Cytoskeleton regulators CAPZA2 and INF2 associate with CFTR to control its plasma membrane levels under EPAC1 activation

Cystic Fibrosis (CF), the most common lethal autosomic recessive disorder among Caucasians, is caused by mutations in the gene encoding the Cystic Fibrosis Transmembrane conductance Regulator (CFTR) protein, a cAMP-regulated chloride channel expressed at the apical surface of epithelial cells. Cyclic AMP regulates both CFTR channel gating through a protein kinase A (PKA)-dependent process and plasma membane (PM) stability through activation of the exchange protein directly activated by cAMP1 (EPAC1). This cAMP effector, when activated promotes the NHERF1:CFTR interaction leading to an increase in CFTR at the PM by decreasing its endocytosis. Here, we used protein interaction profiling and bioinformatic analysis to identify proteins that interact with CFTR under EPAC1 activation as possible regulators of this CFTR PM anchoring. We identified an enrichment in cytoskeleton related proteins among which we characterized CAPZA2 and INF2 as regulators of CFTR trafficking to the PM. We found that CAPZA2 promotes wt-CFTR trafficking under EPAC1 activation at the PM whereas reduction of INF2 levels leads to a similar trafficking promotion effect. These results suggest that CAPZA2 is a positive regulator and INF2 a negative one for the increase of CFTR at the PM after an increase of cAMP and concomitant EPAC1 activation. Identifying the specific interactions involving CFTR and elicited by EPAC1 activation provides novel insights into late CFTR trafficking, insertion and/or stabilization at the PM and highlighs new potential therapeutic targets to tackle CF disease.


Introduction
Cystic Fibrosis (CF) is the most common lethal and life-limiting autosomic recessive disorder among Caucasian population, affecting about 1:2500-6000 new borns (1) and it is caused by absence or dysfunction of the Cystic Fibrosis Transmembrane Conductance Regulator (CFTR) protein. This is a member of the ABC (ATP-binding cassette) transporter superfamily and functions as a cAMP-regulated chloride (Cl -) and bicarbonate (HCO 3 -) channel in the apical membrane of epithelial cells to maintain ion and fluid homeostasis (2,3). CFTR is composed by two membrane spanning domains (MSD1 and MSD2), forming the pore of the channel; two cytosolic nucleotide binding domains (NBD1 and NBD2), where ATP binds and is hydrolysed regulating channel gating and one CFTR-exclusive regulatory domain (RD) that contains multiple phosphorylation sites essential for channel activity (4).
The most common CF-causing mutation corresponds to a deletion of phenylalanine (Phe) at position 508, located in NBD1 (F508del), which occurrs in ~85% of CF patients in at least one allele and leads to CFTR misfolding and endoplasmic reticulum (ER) retention. The mutant protein is prematurely degraded precluding its delivery to the cell surface (3).
Levels of CFTR at the PM have been shown to be determined by a balance between three main processes -1) anterograde transport, 2) endocytosis and 3) recycling (5, 6)all of which rely on CFTR interaction with several proteins among which PDZ domain-containing proteins (PDZ proteins) play an essential role. Indeed, NHERF1 anchors CFTR to the PM and to the actin cytoskeleton (5,7). In the PDZ-dependent CFTR-NHERF1 complex, NHERF1 interacts with ezrin and this NHERF1-ezrin interaction locks CFTR in an immobile and actin-tethered complex preventing its endocytosis (8,9). NHERF1 not only regulates wt-CFTR at the PM but also of F508del-CFTR by protecting it from lysosomal degradation (10). Recently, it was shown that activation of the exchange protein directly activated by cAMP 1 (EPAC1) promotes its interaction with NHERF1 and CFTR, leading to an increase in the amount of CFTR at the PM at steady-state (11).
EPAC proteins function as guanine nucleotide exchange factor (GEF) for both Rap1 and Rap2 (12) and are involved in the regulation of cell-cell and cell-matrix adhesion, cytoskeleton rearrangements and cell polarization (13). Upon activation by cAMP, EPAC1the major EPAC expressed in the lung (14) is targeted to the PM where it tethers to phosphatidic acid or to phosphorylated proteins in the ezrin/radixin/moesin (ERM) family inducing its downstream effectors such as Rap (15). Identification of NHERF1-dependent CFTR-EPAC1 interaction (11) highlighted a two-level regulation of the channel by cAMPlow concentrations of cAMP activate PKA to regulate CFTR function whereas high cAMP levels promote EPAC1-dependent increase of its PM levels (11).
Thus, EPAC1 appears to be a hub in the cAMP/Ca 2+ crosstalk. Despite these recent advances the mechanism and the macromolecular complexes elicited by EPAC1 activation at the PM are still largely uncharacterized.
We used a proteomic interaction profiling approach coupled to global bioinformatic analysis to identify specific CFTR interactions elicited by EPAC1 activation. We identified several novel interactors and characterized two of them involved in its PM anchoring, CAPZA2 and INF2. Our data show that while CAPZA2 downregulation, leads to an effect by EPAC1 on wt-CFTR by enhancing PM levels, in contrast, INF2 downregulation, leads to the opposite effected i.e., reduced levels of wt-CFTR at the PM. We report here for the first time the acting cytoskeleton dynamics regulators CAPZA2 and INF2 as modulators of CFTR anchoring at the PM.
Considering that the therapeutic opportunities of restoring CFTR PM stability are still poorly explored, characterization of the molecular mechanism of CFTR anchoring as described here may contribute to identify novel targets that can be used in combination with other therapeutic strategies.
Downloaded from https://portlandpress.com/biochemj/article-pdf/doi/10.1042/BCJ20200287/885453/bcj-2020-0287.pdf by guest on 02 July 2020 Biochemical Journal. This is an Accepted Manuscript. You are encouraged to use the Version of Record that, when published, will replace this version. The most up-to-date-version is available at https://doi.org/10.1042/BCJ20200287

