Structural Basis of the Na+/H+ Exchanger Regulatory Factor PDZ1 Interaction with the Carboxyl-terminal Region of the Cystic Fibrosis Transmembrane Conductance Regulator*

The PDZ1 domain of the Na+/H+ exchanger regulatory factor (NHERF) binds with nanomolar affinity to the carboxyl-terminal sequence QDTRL of the cystic fibrosis transmembrane conductance regulator (CFTR) and plays a central role in the cellular localization and physiological regulation of this chloride channel. The crystal structure of human NHERF PDZ1 bound to the carboxyl-terminal peptide QDTRL has been determined at 1.7-Å resolution. The structure reveals the specificity and affinity determinants of the PDZ1-CFTR interaction and provides insights into carboxyl-terminal leucine recognition by class I PDZ domains. The peptide ligand inserts into the PDZ1 binding pocket forming an additional antiparallel β-strand to the PDZ1 β-sheet, and an extensive network of hydrogen bonds and hydrophobic interactions stabilize the complex. Remarkably, the guanido group of arginine at position −1 of the CFTR peptide forms two salt bridges and two hydrogen bonds with PDZ1 residues Glu43 and Asn22, respectively, providing the structural basis for the contribution of the penultimate amino acid of the peptide ligand to the affinity of the interaction.

The PDZ1 domain of the Na ؉ /H ؉ exchanger regulatory factor (NHERF) binds with nanomolar affinity to the carboxyl-terminal sequence QDTRL of the cystic fibrosis transmembrane conductance regulator (CFTR) and plays a central role in the cellular localization and physiological regulation of this chloride channel. The crystal structure of human NHERF PDZ1 bound to the carboxyl-terminal peptide QDTRL has been determined at 1.7-Å resolution. The structure reveals the specificity and affinity determinants of the PDZ1-CFTR interaction and provides insights into carboxyl-terminal leucine recognition by class I PDZ domains. The peptide ligand inserts into the PDZ1 binding pocket forming an additional antiparallel ␤-strand to the PDZ1 ␤-sheet, and an extensive network of hydrogen bonds and hydrophobic interactions stabilize the complex. Remarkably, the guanido group of arginine at position ؊1 of the CFTR peptide forms two salt bridges and two hydrogen bonds with PDZ1 residues Glu 43 and Asn 22 , respectively, providing the structural basis for the contribution of the penultimate amino acid of the peptide ligand to the affinity of the interaction.
The cystic fibrosis transmembrane conductance regulator (CFTR) 1 is an ATP-regulated chloride channel that determines the rate of electrolyte and fluid transport in the apical mem-brane of epithelial cells (1)(2)(3). Abnormal CFTR function is associated with the pathogenesis of cystic fibrosis and secretory diarrhea (1)(2)(3). The CFTR activity is modulated through interactions with other proteins; however the regulatory mechanisms remain unknown. One protein that interacts with the carboxyl terminus of CFTR is the Na ϩ /H ϩ exchanger regulatory factor (NHERF), a cytoplasmic protein originally cloned as an essential cofactor for the cAMP-dependent protein kinasemediated inhibition of the Na ϩ /H ϩ exchanger 3 (4 -6). NHERF is also known as EBP50 (ezrin-radixin-moesin-binding phosphoprotein-50), a membrane-cytoskeleton linking protein that binds to membrane proteins through its two PDZ (PSD-95/ Discs-large/ZO-1) domains and to the cortical actin cytoskeleton through its carboxyl-terminal domain (7). The NHERF PDZ1 and PDZ2 domains (Fig. 1A) bind with nanomolar affinity to the CFTR carboxyl-terminal sequence QDTRL and play a critical role in the regulation of channel gating (8 -10). In addition, the NHERF-related protein, NHERF2, also binds to the carboxyl-terminal tail of CFTR through its two PDZ domains (11) (Fig. 1A). Interestingly, CAP70, the murine homolog of the PDZK1 protein (12,13), also interacts with the CFTR carboxyl terminus through its PDZ3 domain and modulates the channel activity (12). These findings corroborate previous studies in establishing the essential role of the CFTR carboxylterminal motif DTRL for the functional expression of this channel in the apical plasma membrane (14 -16).
