Purification and amino-terminal sequencing of the high affinity phenylalkylamine Ca2+ antagonist binding protein from guinea pig liver endoplasmic reticulum.

A high affinity phenylalkylamine Ca2+ antagonist binding polypeptide (Moebius, F. F., Burrows, G. G., Striessnig, J., and Glossmann, H. (1993) Mol. Pharmacol. 43, 139-148) was purified to homogeneity from the endoplasmic reticulum of guinea pig liver with the aid of [3H]emopamil, an antiischemic agent, and [3H]azidopamil, a photoaffinity label. The purified protein retained its high affinity for the antiischemic drugs emopamil (Kd = 4 nM), opipramol (IC50 = 15 nM), trifluoperazine (IC50 = 2 nM), and for Zn2+ (IC50 = 2 microM). Ferguson plots revealed a molecular mass of 27.2 kDa. Partial amino acid sequence information was obtained by Edman degradation and revealed no homology to known protein sequences. Antibodies raised against a synthetic peptide corresponding to the first 25 NH2-terminal amino acid residues specifically immunoprecipitated the [3H]azidopamil photoaffinity-labeled polypeptide and recognized the protein in Western blots. Cross-linking with a variety of homo- and heterobifunctional agents lead to the formation of dimers. Since in the purified preparation no other subunit could be identified with different protein stains, our results indicate that the [3H]emopamil binding site is formed by the homodimer of a novel membrane protein.

Organic Ca2+ channel blockers (like verapamil, nifedipine, and diltiazem) are widely used clinically to treat cardiovascular disorders like angina, hypertension, and certain arrhythmias. Their pharmacological effects are exerted by blocking depolarization-induced Ca2+ entry into smooth and cardiac muscle through L-type voltage-gated Ca2+ channels after high affinity interaction with binding domains on the a1 subunit (1).
In addition to their cardiovascular effects, tissue protective properties have been reported for many Ca2+ antagonists. In various animal models Ca2+ antagonists were found to protect from ischemic (2) as well as toxic cell injury (3). Such drugs may therefore be useful to minimize ischemic damage during stroke and myocardial infarction as well as for the prevention of damage after organ transplantation. An extensive review of the literature is given in Ref.

