Inc. (San Diego, CA), following the provided protocol with minor modifications. Briefly, control or treated cells were lysed, and the cell lysates were incubated with 20 μl RalGDS RBD agarose beads that binds the GTP-bound active Rap1 specifically. Positive and negative controls were created by pre-loading Rap1 in the cell lysate with GTP or GDP, respectively. After precipitating the RBD beads and washing off the non-specific binding, the co-precipitated GTP-bound Rap1, which is the active form, was detected in Western blotting using the antibody against Rap1.

2.6. Phase imaging and immunofluorescence

Light and fluorescent microscopy imaging were performed similarly as we did previously (14,15,21,24). For staining the actin cytoskeleton, cells were plated on fibronectin-coated, glass bottom 6-well plates (Mountain View, CA), fixed with 4% paraformaldehyde for 15 minutes, washed twice with PBS (Phosphate-Buffered Saline), and permeabilized with PBS containing 0.1% Triton X-100 for 5 minutes. Cells were then incubated with Alexa-Fluor 488 Phalloidin for 1 hour, washed three times each for 5 minutes with PBS, and mounted using Vectashield mounting medium from Vector Laboratories Inc. (Burlingame, CA). Images were taken with florescent microscopy using the Bio Tek Cytation 5 imaging system. For measuring cell areas, cells were plated on 6-well plates pre-coated with fibronectin overnight, followed by phase imaging using 10x objective lens with the Bio Tek Cytation 5 imaging system. The ImageJ software (rsb.info.nih.gov/ij) was used to measure areas of the selected 15 cells per field in each of three independent experiments. For immunostaining of the re-expressed WTCAP1 or the S307A/S309A mutant with Rap1 in HeLa cells, the rabbit polyclonal CAP1 antibody (Catalog# 16231–1-AP) from ThermoFisher (Waltham, MA) and a mouse monoclonal Rap1 antibody (Catalog# 610196) from BD Biosciences (San Jose, CA) were used, followed by detection with an Alexa Flour 594 and Alexa Flour 488 secondary antibody, respectively. Images were taken using the confocal (Z-Stack) function of the Bio Tek Cytation 5 imaging system.

2.7. Cell adhesion assays and wound healing assays

Cell adhesion and wound healing assays were conducted similarly as we did previously (14,15,21). For cell adhesion assays, approximately 5×10⁴ cells were plated on each well of a 6-well plate coated with fibronectin and incubated at 37° C for the indicated time durations. After incubation, the unattached cells were washed off with 1x PBS and the attached cells were scored in three representative fields in each well. Images were taken using Bio Tek Cytation 5. For wound healing assays, cells were plated on 6-well plates overnight to reach full confluences, and a wound was then introduced to the cell monolayer and cells were further incubated for 18 hours at 37° C. Phase images were taken using Bio Tek Cytation 5 following introduction of the wound as well as at the end of the assay. The width of wounds was measured for calculating the cell motility using the following formula: [cell motility = (wound width a 0 hr-wound width at n hr)/wound width at 0 hr].

2.8. Cellular senescence assays

Cellular senescence assays were performed by detecting the SA-β-gal activity, following the protocol provided with the Cellular Senescence assay kit (KAA002) from Millipore Sigma (Burlington, MA). Cells were plated overnight on a 6-well plate and were washed twice with PBS and fixed the next day with 1x Fixing Solution for 10 minutes at room temperature. Following the fixation, cells were washed twice with PBS then 1x SA- β-gal detection solution was added to cells and incubated overnight (~17 hrs) at 37° C and away from the light. Cells were then washed twice with PBS and the blue stained cells were counted and imaged under the color brightfield microscopy using Bio Tek Cytation 5.

2.9. Transient CAP1 knockdown mediated by siRNA and BrdU cell proliferation assays

Transient knockdown of CAP1 was performed using CAP1 siRNA and required reagents from Santa Cruz Biotechnology (Dallas, TX), following the protocol provided by the manufacturer. The reagents include Control siRNA (sc-37007), CAP1 siRNA (sc-88068), siRNA Transfection Reagent (sc-29528), and siRNA Transfection Medium (sc-36868). 48 hours after siRNA transfection, the BrdU cell proliferation assay was performed using the BrdU Cell Proliferation ELISA Kit from Cell BioLabs, Inc. (San Diego, CA), following the manufacturer’s protocol. Approximately 2×10⁴ cells were plated in triplicate onto wells of a 96-well plate for both treated and untreated cells, 24 hours after siRNA transfection. Cells were maintained overnight at 37°C and 5% CO₂ before proceeding to the BrdU assay. 10x BrdU Solution was added to the treated wells but not to the untreated wells, and cells were further incubated at 37°C and 5% CO₂ for 4 hours. Following incubation with the BrdU Solution, cells were fixed and incubated with Anti-BrdU, the HRP-conjugated secondary Antibody, and Substrate Solution following the suggested time duration between the steps. Stop Solution was then added after cells were incubated with the Substrate Solution for 5 min at room temperature on an orbital shaker. The absorbance was measured at 450 nm using Bio Tek Epoch 96-Well Plate Reader, and analyzed using Student’s t-test.

3. Results

3.1. No evidence for up-regulation of CAP1 in colon cancer cells

Concerning the role of CAP1 in cancer cell invasiveness, a speculation was that up-regulation of CAP1 would enhance the rate of actin filament turnover and thus leads to elevated motility and invasiveness of cancer cells. Our previous studies, however, did not find any supporting evidence for the case in breast or pancreatic cancer cells (15,21). A public database from comparison of CAP1 expression levels in tissues of different cancer types, http://www.proteinatlas.org/ENSG00000131236-CAP1/cancer, suggests colorectal cancer samples may have enhanced CAP1 staining. We decided to examine CAP1 expression levels in colon cancer cells, by detecting that in a panel of four commonly employed colon cancer cell lines, including SW480 that was derived from a Dukes’ Type B colon adenocarcinoma (41), SW620 that was derived from the same patient where SW480 is derived, although cancer had spread to nearby site (42), HCT116 that is a rare cell line derived from carcinoma (43–46) that belongs to Dukes’ Type D and Caco-2 (47) cells. As a control cell line, we used the untransformed fetal colon cell line FHC that has frequently been used in similar studies (48–50). As shown in Figure 1A (left), Western blotting results show no up-regulation of CAP1 in any of the four cancer cell lines as compared to that in the control cells. All the four cancer cell lines express readily detectable levels of CAP1. However, the FHC cells had even noticeably higher CAP1 levels, as normalized by GAPDH. We also detected expression of the other CAP isoform, CAP2, and found overall modest expression levels in the cancer cell lines, where the control FHC cells again had noticeably higher expression of CAP2 (Figure 1A; left). Lastly, the CAP1 expression levels in the colon cancer cells were overall comparable to those in a number of other cell types that we previously found to express abundant CAP1 levels, including HeLa, HEK293T, and MCF-7 and MDA-MB-231 breast cancer cells (Figure 1A; right).

Figure 1. CAP1 knockdown caused actin-related phenotypes in colon cancer cells, including multi-nucleated SW480 cells, whereas no evidence supports up-regulation of CAP1 in the colon cancer cells.

(A) Western blotting reveals no up-regulation of CAP1 in colon cancer cells compared to the control FHC cells. Comparable CAP1 expression levels were detected in colon cancer cells to that in a number of other cell lines. We also detected modest expression levels of CAP2, again with no up-regulation in the cancer cells. GAPDH serves as the loading control. (B) Confirmation of CAP1 knockdown by Western blotting in the SW480 and HCT116 stable clones derived from two independent shRNA constructs, S2 and S3, that target separate nucleotide sequences. “*” indicates the clones with efficient CAP1 silencing that were selected for further studies. (C) CAP1-knockdown SW480 cells had enhanced actin stress fibers. Cells cultured on glass bottom plates overnight were fixed, permeabilized and stained with Alexa Fluor 488 Phalloidin. Images were taken under fluorescent microscopy using Bio Tek Cytation 5 imaging system, and arrows indicate the enhanced stress fibers, or lack thereof by arrowheads. (D) Knockdown of CAP1 in HCT116 cells led to increased cell size as well as enhanced lamellipodia. Cells were plated on fibronectin for 20 hrs and phase images were taken. For quantifying cell size, areas of 15 cells were measured using ImageJ in each of the three independent experiments. The quantified results were statistically analyzed using one-way ANOVA where error bars represent S.E.M. For quantifying lamellipodia, three areas each with 100 cells were counted and cells harboring two or more distinct lamellipodia were scored (indicated with arrows). (E) Knockdown of CAP1 in SW480 cells led to increased cell size, as well as development of multi-nucleated cells (indicated with arrows). Cell areas were measured similarly as that for HCT116 cells. The number of multi-nucleated cells was scored in cells cultured for ~72 hours, in representative areas each having 100 cells from three independent experiments. The quantified results were analyzed using one-way ANOVA and plotted in the graph where the error bars represent S.E.M. (F) SA-beta-gal staining experiments show that the multi-nucleated CAP1-knockdown SW480 cells were not derived from cellular senescence. No significant difference in the percentage of beta-galactosidase positive cells were detected in the control and CAP1-knockdown SW480 clones. The quantified results were analyzed using one-way ANOVA and plotted in the graph where the error bars represent S.E.M. “*” indicates P < 0.05, “**” indicates P < 0.01, “***” indicates P < 0.001 and “****” indicates P < 0.0001.

