Dynamic Regulation of Cystic Fibrosis Transmembrane Conductance Regulator by Competitive Interactions of Molecular Adaptors*

Disorganized ion transport caused by hypo- or hyperfunctioning of the cystic fibrosis transmembrane conductance regulator (CFTR) can be detrimental and may result in life-threatening diseases such as cystic fibrosis or secretory diarrhea. Thus, CFTR is controlled by elaborate positive and negative regulations for an efficient homeostasis. It has been shown that expression and activity of CFTR can be regulated either positively or negatively by PDZ (PSD-95/discs large/ZO-1) domain-based adaptors. Although a positive regulation by PDZ domain-based adaptors such as EBP50/NHERF1 is established, the mechanisms for negative regulation of the CFTR by Shank2, as well as the effects of multiple adaptor interactions, are not known. Here we demonstrate a physical and physiological competition between EBP50-CFTR and Shank2-CFTR associations and the dynamic regulation of CFTR activity by these positive and negative interactions using the surface plasmon resonance assays and consecutive patch clamp experiments. Furthermore whereas EBP50 recruits a cAMP-dependent protein kinase (PKA) complex to CFTR, Shank2 was found to be physically and functionally associated with the cyclic nucleotide phosphodiesterase PDE4D that precludes cAMP/PKA signals in epithelial cells and mouse brains. These findings strongly suggest that balanced interactions between the membrane transporter and multiple PDZ-based adaptors play a critical role in the homeostatic regulation of epithelial transport and possibly the membrane transport in other tissues.

Disorganized ion transport caused by hypo-or hyperfunctioning of the cystic fibrosis transmembrane conductance regulator (CFTR) can be detrimental and may result in life-threatening diseases such as cystic fibrosis or secretory diarrhea. Thus, CFTR is controlled by elaborate positive and negative regulations for an efficient homeostasis. It has been shown that expression and activity of CFTR can be regulated either positively or negatively by PDZ (PSD-95/discs large/ZO-1) domain-based adaptors. Although a positive regulation by PDZ domain-based adaptors such as EBP50/NHERF1 is established, the mechanisms for negative regulation of the CFTR by Shank2, as well as the effects of multiple adaptor interactions, are not known. Here we demonstrate a physical and physiological competition between EBP50-CFTR and Shank2-CFTR associations and the dynamic regulation of CFTR activity by these positive and negative interactions using the surface plasmon resonance assays and consecutive patch clamp experiments. Furthermore whereas EBP50 recruits a cAMP-dependent protein kinase (PKA) complex to CFTR, Shank2 was found to be physically and functionally associated with the cyclic nucleotide phosphodiesterase PDE4D that precludes cAMP/PKA signals in epithelial cells and mouse brains. These findings strongly suggest that balanced interactions between the membrane transporter and multiple PDZ-based adaptors play a critical role in the homeostatic regulation of epithelial transport and possibly the membrane transport in other tissues.
The cystic fibrosis transmembrane conductance regulator (CFTR) 2 is a Cl Ϫ channel and a key regulator of fluid and ion transport in the gastrointestinal, respiratory, and genitourinary epithelia (1,2). Hypofunctioning of CFTR due to genetic defects causes cystic fibrosis, the most common lethal genetic disease in Caucasians (3,4), whereas hyperfunctioning of CFTR resulting from various infections evokes secretory diarrhea (5), the leading cause of mortality in early childhood in the world (World Health Organization statistics). Therefore, maintaining and regulating a dynamic balance between CFTR-activating and CFTR-inactivating machineries is an important mechanism for maintaining body homeostasis.
