Rational redesign of neutral endopeptidase binding to merlin and moesin proteins

Merlin and NEP do not coassociate

Our previous studies have shown that ERM proteins coimmunoprecipitate with NEP in prostate cancer cell lines9 and NEP coassociates with tumor-suppressor proteins such as PTEN.11 We considered whether NEP might also interact with the tumor-suppressor ERM-like protein merlin. For these experiments, we used (a) NEP-expressing LNCaP prostate cancer cells; (b) WT-5 cells, a derivative of Tsu-Pr1 bladder cancer cells, in which NEP expression is regulated by a tetracycline-repressible system (Tet-off)12; and (c) DU-145 prostate cancer cells that lack NEP expression. As shown in Figure 1, merlin protein was expressed in all three cell types. Western blotting of NEP following immunoprecipitation using anti-merlin antibody of cell lysates derived from WT-5 cells cultured without tetracycline (NEP expressed) or LNCaP cells indicated that NEP and merlin proteins do not associate with each other [Fig. 2(A), upper panel]. Similar results were obtained with Western blotting of merlin following immunoprecipitation using anti-NEP antibody [Fig. 2(A), lower panel]. Studies in mammalian cells have shown that phosphorylation of S518 in merlin is essential for intramolecular interactions between its C-terminal tail and the FERM domain. Hypophosphorylated merlin is localized to the membrane–cytoskeleton interface and interacts with many membrane-associated ERM partners via its FERM domain. It also interacts either directly or indirectly with the actin cytoskeleton.13-15 We therefore assessed whether phosphorylation of S518 influences merlin's interaction with NEP. We cotransfected NIH3T3 cells with vectors expressing NEP, and either the nonphosphorylatable mutant of merlin (S518A) or its phosphomimicking mutant (S518D). Western blotting of NEP immunoprecipitates showed that NEP and merlin proteins do not associate, regardless of the phosphorylation state of S518 [Fig. 2(B), lower panel]. Similar results showing no association between merlin and NEP were detected in merlin- and NEP-expressing NIH3T3 cells immunoprecipitated with an antibody to Flag and Western blotted with an antibody to NEP [Fig. 2(B), upper panel]. Finally, as we had previously shown that NEP associates with ERM proteins using an in-vitro binding assay,11 we performed a similar experiment using the N-terminal domain of merlin. Again, we could not demonstrate any association between merlin and NEP [Fig. 2(C)]. Together, these data show that in contrast to ERM proteins, merlin does not associate with NEP.

Prediction of moesin region involved in NEP binding

We previously demonstrated that the interaction between NEP and ERM proteins occurs via the positively charged amino acids within the cytoplasmic domain of NEP9: NEP and ERM proteins associated in cell lysates expressing wild-type NEP, but not in lysates from cells expressing cytoplasmic-domain-truncated or cytoplasmic-domain-mutated NEP containing amino-acid substitutions K¹⁹K²⁰K²¹→Q¹⁹N²⁰I²¹. To determine why ERM proteins bound NEP whereas merlin did not, we used computational analysis to predict which region in moesin is responsible for binding NEP. We used computational threading against another complex of moesin with a long peptide (its own C-terminal tail)16 and against a complex with ICAM-2 peptide17 to find the most feasible conformation for the NEP cytoplasmic domain interaction with this protein. The resulting predicted model of the protein–peptide complex suggested an association between NEP K21 and moesin E¹⁵⁹Q¹⁶⁰ (Supporting Information Fig. 2). The E¹⁵⁹Q¹⁶⁰H¹⁶¹K¹⁶²L¹⁶³ region is conserved in ERM proteins, but has a different sequence in merlin (see Fig. 3). Therefore, this region might be responsible for determining binding (as in ERM) or lack of binding (as in merlin) to NEP. We tested this hypothesis by mutagenesis of this region in both moesin and merlin proteins.