3.2. Depletion of CAP1 led to actin-related phenotypes, including development of multi-nucleated SW480 cells

Studies including ours have established CAP1 as a cytoskeletal protein that facilitates the cofilin-driven actin dynamics in mammalian cells. Depletion of CAP1 in cells leads to enhanced actin stress fibers across cell types (14,15). To determine the CAP1 functions in colon cancer cells, we next pursued RNAi-mediated silencing of CAP1. Using two shRNA constructs, S2 and S3, that we previously developed that target independent nucleotide sequences on CAP1 and effectively silenced CAP1 in multiple cell types (14,15,21,40), we were able to generate CAP1-knockdown stable clones for both SW480 and HCT116 cells. Figure 1B shows Western blotting results confirming efficient knockdown of CAP1 in multiple stable clones derived from each construct and cell line. We did not pursue the shRNA silencing in SW620 and Caco-2 cells, because certain characteristics of these cells, such as the extremely weak adherence of SW620 cells (not shown), make it very unlikely for these cells to yield stable clones.

When we stained the actin cytoskeleton in the knockdown cells, as expected, we detected remarkably enhanced stress fibers in the CAP1-knockdown SW480 and HCT116 cells, a phenotype we observed in HeLa (14) and pancreatic cancer (15) cells. This phenotype is believed to derive from the reduced rate of actin filament disassembly and turnover due to the loss of the CAP1 function in promoting actin dynamics, as well as the loss of the CAP1 capacity in sequestering actin monomers in helping maintain a high concentration of monomeric actin in the cell by preventing polymerization of them into filaments (23). As shown in Figure 1C, actin stress fibers were considerably enhanced in the CAP1-knockdown SW480 cells stained with fluorescent Phalloidin, as compared to the cells harboring an empty knockdown vector. It is noted that colon cancer cells have in general less organized stress fibers than that in some cell types, such as NIH3T3 and HeLa cells (14,24) that are commonly used for actin cytoskeleton studies. In SW480 cells, out of the 30 control (Vec) cells examined, only 3 of them (10%) had a modest presence of stress fibers. In comparison, 21 out of 37 (56.8%) of the S2#2 and 19 out of 39 (48.7 %) cells of the S3#2 CAP1-knockdown clones examined had remarkably enhanced stress fibers. Similar observations were made in HCT116 cells, in which out of the 23 control (Vec) cells examined, only 2 of them (8.7%) had a modest presence, whereas the remaining cells had a minisule amount, of stress fibers. In comparison, 26 out of 38 (68.4%) of the S2#2 and 21 out of 33 (63.6 %) of the S3#1 CAP1-knockdown cells examined had remarkably enhanced stress fibers.

The actin cytoskeleton is critical in maintaining the integrity and morphology of the cell. Deletion of the CAP gene in both the budding yeast S. cerevasiae and fission yeast S. pombe led to cell morphological changes accompanied by a disrupted actin cytoskeleton (1,51,52). In mammalian cells, knockdown of CAP1 led to increased size in HeLa and metastatic breast cancer cells (14,21), in which the enhanced matrix adhesion is believed to have contributed to the phenotype. In contrast, depletion of CAP1 did not cause remarkable change in the size of pancreatic cancer cells, in which the FAK activity was reduced (15). We measured the cell areas in the CAP1-knockdown colon cancer cells, and the statistical analyses reveal that both the CAP1-knockdown HCT116 (Figure 1D) and SW480 (Figure 1E) cells had significantly increased cell size. Additionally, remarkably enhanced peripheral protrusions (lamellipodia) were observed in the CAP1-knockdown HCT116 cells, and Figure 1D shows the statistically analyzed data of the enhanced lamellipodia development. In contrast, quantification and statistical analyses reveal that CAP1 knockdown did not promote lamellipodia development in SW480 cells.

Interestingly, we also observed unusually high numbers of the CAP1-knockdown SW480 cells were multi-nucleated, as shown in Figure 1E. Statistical analyses of the quantified data reveal significantly increased percentage of multi-nucleated cells in the knockdown cells. Moreover, in order to exclude the possibility that the multi-nucleated phenotype was a result of senescence, we conducted SA-beta-gal staining assays, and our results show no significant difference in the staining (Figure 1F). These results suggest that the CAP1 function is required for the actin cytoskeletal rearrangement in the cytokinesis of SW480 cells. Indeed, we previously observed a similar phenotype in the CAP1-knockdown HeLa cells, albeit not as robust, for which we did not pursue further (unpublished results from our laboratory).

3.3. CAP1 knockdown caused opposite alterations in matrix adhesion and FAK activity in SW480 and HCT116 cells

We previously reported that knockdown of CAP1 caused opposite matrix adhesion phenotypes in different cell types, accompanied by consistently altered FAK activities (14,15,21). As shown in Figure 2A (left panel), we found that both the CAP1-knockdown SW480 clones derived from the S2 and S3 shRNA constructs had significantly reduced adhesion on fibronectin. When we tested effect of the CAP1 depletion on SW480 cell motility in the wound healing assays, the knockdown SW480 cells actually had significantly increased motility (Figure 2B; left panel). In contrast, both the CAP1-knockdown HCT116 clones derived from S2 and S3 had significantly increased adhesion on fibronectin (Figure 2A; right panel). When tested in the migration assays, the CAP1-knockdown HCT116 cells also had significantly increased motility (Figure 2B; right panel). While cell adhesion provides the traction force essential for pulling the cell body forward during migration, too strong a cell adhesion also hinders cell migration by making the turnover of adhesions difficult. We speculate that the exceptionally strong adherent property of the SW480 cells had a negative impact on motility, and the compromised adhesion from CAP1 depletion actually helped in relieving the negative effect.

Figure 2. Knockdown of CAP1 caused opposite alterations in FAK activity and matrix adhesion, and distinct alterations in cofilin activity in SW480 and HCT116 cancer cells, while increased motility in both cell types.

(A) Knockdown of CAP1 in SW480 cells reduced cell adhesion on fibronectin whereas it stimulated adhesion in HCT116 cells. Approximately 5×10⁴ cells were plated onto each well of a 6-well plate, and cells remain unattached at 120 minutes were washed off with PBS. The numbers of attached cells were scored in three representative fields under the microscope and images were taken using the Bio Tek Cytation 5 system. Numbers from three independent assays each with three replicates were quantified and statistically analyzed using one-way ANOVA, and plotted on the graph where error bars represent S.E.M. (B) Depletion of CAP1 led to significantly increased motility in both SW480 and HCT116 colon cancer cells. Cells were cultured overnight to confluent on 6-well plates followed by introduction of a wound, cells were further cultured for 18 hours and images were taken at both time points. Quantified cell motility was statically analyzed using one-way ANOVA and plotted in the graph where error bars represent S.E.M. (C) CAP1-knockdown SW480 cells had reduced FAK activity, whereas the CAP1-knockdown HCT116 cells had elevated FAK activity in two out of the three knockdown clones. The FAK activity was assessed using a phosphor-specific antibody against Tyr397 on FAK in Western blotting. In one out of the three SW480 CAP1-knockdown clones, there was also a significant decrease in total FAK levels as compared to that in the control (Vec) cells when normalized by GAPDH. The signals were quantified using ImageJ, followed by analysis using one-way ANOVA before plotted in the graphs where the error bars represent S.E.M. (D) Knockdown of CAP1 did not cause remarkable alterations in the cofilin activity in SW480 cells, whereas it significantly stimulated the cofilin activity in HCT116 cells. The cofilin activity was assessed by detecting the inhibitory phosphorylation at Ser3 on cofilin in Western blotting. Signals from three independent experiments were quantified using ImageJ, analyzed using one-way ANOVA and plotted in the graph where the error bars represent S.E.M. “*” indicates P < 0.05, “**” indicates P < 0.01, “***” indicates P < 0.001 and “****” indicates P < 0.0001.