Accumulating evidence suggests that protein-protein interactions play a critical role in the regulation of CFTR and other epithelial transporters (6 -8). PDZ (PSD-95/discs large/ZO-1) domain-based adaptors, best studied in the postsynaptic density (PSD) region of neurons, have emerged as a large group of proteins that sequester functionally related groups of transporters, receptors, and other effector proteins into integrated molecular complexes (9,10). Epithelial cells also utilize specific PDZ proteins to direct the polarized activities in their apical and basolateral membranes. Recently we reported functional and physical associations between the PDZ domain-containing protein Shank2 and two epithelial transporters, CFTR and the Na ϩ /H ϩ exchanger 3 (NHE3) (11,12). Interestingly Shank2 attenuated the cAMP-dependent regulation of CFTR and NHE3. Conversely it has been shown that PDZ domain-based adaptors, such as EBP50/NHERF1 and E3KARP/NHERF2, can enhance the effects of cAMP on these transporters by recruiting a cAMP-dependent protein kinase (PKA)/PKA-anchoring protein (AKAP) complex (13,14). Physiological significance of these adaptors in the regulation of CFTR function has also been verified in many other studies (6). For example, recent evidence demonstrates that native CFTR and EBP50 are co-immunoprecipitated in human airway epithelial cells (15) and that CFTRdependent anion current is reduced in the intestine of EBP50 knock-out mice (6).
The PDZ domain of Shank proteins has a three-dimensional structure very similar to the PDZ domains of EBP50/NHERF1 (16). In particular, these PDZ domains all contain a negatively charged amino acid at the end of the ␤ C strand of the PDZ domain structures (Glu 43 in hEBP50, Asp 634 in rShank1, and Asp 80 in hShank2) that preferentially interacts with a positively charged residue at the Ϫ1 position in the C terminus of the membrane transporters, such as ϪTRL in CFTR. Therefore, the question arises whether EBP50 and Shank2 have distinct CFTR binding sites or mutually compete for binding of a single site in cells expressing both adaptor proteins.
In the present study, we determined the kinetic properties and physiological significance of the interactions between CFTR and the PDZ-based adaptors EBP50 and Shank2. Using surface plasmon resonance (SPR) assays, we found that the dissociation constant (K D ) of CFTR-Shank2 binding was similar to that of CFTR-EBP50 binding and that both proteins apparently compete for binding at the same site. Consecutive patch clamp studies revealed that CFTR Cl Ϫ channel activity was dynamically regulated by the competition of Shank2 and EBP50 binding. Notably in contrast to the PKA/AKAP recruitment by EBP50 (6), Shank2 was found to tether PDE4D to the CFTR complex, thus attenuating cAMP/PKA signals. Our results suggest that the competitive balance between Shank2 and EBP50 binding to the CFTR channel may maintain homeostatic regulation of epithelial ion and fluid transport.
SPR Measurements and Kinetic Analysis of Sensorgrams-PDZ1-(1-139), PDZ2-(132-299), and PDZ1ϩ2-(1-299) domains of hEBP50/NHERF1 and the PDZ domain of rShank2/CortBP1-(1-142) were PCR-amplified and cloned into the pRSET A vector (Invitrogen) to create His-tagged constructs. The C-terminal 29 amino acid residues of hCFTR (GenBank TM accession number NM_000492) and PDZ1ϩ2 of hEBP50 (for competition assay, see Fig. 1D) were PCR-amplified and cloned into the pGEX-4T-1 vector (Amersham Biosciences) to create GST fusion constructs. All constructs were confirmed by sequence analysis. All fusion proteins were expressed in Escherichia coli BL21 (DE3) cells and purified on Ni 2ϩ -nitrilotriacetic acid (NTA) resin (Qiagen, Venlo, The Netherlands) and glutathione-Sepharose 4B (Amersham Biosciences) as appropriate. The purified proteins were quantified using Bio-Rad protein assay reagent, and their purity was assessed as Ͼ90% by Coomassie staining of SDS-PAGE gels. An SPR system equipped with an NTA chip (BIAcore 3000; Biacore AB, Uppsala, Sweden) was used to capture His-tagged PDZ fusion proteins. Analytes (GST-tagged proteins) at various concentrations in HEPES-buffered saline-EP (0.01 M HEPES, 0.15 M NaCl, 50 M EDTA, 0.005% Tween 20, pH 7.4) were perfused at a flow rate of 30 l/min. The sensor chip was regenerated between each analysis using successive injections of solutions containing 0.15 M NaCl and 0.35 M EDTA. Response curves were generated by subtracting the background signal generated simultaneously by the control flow cell. Background-subtracted curves were prepared for fitting by subtracting the signal generated by buffer alone on experimental flow cells. Sensorgram curves and kinetic parameters were evaluated by the BIAevaluation 3.1 software (Biacore AB), which uses numerical integration algorithms.