Cell culture and compound treatment
Parental CFBE41o-cells and CFBE stably transduced with wt-CFTR were grown in EMEM (Lonza, #BE12-611F) supplemented with 10% (v/v) FBS (GIBCO® Life Technologies, #10270-106) and 5 µg/ml puromycin (Sigma, #P8833). CFBE41o-cells expressing double tagged mCherry-FLAG-wt-CFTR, under a doxycycline inducible promoter were grown in DMEM with 4.5 g/L glucose and L-Glutamine supplemented with 10% (v/v) FBS, 2 µg/ml puromycin and 10µg/ml blasticidin (InvivoGen, #ant-bl). CFTR expression was induced with doxycycline 1 µg/ml (Sigma #9891) for 24h. All cell lines were grown at 37ºC in 5% CO2. ProtScore and the number of replicates where the protein was identified in both samples and controls. R programming was used to calculate the Confidence score for each protein 1 . Lists of proteins identified in each of the three conditions tested were compared to identify common and specific proteins for each condition.

Networks and gene ontology
The Database for Annotation, Visualization and Integrated Discovery (DAVID) 2 (19,20) was used to analyse the obtained data set and identify the Gene Ontology (GO) terms which were enriched in our data. The GO terms considered were the Biologic process (BP) and Cellular Component (CC) using version 6.8.
Data were also analysed with Agile Protein Interactomes DataServer (APID) in combination with R programming (21). APID human interactomes of quality level 0 (all reported interactions) or level 1 (interactions proven by 2 or more experimental evidences) were used to assess the distance (1 to 5 edges distance) of the identified proteins (proteins interacting with CFTR specifically after cell incubation with 007-AM) to proteins of interest (CFTR and EPAC1). APID level 0 was also used to generate protein networks. Finally, the shortest path in the network between two proteins of interest was determined using APID level 0 or APID level 1. Networks generated using APID (22) were accessed and visualized/analyzed using Cytoscape (23). To generate the wt-CFTR general network, we used the proteins detected in our assays with a Confidence score of 2 or higher (Table S1) for all sample-control combinations, using APID level 0 as the basis for the PPI.
siRNA transfection siRNA transfection was performed as previously described (24). Transfection mixture using  Finally, cells were washed three times in cold PBS +/+ and immersed in PBS +/+ until cell imaging.
All solutions were prepared in PBS +/+ and primary and secondary antibodies were diluted in For image acquisition, cell imaging was performed with inverted widefield fluorescence microscope Leica DMI6000 equipped with a DFC360FX camera (12-bit, 1344x1024 pixel resolution) (Leica) and a 10x objective. The exposure times at maximum light brightness for Then, the deviation score was calculated according to the following formula: The Traffic Efficiency Test corresponds to the average of CFTR traffic efficiency for all images transfected with the same siRNAs or treated with the same compound which passed the QC. The Traffic Efficiency control corresponds to the average of CFTR traffic efficiency for all the conditions in the plate or in case of compound incubation the DMSO -All Conditions or DMSO. The standard error of the mean of the control (SEM control ) corresponds to the SEM recorded for the control condition -All Conditions or DMSO. Deviation score identifies traffic enhancers (when it is positive) or inhibitors (when it is negative).