PDZ domains are protein modules that mediate specific interactions between proteins and participate in the assembly of membrane receptors, ion channels, and other signaling molecules into specific signal transduction complexes (17,18). PDZ domains bind to short carboxyl-terminal peptides and have been categorized into two classes based on target sequence specificity. Class I domains bind to peptides with the consensus sequence (S/T)X(V/I/L) (X denoting any amino acid), whereas class II domains recognize the motif (F/Y)X(F/V/A) (19). The PDZ fold comprises a six-stranded antiparallel ␤-barrel capped by two ␣-helices (20 -27). Peptide ligands interact with PDZ domains by a ␤-sheet augmentation process, in which the peptide forms an additional antiparallel ␤-strand in the PDZ ␤-sheet (28). It is believed that the specificity and affinity of the PDZ-peptide interaction is achieved by the residues at positions Ϫ3, Ϫ2, and 0 of the peptide (position 0 referring to the carboxyl-terminal residue), whereas residue Ϫ1 does not play an important role in the interaction.
To elucidate the structural determinants of the NHERF PDZ1-CFTR interaction, we solved the crystal structure of the PDZ1 domain in complex with the CFTR carboxyl-terminal peptide QDTRL. The structure reveals for the first time that the arginine at position Ϫ1 of the peptide ligand interacts with PDZ1 residues, thus contributing to the affinity of the NHERF PDZ1-CFTR interaction.

EXPERIMENTAL PROCEDURES
Protein Purification and Crystallization-A DNA fragment encoding the human NHERF PDZ1 (residues 11-94), and having the carboxylterminal extension Q 95 DTRL 99 that corresponds to residues 1476 -1480 of human CFTR (1), was amplified using the polymerase chain reaction and cloned in the vector pGEX-2TJL (27). PDZ1 was expressed in Escherichia coli BL21 (DE3) cells as a glutathione S-transferase fusion protein, purified using glutathione-Sepharose 4B resin, and the PDZ1 was released by digestion with thrombin, as described previously (27). PDZ1 protein (18 mg/ml) was crystallized using the sitting drop vapor * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The  1 The abbreviations used are: CFTR, cystic fibrosis transmembrane conductance regulator; NHERF, Na ϩ /H ϩ exchanger regulatory factor; PDZ, PSD-95/Discs-large/ZO-1 homology. diffusion method in 0.1 M sodium acetate, pH 4.6, 2 M sodium chloride, at 20°C. Diffraction data were collected at room temperature using an R-AXIS IV detector and CuK ␣ radiation. The data were processed using the programs DENZO and SCALEPACK (29) (Table I). Crystals belong to space group P3 1 21 with unit cell parameters a ϭ b ϭ 51.7 Å, c ϭ 67.0 Å, and one molecule in the asymmetric unit.
Structure Determination and Refinement-The structure was solved by molecular replacement using the program MOLREP (30) and the human NHERF PDZ1 (Protein Data Bank code 1g9o) as the search model. The rotation function search in the 20 -3 Å resolution range produced a clear solution with a peak height of 5.9 sigma. The translation function indicated that the correct space group was P3 1 21 with a correlation coefficient of 0.36 and an R cryst of 0.50, compared with its enantiomorphic mate P3 2 21, which had a correlation coefficient of 0.23 and an R cryst of 0.56. The model was built using the program O (31) and was refined by the maximum likelihood method using REFMAC5 (32). The structure is well ordered except for the loop regions 31-35 and 81-85, which are disordered. The PDZ1 also contains at its amino terminus the vector-derived residues GSSRM, from which only methionine is ordered and included in the final model.

RESULTS AND DISCUSSION
Structure Determination-We recently determined the crystal structure of the human NHERF PDZ1 domain (residues 11-99) at 1.5-Å resolution (27). The crystal structure produced a dimeric arrangement of PDZ1 domains with the carboxylterminal region T 95 DEQL 99 of one PDZ1 molecule bound to a neighboring PDZ1 because of its resemblance to the PDZ1 ligand consensus (27). We exploited this intermolecular association of NHERF PDZ1 in the crystalline state to facilitate the co-crystallization of this domain with the CFTR ligand by converting the PDZ1 sequence T 95 DEQL 99 to Q 95 DTRL 99 , which corresponds to the CFTR carboxyl-terminal tail. Recombinant NHERF PDZ1 was crystallized, and its structure was determined by molecular replacement. The model was refined to an R cryst of 18.7% and an R free of 21.7% (Table I), and the evaluation of its stereochemistry using PROCHECK (33) showed that 89.2% of the residues are in the most favored, 8.1% in the additional allowed, and 2.7% in the generously allowed regions.