We next examined if alterated FAK activity underlies the opposite matrix adhesion phenotypes in the CAP1-knockdown SW480 and HCT116 cells, by assessing FAK activity using a phosphor-specific antibody against Tyr397 that detects activated FAK. Indeed, significantly reduced FAK activity was detected in the CAP1-knockdown SW480 clones that have reduced adhesion, as indicated by reduced ratio of phosphorylated FAK to total FAK (Figure 2C; left panel). Since the total FAK levels in the knockdown clones also appeared to be lowered, we furtehr quantified the FAK/GAPDH ratio, and statistical analyse results show that in two out of the three clones, the down-regulation did not reach statictical significance. In contrast, knockdown of CAP1 actually stimulated FAK activity in HCT116 cells (Figure 2C; right panel), with two out of the three knockdown clones (S2#2 and S3#1; the clones with more complete CAP1 knockdown) showing significantly elevated FAK activities. These opposite alterations in FAK activity are consistent with the opposite matrix adhesion phenotypes in the CAP1-knockdown SW480 and HCT116 cells, similar to the case in the breast cancer cells (21).

3.4. CAP1 depletion led to cofilin activation in HCT116 cells but had no effect on colon cancer cell proliferation

We previously reported that CAP1 knockdown in HeLa, pancreatic cancer, and breast cancer cells also caused cell type-specific alterations in the activity of cofilin (14,15,21), which is a key ADF (Actin Depolymerization Factor). We next looked into possible alterations in coflin in the knockdown cells. As shown in Figure 2D (left panel), no remarkable alterations in the expression level or the activity of cofilin were detected in CAP1-knockdown SW480 cells, with the latter assessed using an antibody that detects the inhibitory phosphorylation at Ser3 (53). In contrast, CAP1-knockdown HCT116 cells had significantly elevated cofilin activity, as indicated by the reduced ratio of phosphorylated cofilin to total cofilin (Figure 2D; right panel). Thus, knockdown of CAP1 caused cell type-specific alterations in the activity of cofilin in the colon cancer cells, although the case is somewhat different from the opposite alterations of cofilin in breast cancer cells (21). The elevated cofilin activity in the CAP1-knockdown HCT116 cells is believed to have contributed to the phenotype of enhanced lamellipodia (Figure 1D) and the elevated motility (Figure 2B; right panle) in these cells.

We did not observe any noticeable changes in the proliferation of the CAP1-knockdown colon cancer cells (not shown). When we examined the activity of ERK (Extracellular signal-Regulated Kinase) using a phosphor-specific antibody that detects activated ERK as we used with the breast cancer cells (21), no remarkable alterations were detected in either the expression levels or the phosphorylation of ERK in the CAP1-knockdown cells (Figure 3A). Quanitified and statistically analyzed results reveal no siginificant alterations in ERK activity, as shown in the graphs (Figure 3A). To address the concern of the stable knockodwn paradigm may have caused a selection bias, we nxet direcetly tested effects of CAP1 knockdown by using a siRNA-mediated transient CAP1 knockdown approach combined with BrdU cell proliferation assays. As shown in Figure 3B, we were able to achieve efficient CAP1 depletion in both SW480 and HCT116 cells, and consistent with the ERK results from the stable knockdown clones, transient CAP1 knockdown did not cause significant alterations in cell proliferation. Taken together, our findings consistently support that CAP1 does not play a significant role in the proliefartion of colon cancer cells.

Figure 3. Knockdown of CAP1 did not have a significant effect on ERK activity or cell proliferation.

(A) Depletion of CAP1 did not cause noticeable alterations in the activity or expression levels of ERK in SW480 or HCT116 cells. A phosphor-specific antibody against Thr202/Tyr204 was used to detect activated ERK. Results were quantified using ImageJ as and statistically analyzed using one-way ANOVA. (B) Transient CAP1 knockdown mediated by siRNA in SW480 and HCT116 cells had no effect on cell proliferation, as detected in BrdU proliferation ELISA assays. The transient CAP1 knockdown was confirmed through Western blotting, and cell proliferation was assessed through BrdU assays. The absorbance was measured at 450 nm and results for three replicates were analyzed using Student’s t-test and plotted on the graphs where the error bars represent S.E.M.

3.5. Altered Rap1 activities consistent to the adhesion phenotypes were detected in the CAP1-knockdown cells

Our findings thus far support that FAK and talin, which facilitate integrin activation (29,54), are involved in facilitating the CAP1 function in adhesion. The recent finding that CAP1 interacts with Rap1 (34) suggests Rap1 may also facilitate the CAP1 function in adhesion, since the small G-protein is also a key adhesion regulator (35,55). We found that colon cancer cells express readily detectable levels of Rap1, roughly comparable to that in a few other cell types (Figure 4A; upper panel). Notably, HCT116 cells express substantially higher levels of Rap1 than in SW480 cells, while the Rap1 expression in SW620 cells was between those in SW480 and HCT116 cells. We also found that knockdown of CAP1 in SW480 or HCT116 cells did not cause noticeable changes in the expression levels of Rap1 (Figure 4A; lower panel). Interestingly, remarkably reduced Rap1 activity was detected in both the CAP1-knockdown SW480 cells, derived from S2 or S3, in the Rap1 activation assays based on RalGDS RBD (56), as compared to that in the control cells (Figure 4B; upper panel). Quantified and statistically analyzed results show that the Rap1 activity was significantly reduced in the S2#2 clone we examined, as shown in the graph in Figure 4B (upper panel). In contrast, remarkably elevated Rap1 activity was detected in both the CAP1-knockdown HCT116 clones that had enhanced matrix adhesion (Figure 4B; lower panel). Again, quantified and statistically analyzed results show that the Rap1 activity was significantly elevated in the S2#2 CAP1-knockdown HCT116 clone examined, as shown in the graph in Figure 4B (lower panel). These results support that while knockdown of CAP1 did not alter Rap1 expression levels, Rap1 activities were altered oppositely in the CAP1-knockdown SW480 and HCT116 cells and the altered Rap1 activities underlie the opposite matrix adhesion phenotypes in the knockdown cells, along with the altered FAK activities. The reduced activities in both FAK and Rap1 led to the reduced matrix adhesion in the CAP1-knockdown SW480 cells, whereas the elevated activities in FAK and Rap1 stimulated matrix adhesion in the CAP1-knockdown HCT116 cells. These findings also suggest a possible co-regulation between FAK and Rap1, for facilitating the CAP1 function in adhesion in a collaborative manner.

Figure 4. CAP1-knockdown SW480 and HCT116 cells had oppositely altered Rap1 activities that are consistent with the opposite matrix adhesion phenotypes.

(A) HCT116 cells express considerably higher levels of Rap1 than in SW480 cells (upper panel). Rap1 expression levels were detected in the colon cancer cells along with that in a number of other cell lines. Knockdown of CAP1 did not cause noticeable alterations in Rap1 expression in HCT116 or SW480 cells (lower panel). (B) CAP1 knockdown in SW480 and HCT116 cells caused opposite alterations in Rap1 activity, as detected in the RalGDS RBD-based Rap1 activation assays. Reduced Rap1 activity was detected in the SW480 knockdown clones (upper panel), while the CAP1-knockdown HCT116 cells actually had elevated Rap1 activity (lower panel) than the control cells. An irrelevant lane has been removed from the original Western blot image in the SW480 results (dashed line indicates the position during figure composition). Comparable levels of total Rap1 in the tested samples were confirmed, as in the signals labeled “Total Rap1.” “Vec” indicates lysates from cells harboring an empty vector for the shRNA in silencing CAP1. GTP and GDP indicate the samples where the cell lysates were loaded with GTP or GDP for serving as positive and negative control, respectively. Signals were quantified using ImageJ, and the ratios of active Rap1 versus total Rap1 were analyzed using Student’s t-test, plotted on the graphs where error bars represent S.E.M. “*” indicates P < 0.05, “**” indicates P < 0.01.

To summarize, knockdown of CAP1 in SW480 and HCT116 cells caused opposite phenotypes in matrix adhesion as well as opposite alterations in FAK and Rap1 activities. The depletion of CAP1 also caused phenotypes in some actin-dependent cell functions in both cell lines, including the novel function in cytokinesis. Our results do not support a significant role for CAP1 in the proliferation of these cancer cells. Table 1 provides a summary of whether CAP1 is involved in the known cell functions for the protein in the colon cancer cells, as well as alterations in the activities of the regulators in the CAP1-knockodwn cells.

Table 1.

Cell type-specific effects of CAP1 knockdown on colon cancer cell functions and their regulators

CAP1 KD effect	SW480	HCT116
Rapl activity	↓	↑
FAK activity	↓	↑
Matrix adhesion	↓	↑
cofilin activity	—	↑
Lamellipodia	—	↑
Cell migration	↑	↑
ERK activity	—	—
Cell proliferation	—	—

The symbol (−) indicates no change. The upward facing arrow (↑) indicates increase. The downward facing arrow (↓) indicates decrease.