Cell-surface biotinylation assay
Cell surface biotinylation was performed as previously described (11,27). Cells were grown in P60 plates until pre-confluence. After three washes with PBS +/+ cells were incubated at 4ºC with 3mL of Sulfo-NHS-SS-Biotin buffer (0.1mg biotin (Thermo Scientific, #21331)/1 mL PBS with 1mM MgCl2, 0.1mM CaCl2, pH 8.2) for 30 min and then washed three times with PBS ++ . Cells were lysed with 1.5mL of BL buffer (HEPES 25mM, Triton X-100 1% (v/v), glycerol 10% (v/v), supplemented with complete protease inhibitor cocktail, pH 8.2) and incubated for 15min at 4ºC. Cell membranes and debris were pelleted by centrifugation at 14000g for 10min at 4ºC following by supernatant incubation at 4ºC overnight with streptavidin beads previously washed twice with PBS and once with BL buffer. The beads were pelleted by centrifugation 1min at maximum speed and next washed once with BL buffer and twice with PBS. Then, beads were incubated with 65µL SB suplemented with DTT 1M at 85ºC for 5min.
Finally, samples were loaded into a SDS-PAGE gel for Western blotting (WB) analysis.

Western blotting
Protein extracts were separated on 7% SDS polyacrylamide gel electrophoresis (PAGE) and transferred to PVDF membranes. Membranes were blocked with 5% non-fat milk in PBS the Chemidoc TM XRS system (BioRad). The quantification of band intensity was performed using the Image Lab software (BioRad) and normalized to loading control as appropriate.

Statistical analyses
Data are presented as mean ± standard error of the mean (SEM). Two-tailed Student's ttests were performed to assess statistical significance. In these analyses, p<0.05 was considered as significant.

Proteomic profiling of wt-CFTR under activation of EPAC1
In an attempt to identify the proteins involved in CFTR anchoring/stabilization at the PM under EPAC1 activation, CFTR was immunoprecipitated from CFBE wt-CFTR cells under lowstringency conditions so as to capture multiprotein complexes and thus maximize the number of interacting proteins obtained. As we were interested in identifying the differential interactors involved in CFTR stabilization by EPAC1, cells were incubated with: i) forskolin (Fsk); ii) the specific EPAC1 agonist 007-AM (8-pCPT-2´-O-Me-cAMP); or iii) DMSO (vehicle control).
Non-specific interactions ("background") were discarded using an equivalent approach performed in CFBE cells which do not express CFTR (parental CFBE cells) and in CFBE wt-CFTR cells incubated with non-conjugated beads. Isolated proteins were then identified using nanoLC-MS/MS.
We were able to pull-down a total of 1,599 proteins ( Removal of proteins identified in the controls resulted in a total of 207 different proteins (Table S4A -Supplementary Data) for the sum of the three conditions tested, that were then crossed to identify the specific proteins for each condition (Table 1, Figure S1). CFTR is one of the three proteins that is common to all the conditions tested. Interestingly, the higher number of specific wt-CFTR-interacting proteins (CIPs) was identified when EPAC1 is activated (CFBE cells incubated with 007-AM) -101 proteins (Table S4B -Supplementary Data). In addition, 23 CFTR interacting proteins were identified specifically when adenylyl cyclase (AC) is activated (CFBE cells incubated with Fsk) and 62 CFTR interacting proteins were identified exclusively for the incubation with DMSO control condition ( Figure S1; Table S4B -Supplementary Data).