Overview of the Structure-The present NHERF PDZ1 crystal structure produces infinite head-to-tail polymers of PDZ1 molecules along the z axis, with the carboxyl-terminal extension Q 95 DTRL 99 of one PDZ1 molecule serving as a ligand for a neighboring PDZ1 (Fig. 1B). The overall topology of NHERF PDZ1 is similar to other PDZ structures (20 -27), consisting of six ␤-strands (␤1-␤6) and two ␣-helices (␣1 and ␣2) (Fig. 1,   FIG. 1. Structure of the NHERF PDZ1 domain bound to the CFTR sequence QDTRL. A, sequence comparison of PDZ domains that bind to CFTR. The indicated PDZ domains from human NHERF (5), human NHERF2 (8), and murine PDZK1/CAP70 (12) were aligned using MACAW (36). Absolutely conserved residues are shown as white letters on blue background. Identical residues in four domains are shaded in cyan. The secondary structure of NHERF PDZ1 is indicated at the top. Conserved acidic residues proposed to interact with Arg Ϫ1 of the CFTR ligand are denoted by an asterisk. B, stereo view of the NHERF PDZ1 crystal packing. Each carboxyl terminus serves as a ligand for a neighboring PDZ1 molecule. C, ribbon diagram of the NHERF PDZ1 domain bound to the QDTRL peptide. The strands ␤1-␤6 are shown in yellow, and the helices ␣1 and ␣2 are shown in green. The peptide ligand QDTRL is shown in pink. The figure was made using MOLSCRIPT (37) and Raster3D (38). D, surface topology of the NHERF PDZ1 bound to the peptide QDTRL. The figure was generated using GRASP (39).

A and C).
Structural Basis for the Specificity of the NHERF PDZ1-CFTR Interaction-The peptide ligand Q 95 DTRL 99 inserts into the PDZ1 binding pocket antiparallel to the ␤2 strand and extends the ␤-sheet of PDZ1 (Fig. 1, C and D). In this arrangement, the invading pentapeptide is highly ordered, as indicated by the high quality electron density map ( Fig. 2A). The carbonyl oxygen of Gln Ϫ4 hydrogen bonds with the amide nitrogen of Gly 30 (Fig. 2, B and C), indicating that Gln Ϫ4 does not contribute to the specificity of the interaction. By contrast, Asp Ϫ3, Thr Ϫ2, and Leu 0 are engaged in numerous interactions with PDZ1, consistent with biochemical evidence on the central role of these residues in the specificity and affinity of the NHERF PDZ1-CFTR interaction (8 -10). Specifically, the O ␦1 atom of Asp Ϫ3 hydrogen bonds with N ␦1 of His 27 , and the O ␦2 atom of Asp Ϫ3 forms a salt bridge with N 1 of Arg 40 (Fig. 2, B  and C). Similarly, the amide nitrogen and carbonyl oxygen of Thr Ϫ2 hydrogen bond with the carbonyl oxygen and amide nitrogen of Leu 28 , respectively, while the O ␥1 atom of Thr Ϫ2 hydrogen bonds with the N ⑀2 atom of the conserved His 72 (Fig.  2, A-C).