4.
One of the most extensively studied tissue-protective drugs F. F. M.), the Dr. Legerlotz Foundation (to J. S.), and Fonds zur Forde-* This work was supported by a Boehringer-Ingelheim fellowship (to rung der wissenschaftlichen Forschung, Austria, Grants S6601 (to H. G.) and P9351 (to J. S.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must U.S.C. Section 1734 solely to indicate this fact. therefore be hereby marked "advertisement" in accordance with 18 $ Both authors contributed equally to this work.
5 To whom correspondence should be addressed.
is the phenylalkylamine (PAA)l emopamil that is structurally closely related to the Ca2+ antagonist verapamil. Although it is a weaker Ca2+ channel blocker than verapamil it exerts more potent antiischemic effects in animal models of global and cerebral ischemia (see Ref. 5, for references; Refs. 6-10). We have recently identified with radiolabeled [3H]emopamil possible molecular targets of this drug that could mediate its antiischemic effects. Indeed we characterized a novel high affinity (Kd = 10 nM), sodium-and zinc-sensitive binding site for [3H]emopamil in guinea pig liver, brain, kidney, adrenal gland, and lung (5, 11). Further biochemical studies and photoaffinity labeling with the PAA [3H]azidopamil revealed that the new binding site is located on a polypeptide with an apparent molecular mass of 22.5 kDa (5). This emopamil-binding polypeptide (EBP) is preferentially localized in the endoplasmic reticulum as shown by subcellular fractionation of guinea pig liver membranes (5). More recently, we demonstrated that EBP shares many properties of so-called u-binding sites (SBSs) (12). SBSs have been discussed as mediators for a variety of pharmacological actions of so-called u ligands (for review, see Ref. 13) including antiischemic effects (14). Despite considerable efforts their functional role has, however, not yet been determined. Thus, similar to other high affinity drug-binding polypeptides with still undefined function (i.e. imidazoline-binding sites (151, peripheral benzodiazepine receptors (16)), the assessment of the physiological role of EBP and u sites may be greatly facilitated by analysis of their molecular structure.
Here we report the purification, partial amino acid sequencing, and analysis of the oligomeric structure of the EBP. Our experiments reveal that this protein is structurally unrelated to the PAA-binding a1 subunit of L-type Ca2+ channels or any other known protein. We also rule out that this protein represents any known drug-metabolizing enzyme including cytochrome P450, as has been suggested for SBSs (17, 18).
The purification of EBP will permit more direct studies on this class of drug-binding proteins and eventually help to reveal their role in mediating responses elicited by antiischemic CaZ+ antagonists and u ligands. Dr. Traber (Tropon, Cologne, Germany). Other chemicals were obtained from the following sources: opipramol, Ciba-Geigy (Vienna, Austria); hydroxylapatite, Bradford protein reagent, electrophoresis reagents, and molecular weight markers, Bio-Rad; DEAE-, SP-and chelating Sepharose, Pharmacia Biotech Inc.; all other chemicals, Sigma (Deisenhofen, Germany).
Binding A~says- [~H]Emopamil binding experiments with membrane bound and solubilized EBP were carried out as described previously (5). Briefly, 0.8-2.3 n M [3H]emopamil were incubated with protein in 0.1% (w/v) digitonin, 10 rm Tris-HC1, pH 7.4 (37 "C), 0.1 m~ phenylmethylsulfonyl fluoride in a final volume of 0.25 or 0.5 ml at 22 "C for 2 h. Nonspecific binding defined by 1 p~ emopamil was subtracted from total binding to yield specific binding. For inhibition experiments serial drug dilutions were made in dimethyl sulfoxide (Me,SO) and added directly to the incubation mixture. The final Me,SO concentration was 1% (v/v), which did not affect radioligand binding. Separation of bound from free ligand by filtration over GFIC glass fiber filters was carried out as described (5). For saturation studies with [3Hlemopamil, its specific activity was decreased by the addition of increasing concentrations of unlabeled ligand. Protein concentrations were determined according to Bradford (191, using bovine serum albumin as a standard. EBP Purification-All purification procedures were carried out at 5 "C. Guinea pig liver microsomes were prepared as described ( 5 ) with the following modifications. The supernatant obtained after centrifugation at 8,000 x g was centrifuged for l h at 100,000 x g. The pellet was resuspended in 0.5 M KCl, 0.15 M Tris-HC1, pH 8.0 (4 "C), and centrifuged as described above. After resuspension in 10 IILM Tris-HC1, pH 9.1 (4 "C), and recentrifugation the pellet was solubilized in 40 ml of 1% (w/v) digitonin in buffer A (5% (v/v) glycerol, 40 m M Tris-HC1, pH 9.1 (4 "C), 5 rm NaCl, 2 m M dithiothreitol, 1 m M EDTA, 0.1 m M phenylmethylsulfonyl fluoride) at a protein concentration of 5-6 mg/ml. The mixture was centrifuged as above and the supernatant diluted in buffer A giving a final digitonin concentration of 0.5% (w/v).
Solubilized microsomes were applied (100 mlh) to a DEAE-Sepharose column (2.5 x 11.5-cm packed resin) equilibrated with 0.5% (w/v) digitonin in buffer A. Bound protein was step eluted with buffer A containing 0.1% (w/v) digitonin and 170 or 500 m M NaCl, respectively. Activity was only recovered in the flow-through. The flow-through from the DEAE-Sepharose was diluted with 3 M &HP04, pH 6.45, to a final concentration of 0.5 M and applied (80 mlh) to a hydroxylapatite column (2.5 x 1.5 cm) equilibrated with 0.1% (w/v) digitonin, 0.5 M &HPO,, pH 6.45, in buffer A. Bound binding activity was step-eluted with 30-ml portions of 0.1% (w/v) digitonin containing 0.7, 0.9, 1.2, and 1.5 M &-HPO,, pH 6.45. The activity containing fractions (0.9-1.5 M &€€,PO,) were pooled and dialyzed overnight (molecular weight cutoff of 6,000-8,000) against 20 m~ &H;PO,, pH 6.05. The dialysate was applied (150 ml/h) to a SP-Sepharose column (1.5 x 3.5 cm) equilibrated in buffer B (0.1% (w/v) digitonin, 20 m M KJI,,PO,, pH 6.05). Binding activity eluted in the flow-through was collected and bound contaminating protein was eluted with 1 M NaCl in buffer B. The flow-through from the SP-Sepharose column was adjusted to a final concentration of 0.065 M NaCl and passed through (150 mlh) green-agarose (1.5 x 2.5 cm) equilibrated with 0.065 M NaCl in buffer B. The activity containing flow-through was collected and bound contaminating protein eluted with 1 M NaCl in buffer B.
The green-agarose column flow-through was adjusted to 0.5 M NaCl and 1 m M imidazol and applied (120 ml/h) to chelating Sepharose (0.7 x 4 cm, loaded with CuCl, as described by the manufacturer) equilibrated with buffer B containing 0.5 M NaCl and 1 rm imidazol. Adsorbed binding activity was eluted with buffer B containing 15 m M imidazol and 15% (v/v) glycerol.
Photoafinity Labeling and SDS-PAGE-[3HlAzidopail was incubated in the dark with purified EBP in 0.1% (w/v) digitonin, 10 m M Tris-HC1, pH 7.4 (37 "C), 0.1 rm phenylmethylsulfonyl fluoride in the absence or presence of other drugs for 1 h at 22 "C. Samples were then irradiated for 55 s with an ultraviolet lamp (Sylvania GTE germicide) at 10 cm distance. Photolyzed protein was dialyzed against 0.05% (w/v) SDS, lyophilized, resuspended in sample buffer containing 10 m M Nethylmaleimide (nonreducing conditions) (51, and separated on SDSpolyacrylamide gels as described (5). For fluorography, Coomassie Bluestained gels were equilibrated in AmplifyR (Amemham), dried, and exposed to Kodak X-Omat AR5 films for the indicated times (at -80 "C). Ferguson plot analysis of purified EBP was carried out as described in the legend to Fig. 2.
Partial Amino Acid Sequencing of EBP-The Cu2+-chelating Sepharose eluate was dialyzed (molecular weight cutoff of 12,000-14,000) for 48 h against 0.05% (w/v) SDS. The lyophilized sample was resuspended in sample buffer containing 10 rm dithiothreitol (reducing conditions) and separated by preparative SDS-PAGE. After negative staining of the gel with CuCl, (201, the 22-kDa band was cut out, the gel pieces placed in a Bio-Rad electroelution apparatus in 0.01% (w/v) SDS, 25 m M Tris, 192 m M glycine, and electroeluted for 15 h at 100 V. The electroeluted protein formed an insoluble precipitate in the electroeluate (0.6 ml). The precipitate was collected by centrifugation at 12,500 x g and washed twice by resuspension and centrifugation with distilled water (1.0 ml).
After lyophilization the protein was dissolved in hexafluoroacetone trihydrate and insoluble material was removed by centrifugation. 10 p1 of the supernatant were loaded on Porton protein support and subjected to automated Edman degradation in a Porton Instruments PI2090E microsequencer (see also Table 111). To obtain internal sequence information, 350 pmol of protein were digested with TPCK-trypsin (Promega, sequencing-grade, final concentration 0.4 mg/ml) in 0.5% (w/v) CHAPS, 100 m M Tris-HC1, pH 8.7, at 37 "C. Tryptic fragments were isolated from the digest by reversed-phase high performance liquid chromatography on a Vydac C, column (2.1 x 150 mm, 35 "C) eluted with a linear gradient from 2 (v/v) to 50% (v/v) acetonitrile over 70 min at a flow of 0.1 mllmin. The individual peaks were collected directly onto Porton peptide support which were dried and subjected to automated Edman degradation.
Zmmunological Techniques-A polyclonal antiserum was raised in New Zealand White rabbits against a synthetic peptide corresponding to the first 25 NH,-terminal amino acid residues of EBP with an additional NH,-terminal lysine residue (peptide EBP,_,,). Peptide synthesis, coupling to bovine serum albumin, immunization, enzyme-linked immunosorbent assay, and affinity purification of antisera were performed as described (21).