Bioinformatic and comparative analysis of CFTR interacting proteins
We used available bioinformatics tools to analyse the lists of CIPs upon activation of either AC or EPAC1, i.e., in cells incubated with either Fsk or 007-AM, respectively. term -Cellular Compartment (CC), adherent junction, anchoring junction and extracellular vesicles are the most represented terms. At the CC level, the enrichment in categories of cytoskeleton and actin filaments-associated is even more evident since more than 2/3 of all categories represent cellular components which are strongly connected with cytoskeleton and actin filaments as shown in Figure 1B.
We then used APID to calculate the distance between the 110 proteins specifically identified upon EPAC1 activation to either CFTR or EPAC1. The proteins were analysed with APID level 0 (accounts for all known interactions) and APID level 1 (accounts for interactions proven by two or more experiments). Among these 110 proteins, 6 (based on both APID level 0 and APID level 1 - Figure S2A) had been previously reported as directly interacting with CFTR, namely: AHSA1, CSE1L, HSPA2, PPP2R1A, PSMD2 and NHERF1. Moreover, we found that more than 85% of the 110 proteins are within 2 to 3 edges of distance from CFTR (with the more stringent APID level 1). When measuring the distance of the 110 proteins to EPAC1, there was no protein detected in APID previously reported to interact directly with EPAC1 ( Figure S2B). The fact that we pulled-down CFTR and not EPAC1 is a plausible explanation for this observation. However, more than 89% of these proteins are within 3 to 4 edges of distance from EPAC1.
To analyse the robustness and confidence for the identified CIPs, a scoring system (named Confidence score and detailed in Table S1) was developed. This Confidence score takes into consideration the Unused ProtScore and the number of replicates in which each protein was detected (comparing samples and controls). 5 proteins were identified with a high robustness and confidence (score -5) for their interactions upon EPAC1 activation (Table S5 -Supplementary Data). CFTR is one of these proteins supporting the approach used as well as the reliability of the scoring methodology. Overall, for the proteins detected under EPAC1 activation, 27% were scored above or equal to 3.
Comparison with previously published CFTR interactomes revealed a very low overlap (around 2% with 20 and 13 proteins in common respectively) (28, 29) -suggesting that the 110 CIPs correspond to proteins that specifically interact with CFTR when EPAC1 is activated and thus do not belonging to the "core" CFTR interactome.
In order to select the hits for validation among the 110 interactors, the above-described data were integrated (Table S6 -Supplementary Data) and complemented with extensive literature-mining. We then selected for validation: i) 3 proteins with a Confidence score of 5; ii) 7 proteins with a score of 4; and iii) 3 proteins with a score of 3 (Table S6 -Supplementary   Data). In addition, we also selected 5 proteins with a score of 2 (lower robustness and confidence) in order to evaluate the strength of the selection methodology used. Interestingly, among the 18 selected hits, 16 are within 2 edges of distance to CFTR and, from these 16, 15 Downloaded from https://portlandpress.com/biochemj/article-pdf/doi/10.1042/BCJ20200287/885453/bcj-2020-0287.pdf by guest on 02 July 2020 proteins are within 3 edges of distance to EPAC1 showing the high proximity of the selected hits to CFTR and EPAC1 (Table S6 -Supplementary Data).

Generation of protein networks
To assess the possible connections between the identified CIPs, we generated a network using all proteins with a Confidence score of 2 or higher. For this, proteins identified in the controls were removed from the lists identified for each sample (Fsk, DMSO or 007-AM) resulting in a total of 260 proteins that were used for the network generation (Table S7  As the 18 hits selected for validation seem to be in close proximity among themselves, we built a minimum network needed to link them to CFTR and EPAC1 ( Figure 2) with APID level 1 as the source of interactions. This confirms that they can be linked through a minimal network to which only 8 additional proteins (shown in blue) are needed to connect them to CFTR and EPAC1 (shown in yellow) (Figure 2).

Impact of hit knockdown on CFTR trafficking
We next assessed the impact upon CFTR trafficking of knocking down by siRNA the 18 hits selected for validation using a previously described assay (25).
We started by validating the assay for the previously reported increase of CFTR trafficking upon EPAC1 activation (as reported by us in (11)). For that, we used CFBE cells expressing mCherry-Flag-wt-CFTR under the control of a doxycycline-inducible promoter, that were incubated with either the EPAC1 agonist 007-AM or DMSO (vehicle control).
Immunostaining assay was performed and CFTR at the PM was monitored by the Flag (Cy5) signal. CFTR trafficking (given by the ratio of PM CFTR to total CFTR) and the deviation score was also calculated as described (25). Results show that EPAC1 activation leads to an increase in wt-CFTR at the PM corresponding to a statistically significant increase in traffic efficiency ( Figure S4). This result confirms that this trafficking assay is robust to assess the impact of hit knockdown in combination with EPAC1 activation.  (FLG2, PSMB6, RPS14 and RPS18) supporting the selection strategy adopted. Being this a screening assay, we cannot of course exclude the fact that lack of an effect may correspond to low knock-down efficiency by the siRNA(s) used.