E¹⁵⁹Q¹⁶⁰ residues in moesin are required for binding of NEP in vitro

To test our hypothesis that moesin residues E¹⁵⁹Q¹⁶⁰ constitute the determinants for NEP binding, we carried out a binding assay using in-vitro-translated wild-type moesin protein or the identical sequence mutated to the residues found in merlin (N¹⁵⁹L¹⁶⁰), and the GST-fusion cytoplasmic domain of NEP. We had previously demonstrated that the GST-fusion cytoplasmic domain of NEP binds in-vitro-translated wild-type ERM proteins.11 Glutathione-S-transferase GST alone, used as a control (Supporting Information Fig. 1, lanes 1 and 2), possessed negligible binding affinity for the recombinant N-terminal wild-type or mutant moesin. The N-terminal wild-type moesin bound to the GST-fusion protein of NEP cytoplasmic domain (GST-NEP1-19), whereas the moesin double-mutant (E159NQ160L) demonstrated weak binding, similar to the GST control (Supporting Information Fig. 1). These results suggested that residues E¹⁵⁹Q¹⁶⁰ in moesin are essential for the binding of NEP and ERM proteins in vitro. To confirm the importance of these residues for the interaction, we constructed an isopropyl-β-D-thiogalactopyranoside (IPTG)-inducible GST fusion with a merlin mutant in which residues N¹⁷⁵L¹⁷⁶ were replaced with EQ (merlin 2a.a mut) and residues N¹⁷⁵L¹⁷⁶Y¹⁷⁷Q¹⁷⁸M¹⁷⁹ were replaced with EQHKL, the corresponding sequence in the ERM proteins [merlin 5a.a mut; Figs. 3 and 4(A)]. We then performed a binding assay using wild-type and mutant merlin GST fusion proteins and lysates from WT-5 cells grown in the presence (NEP not expressed) or absence (NEP expressed) of tetracycline. Wild-type merlin and the merlin 2a.a mut (not shown) exhibited no binding affinity for NEP protein. However, as shown in Figure 4(B,C), immunoprecipitation using NEP antibody and western blotting for merlin showed that NEP and the merlin 5a.a mut had coassociated. Importantly, the NEP protein containing the QNI mutation in the cytoplasmic domain did not bind to either the wild-type or mutant merlin proteins [Fig. 4(C)]. Our results suggest that the EQHKL motif in moesin and merlin is necessary and sufficient for in-vitro interaction with NEP and that the binding involves the basic cluster of NEP in the cytoplasmic domain.

Cells and Antibodies

LNCaP, DU145, and NIH3T3 were maintained in RPMI 1640 medium supplemented with 2 mM glutamine, 1% nonessential amino acids, 100 units/mL streptomycin, and penicillin containing 10% fetal calf serum. WT-5 cells or QN-12 cells, containing full-length (starting from the 8th amino acid in the sequence NP_009220) NEP cDNA or mutated NEP cDNA in which the positively charged amino acids in the cytoplasmic domain, K¹⁹K²⁰K²¹, were replaced with Q¹⁹N²⁰I²¹, subcloned into the tetracycline-repressible, transactivator-protein-responsive plasmid tel-regulated expression (pTRE/NEP), which were derived and maintained as described previously.13 The cells were cultured in medium with 1 μg/mL tetracycline to repress expression of NEP. Antibodies used included anti-actin, anti-merlin, anti-phospho-NF2 and anti-Flag antibodies (Santa Cruz Biotechnology, Inc.), mouse anti-NEP mAb NCL (Novocastra Laboratories, Ltd.), and mouse anti-NEP mAb J5 (Beckman Coulter, Inc.).

Immunoprecipitation, SDS-PAGE, and immunoblotting

Confluent monolayer cultures of cells on 10-cm dishes were washed twice with cold phosphate buffered saline (PBS) and then lysed in 1 mL of radioimmune precipitation assay buffer (10 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1% Nonidet P-40, 5 mM ethylenediaminetetraacetic acid (EDTA), 1% sodium deoxycholate, 0.1% sodium dodecyl (lauryl) sulfate (SDS), 1.2% aprotinin, 5 μM leupeptin, 4 μM antipain, 1 mM phenylmethylsulfonyl fluoride, and 0.1 mM Na₃VO₄) for 30 min. After centrifugation at 12,000g for 30 min, the supernatant was collected, and protein concentration was determined using BCA protein assay reagent (Pierce). Lysate (500 μg) was incubated 1 h to overnight with 2 μg of primary antibody and 30 μL of protein G-Sepharose beads (Amersham Biosciences) at 4°C. Immune complexes were washed three times with radioimmune precipitation assay buffer and then diluted in 2× sample buffer (125 mM Tris-HCl, pH 6.8, 4% SDS, 10% glycerol, 0.2% bromphenol blue, 4% 2-mercaptoethanol). The samples were separated by 10% SDS-PAGE, transferred to a nitrocellulose membrane (OPTITRAN; Schleicher and Schuell), and immunoblotted as described previously.32