The side chain and carboxylate group of Leu 0 enter into a deep cavity formed by Tyr 24 , Gly 25 , Phe 26 , Leu 28 , Val 76 , and Ile 79 residues (Fig. 1D). The C ␦1 atom of Leu 0 makes hydrophobic contacts with the atoms C ⑀2 and C of Phe 26 and C ␦1 of Ile 79 (Fig. 2C). In addition, the carboxyl oxygen of Leu 0 hydrogen bonds with the amide nitrogens of Gly 25 and Phe 26 , whereas the carbonyl oxygen of Leu 0 hydrogen bonds directly with the amide nitrogen of Tyr 24 and indirectly with the N atom of Arg 80 in the ␣2 helix through two ordered water molecules (Fig. 2, B and C). The involvement of Arg 80 in carboxylate binding through ordered water molecules represents a novel feature of the PDZ-ligand interaction and differs from other PDZ structures where this function is mediated by an arginine residue in the ␤1-␤2 loop (20), corresponding to NHERF PDZ1 Lys 19 . In the present structure, Lys 19 does not appear to be involved in hydrogen bonding with the terminal carboxylate group. The isobutyl side chain of Leu 0 fits tightly in the hydrophobic cavity of PDZ1, suggesting that this stereochemical complementarity may underlie the strict requirement for carboxyl-terminal leucine in all the high affinity ligands of NHERF PDZ1 (8 -10). Conceivably, smaller side chains would leave vacated spaces within this hydrophobic cavity that would be energetically unfavorable (34), whereas bulkier side chains would not readily fit within this pocket. Moreover, the hydro-  phobic character of the cavity would likely exclude polar and charged side chains. It therefore appears that the volume, shape, and hydrophobicity of the PDZ pocket provide the structural determinants for the selection of stereochemically complementary hydrophobic carboxyl-terminal side chains for high affinity binding.
The Importance of Arg Ϫ1 for the Affinity of the NHERF PDZ1-CFTR Interaction-Strikingly, the guanido group of Arg Ϫ1 forms two salt bridges with O ⑀2 of Glu 43 and two hydrogen bonds with the carbonyl oxygen of Asn 22 (Fig. 2, A-C). This finding was unexpected because the residue Ϫ1 of the peptide has been considered to be unimportant for the PDZ-ligand interaction. Indeed, in other PDZ structures the side chain of the penultimate residue is oriented toward the solution and does not interact with PDZ residues (20,22). Nevertheless, previous biochemical studies demonstrated that arginine is the preferred residue at position Ϫ1 for optimal binding to NHERF PDZ1 (8,9). Affinity selection experiments showed that NHERF PDZ1 selected almost exclusively ligands with arginine at position Ϫ1 from random peptides (9). Furthermore, point mutagenesis of the penultimate arginine to alanine, phenylalanine, leucine, or glutamic acid decreased the affinity of the PDZ1-ligand interaction by 2-10-fold (8). The multiple interactions between the Arg Ϫ1 guanido group and PDZ1 residues Glu 43 and Asn 22 observed in our structure explain the remarkable preference for a penultimate arginine by NHERF PDZ1. Taken together, these observations indicate that although the peptidic residue Ϫ1 is not important for specificity, it may contribute to the affinity of the PDZ-ligand interaction. Consequently, PDZ domains have a preference for specific side chains at position Ϫ1 and interact optimally with peptide ligands having the corresponding penultimate residues. In support of this conclusion, it was found that the MAGI3 PDZ2 domain exclusively selected ligands with Trp Ϫ1 from random sequences, and it was predicted that tryptophan at this position interacts with Leu 40 for high affinity binding (35). Furthermore, the solution structure of the ␣-syntrophin PDZ domain showed that Leu Ϫ1 of the peptide ligand makes hydrophobic contacts with Phe 34 (23). Remarkably, both MAGI3 Leu 40 and ␣-syntrophin Phe 34 residues correspond to NHERF Glu 43 , suggesting that the amino acid at this position may play a critical role in the PDZ-ligand affinity through interaction with residue Ϫ1 of the peptide. Interestingly, the NHERF, NHERF2, and PDZK1/CAP70 PDZ domains that bind to the CFTR tail (8)(9)(10)(11)(12)(13) have either glutamate or aspartate at the position corresponding to NHERF Glu 43 (Fig. 1A), suggesting that Arg Ϫ1 of the CFTR tail may form similar salt bridges with these residues.
Perspective-The present work reveals the specificity and affinity determinants of the NHERF PDZ1-CFTR interaction and provides insights into carboxyl-terminal leucine recognition by class I PDZ domains, particularly those of NHERF, NHERF2, and PDZK1/CAP70. The sequence similarity shared among the aforementioned PDZ domains (Fig. 1A) suggests similar modes of interactions with CFTR. Elucidation of the molecular mechanisms underlying the interaction between these proteins and CFTR may facilitate the design of potent and specific modulators of CFTR activity with important clinical applications in the treatment of secretory diarrhea and cystic fibrosis.