In this issue of the Biochemical Journal, Zhang et al. reveal a new strategy for modifying the regulated function of CFTR (cystic fibrosis transmembrane conductance regulator) on the apical surface of epithelial cells. Simply stated, these authors tested the idea that the cAMP-dependent channel activity of CFTR could be effectively enhanced by disruption of a protein–protein interaction which is normally inhibitory for the production of cAMP. This particular protein–protein interaction [between the PDZ motif of LPA2 (type 2 lysophosphatidic acid receptor) and the scaffold protein Nherf2 (Na+/H+ exchanger regulatory factor 2)] is localized in the CFTR interactome on the apical membrane of epithelial cells. Hence disruption of the LPA2–Nherf2 interaction should lead to a localized elevation in cAMP and, consequently, increased cAMP-dependent CFTR activity on the surface of epithelial cells. Zhang et al. confirmed these expectations for a small-molecule compound targeting the LPA2–Nherf2 interaction using relevant cultures and tissues thought to model the human respiratory epithelium. The success of this strategy depended on previous knowledge regarding the role for multiple PDZ-motif-mediated interactions in signalling (directly or indirectly) to CFTR. Given the number and diversity of such PDZ-mediated interactions, future structural and computational studies will be essential for guiding the design of specific pharmacological interventions.
IMPORTANCE OF PROTEIN–PROTEIN INTERACTIONS IN BIOLOGY
Protein–protein interactions regulate virtually all cellular processes by promoting appropriate cellular localization of regulatory partners and/or by facilitating the channelling of protein substrates or products through reaction pathways ensuring exquisite temporal and spatial control. There has been extensive study of the role of protein–protein interactions in regulating cAMP-mediated signal transduction. We know now that discrete signalling complexes can contain not only proteins which act to generate cAMP (i.e. adenylate cyclase), but also contains proteins which control its degradation (i.e. phosphodiesterase 4) . The co-ordinated function of the protein constituents of these complexes ensure tight regulation of the concentration of this critical second messenger. Not only is the localized concentration of cAMP regulated by interacting proteins, but also the downstream effector molecules, including PKA (protein kinase A) and EPAC (exchange protein directly activated by cAMP) are specifically integrated and regulated by such protein complexes.
Cystic fibrosis is caused by loss of activity of the CFTR (cystic fibrosis transmembrane conductance regulator) in the apical membrane of epithelial cells, and this loss of function leads to respiratory and gastrointestinal disease with early morbidity and mortality. CFTR is an effector molecule of cAMP action and its activity as a phosphorylation-regulated anion channel is dependent on the interactome of regulatory proteins which control cAMP concentration and cAMP-dependent PKA activity. Interactions vital for the phosphorylation-dependent regulation of CFTR are mediated by PDZ-domain-containing scaffold proteins, and the paper by Zhang et al.  in this issue of the Biochemical Journal describes their investigation of the potential for targeting specific PDZ-mediated interactions as a means of modifying CFTR activity.
IMPORTANCE OF PDZ-MOTIF-MEDIATED INTERACTIONS IN BIOLOGY
A PDZ motif is a specific protein-binding motif or -binding sequence that is recognized by many interacting partners. A PDZ motif is so named for the first several proteins identified to possess such motifs [PSD-95 (postsynaptic density 95), Drosophila junctional protein Dlg (Discs large) and the epithelial tight junction protein ZO-1 (zonula occludens 1)]. PDZ motifs are recognized by a broad array of PDZ scaffolding proteins, which typically possess a hydrophobic cleft into which the PDZ motif can bind. These scaffold proteins often possess multiple PDZ-binding domains capable of simultaneously integrating a complex of proteins which interact to mediate biological signals.
The canonical mode for interaction with PDZ-binding domains in scaffold proteins is via the extreme C-terminus of bait proteins. Importantly, however, PDZ-binding domains show a range of selectivities for different C-terminal residues, and various classification schemes have been developed to help in predicting target recognition. For instance, the hydroxy side chain of the serine or threonine residue as position P-2 of the class 1 PDZ-binding mode (X-S/T-X-Φ, with Φ representing a hydrophobic amino acid) typically binds to a histidine residue located in a helical element in the PDZ domain . To date, the determinants for specificity of target protein recognition remain unclear. Recently, computational methods have been applied to mine PDZ-interaction databases generated using large-scale proteomic approaches. The results of one such study emphasizes the propensity for cross-reactivity within the ‘PDZome’ and highlights the importance of auxiliary factors (as yet incompletely defined) in preventing such cross-reaction . The computational and proteomic studies of Macbeath and co-workers  support the view that PDZ-mediated interactions do not fall into discrete classes, but rather are evenly distributed throughout ‘selection space’ in the proteome. Together, these studies emphasize the need for care in designing ligands to target specific interactions.