Site-directed mutagenesis and vector constructs

Expression vectors for the Flag mutants of merlin (S518A and S518D) were kindly provided by Dr V. Ramesh (Massachusetts General Hospital, Boston, MA). Expression vectors for N-terminal and full-length moesin were kindly provided by Dr. Heinz Furthmayr (Stanford University, Stanford, CA).33 Moesin and merlin cDNA with mutations were constructed by site-directed mutagenesis (QuikChange; Stratagene). To substitute amino acids E¹⁵⁹Q¹⁶⁰ in the N terminus of moesin with N¹⁵⁹L¹⁶⁰, PCR-based site-directed mutagenesis was performed using a set of primers including point mutations (underlined) as follows: 5′-CG CAG AGA GTC CTG AAT CTG CAC AAA CTC AAC AA-3′ and 5′-TT GTT GAG TTT GTG CAG ATT CAG GAC TCT CTG CG-3′. To substitute amino acids N¹⁷⁵L¹⁷⁶ in merlin with E and Q, respectively, PCR-based site-directed mutagenesis was performed using: 5′-CTT CCA AAA AGG GTA ATA GAG CAG TAT CAG ATG ACT CCG GAA-3′ and 5′-TTC CGG AGT CAT CTG ATA CTG CTC TAT TAC CCT TTT TGG AAG-3′. To substitute amino acids N¹⁷⁵L¹⁷⁶Y¹⁷⁷Q¹⁷⁸N¹⁷⁹ in the N-terminus of merlin with E¹⁷⁵Q¹⁷⁶H¹⁷⁷K¹⁷⁸L¹⁷⁹, PCR-based site-directed mutagenesis was performed using: 5′-AA TTG CTT CCA AAA AGG GTA ATA GAG CAG CAC AAG CTG ACT CCG GAA ATG TGG GAG GAG-3′ and 5′-CTC CTC CCA CAT TTC CGG AGT CAG CTT GTG CTG CTC TAT TAC CCT TTT TGG AAG CAA TT-3′. PCR products were confirmed by DNA sequencing.

Production and purification of GST fusion of NEP cytoplasmic domain

cDNA fragments containing the cytoplasmic domain starting from Met⁸ (amino acids 8–26) were subcloned into pESP-1 plasmid (Stratagene) to produce GST fusion proteins (pESP-1/NEP1-19), as described previously.9Schizosaccharomyces pombe yeast cells SP-Q01 were transformed with pESP-1 or pESP-1/NEP1-19 using lithium acetate.34 Synthesis of the GST fusion proteins was induced by incubating transformed yeast cells in essential minimal medium (EMM) broth (ESP Yeast Protein Expression and Purification System; Stratagene) for 20 h, and then sedimenting them by centrifugation at 1000g for 5 min and resuspending in PBST-PI (PBS containing 1% Triton X-100, 1 mM phenylmethylsulfonyl fluoride, 1 μg/mL aprotinin, 1 μM pepstatin A, 100 μM leupeptin, 1 μg/mL chymostatin). Yeast cells were lysed by vortexing for 10 min at 4°C with glass beads in PBST-PI,35 and the cell debris was removed by centrifugation (12,000g, 5 min, at 4°C). The supernatant was then mixed with G-Sepharose 4B (Amersham) and incubated for 30 min at room temperature, and the pellet was collected by centrifugation (500g, 5 min), washed three times with PBS, and resuspended in GST elution buffer (10 mM reduced glutathione, 50 mM Tris-HCl, pH 8.0). This mixture was incubated at room temperature for 10 min, and the supernatant was collected as a GST fusion protein solution by centrifugation (500g, 5 min). The concentration of GST fusion protein was estimated by SDS-PAGE.

Production and purification of GST fusion of merlin

NEB Express I^q Competent Escherichia coli cells were transformed with pGEX-2T, pGEX-2T/merlin wild-type, pGEX-2T/merlin 2a.a mut, or pGEX-2T/merlin 5a.a mut. Synthesis of the GST fusion proteins was induced by incubating transformed cells in 0.1 mM IPTG for 4 h at 37°C. Then the cells were sedimented by centrifugation at 7700g for 10 min and the cell pellet was completely suspended by adding 50 μL ice-cold 1× PBS per milliliter culture. The suspended cells were disrupted by sonication on ice in short bursts (Triton, Promega Corp.) as described previously.9 The amount of labeled moesin protein used in the binding (20%) was added to a final concentration of 1% and mixed gently for 30 min to aid in solubilizing the fusion protein. Then the cells were sedimented by centrifugation at 12,000g for 10 min at 4°C and the supernatant was transferred to a fresh container. The concentration of GST fusion protein was estimated by BCA protein assay kit (Pierce).

In-vitro binding assay

In-vitro-translated, [³⁵S]methionine-labeled N-terminal wild-type or mutant moesin proteins were generated using TNT Quick Coupled Transcription/Translation Systems (Promega) as previously described.9 The amount of labeled moesin protein used in the binding assays was determined by normalizing the intensity of the autoradiography signal for each labeled protein. In-vitro translation product was incubated with 1–2 μg of GST fusion proteins immobilized on G-Sepharose beads and washed with binding buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM EDTA, 0.1% Nonidet P-40) at 4°C for 1 h with rocking. Complexes were pelleted at 10,000g for 2 min, washed three times in 0.5 mL of binding buffer, and subjected to SDS-PAGE. The gels were fixed, dried, and subjected to autoradiography. To assay NEP-merlin binding, 200 μg of WT-5 cell lysate with (+) or without (−) tetracycline or QN-12 was incubated for 2 h with 2 μg of anti-NEP mAb J5 and 30 μL of protein G-Sepharose beads at 4°C. Immune complexes were washed three times with cold PBS. After 25 μg of GST-merlin fusion protein was mixed and incubated for 2 h at 4°C, the complexes were washed three times with cold PBS and subjected to SDS-PAGE.

Computational analysis