3.6. Activated cAMP signaling induced CAP1 dephosphorylation in SW480 and HCT116 cells

We previously identified a phosphor-regulation mechanism for CAP1 through the S307/S309 tandem site (24,25), in which multiple cell signals function in concert to facilitate the transient phosphorylation of CAP1. The phosphor-regulation is critical for CAP1 functions in the actin cytoskeleton and adhesion (24,25), as well as that in cell migration, invasiveness and proliferation (15,21). We reported that activated cAMP and PKC signals both induce CAP1 dephosphorylation, and our findings support that the dephosphorylated CAP1 is the active form for the protein functions in the actin cytoskeleton, cell adhesion and proliferation (15,21,24,25). Additionally, we also found that PDGF (Platelet-Derived Growth Factor) induced CAP1 dephosphorylation in pancreatic cancer cells as well (15). We next tested if these cell signals also induce CAP1 dephosphorylation in colon cancer cells, with an emphasis on the cAMP signals due to the hypothesized regulation of the CAP1/Rap1 axis by cAMP in cell adhesion as mentioned earlier. As shown in Figure 5A, the cAMP activator forskolin (IC₅₀ at 41 nM), which directly activates adenylyl cyclase (57), effectively induced CAP1 dephosphorylation in both SW480 and HCT116 cells. Statistical analyses of the quantified data show that the forskolin effect was significant. Importantly, we found that the inactive forskolin derivative (1,9-Dideoxyforskolin; IC₅₀ at 1.6 μM), which serves as a negative control for forskolin, did not induce CAP1 dephosphorylation. Moreover, isoproterenol (IC₅₀ at 12 nM), which is a physiological activator of the cAMP signaling that functions through the β-adrenergic receptor (58), also significantly induced CAP1 dephosphorylation in these cells (Figure 5B). Lastly, the cAMP analogue 8-Br-cAMP (IC₅₀ at 0.84 mM) also induced CAP1 dephosphorylation in these cells (Supplementary Figure 1A). Together, these results firmly establish the role of cAMP signaling in phosphor-regulating CAP1 functions in colon cancer cells, by inducing CAP1 dephosphorylation. We next tested the PKC activator PMA (Phorbol-12-Myristate-13-Acetate; IC₅₀ at 16–20 nM for several characterized isoforms), and found that it also induced CAP1 dephosphorylation in SW480 and HCT116 cells (Supplementary Figure 1B; not shown but in SW620 cells as well). Similarly, PDGF also induced CAP1 dephosphorylation in SW480 and HCT116 cells (Supplementary Figure 1C), but had no effect in SW620 cancer cells (not shown), consistent to the report that SW620 cells do not express PDGF receptors (59). Notably, the strongest PDGF effect was detected at the 10-minute time point of the treatment, which is a comparatively delayed response than at 5 minutes in pancreatic cancer cells (15). Lastly, serum also induced CAP1 dephosphorylation in these cells (Supplementary Figure 1D), with the effect started from the 1-hr time point and lasted throughout 4 hrs, the longest time point examined.

Figure 5. Activated cAMP signaling induced CAP1 dephosphorylation in SW480 and HCT116 cells, and CAP1 is required for cAMP signals to activate Rap1.

(A) Activation of cAMP signaling by forskolin (Forsko) treatment induced CAP1 dephosphorylation at the S307/S309 tandem site in both SW480 and HCT116 cells. In contrast, the biologically inactive forskolin derivative (1,9-Dideoxyforskolin), which serves as a negative control [(−) Ctrl], did not have the effect. Three independent western blot signals were quantified using ImageJ, analyzed using one-way ANOVA and plotted on the graphs where the error bars represent S.E.M. (B) Treatment of the cells with isoproterenol, a physiological activator of cAMP signaling, also induced CAP1 dephosphorylation in SW480 and HCT116 cells. Treatment with isoproterenol for 30 minutes induced remarkable CAP1 dephosphorylation in SW480 cells, while in HCT116 cells the isoproterenol effect was the most remarkable at the120-minute time point. Statistical analyses were performed similarly as in panel A. (C) Forskolin treatment stimulated Rap1 activity in SW480 cells, with the strongest effect appearing at the 60-minute time point (upper panel). The specific activator of Epac, 8PCT-2Me-cAMP (Epac Act.), also activated Rap1 in SW480 cells, with the stimulatory effect compromised in the CAP1-knockdown cells (lower panel). Cells were treated with 8PCT-2Me-cAMP (100 μM) for 30 minutes. One irrelevant lane has been removed from the original Western blot image (dashed line indicates the position). Statistical analyses were performed similarly as in panel A. (D) Isoproterenol (Iso) stimulated Rap1 activity in HCT116 cells, with the strongest effect showing at the 1-hr time point, which was abolished in the CAP1-knockdown cells (top panel). Forskolin and the specific Epac activator also stimulated Rap1 activity in HCT116 cells, and the stimulatory effects were again compromised in the CAP1-knockdown cells (middle panel). In a separate Rap1 activation assay, 2-hr forskolin treatment significantly activated Rap1 (bottom panel). The concentrations of forskolin and isoproterenol used in Rap1 activation assays were the same as used for the respective cell lines in panels A-B. Signals from three independent experiments were quantified using ImageJ, and the ratios of active Rap1 versus total Rap1 were analyzed using Student’s t-test, and plotted on the graph where error bars represent S.E.M. “*” indicates P < 0.05, “**” indicates P < 0.01 and “***” indicates P < 0.001.

3.7. CAP1 is required for cAMP signals to activate Rap1 in SW480 and HCT116 cells

Literature and our own findings support a hypothesized model that dephosphorylation of CAP1 mediates the cAMP signals to activate Rap1 in stimulating matrix adhesion in colon cancer cells. We next tested if CAP1 is required for cAMP activation of Rap1. Treatment of SW480 cells with the cAMP activator forskolin led to remarkable activation of Rap1, with the strongest effect at the 30-minute and at 60-minute time points (Figure 5C; upper panel). When we tested the cAMP analogue 8PCT-2Me-cAMP, which specifically activates the cAMP effector Epac but not PKA, it also activated Rap1 (Figure 5C; lower panel). Interestingly, the activation of Rap1 by the Epac activator was abolished in the CAP1-knockdown SW480 cells. Similarly, the specific activator of PKA 6-Bnz-cAMP (IC₅₀ at 75 nM) also activated Rap1 in SW480 cells (Supplementary Figure 2). These results suggest that both Epac and PKA, which we previously reported to relay the cAMP signals leading to CAP1 dephosphorylation in HeLa and HEK293T cells, also relay the cAMP signals that activate Rap1 in SW480 cells. We also found that forskolin, isoproterenol, and the Epac activator 8PCT-2Me-cAMP all caused Rap1 activation in HCT116 cells (Figure 5D). Again, the stimulatory effects of these cAMP activators on Rap1 were all compromised by the knockdown of CAP1 in HCT116 cells (Figure 5D). We further quantified and statistically analyzed the Rap1 activation by forskolin in HCT116 cells, and as shown in the graph in Figure 5D (bottom panel), forskolin significantly stimulated Rap1 activity. Taken together, our results support a required role for CAP1 in mediating the cAMP signaling, relayed by its effectors Epac and PKA, to activate Rap1 in both SW480 and HCT116 cells.

3.8. CAP1 is indispensable for activated cAMP signals to stimulate matrix adhesion in SW480 and HCT116 cells

We hypothesize that cAMP signals induce CAP1 dephosphorylation, and through the CAP1/Rap1 axis to stimulate matrix adhesion. Activated cAMP signaling has been reported to stimulate matrix adhesion (35,37), and we found forskolin treatment stimulated matrix adhesion in colon cancer cells as well. As shown in Figure 6A (left panel), forskolin significantly stimulated matrix adhesion in both SW480 and HCT116 cells. Importantly, the inactive forskolin analog, which failed to induce CAP1 dephosphorylation (Fig.5A and Figure 6A; right panel), also failed to stimulate adhesion in these cells. Moreover, pre-treatment of the cells with KT5823, an inhibitor for Protein Kinase G (IC₅₀ at 60 nM), did not block the stimulatory effect of forskolin on the matrix adhesion or the forskolin-induced CAP1 dephosphorylation in SW480 and HCT116 cells (Figure 6A; right panel). Together, these results strongly support that the forskolin effects in inducing CAP1 dephosphorylation and stimulating matrix adhesion are specific to the activation of the cAMP signaling and mediated by cAMP effectors Epac and PKA, since perturbation of the similar cGMP signaling did not interrupt these effects. These results also support that the role in inducing CAP1 dephosphorylation underlies the stimulation of matrix adhesion by activated cAMP signaling. Our findings also support that Rap1 localization to the cell membrane, facilitated by the cAMP-induced CAP1 dephosphorylation, plays a key role in the Rap1 activation by cAMP. As we previously reported, dephosphorylated CAP1 shows a predominant cell peripheral (membrane) localization, whereas phosphorylated CAP1 predominantly localizes to the cytosol (24). Further supporting this scenario is that in the CAP1-knockdown HeLa cells stably re-expressing the S307A/S309A (AA) mutant that mimics the dephosphorylated CAP1, remarkably enhanced co-localization of Rap1 to the cell periphery along with the AA mutant was detected, as compared to that in the cells re-expressing WTCAP1 (Supplementary Figure 3).