Validation of INF2 and CAPZA2 as CFTR interactors under EPAC1 activation
As we aim to identify proteins involved in CFTR PM stabilization by EPAC1 activation, we then validated and characterized the role of the two hits identified, namely CAPZA2 and INF2. For that, we validated the interaction between CAPZA2 or INF2 proteins and CFTR and assessed the effect of knocking down CAPZA2 and INF2 at 3 levels: i) the amount of wt-CFTR at the PM; ii) amount of EPAC1 interacting with CFTR and iii) the cellular localization of CFTR and EPAC1 vs that of INF2 and CAPZA2. Interaction between EPAC1 and CAPZA2 or INF2 was also assessed to get additional information on its mechanism of action.
Firstly, we assessed the efficiency of KD of both siRNAs used against CAPZA2 or INF2 ( Figure S5). Results show that both siCAPZA2 and siINF2 decrease the protein levels of CAPZA2 or INF2 by more than 50% or 40-60% , respectively as compared to siRNA against EGFP as a negative control ( Figure S5B and C, respectively).
We then assessed the interaction between CAPZA2 and CFTR by co-IP to validate the MS results. For that, CFTR was immunoprecipitated from CFBE wt-CFTR cells incubated with 007-AM or DMSO. Lysates incubated with beads only were used as a control. WB analysis shows that CAPZA2 is detected after CFTR IP when EPAC1 is activated, but not under DMSO ( Figure 4A). The quantification of these data shows a significant increase in the amount of CAPZA2 in the CFTR co-IP when cells were incubated with 007-AM vs DMSO ( Figure 4B

CFTR levels at the PM after modulation of CAPZA2 and INF2
Results from the trafficking assay (Fig.3)  On the other hand, for CAPZA2 knockdown, results show that it leads to a statistically significant decrease in CFTR PM levels ( Figure 5B) when compared to the control siRNA EGFP.

So, cell surface biotinylation results further support the previous observations that INF2
and CAPZA2 influence the amount of CFTR at the PM, with opposite effects.

Interaction between CAPZA2, INF2 and EPAC1
Since both INF2 and CAPZA2 influence CFTR PM levels upon EPAC1 activation, we then evaluate whether these two proteins interact with EPAC1. Co-IP of EPAC1 was performed in CFBE wt-CFTR cells with a specific antibody and using DMSO or beads only as controls.
Presence of INF2 and CAPZA2 was asssessed by WB.
We were able to detect CAPZA2 after EPAC1 co-IP mainly after 007-AM treatment ( Figure 6A). Results show that the EPAC1:CAPZA2 interaction is increased by 2-fold upo EPAC1 is activated ( Figure 6B), thus, indicating that CAPZA2 interacts with EPAC1 as well as with CFTR. For INF2, we were able to detect it after treatment with either 007-AM or DMSO, and observed a slight, albeit not significant, increase when cells are incubated with 007-AM ( Figure 6A).
In summary, our results show a strong EPAC1:CAPZA2 interaction promoted by 007-AM and also an interaction between INF2:EPAC1 (but in this case independent of EPAC1 activation). Thus, next we assessed the impact of CAPZA2 or INF2 knockdown on the EPAC1:CFTR reported interaction. For that, we used CFTR co-IP followed by detection of