POTENTIAL ROLE OF BIOLOGICAL AGENTS OR SMALL MOLECULES WHICH SPECIFICALLY TARGET PDZ INTERACTIONS IN THERAPEUTIC INTERVENTION
Several previous studies have revealed the potential of small molecules or biological agents (such as peptides) to disrupt a particular PDZ-mediated interaction and modify a specific biological process . Recently, a small-molecule compound disrupting the PICK1 (protein that interacts with protein C-kinase 1) PDZ domain interaction with GluR2 (glutamate receptor 2) was described . As this PDZ interaction has been implicated in the generation of neuropathic pain, excitotoxicity and cocaine addiction, the authors anticipate that this small molecule may be useful for therapeutic intervention in these pathological conditions.
In this particular case, the small-molecule inhibitor was discovered using the following strategy . First, a high-throughput screening platform was developed to identify chemical compounds (44000 screened in this instance) which bind to the PDZ domain of PICK1. The authors of this study selected a chemical ligand which binds with an affinity (10 μM) similar to that of the natural ligand (i.e. the isolated PDZ domain of GluR2). Secondly, computational docking methods were employed to define the small-molecule-binding site on the PICK1 PDZ domain crystal structure. Importantly, this interaction was confirmed by mutagenesis, thereby defining the molecular basis for specificity of the small-molecule interaction with the protein of interest. Finally, this specificity was established in comparative biochemical studies with other biologically relevant proteins containing PDZ domains, i.e. PSD-95 and GRIP1 (glutamate-receptor-interacting protein 1). Clearly, the availability of high-resolution structural information regarding PICK1 was invaluable in defining the binding site of the ‘lead’ small molecule in this case.
CFTR FUNCTIONS IN A MACROMOLECULAR COMPLEX ON THE APICAL MEMBRANE OF EPITHELIAL CELLS: A COMPLEX MEDIATED VIA PDZ INTERACTIONS
The phosphorylation and localization of CFTR to the apical membrane of epithelial cells lining the airways and other tubular organs is crucial to its function in salt and water transport across the relevant organ surface. Both its phosphorylation, which modulates its activity, and apical surface localization are achieved, at least in part, through interaction with a variety of proteins in a complex protein–protein interaction network known as the CFTR interactome.
CFTR is known to interact with a wide and growing array of protein partners in vivo. Such partner proteins include cytoskeletal proteins, transporters and channels, receptors, kinases and phosphatases (see Li and Naren  for an excellent review). In the C-terminal region of CFTR, the interactions occur via a PDZ motif, which in the case of human CFTR, possesses the sequence D1477TRL1480.
CFTR interacts directly with at least six different PDZ scaffolding proteins via its PDZ motif. These scaffolding proteins include Nherf1 [Na+/H+ exchanger regulatory factor 1; EBP50 (ezrin/radixin/moesin-binding phosphoprotein 50)], Nherf2, PDZK1 [PDZ domain-containing protein in the kidney 1; Cap70 (CFTR-associated protein 70); Nherf3], PDZK2 [IKEPP (intestinal and kidney-enriched PDZ protein); Nherf4], CAL [CFTR-associated ligand; GOPC (Golgi-associated PDZ and coiled-coil domain-containing)] and Shank2 (Src homology 3 and ankyrin repeats-containing protein 2). CAL is localized mainly to the trans-Golgi network and binds directly to CFTR .
Nherf1 and Nherf2 interact with the cytoskeleton via ezrin , suggesting that their interaction with CFTR may help anchor the protein to the cytoskeleton at the apical membrane, although the PDZ motif is not strictly required for apical expression of the protein . Nherf1 competes for binding to the PDZ domain of CFTR with CAL, and its overexpression stimulates CFTR apical expression and therefore activity . CAL appears to retain CFTR and target it for degradation [12,13], as its knockdown increases surface expression of the mutant ΔF508-CFTR .
LPA2 [type 2 LPA (lysophosphatidic acid) receptor] binds to the signalling lipid LPA, and resides in the apical membrane region of epithelial cells where CFTR is localized. LPA2 contains a C-terminal PDZ-binding motif and is known to interact with Nherf2 [15–17]. Upon interaction of LPA with LPA2, a Gi signalling pathway is activated which inhibits adenylate cyclase, reducing local cAMP levels in the vicinity of CFTR and reducing PKA-stimulated phosphorylation of CFTR. Therefore disruption of the LPA2–Nherf2 interaction would be anticipated to increase the channel function of CFTR in the apical membrane.
A NEW APPROACH TO ENHANCE ACTIVITY OF CFTR BY TARGETING SIGNALLING THROUGH LPA2
The study by Zhang et al.  in this issue of the Biochemical Journal shows that it possible to modify the regulation of CFTR in the apical membrane of epithelial membranes, not by targeting its own PDZ-motif-mediated interactions with Nherf2, but rather by targeting Nherf2 interactions with a distinct membrane protein, one which acts to inhibit CFTR. Zhang et al.  show that a small molecule (compound CO-068) which targets LPA2–Nherf2 interaction is effective in enhancing CFTR-dependent Cl− secretion because it leads to an elevation in the concentration of cAMP in the local microenvironment surrounding CFTR. This strategy is novel and highlights the potential to fine-tune the regulated function of CFTR by modulating indirect functional interactions rather than by modifying direct interactions with CFTR.