Figure 6. Evidence supporting a required role for CAP1 dephosphorylation in mediating the cAMP signals that stimulate matrix adhesion in colon cancer cells.

(A) Activation of cAMP signaling by forskolin (Forsko) induced CAP1 dephosphorylation (right panels), and also significantly stimulated matrix adhesion in SW480 and HCT116 cells. The biologically inactive forskolin analog (− Ctrl Forsko) had neither of the effects. Pre-inhibition of PKG with KT5823 (PKG inhi.) did not block the effects of forskolin. Cells were treated with forskolin (10 μM) or the inactive forskolin analog (10 μM) for 120 minutes, or pre-treated with KT5823 (234 nM) for 60 minutes followed by the forskolin treatment, cells were then used for Western blotting or the cell adhesion assays. Data from three independent assays each having three replicates were statistically analyzed using one-way ANOVA and plotted on the graphs where the error bars represent S.E.M. (B) The stimulatory effects of cAMP activators forskolin and isoproterenol on matrix adhesion were abolished by knockdown of CAP1 in SW480 cells. Forskolin (Forsko) and isoproterenol (Iso) both significantly stimulated adhesion of SW480 cells on fibronectin, but their effects were abolished in the CAP1-knockdown clones. Cells were treated with forskolin (10 μM) for 120 minutes or isoproterenol (5 μM) for 60 minutes before they were detached for adhesion assays. Data from three independent assays each having three replicates were statistically analyzed using two-way ANOVA and plotted on the graphs where error bars represent S.E.M. (C) Transient CAP1 knockdown through siRNA in the SW480 cells reduced cell matrix adhesion, and also abolished the stimulatory effect of isoproterenol on the matrix adhesion of SW480 cells. Efficient CAP1 knockdown was confirmed in Western blotting, and cell adhesion assays as well as statistical analyses were performed similarly as in panel B. (D) The stimulatory effects of cAMP activators forskolin and isoproterenol on matrix adhesion was abolished by knockdown of CAP1 in HCT116 cells. Cells were treated with forskolin (10 μM) for 120 minutes or isoproterenol (10 μM) for 60 minutes before they were detached for cell adhesion assays. Data from three independent assays each having three replicates were quantified and statistically analyzed using two-way ANOVA, and plotted on the graphs where error bars represent S.E.M. “*” indicates P < 0.05, “***” indicates P < 0.001 and “****” indicates P < 0.0001.

We next determined if depletion of CAP1 compromises the stimulatory effect of activated cAMP signaling on matrix adhesion. As shown in Figure 6B, the forskolin treatment significantly stimulated matrix adhesion in the control (Vec) SW480 cells. Importantly, the stimulatory effect was abolished in both CAP1-knockdown stable clones. Similarly, knockdown of CAP1 also abolished the stimulatory effect of the physiological cAMP activator isoproterenol on the matrix adhesion in SW480 cells (Figure 6B; graph to the right). We further established the required role for CAP1 in mediating the cAMP stimulation of matrix adhesion in SW480 cells, again by using the transient CAP1 knockdown approach. As shown in Figure 6C, in SW480 cells with efficient CAP1 depletion transiently, isoproterenol failed to stimulate matrix adhesion. Our results also show that transient CAP1 knockdown also significantly reduced the matrix adhesion in SW480 cells, which further support our findings from the stable knockdown approach that depletion of CAP1 reduces SW480 cell adhesion. Moreover, similar effects were detected in the CAP1-knockdown HCT116 cells in terms of the abolished stimulatory effects of forskolin and isoproterenol (Figure 6D). Therefore, whereas knockdown of CAP1 led to opposite alterations in the Rap1 activity and matrix adhesion phenotypes in SW480 and HCT116 cells, CAP1 is required for mediating cAMP activation of Rap1 in stimulating matrix adhesion in both these cell types. Our studies thus identify a novel signaling cascade, cAMP/CAP1/Rap1, in which CAP1 plays a pivotal role in linking the cAMP signals to activate Rap1 in stimulating matrix adhesion.

We attempted to adopt a rescue strategy, by re-expressing WTCAP1 or the CAP1 phosphor mutant in the knockdown cells, to further verify that the abolished stimulatory effects of the cAMP activators were indeed caused by depletion of CAP1. We have successfully done so in multiple cell types including HeLa, breast cancer and pancreatic cancer cells. However, unexpectedly, we encountered great difficulty in doing so in SW480 and HCT116 cells, and the effort completely failed in HCT116 cells while we screened large numbers of clones for each construct. We were able to establish a few SW480 stable clones that re-express either WTCAP1 or the DD (S307D/S309D) mutant, as shown in Supplementary Figure 4. Our results from using these clones show that re-expressed WTCAP1, but not the phosphor-mimetic DD mutant that resists dephosphorylation, rescued the stimulatory effect of isoproterenol on matrix adhesion in the CAP1-knockdown SW480 cells. These results support our conclusion that CAP1, through its dephosphorylation, mediates the cAMP signals to stimulate matrix adhesion.

3.9. Both Epac and PKA mediate the cAMP signals that stimulate matrix adhesion through the CAP1/Rap1 axis

Both Epac and PKA relay the cAMP signals to induce CAP1 dephosphorylation in HeLa and HEK293T cells (25). Our results in Figure 5 also suggest roles for Epac and PKA in relaying the cAMP signals that regulate CAP1 to activate Rap1 in SW480 and HCT116 cells. We next tested if activation of Epac or PKA induces CAP1 dephosphorylation to stimulate matrix adhesion. Indeed, specific activators of Epac or PKA, in 8PCT-2Me-cAMP and 6-Bnz-cAMP respectively, significantly induced CAP1 dephosphorylation in SW480 and HCT116 cells (Figures 7A&7B). We next tested effects of pre-inhibition of Epac or PKA in compromising the forskolin-induced CAP1 dephosphorylation. Cells were pre-treated with the Epac inhibitor ESI 09, namely α-[(2-(3-Chlorophenyl) hydrazinylidene]-5-(1,1-dimethylethyl)-β-oxo-3-isoxazolepropanenitrile (IC₅₀ at 3.2 μM and 1.4 μM for Epac1 and Epac2), or the specific PKA inhibitor PKI 14–22 amide (myristoylated and cell-permeable; IC₅₀ at 3 μM) for two hours followed by the forskolin treatment. We found that the pre-inhibition of Epac or PKA noticeably compromised the forskolin effect in inducing CAP1 dephosphorylation, with the most remarkable effect at the 2-hr forskolin treatment (Figure 7C). Pre-treatment with another commonly used PKA inhibitor H89 also had a similar effect (not shown). These results further support involvement of both Epac and PKA in relaying the cAMP signals that induce CAP1 dephosphorylation in colon cancer cells.

Figure 7. Evidence supporting roles for the cAMP effectors Epac and PKA in relaying the cAMP signals that stimulate matrix adhesion in SW480 and HCT116 cells through CAP1.

(A) Treatment with the specific Epac activator 8PCT-2Me-cAMP significantly induced CAP1 dephosphorylation in SW480 and HCT116 cells. Results from three independent experiments were quantified using ImageJ, analyzed using one-way ANOVA, and plotted on the graphs where the error bars represent S.E.M. (B) The specific PKA activator 6-Bnz-cAMP also induced CAP1 dephosphorylation in SW480 and HCT116 cells. Statistical analyses were performed similarly as in panel A. (C) Pre-inhibition of Epac or PKA, respectively, partially blocked the forskolin-induced CAP1 dephosphorylation. Cells were pre-treated with the Epac inhibitor ESI 09 (10 μM), or the PKA inhibitor PKI 14–22 amide (2 μM) for 60 minutes, followed by forskolin treatment for the indicated time durations before cells were harvested for Western blotting. (D) Treatment of cells with the specific Epac activator 8PCT-2Me-cAMP stimulated matrix adhesion in both SW480 and HCT116 cells, which was abolished in the CAP1-knockdown clones. SW480 and HCT116 cells were treated with 100 μM 8PCT-2Me-cAMP for 60 minutes before being plated on fibronectin-coated plates and allowed to attach for 120 minutes and 45 minutes, respectively. Data from three independent assays each having three replicates were quantified and statistically analyzed using two-way ANOVA and plotted on the graphs where the error bars represent S.E.M. (E) Activation of PKA stimulated adhesion in both SW480 and HCT116 cells, which was again abolished in the CAP1-knockdown clones. SW480 and HCT116 cells were treated with 100 μM 6-Bnz-cAMP for 60 minutes before cells were used in the adhesion assays similarly as in panel D. Data from three independent assays each having three replicates were quantified and statistically analyzed using two-way ANOVA and plotted on the graphs where the error bars represent S.E.M. “*” indicates P < 0.05, and “****” indicates P < 0.0001.