Single channel activity of CFTR was measured in inside-out configurations using fire-polished pipettes with a resistance of 20 -25 megaohms. The pipette solution contained 140 mM NaCl, 5 mM KCl, 1 mM MgCl 2 , 1 mM CaCl 2 , and 10 mM HEPES (pH 7.4), and the bath solution contained 140 mM NaCl, 5 mM KCl, 1 mM MgCl 2 , and 10 mM HEPES (pH 7.4). Following patch excision, channels were activated by adding the catalytic subunit of PKA (40 unit/ml; Promega, Madison, WI) and 1 mM MgATP. After channel activation, EBP50 and/or Shank2-PDZ at desired concentrations was added to the bath. Holding voltage used in the single channel recording was ϩ60 mV. The pCLAMP software package (version 9.2, Axon Instruments) was used for data acquisition and analysis. The voltage and current data were low pass-filtered at 1 kHz during the recordings, and the single channel data were further digitally filtered at 25 Hz. The number of active channels in a patch was determined from the number of simultaneously open channels during at least 15 min of recording.
Reverse Transcription-PCR-Reverse transcription-PCR analysis was performed to identify the phosphodiesterase (PDE) isoforms in NIH 3T3 cells and T84 cells using protocols reported previously (12). The primer sequences specific to PDE3 and PDE4 were selected from regions common to both mouse and human PDEs Immunoprecipitation (IP), Immunoblotting, and Immunohistochemistry-CHO-K1, COS-7, T84 cells, and rat cerebellum were lysed with lysis buffer (20 mM HEPES, pH 7.4, 200 mM NaCl, 5 mM EDTA, 1% Triton X-100, 1 mM NaVO 4 , and 1 mM ␤-glycerophosphate) containing Complete protease inhibitor mixture (Roche Applied Science). After lysis, cell debris were removed by centrifugation, and cleared lysates were mixed with the appropriate antibodies and incubated overnight at 4°C. Immune complexes were collected by incubation for 2 h at 4°C with protein A/G PLUS-agarose and washed four times with lysis buffer prior to electrophoresis. The immunoprecipitates and cell lysates were suspended in 2ϫ SDS sample buffer and boiled for 5 min and then separated by SDS-PAGE. The separated proteins were transferred to nitrocellulose membranes, and the membranes were blocked by incubation for 1 h in a solution containing 5% nonfat dry milk in 20 mM Tris-HCl, pH 7.5, 150 mM NaCl, and 0.05% Tween 20. The membranes were then incubated with the appropriate primary and secondary antibodies, and protein bands were detected with enhanced chemiluminescence solutions. Rabbit polyclonal antibodies anti-Shank2 1136 (17) and anti-PDE4D5 ␣-4D5 (18) and the mouse monoclonal pan-PDE4D antibody M3S1 (20) were described previously. Rabbit polyclonal pan-PDE4D antibody ab14613 was purchased from Abcam Inc., Cambridge, UK.
Immunostaining of frozen sections was performed as reported previously (12). Briefly colon tissues from Sprague-Dawley rats were embedded in OCT (Miles, Elkhart, IN), frozen in liquid N 2 , and cut into 4-m sections. The sections were stained by incubation with polyclonal anti-Shank2 1136 and monoclonal anti-pan-PDE4D M3S1 antibodies followed by appropriate secondary antibodies tagged with fluorophores. Images were obtained with a Zeiss LSM 510 confocal microscope.
