The plasmamembrane calmodulin–dependent calcium pump : a major regulator of nitric oxide synthase I

In initial attempts to gain insight into the unknown regulatory function of the plasma membrane calcium/calmodulin-dependent calcium ATPase (PMCA)^* (Carafoli, 1997; Shull and Greeb, 1988; Verma et al., 1988), we have shown previously that the PMCA is controlled by β-adrenergic stimulation (Neyses et al., 1985), that it regulated differentiation in a skeletal muscle cell line (Hammes et al., 1994, 1996), modulated hypertrophic stimuli in the myocardium of transgenic animals expressing PMCA under the control of the ventricle-specific myosin light chain-2 promoter, and was not involved in beat-to-beat regulation of heart contraction (Hammes et al., 1998). Furthermore, we and others have shown that PMCA 4 is localized to caveolae (Fujimoto, 1993; Hammes et al., 1998), membrane structures (sometimes also called “signalosomes” or microdomains) containing a wide variety of effectors of cellular signal transduction (Sargiacomo et al., 1993; Chang et al., 1994; Lisanti et al., 1994; Robbins et al., 1995; Schnitzer et al., 1995; Song et al., 1996a) including nitric oxide synthase (NOS)-I (Venema et al., 1997; Segal et al., 1999) and the cytoskeletal protein dystrophin (Song et al., 1996b). Because PMCA 4b has recently been shown to bind to members of the membrane-associated guanylate kinase protein family via their PDZ domains (Kim et al., 1998), we tested whether it interacts with NOS-I, which has been demonstrated recently to be part of a multiprotein complex largely organized by PDZ domains (through direct interaction or via adapter proteins; Brenman et al., 1995, 1996; Hillier et al., 1999) and whether this interaction was physiologically relevant.

Physical interaction of the PMCA and NOS-I was demonstrated by conventional immunoprecipitation (Fig. 1, D and E) as well as glutathione S-transferase (GST) pull-down assays, using the PDZ domain of NOS-I as a bait (Fig. 1 C). “Capping” the PMCA 4b COOH terminus (carrying the PDZ domain binding sequence) with an myc/6×His tag resulted in a loss of physical interaction (Fig. 1 D). Expression of wild-type NOS-I and deletion constructs (overview in Fig. 1 A) was confirmed by Western blotting (Fig. 1 B). This established that PMCA 4b and NOS-I interact physically, either directly or through unknown adapters. Full-length PMCA 4b interacted with purified GST–PDZ fusion protein, directly demonstrating specificity of interaction with the PDZ domain (Fig. 1 C). Similar interaction of PMCA and NOS-I was also observed in extracts of rat brain. Using PMCA-specific antibodies 5F10 and 2A, NOS-I was also coprecipitated from rat brain extracts (Fig. 1 F).

In a simple working model, a PMCA molecule interacting with NOS-I might reduce the local calcium concentration, leading to reduced activity of NOS-I, a calcium/calmodulin-dependent isoform of NOS. This hypothesis was tested using HEK293 cells transfected with expression constructs shown schematically in Fig. 1 A. Increasing PMCA 4b expression reduced NOS-I–dependent cGMP dramatically in a dose-dependent manner (Fig. 2 A, black columns). The model also predicts that it is the calcium transport function of the PMCA rather than some other yet unknown module of the molecule that regulates NOS-I activity: Asp⁶⁷²→Glu point mutation has been shown previously to render PMCA almost (but not totally) inactive with respect to its ATPase and Ca²⁺-transporting activity (reduction of calcium transport activity by ∼90%; Adamo et al., 1995). As predicted, this mutation resulted in a strong reduction of inhibition of NOS-I (Fig. 2 A, grey columns) compared with wild-type PMCA cotransfections (Fig. 2 A, black columns).

To test whether binding of PMCA 4b to the complex via PDZ domains was a prerequisite for its regulatory action, a constitutively active mutant of the pump (“ct120”) with a deletion of both the autoinhibitory and the COOH-terminal PDZ domain binding motif (Enyedi et al., 1993) was cotransfected with NOS-I. No regulation of NOS activity by this construct was observed (Fig. 2 A, last column). An NOS-I mutant carrying a deletion of the PDZ domain showed no regulation by PMCA 4b (Fig. 2 B). Endothelial NOS (eNOS or NOS-III) is also regulated by calcium/calmodulin, but carries no PDZ domain, and we have been unable to coprecipitate NOS-III with PMCA (not shown). In keeping with the absence of a physical interaction between PMCA 4b and NOS-III, the activity of this enzyme was not regulated by PMCA 4b (Fig. 2 C).