We next tested if CAP1 is also required for the activated Epac or PKA to stimulate matrix adhesion in these cells. As shown in Figure 7D, activation of Epac by 8PCT-2Me-cAMP significantly stimulated adhesion in both SW480 and HCT116 control (Vec) cells. Again, depletion of CAP1 abolished the stimulatory effect of the Epac activator in both stable clones. Similarly, the PKA activator 6-Bnz-cAMP also significantly stimulated matrix adhesion in both SW480 and HCT116 control (Vec) cells, with the stimulatory effect again abolished in the CAP1-knockdown stable clones (Figure 7E). These results further support roles for Epac and PKA in relaying the cAMP signals that stimulate matrix adhesion through CAP1 dephosphorylation and subsequent activation of Rap1. It is noted that PKA had been regarded as the primary target of cAMP in the cell, and evidence supports a role for the cAMP/PKA axis in integrin-mediated cell adhesion (60). However, Epac has recently emerged as arguably the more relevant cAMP effector that mediates cAMP effect on matrix adhesion (35,39). In epithelial cells and microvascular smooth muscle cells, cAMP stimulates cell adhesion through Epac but independent of PKA (35,58). Our results support that both Epac and PKA are involved in relaying the cAMP signals that stimulate matrix adhesion in colon cancer cells, through the CAP1/Rap1 axis.

3.10. Rap1 functions downstream of cAMP/CAP1 to facilitate the cAMP stimulation of matrix adhesion

We found that cAMP activates Rap1 in both SW480 and HCT116 cells, for which CAP1 is required, suggesting a cAMP/CAP1/Rap1 signaling cascade. Literature and our findings also point to Rap1 as the adhesion molecule that mediates signals from cAMP/CAP1 to stimulate matrix adhesion in the colon cancer cells. To test this model, we next silenced Rap1 in SW480 and HCT116 cells to determine its effects on the cAMP stimulation of matrix adhesion. Multiple stable clones with efficient Rap1 knockdown were generated in both cell lines, as confirmed in the Western blotting (Figure 8A). Consistent with the role of Rap1 in facilitating integrin-mediated cell adhesion (55), knockdown of Rap1 significantly reduced the basal level matrix adhesion in both SW480 (Figure 8B; upper panel) and HCT116 (Figure 8B; lower panel) cells. More importantly, depletion of Rap1 also abolished the stimulatory effect of forskolin on the matrix adhesion in both SW480 (Figure 8B; upper panel) and HCT116 (Figure 8B; lower panel) cells. These results strongly support that activation of Rap1 by cAMP signaling indeed facilitates the cAMP stimulation of matrix adhesion. We also examined if the knockdown of Rap1 caused alterations in expression levels of both CAP1 and CAP2, the basal level CAP1 phosphorylation as well as the CAP1 dephosphorylation induced by forskolin, and no remarkable effect was detected in any of these in SW480 or HCT116 cells (Figure 8C&D). These results further support that Rap1 functions downstream of cAMP and CAP1 in the signaling pathways where CAP1 mediates the cAMP signals that activate Rap1, through the cAMP/CAP1/Rap1 cascade, in stimulating matrix adhesion.

Figure 8. Rap1 functions downstream of the cAMP/CAP1 axis in the signaling pathway that facilitates cAMP stimulation of matrix adhesion in SW480 and HCT116 cells.

(A) Stable knockdown of Rap1 in SW480 and HCT116 cells as confirmed in Western blotting. “*” indicates the clones with efficient Rap1 silencing that were selected for further studies. (B) Depletion of Rap1 reduced the basal level matrix adhesion on fibronectin in both SW480 (upper panels) and HCT116 (lower panels) cells. Importantly, depletion of Rap1 also abolished the stimulatory effect of forskolin on matrix adhesion in these cells as shown in the graphs. Cells were treated with forskolin (10 μM) for 120 minutes. The treated SW480 and HCT116 cells were plated on fibronectin-coated plates and allowed to attach for 120 and 45 minutes, respectively, before the unattached cells were washed off. Data from three independent assays each having three replicates were quantified and statistically analyzed using two-way ANOVA and plotted on the graphs. Error bars represent S.E.M. (C) Knockdown of Rap1 in SW480 and HCT116 cells did not cause remarkable alterations expression levels of CAP1 or its basal level phosphorylation. (D) Depletion of Rap1 did not block the CAP1 dephosphorylation induced by forkolin in SW480 and HCT116 cells. “****” indicates P < 0.0001.

Our studies thus identify a novel signaling pathway in which cAMP, which is a major second messenger activated by a wide variety of extracellular stimuli, regulates the CAP1/Rap1 axis in leading to Rap1 activation and subsequent stimulation of matrix adhesion. The cytosolic effectors of cAMP, Epac and PKA, both relay the cAMP signals to regulate the CAP1/Rap1 axis. Figure 9 depicts a schematic model where CAP1 plays a pivotal role in mediating the cAMP signals to stimulate matrix adhesion through the cAMP/Epac-PKA/CAP1/Rap1 signaling pathway. Identification of this novel pathway represents a major advance towards a comprehensive and in-depth picture on the CAP1 function in adhesion as well as its regulation by cell signals, which may potentially also regulate the CAP1 function in the proliferation depending on the cell type.

Figure 9. A schematic model depicting that activated cAMP signals, relayed by cAMP effectors Epac and PKA, induce CAP1 dephosphorylation in leading to Rap1 activation and subsequent stimulation of matrix adhesion in colon cancer cells.

Extracellular stimuli, such as hormones and growth factors that bind G-protein coupled receptors, trigger the activation of adenylyl cyclase, which converts ATP to cAMP in leading to activation of cellular cAMP signaling. The elevated cellular cAMP levels activate the cAMP effectors Epac and PKA, which relay the cAMP signals to induce CAP1 dephosphorylation. The predominant localization of the dephosphorylated CAP1 to the cell periphery (plasma membrane) facilitates membrane translocation of Rap1, which is important for activation of Rap1 and its function in promoting cell matrix adhesion. Note that isoproterenol functions through the receptors to activate adenylyl cyclase whereas forskolin directly activates the enzyme in elevating the cellular cAMP levels. CAP1 plays a pivotal role in mediating the cAMP signals to activate Rap1, and depletion of CAP1 blocks the pathway and thus abolishes the activation of Rap1 and stimulation of matrix adhesion by activated cAMP signaling in the colon cancer cells.

3.11. Molecular insights into the context-dependent CAP1 function in adhesion and its regulation

Our previous studies have revealed a conserved function for CAP1 in matrix adhesion in mammalian cells, although knockdown of CAP1 causes opposite adhesion phenotypes, for which the oppositely altered FAK activities are at least partially responsible (14,15,21). Findings from the current study further support conservation of this CAP1 function as well as its context-dependent nature across mammalian cells. Moreover, our findings reveal that altered activities of Rap1 also underlie the opposite matrix adhesion phenotypes in the CAP1-knockdown cells, along with that of FAK. Interestingly, in MCF-7 and MDA-MB-231 breast cancer cells where knockdown of CAP1 also caused opposite adhesion phenotypes and altered FAK activities (21), we also detected oppositely altered Rap1 activities that are similar to the SW480 and HCT116 cells (not shown). However, little is known as to how CAP1 regulates the adhesion molecules including FAK and Rap1 to facilitate the cell type-specific function in matrix adhesion, and how CAP1 phosphorylation may regulate these.