The strategy used to identify a small-molecule inhibitor of the LPA2–Nherf2 interaction was different from the strategy described previously for identification of a chemical inhibitor of the PICK1–GluR2 interaction . First, Zhang et al.  employed a luminescence-based assay to define the affinity for direct binding between GST (glutathione transferase)-tagged Nherf2 and a biotin-tagged LPA2 peptide (a ten-residue peptide). The affinity of this interaction was estimated as 10 μM. Then Zhang et al.  screened a library of compounds designed previously to inhibit PDZ based protein–protein interactions [18–21] for the ability of these compounds to inhibit the above interaction. Compound CO-068 was found to inhibit the LPA2–Nherf2 interaction with an IC50 of approximately 65 μM. Furthermore, the potential for this small molecule to modify CFTR function indirectly by disrupting LPA2–Nherf2 interactions was validated in meaningful biological systems such as cultures of human respiratory epithelia and pig tracheal tissues. Unfortunately, the IC50 for compound CO-068 is relatively high, thereby limiting its application to in vivo studies. However, we suggest that this problem will be resolved and higher-affinity probes can be defined once structural models for the LPA2–Nherf2 interaction have been determined and the binding site for compound CO-068 has been defined.
POTENTIAL THERAPEUTIC APPLICATIONS OF SMALL MOLECULES THAT TARGET CFTR PDZ INTERACTIONS
The most common cystic fibrosis mutation, ΔF508-CFTR, is misfolded and much of the protein is retained and degraded in the biosynthetic compartment ( and reviewed in ). In addition to the trafficking defect, ΔF508-CFTR possesses channel activity that, although measurable, is compromised in comparison with wild-type CFTR (reviewed in ). Corrector molecules are being developed to help overcome the trafficking defect for ΔF508-CFTR, but its activity may still be compromised at the cell surface. A small molecule similar to compound CO-068 (described in the paper by Zhang et al. ) may be clinically useful in combination with a corrector molecule which causes biosynthetic rescue, possibly by enhancing phosphorylation-dependent gating. However, it remains to be shown whether treatments which disrupt the LPA2–Nherf2 interaction and cause an increase in cytosolic cAMP and cAMP-dependent kinase activity will have an effect on this or other mutant versions of CFTR.
Although loss of CFTR function leads to cystic fibrosis, it is well known that excessive gain of CFTR function also leads to disease. Cholera toxin released by pathogenic bacteria activate membrane-associated secondary messengers such as cAMP, leading to phosphorylation of CFTR and excessive CFTR channel activity in the gut (reviewed in ). This induces massive Cl− secretion and concomitant Na+ and water loss that is associated with secretory diarrhoea in cholera infection. The selection of small molecules to disrupt the CFTR–Nherf1 interaction in a manner analogous to that used for the LPA2–Nherf2 interaction by Zhang et al.  would be anticipated to help to spatially isolate the channel from the aberrant activating signals and attenuate the effects of the toxin.
NEXT STEPS IN OPTIMIZING SMALL MOLECULES WHICH TARGET PDZ-MEDIATED INTERACTIONS WHICH MODIFY CFTR
The elegant study by Zhang et al.  reveals a new strategy for modifying the regulated function of CFTR on the surface of epithelial cells. These authors provided proof for the principle that the regulation of CFTR can be modified by the introduction of a small molecule selected for its ability to disrupt an inhibitory interaction located in the apical membrane–CFTR interactome. The small molecule selected showed a relative specificity for interfering with the LPA2–Nherf2 interaction over the CFTR–Nherf2 or PLC (phospholipase C)-β3 interaction, thereby assuring its competition of a negative interaction and overall stimulation of cAMP-dependent channel activity. These findings are exciting and will probably set the stage for related drug discovery strategies for regulating CFTR.
On the basis of the number and diversity of PDZ-domain-mediated interactions, we predict that the development of small-molecule compounds which specifically target a particular interaction will continue to be challenging. As the PDZome expands, so will the number of possible protein partnerships that may be affected by a single intervention. On the other hand, the development of specific interventions will be enhanced through the generation of high-resolution structural information defining the molecular details of each of the PDZ domain interactions affecting CFTR in the apical membrane. Furthermore, the affinity of small-molecule interactions may be improved with innovations in the field of chemical biology. For example, as most scaffold proteins contain multiple PDZ domains, the invention of bivalent ligands are likely to exhibit improved affinity. The first example of a bivalent peptide-based ligand was recently described with relatively high affinity (10 μM) for the AMPA (α-amino-3-hydroxy-5-methylisoxazole-4-propionic acid) receptor regulatory protein, TARP . Finally, we are optimistic that the results of the study by Zhang et al.  will drive discovery in cystic fibrosis research and in the field of epithelial cell biology.
This work was supported by Operating Grants awarded to C.E.B. by Cystic Fibrosis Canada, the Canadian Institutes of Health Research and the U.S. Cystic Fibrosis Foundation. P.D.W.E. was supported by a Restracomp (Research Training Competition) award from the Hospital for Sick Children and a postdoctoral fellowship from Cystic Fibrosis Canada.