Following overnight incubation at 4°C, samples were supplemented with 80 l of glutathione-Sepharose (Amersham Biosciences) and incubated for an additional 4 h at 4°C. The glutathione-Sepharose was pelleted and washed (3 ϫ 5 min) at 4°C with wash buffer (phosphate-buffered saline containing 0.1% Triton X-100 and 100 mM ␤-mercaptoethanol) prior to resuspension in SDS sample buffer and SDS-PAGE.
Measurements of PDE Activity-Whole brain tissues from wild type and PDE4D knock-out mice (21) were lysed in buffer containing 50 mM HEPES, pH 7.4, 250 mM NaCl, 10% glycerol, 0.5% Nonidet P-40, 1 mM EDTA, 0.2 mM EGTA, 10 mM NaF, 10 mM Na 4 P 2 O 7 , 1 M microcystin, 2 mM 4-(2-aminoethyl) benzenesulfonylfluoride, and a mixture of protease inhibitors. Cell lysates were centrifuged at 14,000 ϫ g, and supernatants were immunoprecipitated using 5 l of anti-Shank2 antibody coupled to 25 l of protein G-Sepharose for 4 h. After centrifugation, pellets were washed three times, and PDE activity was then measured as described previously (18). In brief, samples were assayed in a reaction mixture of 200 l containing 40 mM Tris-HCl, pH 8.0, 10 mM MgCl 2 , 5 mM ␤-mercaptoethanol, 1 M cAMP, 0.75 mg/ml bovine serum albumin, and 0.1 Ci of [ 3 H]cAMP for 10 min at 33°C. The reaction was terminated by the addition of 200 l of 10 mM EDTA in 40 mM Tris-HCl, pH 8.0, followed by heat inactivation in a boiling water bath for 1 min. The PDE reaction product 5Ј-AMP was then hydrolyzed by incubation of the assay mixture with 50 g of Crotalus atrox snake venom for 20 min at 33°C. The resulting adenosine was separated by anion exchange chromatography using 1 ml of AG1-X8 resin and quantitated by scintillation counting.

Shank2 and EBP50 Compete for Binding on CFTR-To
understand the dynamics of CFTR positive and negative regulations, we determined the binding kinetics of the association between CFTR and the PDZ domains of EBP50 and Shank2 using SPR analyses (Fig. 1). As EBP50 has two PDZ domains, three His-tagged proteins, PDZ1-(1-139), PDZ2-(132-299), and PDZ1ϩ2-(1-299), were used to analyze their kinetics separately. His-tagged PDZ domains were captured on NTA chips, and their binding to the GST-tagged C terminus of CFTR was sensored by SPR. As shown in the sensorgrams, the PDZ domains of Shank2 and EBP50 specifically bind the C terminus of CFTR in a dose-dependent fashion (Fig. 1, A and B). The overall dissociation constants at the equilibrium state (K D values) of Shank2 PDZ and EBP50 PDZ1ϩ2 were within comparable ranges at the tens of nanomolar levels (Fig. 1C). Interestingly EBP50 PDZ1ϩ2 has approximately a 1 order of magnitude faster association (K a ) and dissociation (K d ) kinetics than Shank2 PDZ. Further analyses of EBP50 PDZs using separate PDZ1 and PDZ2 proteins revealed that the overall EBP50 binding to CFTR is driven by PDZ1, which has a ϳ3 times higher affinity than PDZ2 (Fig. 1C). A solution competition assay was performed to examine the possibility of mutual competition between the PDZs of Shank2 and EBP50 in binding to CFTR. After capturing His-tagged Shank2 PDZ on NTA chips, the SPR system was perfused with a fixed dose of the C terminus of CFTR (300 nM) and increasing doses of GST-tagged EBP50 PDZ1ϩ2. Importantly the addition of EBP50 PDZs dose-dependently decreased the association between Shank2 PDZ and CFTR with an IC 50 value of 91 nM (Fig. 1D).