The potential physiological relevance of NOS-I regulation by PMCA 4b was tested in neuro-2a neuroblastoma cells, a commonly used model system in neuronal biology (Olmsted et al., 1970). Similarly to HEK293 cells, tight functional coupling of PMCA4b and NOS-I was observed: NOS-I activity was strongly downregulated by the PMCA4b (Fig. 3 A). This effect was reversed by the NO donor NOC-18 (2,2′-[hydroxynitrosohydrazino]bis-ethanamine), suggesting that interaction of these proteins not only occurs in HEK293 cells, but also in a neuroblastoma-derived cell line.

These results show that the plasma membrane calmodulin-dependent calcium pump 4b is an interaction partner and a major regulator of neuronal NOS-I and also that this regulation very likely is of physiological relevance. The strategic localization to caveolae (Fujimoto, 1993; Hammes et al., 1998) also suggests that local control of calcium and/or NO might have further regulatory effects in caveolae-mediated signal transduction.

The pivotal role of NOS-I in neuronal tissue is well established, exemplified by the observation that NOS-I deficiency leads to reduced susceptibility to cerebral ischemic damage (Huang et al., 1994). Its function in other excitable cells has been extensively characterized in recent years (Christopherson and Bredt, 1997). In contrast, the specific role of the PMCA, beyond the general concept of a calcium transporter, has been elusive. Our present results assign a function to isoform PMCA 4b, i.e., regulation of NOS-I activity. In a simple model (see Fig. 4), NOS-I, a Ca²⁺/calmodulin-dependent isoform of NOS, is regulated by the PMCA through adjustment of local calcium concentration. Both interact either directly or through an adapter protein via the PDZ domain of NOS-I and the COOH-terminal amino acids of PMCA 4b. Our results are in line with two previous reports showing the interaction of COOH-termini of different PMCA “b” splice variants (PMCA 2b and 4b) with PDZ domains and PDZ domain containing proteins of the MAGUK (membrane-associated guanylate kinase) protein family, whereas the splice variants 2a and 4a failed to bind to PDZ domains (Kim et al., 1998; DeMarco and Strehler, 2001). Indeed, it is conceivable that the predominantly neuronal isoform 2b regulates NOS-I in the context of (at least some specialized) neurones in vivo. This important issue will need to be addressed in further studies. These results and the reduced inhibition of NOS-I activity by the mutated PMCA (PMCAmut, Fig. 2, in which the Ca²⁺-transporting and ATPase activity are reduced) strongly suggest that it is the calcium transport function of the PMCA which affects regulation of NOS-I. However, it cannot be entirely excluded that the COOH-terminal peptide of the PMCA 4b, interacting with the PDZ domain of NOS-I, mediates inhibition directly; in this case one would have to postulate that the PMCAmut (Asp⁶⁷² mutation) undergoes a conformational change conducive to reduced interaction with NOS-I. A definitive answer to this and other unlikely possibilities may have to await direct measurements of calcium concentrations in the vicinity of the pump which is currently not possible.

The coupling of interaction and function of the two proteins is in close analogy to the observation that loss of the PDZ-binding COOH-termini of NMDA receptors results not only in loss of interaction, but also in complete loss of receptor function (Sprengel et al., 1998). Different PMCA isoforms and their splice variants (overview in Strehler and Zacharias, 2001) are expressed ubiquitously and NOS-I is a widespread isoform of NOS in neurons and in excitable tissues (Christopherson and Bredt, 1997). Therefore, the physical and functional interaction of PMCA isoforms and NOS-I or other PDZ domain–containing proteins is most likely relevant in a large variety of cells. These enzymes may define a modular signal transduction system which subserves different functions in different cellular contexts.

Samples were lysed in RIPA buffer (1× PBS containing 1% Igepal CA-630, 0.5% sodium deoxycholate, 0.1% SDS and Complete^TM protease inhibitor (Roche Diagnostics). Debris was removed by centrifugation (10,000 g, 5 min, 4°C). After a preclearing step with 0.5 μg irrelevant antibody (anti-FLAG epitope, Neomarkers) and 25 μl of protein G–Sepharose^TM (∼50% slurry; Amersham Pharmacia Biotech) 500 μg remaining protein lysate was coincubated together with 3 μg of precipitating antibodies (polyclonal rabbit anti–NOS-I, Affinity Bioreagents catalog no. PA3-032A; monoclonal mouse anti-hPMCA4b, clone JA3; and monoclonal mouse anti-PMCA, clone 5F10, Neomarkers, catalog no. MS-1305-P1 and MS-1304-P1; polyclonal rabbit anti-hPMCA4b was generated using a peptide spanning the 15 COOH-terminal amino acids of the PMCA4b; polyclonal rabbit anti-PMCA (“2A”) was a gift from D. Guerini, ETH Zürich, Switzerland) and 25 μl protein G–Sepharose^TM at 4°C for 2 h. Calmodulin-Sepharose beads (CaM beads; Amersham Pharmacia Biotech) served as control. After washing four times with RIPA buffer (see above) beads were resuspended in 100 μl sample buffer and boiled. 20 μl of each sample was loaded and 5 μg (1/100 input) protein extract served as control. NOS-I was detected using a monoclonal antibody (clone “16”; Transduction Laboratories, catalog no. N31020) in a dilution of 1:1,000, and transfected hPMCA4b was detected using the PMCA 4 isoform-specific antibodies JA3 (1:1,000; Figs. 1, C and E, 2 A, and 3 B) and 5F10 (dilution 1:1,000, Neomarkers), necessary to detect appropriate expression of PMCA(ct120). Immunodetection of proteins on membrane was done using the ECL^TM system according to the manufacturers protocol (Amersham Pharmacia Biotech).