Our studies have revealed involvement of adhesion molecules FAK, Rap1 and talin in facilitating the CAP1 function in adhesion. Each of these molecules is believed to bind or associate with CAP1, and for FAK and Rap1, we have also identified functional interactions with CAP1. CAP1 likely forms a large protein complex with the adhesion molecules in the cell. FAK interacts with talin, although whether FAK recruits talin to integrin adhesions or vice versa likely differs in nascent adhesions and mature adhesions (61,62). Rap1 also binds talin in leading to integrin activation (30,63). CAP1 directly binds Rap1 (34) and most likely talin as well (14), whereas FAK may join the complex through talin, and possibly through direct binding with CAP1 as well. We speculate that the protein complex facilitates the CAP1 function in matrix adhesion, likely through the dynamic assembly and disassembly of the complex in a tempo- and spatially controlled manner. In this model, the relative abundance of one or more of the adhesion molecules will likely impact how the loss of CAP1 alters the dynamics of the adhesion complex, and consequently cell matrix adhesion. Therefore, vastly different expression levels of one or more of these adhesion molecules may underlie the cell type-specific alterations in matrix adhesion caused by depletion of CAP1. Moreover, regulation of the CAP1 interaction with one or more of the adhesion molecules likely also underlie the phosphor-regulation of the CAP1 function in adhesion (25), similar to the case where phosphorylation on CAP1 controls its binding with cofilin and actin in facilitating the CAP1 function in the actin cytoskeleton (24).

Depletion of CAP1 caused opposite adhesion phenotypes in HCT116 and SW480 cells. Indeed, HCT116 cells express substantially higher Rap1 levels than in SW480 cells (Figure 4A). Similarly, HCT116 cells also express much higher levels of FAK than in SW480 cells, as shown in Figure 10A. Interestingly, similar to the case of Rap1, the FAK expression in SW620 was between that in SW480 and HCT116 cells. In immunoprecipitation assays, we detected FAK co-precipitation with CAP1 from the lysates of HCT116 cells (Figure 10B), which is stronger than that from the lysates of SW480 cells (not shown), to which the lower FAK expression levels in SW480 likely contributed. We were unable to detect co-precipitation of Rap1 and CAP1 (not shown), for which the full-length CAP1 had been reported to have a much reduced capacity in binding Rap1 than the CAP1 truncate with the short C-terminal region deleted (34). These results suggest that the Rap1 binding site on CAP1 may be masked, and unmasking it by a conformational change (e.g. controlled by CAP1 phosphorylation) may be involved in regulating the CAP1 binding with Rap1 in facilitating the CAP1 function in matrix adhesion. The vastly different expression levels of Rap1 and FAK in SW480 and HCT116 cells likely have played a role in the opposite alterations in the activities of these adhesion molecules caused by CAP1 depletion, leading to the opposite adhesion phenotypes.

Figure 10. Molecular mechanisms that may underlie the cell type-specific CAP1 function in matrix adhesion and the phosphor-regulation of CAP1 function.

(A) HCT116 cells express much higher levels of FAK than in SW480 cells. FAK levels were detected in these cells along with several other cell lines in Western blotting. (B) FAK co-precipitated with CAP1 from the lysates of HCT116 cells. CAP1 in the cell lysate was immuno-precipitated with a beads-conjugated CAP1 antibody, and the precipitated endogenous CAP1 (indicated by “⁺”) and co-precipitated FAK (indicated by “*”) were detected in Western blotting. Cell lysates were incubated for 2 hours with the beads-conjugated mouse anti-CAP1 antibody followed by detection of the proteins in Western blotting. WCL indicates the total whole-cell lysate. (C) Results supporting that phosphorylated CAP1 has increased association with FAK in SW480 cells. Cells were treated with lithium chloride (LiCl), or cultured shortly on fibronectin-coated plates (FN), to induce dephosphorylation of the cellular CAP1 pool, as confirmed in the Western blotting (upper panel). Co-immunoprecipitation of FAK with CAP1, their detection and labeling of the signals were done similarly as in panel B. Cells were treated with 100 mM LiCl for 20 hrs or plated on fibronectin for 1 hr before they were harvested for making the cell lysates used in Western blotting.

We next explored possible regulation of the CAP1 association with FAK by phosphorylation. For this, we induced dephosphorylation of the cellular CAP1 through two separate approaches, by treating cells with LiCl that inhibits GSK3 (24), or by culturing cells shortly on fibronectin for cells to undergo active attachment and spreading (24). Both approaches effectively induced CAP1 dephosphorylation in SW480 and HCT116 cells (Figure 10C; upper panel). Interestingly, enhanced co-immunoprecipitation between FAK with CAP1 was detected from the lysate of the control SW480 cells that harbor a more phosphorylated CAP1 pool (Figure 10C; lower panel). These results suggest that phosphorylation on CAP1, controlled by the phosphor-regulatory cell signals of CAP1, may positively regulate its association with FAK. Thus, phosphorylation may control the dynamics of the adhesion complex, which underlies the phosphor-regulation of the CAP1 function in cell adhesion. However, only marginally enhanced co-precipitation between CAP1 and FAK was detected in the control HCT116 cells compared to that in the treated cells (not shown). Further studies are required in order to obtain a more comprehensive and mechanistic picture on the CAP1 function and its regulation.

4. Discussion

CAP is best established as a key actin-regulating protein in eukaryotes, while CAP1 also regulates mammalian cell adhesion, migration and proliferation. The CAP1 functions, except that in actin dynamics, are all cell type-specific, with molecular mechanisms poorly defined. Cell adhesion is not only essential for normal physiological processes, but also plays key roles in pathological conditions such as cancer metastasis, where it collaborates with actin dynamics to control cell migration in leading to cancer invasiveness. Molecular insights into CAP1 functions in adhesion and proliferation, and especially their cell type-specific regulation, remain a paucity. Moreover, while we previously reported that phosphorylation regulates the CAP1 functions in cell adhesion and proliferation, it remains largely unknown how it is achieved. Mechanistic insights into these CAP1 functions and their regulation will thus not only vertically extend our knowledge on CAP biology, but also carry important translational potential.

We report here our findings on CAP1 functions in colon cancer cells. As expected, CAP1 regulates the actin cytoskeleton and actin-dependent cell functions, including the novel one in the cytokinesis of SW480 cells. CAP1 also regulates matrix adhesion in colon cancer cells, again in a cell type-dependent manner. Knockdown of CAP1 reduced matrix adhesion in SW480 cells while it stimulated adhesion in HCT116 cells. Interestingly, we found that oppositely altered Rap1 activities also underlie the opposite adhesion phenotypes in the CAP1-knockdown cells, along with the altered FAK activities. The opposite effects of CAP1 depletion on adhesion molecules leading to opposite adhesion phenotypes in the knockdown SW480 and HCT116 colon cancer cells, which are of different origin and stage, can be potentially interesting, although further studies are needed to determine whether this is also the case with other colon cancer cell lines in their respective groups. The findings that expression levels of both Rap1 and FAK increase in the order of SW480 (Dukes’ Type B), SW620 (Dukes’ Type C) and HCT116 (Dukes’ Type D) cells may also suggest possible up-regulation of these adhesion molecules when the cancer progresses to a more advanced metastasis stage. Most importantly, we unravel that CAP1 mediates the cAMP signals to activate Rap1 in stimulating matrix adhesion in both cell lines. Our studies identify a novel cell signaling cascade, namely cAMP/Epac-PKA/CAP1/Rap1, where CAP1 plays a pivotal role in mediating cAMP signals to activate Rap1 in stimulating matrix adhesion. Since cAMP is a major second messenger that is activated by a wide variety of extracellular stimuli, CAP1 likely fulfills versatile roles in linking the relevant upstream signals to control matrix adhesion. The pathway may potentially also underlie the role of CAP1 in regulating ERK and cell proliferation in cell types where CAP1 is involved in proliferation, such as breast cancer cells (21), as Rap1 is known to regulate cell proliferation and ERK (38,64–66).

The cAMP signals were relayed by both cytoplasmic cAMP effectors, Epac and PKA, in HeLa and HEK293T cells (25), and this regulation is conserved in colon cancer cells. Activation of cAMP signaling by forskolin that directly acts on adenylyl cyclase, or by isoproterenol that functions through the β-adrenergic receptor, both led to Rap1 activation and stimulated matrix adhesion in these cells. Similarly, specific activators for Epac or PKA also stimulated Rap1 activity and matrix adhesion. Importantly, our results support an indispensable role for CAP1 in mediating the activated cAMP signals to activate Rap1 and stimulate matrix adhesion. Our findings support that cAMP stimulates matrix adhesion through its role in inducing CAP1 dephosphorylation. Further supporting our model is that dephosphorylated CAP1 predominantly localizes to the cell periphery (24), which likely facilitates Rap1 localization to cell membrane in leading to activation of Rap1 for stimulating adhesion (67). Indeed, remarkably enhanced Rap1 co-localization to the cell periphery (plasma membrane) was detected in the CAP1-knockdown HeLa cells with the re-expressed AA mutant that mimics dephosphorylated CAP1. Together, our studies identify a cAMP/Epac-PKA/CAP1/Rap1 pathway, as outlined in Figure 9.