Regulation of CFTR Cl Ϫ Channel Activity by CFTR-EBP50 and CFTR-Shank2 Competitions-The above results imply that Shank2 and EBP50 compete for binding on CFTR. This com-petitive balance may affect CFTR ion transporting activities at the apical membrane of intestine, pancreas, and kidney epithelia, where Shank2 and EBP50 are abundantly expressed (7,12). We therefore determined the effects of the competition between EBP50 and Shank2 PDZ binding on the CFTR Cl Ϫ channel activity using inside-out configurations (Fig. 2). Following patch excision from CFTR-transfected CHO cells, patch membranes were washed with a high bath flow for 5 min to eliminate endogenous activation. When CFTR was activated by the addition of the catalytic subunit of PKA and ATP to the bath solution, an ion channel activity with a single channel conductance of 7.1 Ϯ 0.3 picosiemens and a linear current-voltage relationship was evoked that was absent in mock-transfected cells. It has been shown that adaptors with multiple PDZ domains, such as EBP50 and CAP70/PDZK1, can activate CFTR Cl Ϫ channel activity independently of cAMP signals by altering protein conformation including multimerization (23,24). In agreement with these reports, treatment with EBP50 (100 nM) induced a 2.8-fold increase in the open probability (P o ) of CFTR. Importantly addition of Shank2 PDZ (300 nM) inhibited the EBP50-induced increase in CFTR P o by 75 Ϯ 8% (Fig. 2, A and C). The effect of Shank2 PDZ versus EBP50 competition was then examined in reverse order. Shank2 PDZ treatment alone did not evoke significant changes in CFTR P o at basal levels; however, it greatly inhibited the CFTR activation mediated by EBP50. Addition of EBP50 induced only a minor increase in CFTR P o after Shank2 pretreatment (Fig. 2, B and  D). When performing the above experiments using EBP50 PDZ1ϩ2, which does not bind AKAPs due to the lack of the ezrin-radixin-mosesin (ERM) domain, we obtained results similar to those using the entire EBP50 protein (Fig. 2, C and D).
These findings indicate that Shank2 can reversibly inhibit the EBP50-induced CFTR activation by competing with EBP50 for binding at the C terminus of CFTR and hence by breaching the EBP50-induced active conformation of CFTR. However, this does not explain the observation that Shank2 inhibits the cAMP-induced phosphorylation and activation of CFTR in living cells. In CFTR-expressing NIH 3T3 cells, it has been shown that Shank2 overexpression decreased CFTR channel activity and phosphorylation induced by the adenylyl cyclase activator forskolin (12). It is well known that CFTR is primarily activated by phosphorylation of the R-domain by PKA (2). Recruitment by Shank2 of cyclic nucleotide PDEs, which attenuate cAMP signals, or phosphatases, which offset the action of PKA, might explain these findings. The latter is unlikely, however, because CFTR phosphorylation was decreased in the Shank2-transfected cells even in the presence of a high concentration of the nonspecific phosphatase inhibitor NaVO 4 (12).
Shank2 Associates with PDE4D-An important clue was obtained in an experiment with the non-selective PDE inhibitor IBMX (Fig. 3, A-D). Stimulation of CFTR-expressing CHO cells with 3 M forskolin induced a CFTR Cl Ϫ current that is dependent on the presence of Cl Ϫ in bath and pipette solutions, is inhibited by 5-nitro-2-(3-phenylpropylamino)benzoate, and has linear current-voltage relationships in whole-cell configurations. However, when Shank2 was co-expressed, the same amount of forskolin produced very small or no CFTR currents. Interestingly additional treatment with IBMX evoked a robust CFTR current in Shank2-expressing cells, although its peak amplitude was still 56 Ϯ 7% that of mock-transfected cells (Fig.  3, A and B).