Discussion
The net flow of CFTR activity depends on both the number of these channels at the PM as well as the function of each individual channel. Understanding the mechanisms through which CFTR PM levels are regulated is likely to identify targets amenable to modulation to design potential innovative therapeutic strategies. This is of particular interest because although F508del-CFTR can be rescued to the PM by different mechanisms, the half-life of the rescued mutant protein is significantly reduced due to the high endocytosis rate and/or a reduction in recycling back to the PM (2, 33) and thus its stabilization at the PM can be targeted in a combinatorial therapeutical approach (34).
Here, we aimed at better understanding the mechanism through which EPAC1 activation leads to CFTR stabilization at the PM by identifying the molecular complexes elicited during this process. Using co-IP of CFTR followed by mass spectrometry, we performed a comparative interactomics profiling of wt-CFTR under activation of adenylyl cyclase (by Fsk) or EPAC1 (by 007-AM). From a total of 1,599 CFTR-interacting proteins (CIPs) identified, 110 were exclusively identified after EPAC1 activation ( Figure S1). GO terms enrichment analysis ( Figure 1) in this set of proteins identified a high representation of actin cytoskeleton associated categories. Whereas it is known that CFTR stability and function is correlated with the actin cytoskeleton (35) and that EPAC1 activation is also involved in actin cytoskeleton organization (36), these results suggest that the previously described increase in CFTR:NHERF1 interaction (11) under EPAC1 activation is related to the formation of macromolecular complexes that stabilize the anchoring of CFTR to the actin cytoskeleton.
From the 110 EPAC1-specific CIPs, we selected 18 for validation based on a "Confidence score" (based on confidence levels and robustness of the MS data) and on the distance to CFTR and EPAC1 supported by extensive literature-mining (Table S6). This set of 18 proteins includes mostly proteins related to the actin cytoskeleton, such as: actin related protein 3 homolog B (ACTR3B), which has 89% identity with the Arp2/3 complex member To evaluate the relative proximity between CFTR, EPAC1 and the 18 selected CIPs, we Downloaded from https://portlandpress.com/biochemj/article-pdf/doi/10.1042/BCJ20200287/885453/bcj-2020-0287.pdf by guest on 02 July 2020 refined the analysis to create a minimal network needed to link those 18 CIPs (Figure 2). Only 8 additional proteins (XPO1, RAP1A, EZR, CDK2, PPP1CB, PSMC4, ACTB and PAN2) were needed to connect the selection. Interestingly, EZR, a well known CFTR and NHERF1 interactor (11,44), was one of these added proteins. Among these added proteins, ACTB and XPO1 were also previously described as CFTR interactors (45,46). RAP1A was also added to build the minimal networkdespite not being found as a direct CIP, it is regulated by EPAC as cAMP induces the GEF activity of EPAC towards RAP1A (47).
We then assessed the impact of the KD of the selected proteins on CFTR PM traffic by an immunofluorescence assay. A major impact (statistically significant) was obtained under knockdown of INF2 and CAPZA2 (Figure 3) under 007-AM, while the remaining siRNAs showed only minor impact on CFTR traffic efficiency. While wt-CFTR traffic was significantly increased for both INF2 siRNAs, both CAPZA2 siRNAs showed a negative impact under EPAC1 activation. These results suggest that INF2 impairs and CAPZA2 positively influences CFTR PM traffic and/or stability under EPAC1 activation.
These two CIPs (INF2 and CAPZA2) were both classified with high "Confidence score" (3 or higher) and, according to protein interaction databases, located at 2-and 3-edge distance from CFTR and EPAC1. To our knowledge, this is the first time that these two proteins are identified as CFTR interactors and that their silencing has an effect in CFTR trafficking. We thus proceeded with additional validation/characterization of these 2 hits. CAPZ proteins are highly conserved proteins from yeast to humans and are composed of two unrelated subunits -α and β (48). CAPZ proteins plays a key role in Arp2/3 mediated actin polymerization, influencing several cellular processes such as cell migration and cell invasion (48). They control actin dynamics by blocking the addition of actin monomers to the filament end thus effectively terminating its elongation (48,49). CAPZA2 protein (CAPZ subunit α2) has been associated with cancer aggressiveness (32) and is regulated by different scaffold proteins, such as TNKS1BP1, whose depletion causes a decrease in the association between CAPZA2 and actin filaments.