We employed a combination of activators and inhibitors of the cAMP signaling in dissecting the signaling pathway, and the specificity of these compounds, multiple approaches utilized, as well as the well-controlled experimental designs all contribute to validation of the novel signaling pathway. Firstly, the cAMP activators and cAMP analog all induced CAP1 dephosphorylation, among them include forskolin as well as the physiological cAMP activator isoproterenol. We also included the inactive forskolin analogue as a negative control. Furthermore, by using the PKG inhibitor, we show that the effects in inducing CAP1 dephosphorylation and stimulating cell adhesion are specific to the activation of the cAMP signaling, but not that of the similar cGMP signaling. Secondly, by using specific activators and inhibitors for Epac and PKA in testing the effects on CAP1 dephosphorylation and matrix adhesion, we further establishing roles for Epac and PKA in relaying the cAMP signals to regulate the CAP1/Rap1 axis in stimulating matrix adhesion. Thirdly, we demonstrate that depletion of CAP1 abolished the stimulation of matrix adhesion by cAMP activators forskolin and isoproterenol, as well as that by the cAMP effectors Epac and PKA. Furthermore, despite the difficulties encountered in establishing the stable clones that re-express WTCAP1 or the phosphor mutants in the CAP1-knockdown cells, our confirmation that CAP1 is required for isoproterenol to stimulate matrix adhesion in SW480 cells through the transient CAP1 knockdown further establishes the signaling pathway, while helped addressing the concern of selection bias associated with stable knockdown.

The observation that CAP1 knockdown also compromised the effects of the specific Epac activator in activating Rap1 and stimulating matrix adhesion was somewhat intriguing, because Epac is a GEF of Rap1, which can activate Rap1 independently by promoting GTP binding. These results suggest either CAP1 may also facilitate the GEF function of Epac, or that the role of Epac in relaying the cAMP CAP1 dephosphorylation signals is also significant in activation of Rap1. Also noteworthy, in addition to cAMP, the PKC activator PMA (68) also induces CAP1 dephosphorylation in HeLa and HEK293T cells (24), and we found it is the case in SW480 and HCT116 cells as well. These results suggest conservation of the PKC regulation of CAP1 in colon cancer cells, and a potential role for PKC in also activating Rap1 to regulate cell adhesion. However, only a marginal effect was detected for PMA in activating Rap1 in SW480 cells (not shown). In testing the PMA effect on matrix adhesion in SW480 and HCT116 cells, we actually detected opposite effects in these cells. PMA significantly stimulated matrix adhesion in SW480 cells, whereas it inhibited adhesion in HCT116 cells (Supplementary Figure 5). We speculate two reasons that may underlie these observations. Firstly, functions of PKC family isozymes are known to be extremely complex, due to differential expression of a dozen of PKC isozymes that often fulfill distinct and even conflicting roles in various tissue or cell types (69). Secondly, PKC often executes its cell functions through multiple target molecules or pathways (70), and CAP1 is probably one of the PKC targets involved in adhesion, but a less important one.

Altered activities of Rap1 and FAK in SW480 and HCT116 cells are believed to underlie the opposite matrix adhesion phenotypes in these cells caused by CAP1 depletion. We found that CAP1 knockdown in these cells also caused down-regulation of talin (Supplementary Figure 6), another adhesion molecule that interacts with CAP1 and likely involved in facilitating the CAP1 function. Whereas these findings represent the first lines of evidence that may lead to the elucidation of molecular mechanisms behind the context-dependent CAP1 function in adhesion, they are preliminary and further studies will be required in order to obtain further mechanistic insights.

We did not detect remarkable alterations in the activity of ERK in the CAP1-knockdown colon cancer cells, which is similar to that in pancreatic cancer cells (15) but different from the case in breast cancer cells (21). Moreover, results from the BrdU cell proliferation assays consistently support that CAP1 does not play a significant role in the proliferation of SW480 and HCT116 cells. These results suggest that Rap1, activated by signals from cAMP/CAP1, does not play a significant role in regulating ERK and proliferation in the colon cancer cells. However, the role of CAP1 in mediating the cAMP signals to activate Rap1, which also regulates cell proliferation (38), likely explains how CAP1 fulfills the intriguing role in regulating ERK and cell proliferation in certain cell types such as breast cancer cells (21,28). Indeed, we found that activated cAMP signals induced CAP1 dephosphorylation in MCF-7 and MDA-MB-231 breast cancer cells as well, and caused opposite alterations in ERK activity (not shown). These results are consistent with the opposite alterations in ERK activity and cell proliferation in these cells derived from CAP1 depletion (21). Indeed, the cAMP signaling cross-talks with the MAPK/ERK signaling pathway in regulating cell proliferation through the Rap1/ERK axis (65,66). Furthermore, the reported opposite regulation of ERK and proliferation by the activated Rap1 downstream of cAMP in different breast cancer cell types (65,66) also fits well with the opposite roles for CAP1 in regulating ERK and proliferation in these cells.

Mammals possess two CAP isoforms, CAP1 and CAP2, with expression of CAP2 restricted to a few specific tissue types. CAP2 is believed to complement CAP1 functions in most tissue or cell types, and it was the case in regulating actin dynamics (71,72). However, CAP2 has been greatly understudied compared to CAP1. Modest expression levels of CAP2 were detected in colon cancer cells, suggesting that CAP2 may complement the CAP1 functions, including that in the matrix adhesion of these cells. It is possible that CAP2 may also mediate the cAMP signals in regulating Rap1 and adhesion. However, a phosphor-regulation mechanism like the one we uncovered for CAP1 has yet to be identified for CAP2. Lastly, CAP1 was recently reported to mediate Ras regulation of adenylyl cyclase in the mammalian system as well (73), which had been largely skeptical since the role was found for the yeast homologue. Along with our recent findings on the cAMP regulation of CAP1, a feedback regulation loop between cAMP and CAP1 may exist, at least in some of the mammalian cell types.

Our studies identify a novel signaling pathway where CAP1 plays a pivotal role in mediating the cAMP signals that activate Rap1 in stimulating matrix adhesion. The novel molecular mechanisms from this study by far are the most well defined to date that explain how CAP1 fulfills its function in matrix adhesion. Identification of the role for CAP1 in linking the major second messenger cAMP to activate Rap1, which stimulates cell adhesion and possibly regulates proliferation depending on the cell type, not only provides mechanistic insights into how the living organisms function, but also carries implication for treatment of human cancers. As expected, CAP1 is involved in cancer invasiveness, consistent to its cellular functions in regulating the actin cytoskeleton and matrix adhesion. On the other hand, how CAP1 regulates ERK and cell proliferation is intriguing, and our study may have shed light on that. Given the involvement of CAP1 in cancerous transformation and metastatic progression, CAP1 has been suggested as a therapeutic target for cancer treatment. However, the role of CAP1 in both cancerous transformation and cancer invasiveness is highly complex; and careful considerations must be given in developing therapeutic strategies. Firstly, for some cancer types, CAP1 is involved in cell functions related to cancer cell invasiveness alone, such as in pancreatic cancer and colon cancer cells we report here. In contrast, CAP1 regulates both the invasiveness and proliferative transformation in some other cancer types, such as in breast cancer. Accordingly, we must weigh the benefit in one versus the possible risk in the other, for potentially achieving the ideal outcome of simultaneously suppressing both proliferative transformation and the invasive cycle of cancer. Secondly, CAP1 can fulfill opposing cell functions depending on the type or even the sub-type of cancer, such as in the matrix adhesion in colon cancer and breast cancer. It will be crucial to carefully assess the roles of CAP1 in a particular type or even subtype of cancer in developing the therapeutic strategy through intervening CAP1 or the related cell signals. Careful considerations are also required in developing targeting strategies for CAP1, given the complicated role of adhesion in cell migration that is important for cancer cell invasiveness. In the example of SW480 cells, while the reduced activities of Rap1 and FAK led to reduced matrix adhesion in the CAP1-knockdown cells, these cells actually had increased motility.

5. Conclusion

This study identifies a novel signaling pathway where CAP1 mediates signals from the major second messenger cAMP to activate Rap1 and stimulate matrix adhesion in colon cancer cells. This pathway may also underlie the CAP1 function in the proliferation of certain cell types. Moreover, the study further establishes cell type-specific functions for CAP1 in matrix adhesion, while demonstrating involvement of Rap1 in also facilitating the CAP1 function in adhesion for the first time.

Highlights.

• CAP1 regulates the actin cytoskeleton and adhesion in colon cancer cells

• CAP1 depletion had opposing effects on adhesion in SW480 and HCT116 cells

• Rap1 also facilitates the CAP1 function in matrix adhesion, in addition to FAK

• CAP1 mediates cAMP signals to activate Rap1 in SW480 and HCT116 cells

• The novel cAMP/CAP1/Rap1 signaling pathway stimulates cell matrix adhesion

Data availability

Data will be made available upon request.