These results suggest a functional association between the PDE activity and the Shank2-mediated inhibition of CFTR. We next identified the PDE subtypes involved and determined whether Shank2 directly interacts with PDEs. Of the 11 PDE families, PDE3 and PDE4 are strong candidates for Shank2 association as they are sensitive to IBMX and are known to be expressed in epithelial cells (25). The PDE3 family consists of two genes, 3A and 3B, and the PDE4 family consists of four, 4A-4D. Reverse transcription-PCR results using primers specific to the common regions of mouse and human PDEs revealed that several PDE3 and PDE4 genes, including PDE3A, -4A, and -4D, are expressed in mouse fibroblast NIH 3T3 cells and human colonic T84 cells (Fig. 3E). As both PDE3 and PDE4 are expressed, the patch clamp experiment shown in Fig. 3B was repeated after pretreatment of the cells with the PDE3specific inhibitor trequinsin and the PDE4-specific inhibitor rolipram (Fig. 3, C and D). PDE3 inhibition did not alter the effects of Shank2 (Fig. 3C). Conversely when PDE4 was inhibited by rolipram, 3 M forskolin consistently induced a robust CFTR current in Shank2-expressing cells (Fig. 3D, n ϭ 5), strongly suggesting an association between Shank2 and PDE4 isoforms. Recently two reports suggested that PDE4D plays a major role in forming a cAMP diffusion barrier at the apical regions of epithelial cells (22,26). Consistent with this notion, expression of PDE4D proteins was observed in the immunob- lots of NIH 3T3, CHO, and T84 cells (Fig. 3F) in which the inhibitory effects of Shank2 on cAMP/PKA signals were demonstrated in this and previous studies (11,12).
To probe for a physical association between Shank2 and PDE4D, co-IPs were performed. Human and rat PDE4D loci have multiple transcriptional units that code for at least nine splice variants, PDE4D1 to PDE4D9 (18). Among them, PDE4D5 has been shown to be expressed in epithelial cells and to mediate cAMP hydrolysis at the apical microdomain (22). Thus, Shank2 and the PDE4D splice variant PDE4D5 were overexpressed in CHO cells followed by IP with antibodies against Shank2 and PDE4D5 (Fig. 3G). Shank2 proteins were observed in immunoprecipitates with PDE4D5-specific antibodies. In a converse experiment, PDE4D5 could be detected in Shank2 immunoprecipitates. These co-IPs were specific as they could only be detected in cells expressing both Shank2 and PDE4D5 (Fig. 3G).
To identify its PDE4D-binding domain, several truncated Shank2 constructs were generated and used for pulldown assays. Shank2 has multiple sites for possible protein-protein interactions, including a PDZ domain, a long PR region, and a sterile ␣ motif domain (27). The PR region contains five proline-rich clusters, including a cortactin-binding domain (ppI) (28). In initial pulldown assays using the lysates from the PDE4D5-transfected CHO cells, PDE4D5 showed an interaction with the PR region of Shank2 (Fig. 4B, left). The PR region was then further dissected into two parts, PR1 and PR2, and the PDE4D binding site was mapped to the PR2 region (Fig. 4B,  right) where three proline-rich clusters and a ppI are clustered (Fig. 4A).
Next all PDE4D splice variants were co-immunoprecipitated with Shank2 to identify the PDE4D variants that interact with Shank2 and to identify the putative Shank2 binding site in PDE4D. COS-7 cells were transfected with plasmids encoding Shank2 and the nine PDE4D splice variants, PDE4D1 to PDE4D9, followed by immunoprecipitation using the pan-PDE4D antibody M3S1. All PDE4D long forms (D3, D4, D5, D7, D8, and D9) were found to interact with Shank2, whereas the short forms, D2 and D6, do not (Fig.  4D). A weak interaction was detected in the case of PDE4D1. The PDE4D splice variants are distinguished by the presence or absence of two conserved N-terminal domains called upstream conserved regions 1 and 2 (UCR1 and UCR2; Fig.  4C). The above results imply that UCR1 and the N-terminal part of UCR2 mediate Shank2 binding.