GST pull-down assays were performed using purified GST-PDZ (NOS-I) fusion protein, containing the PDZ domain of NOS-I (Brenman et al., 1995, 1996) bound to glutathione-Sepharose^TM 4B beads (Amersham Pharmacia Biotech). Pelleted beads were washed four times with RIPA buffer, resuspended in electrophoresis sample buffer, boiled for 3 min, and proteins were separated by 10% SDS-PAGE and blotted to Hybond^TM ECL^TM nitrocellulose membrane (Amersham Pharmacia Biotech). Immunodetection was performed as described above.

Expression vector pCMV-hPMCA 4b has been described previously (pCMV-hPMCA4CI; Hammes et al., 1996). The point mutation Asp⁶⁷² →Glu was inserted with the QuikChange^TM site-directed mutagenesis kit (Stratagene) using the hPMCA 4b–modifying primers GTG GGC ATT GAG GAG CCT GTG CGC CCA GAG GTG and CAC CTC TGG GCG CAC AGG CTC CTC AAT GCC CAC. The pCMV-hPMCA4b cap vector was constructed by insertion of hPMCA 4b cDNA in BamHI-KpnI sites of pcDNA3.1(−)/Myc-His vector (Invitrogen) and subsequent mutation of the hPMCA 4b stop codon to obtain a c-myc/6×His tag at the PMCA COOH terminus. The constitutively active form of the PMCA 4b (phPMCA4b[ct120]; Adamo et al., 1995), was a gift of E. Strehler (Mayo Clinic, Rochester, NY). NOS-I expression plasmid (pCMV–NOS-I; Brenman et al., 1996) and pGST-PDZ (of NOS-I; Brenman et al., 1995) were a gift of David S. Bredt (University of California, San Francisco, CA). The NOS-III cDNA (Lamas et al., 1992; EMBL/GenBank/DDBJ accession no. M89952) was cloned in the EcoRI site of pcDNA3 (Invitrogen), similar to the NOS-I expression construct (see above). The ΔNOS-I vector, coding for NOS-I and lacking the PDZ domain, was constructed by excision of nucleotides 1–975 of the rat cDNA (numbering according to EMBL/GenBank/DDBJ accession no. X59949), resulting in a truncated NOS-I, which is expressed in transfected cells and exhibits NO synthase activity.

HEK293 and Neuro-2a cells (DSMZ) were grown in 90% DME/10% fetal bovine serum/1× nonessential amino acids (Sigma-Aldrich) to a density of 50–70% before transfection. HEK293 and neuro-2a cells were transfected using the LipofectaminPLUS^TM system according to the manufacturer's protocol (Life Technologies). After medium replacement cells were incubated overnight. cGMP was measured using cGMP (low pH) colorimetric competitive ELISA (R&D Systems). Cells were saturated with L-arginine (30 min, 1 mM) and, 5 min before lysis, nascent NO was stabilized by addition of superoxide dismutase (100 U/ml culture medium; Roche Diagnostics). In parallel, 3-isobutyl-1-methylxanthine (IBMX, 1 mM; Sigma-Aldrich) was added to the culture 5 min before lysis (to inactivate phosphodiesterase activity) and NO synthesis was induced by addition of 0.5 μM ionomycin 3 min before lysis. Subsequent determination of NO-dependent cGMP production was estimated in relation to untransfected control cells.

We thank D.S. Bredt, E. Strehler, J.T. Penniston, and A. Filoteo for kindly providing reagents, G. Ertl and K. Kochsiek for continuous support, and S. Lohmann and U. Walter for critically reading the manuscript.

This work was supported by the German Research Foundation (Deutsche Forschungsgemeinschaft; SFB 355, TP B6) and by an International Appointee Grant from the Medical Research Council (UK) to L. Neyes.