Our results show that CAPZA2 interacts specifically with wt-CFTR and its KD decreases CFTR PM traffic under EPAC1 activation. We validated these observations using CFTR co-IP followed by detection of CAPZA2, confirming that the interaction is detected only upon EPAC1 activation ( Figure 4A). Also, by cell surface biotinylation we showed that CAPZA2 knockdown in combination with the activation of EPAC1 pathway leads to a statistically significant decrease in wt-CFTR at the PM compared to the non-targeting siEGFP ( Figure 5). This association suggests a role for CAPZA2 in regulating wt-CFTR at the cell surface. It was reported that CFTR anchoring to the actin filaments is associated to the activation of the small GTPase Rac1 through PIP5K and Arp2/3 such that Arp3 knockdown drastically impairs CFTR PM retention (50). Arp2/3 complex acts as an actin nucleator playing a key role in actin Downloaded from https://portlandpress.com/biochemj/article-pdf/doi/10.1042/BCJ20200287/885453/bcj-2020-0287.pdf by guest on 02 July 2020 dynamics (51). In the presence of capping proteins, branching by Arp2/3 complex is favoured, whereas capping protein depletion leads to a formation of actin bundles, thus affecting cell shape (51,52).
The wt-CFTR:CAPZA2 association under EPAC1 activation is observed simultaneously with an increased interaction between EPAC1 and CAPZA2 ( Figure 6A). Interestingly, CAPZA2 knockdown does not influence the EPAC1:CFTR interaction ( Figure 6C). It is known that under EPAC1 activation, CFTR interacts with NHERF1 which then interacts with EPAC1 and ezrin. NHERF1:ezrin interaction locks CFTR in an immobile, stable and actin-tethered complex preventing its endocytosis (8,9,11). Thus, the increased EPAC1:CAPZA2 interaction verified under EPAC1 activation ( Figure 6A) indicates that CAPZA2 further potentiate the CFTR anchoring at the PM. However, the fact that CAPZA2 KD does not influence CFTR:EPAC1 interaction ( Figure 6C) may indicate that the presence of CAPZA2 plays an additive effect on CFTR anchoring at the PM and it is not a key element for the maintenance of the complex. We thus suggest that CAPZA2 promotes the anchoring of wt-CFTR at the PM by stabilizing actin cytoskeleton which consequently leads to a stabilization of the CFTR-NHERF1-Ezrin-EPAC1 complex (Figure 8, Stage I).
INF2 is unique among the formins due to its ability to accelerate depolymerization, in addition to the nucleation and elongation activity characteristic of most formins (31,53).
Our results show that CFTR interacts with INF2, an association that is increased upon EPAC1 activation and that involves mainly the mature form of CFTR (Band C) ( Figure 4C).  (31,54). INF2 depolymerization activity is dependent on its unique C-terminus which contains an actin-binding WASP homology 2 (WH2) motif that sequester actin monomers in 1:1 complex. INF2 C-terminus is also required for fast filament severing (53). INF2 severing ability is directly associated with the increased rate of actin filament depolymerization since it increases the number of depolymerizable ends (53).
This occurs as the FH2 domain of INF2 inhibits the barbed end capping by capping proteins (53).
This triphasic function suggests that the role of INF2 expression may depend on the context/environment, as increased INF2 overexpression has been associated with cell death (55) but also with cell migration, invasion and proliferation (56). To further understand the mechanism through which INF2 impacts on CFTR PM stability under EPAC1 activation, we determined that EPAC1 and INF2 also interact ( Figure 6A) but that, as described above for CAPZA2, INF2 knockdown does not affect the EPAC1:CFTR interaction ( Figure 6C). These results suggest that INF2 is involved in CFTR anchoring at the PM but it may not be a key element in CFTR-EPAC1 complex maintenance. Thus, we suggest that INF2 hampers the anchoring of wt-CFTR at the PM by improving actin cytoskeleton dynamics which interfere with stabilization of the CFTR-EPAC1 complex (Figure 8, Stage II).
Our results, together with reports showing that INF2 is localized mainly in the cytoplasm (57,58), suggest that the impact of INF2 in CFTR regulation may be associated with

Conflict of interest
There is no conflict of interest to disclose.   Statistical analysis was performed using two-tailed unpaired Student's t-test. shown as the mean ± SEM, n=3. * p<0.05. Statistical analysis was performed using two-tailed unpaired Student's t-test. Cells not incubated with biotin solution were used as a control. A) CFTR was detected by WB after streptavidin pulled-down. 20% of the total lysate was analysed as WCL. αTubulin was used as the loading control and Ezrin was used as an intracellular protein control. B) Quantification of CFTR detected after the pull-down was performed and the change (∆) normalized to DMSO was plotted. Data are shown as the mean ± SEM, n=3. $ * p<0.05.
Statistical analysis was performed using two-tailed unpaired Student's t-test.