Shank2 and PDE4D Associate in Vivo-The association between Shank2 and PDE4D was then examined in cells that natively express both proteins to explore its physiological role. CFTR currents in Shank2-transfected cells were restored by treatment with the non-selective PDE inhibitor IBMX (B) or the PDE4-specific inhibitor rolipram (D) but not with the PDE3-specific inhibitor trequinsin (C). E, detection of PDE3 and PDE4 isoforms in NIH 3T3 cells (mouse fibroblast) and T84 cells (human colonic epithelia) by reverse transcription-PCR (RT-PCR). Primer sequences specific to PDE3 and PDE4 isoforms were selected from the regions common to both mouse and human PDEs. Lane M, molecular weight markers. F, expression of PDE4D proteins in NIH 3T3, CHO, and T84 cells was analyzed by immunoblotting using rabbit polyclonal pan-PDE4D antibody. G, Shank2 (Sh2) and the PDE4D splice variant PDE4D5 (4D5) were overexpressed in CHO cells (upper panels) and subsequently immunoprecipitated from cell extracts using anti-PDE4D5 and anti-Shank2 antibodies (lower panels). IB, immunoblot.
Initially we investigated the expression of Shank2 and PDE4D in rat colonic mucosa. Shank2 and PDE4D were highly co-localized at the apical region of colonic crypt cells, where CFTR plays a major role in fluid and ion secretion (Fig. 5A). The Shank2-PDE4D interaction was also verified by immunoprecipitation of the endogenous proteins prepared from T84 human colonic epithelial cells. It is well known that PDE4 can be activated by PKA-induced phosphorylation of UCR1 (29). Interestingly stimulation of the cAMP/PKA pathway using the adenylyl cyclase activator forskolin increased the physical association between Shank2 and PDE4D in T84 cells (Fig. 5B).
Finally the Shank2-PDE4D interaction was examined in mouse brain tissues where multiple PDE isoforms as well as Shank2 are expressed at high levels. In immunoprecipitates using pan-PDE4D antibodies, multiple Shank2-immunoreactive bands were detected in addition to the typical 170-kDa Shank2 (Fig. 5C) due to the presence of distinct Shank2 splice variants in brain tissue (17,28). Mouse brain tissue was then immunoprecipitated with anti-Shank2 antibodies, and the resulting IP pellets were subjected to PDE activity assays. The PDE activity co-immunoprecipitating with Shank2 was inhibited by the PDE4-selective inhibitor rolipram (Fig. 5D). In addi-tion, the co-immunoprecipitation of PDE was ablated in IPs using tissue from PDE4D knock-out mice (Fig. 5E) indicating that Shank2 specifically interacts with PDE4D in mouse brain tissue.

DISCUSSION
In the present study, we provide comprehensive evidence that the activity of CFTR is dynamically regulated by a competitive balance between CFTR-activating and CFTR-inactivating PDZ domain interactions. In addition, activity of Shank2, the CFTR-inactivating PDZ adaptor, was found to be associated with PDE4D, which precludes cAMP/PKA signals. The regulatory mechanisms resulting from the association of CFTR with EBP50 and Shank2 are summarized in Fig. 6. It is known that EBP50 can activate CFTR through two independent mechanisms. First, EBP50 can recruit ezrin, a PKA-anchoring protein, and hence facilitates the cAMP/PKA-mediated phosphorylation of CFTR (7). Second, EBP50 can activate CFTR by conformational changes, possibly by forming a CFTR dimer (23). The results of this study revealed that Shank2 can inhibit both mechanisms by competing with EBP50 at the C terminus of CFTR. The finding that Shank2 inhibited the EBP50-induced increase of CFTR P o in inside-out patches (Fig. 2) demonstrated that Shank2 can antagonize the EBP50 function by breaching the EBP50-induced active conformation of CFTR independently of cAMP signals. Even more interesting, this study shows that Shank2 can ablate cAMP signaling in the vicinity of the CFTR by tethering PDE4D into this macromolecular complex.
It is well known that cAMP signals are compartmentalized in many cell types and that localized cAMP signals play a critical role in various physiological processes (29). For example, localized cAMP accumulation in a region overlapping with the Z band and the T tubules in cardiomyocytes is essential for the regulation of cardiac myocyte contraction by ␤-adrenergic receptors (30). Several recent reports underscored the importance of PDE4D as a cAMP diffusion barrier formed at the apical membrane of epithelia for the proper functioning of CFTR (22,26). However, the molecular mechanisms for PDE accumulation in apical regions remained obscure. A thorough molecular characterization in this study revealed that the apical adaptor Shank2 recruits PDE4D through a direct interaction between the proline-rich region of Shank2 and the UCR1/2 region of PDE4D. The proximal localization of cAMP regulating machinery will induce an increased signaling efficiency in regulating CFTR activity. Recently Shank2E, a splice variant of Shank2 with additional ankyrin repeats and a Src homology 3 domain, was identified in epithelial cells (31). Therefore, we FIGURE 5. Shank2 and PDE4D associate in vivo. A, immunolocalization of Shank2 and PDE4D in rat colon was detected using polyclonal anti-Shank2 antibody 1136 and monoclonal pan-PDE4D antibody M3S1. Shank2 and PDE4D were co-localized in the apical region of colonic crypt cells. B, co-IP of Shank2 and PDE4D from extracts of human colonic T84 cells. Forskolin treatment (10 M for 30 min) enhanced the association between Shank2 and PDE4D. C, detection of Shank2 proteins in PDE4D immunoprecipitates from mouse brain. D and E, measurement of PDE activity in Shank2 immunoprecipitates from mouse brain. In D, PDE activity in Shank2 IPs from wild type mice was determined in the presence or absence of the PDE4-specific inhibitor rolipram (10 M). In E, PDE activity in immunoprecipitates from PDE4D wild type mice (PDE4D-WT) is compared with immunoprecipitates from PDE4D knock-out mice (PDE4D-KO). Ab, antibody; IB, immunoblot. FIGURE 6. A model for the regulation of CFTR through interaction with EBP50 and Shank2. Through their respective PDZ domains, Shank2 and EBP50 compete for binding of the CFTR C terminus. Binding of EBP50 activates CFTR through a conformational change. In addition, EBP50 facilitates the PKA-mediated phosphorylation and activation of CFTR by bringing the AKAP ezrin and PKA into the protein complex. Shank2 inhibits CFTR activity by breaching the CFTR-EBP50 association and by bringing PDE4D, which precludes cAMP/PKA signaling, closer to CFTR. SAM, sterile ␣ motif; R of CFTR, R-domain; R and C of PKA, regulatory and catalytic subunits.
repeated the experiments in Fig. 3 with Shank2E. The results were nearly identical to those obtained using Shank2 suggesting that domains within the structure of Shank2 are important for regulating CFTR function (data not shown).
The physiological significance of the Shank2-PDE4D association is not confined to epithelial cells but can be extended to other organs. For example, Shank2 is widely expressed in many regions of the brain including cortex, hippocampus, and cerebellum. Shank proteins are the key organizer of PSD and interact with Homer (27,32), which is required for efficient signaling of metabotropic glutamate receptors that play a critical role in learning and memory. Effector systems of metabotropic glutamate receptors are not only associated with Ca 2ϩ signaling but also with cAMP signals, and it is believed that cAMP signals play a role in the functional specificity of each metabotropic glutamate receptor isoform (33,34). Therefore, Shank2-associated PDE4D may play a critical role in the regulation of PSD in neuronal cells.
Recent studies with knock-out animals of NHERF family PDZ proteins (EBP50, E3KARP, and PDZK1) have shown that the functions of these adaptors are more complex than previously appreciated and appear to be tissue-specific (6). Because PDZ domains have well defined binding sites, they are promising targets for drug discovery. However, as the present study demonstrates, much is yet to be learned about the function of each PDZ-based adaptor before drugs targeting PDZ interactions can become a reality (35). For example, as the PDZ structures of EBP50 and Shank2 are very similar, small molecules or peptides targeting EBP50 may have untoward effects in vivo by disrupting Shank2 complexes as well.
In conclusion, the present study demonstrates the functional diversity of PDZ-mediated protein-protein interactions and illustrates that opposite signals can be delivered to the same PDZ-binding motif of a given membrane protein by different adaptors. A competitive balance between the signal-conferring and signal-stopping PDZ interactions would be critical in the regulation of many membrane transporters and receptors as demonstrated here for CFTR.