TABLE I
Purification of EBP from guinea pig liver microsomal membranes The binding protein was purified as described under "Experimental Procedures." Data from three independent purifications are given (means + S.D.). [3HlEmopamil binding activity was measured in duplicate at two protein concentrations at a single ligand concentration (2.0-2.3 ma).
Nonspecific binding was measured in the presence of 1 p~ (2)-emopamil. Protein concentrations were determined as described by Bradford (19) using BSA as a standard. Immunostaining of Western blots using aftinity purified anti-EBP,_,, was carried out as described (12).
Sucrose Density Gradient Centrifugation and Cross-linking Experiments-Sucrose-gradient centrifugation was carried out according to published protocols (5) and is described in legend to Fig. 3. For crosslinking experiments, 0.4 M cross-linker stock solutions were prepared in Me,SO and added to purified EBP and liver microsomes solubilized in 1% (w/v) digitonin, 50 mM NaC1, 20 mM &HYPO,, pH 7.5, up to a final concentration of 0.02 M. After incubation for 1 h a t room temperature the reaction was quenched by adding 1 M Tris-HCI, pH 7.4 (37 "C), to a final concentration of 0.05 M. Samples were irradiated with a Sylvania GTE germicide lamp a t a distance of 3 cm for 3 min, subjected to SDS-PAGE, blotted, and immunostained with affinity purified anti-Protein Detection Using Biotin-Streptauidin-0.2 pg of protein eluted from the Cu"-chelating Sepharose were incubated overnight a t 4 "C with 20 mM N-hydroxysuccinimidyl-biotin (Pierce). The sample was separated by SDS-PAGE and transferred to a polyvinylidene difluoride membrane (Immobilon P, Millipore). The membrane was blocked for 2 h a t 22 "C with 10% skim milk in 0.5% (wh) Triton X-100, 0.1% (w/v) Tween 20, 150 mM NaCI, 20 mM Tris-HCI, pH 7.4 (22 "C). After three EBP,-,S. washes streptavidin-conjugated horseradish peroxidase was added a t a 1:2000 dilution and incubated for 1 h in the above buffer without skim milk. After five washes with buffer, biotin-labeled protein was detected with ECL (Amersham) and DuPont Chronex films.

Purification of the EBP from Guinea Pig Liver Microsomes-
Guinea pig liver microsomes possess the highest density of EBP as previously shown by subcellular fractionation experiments (5). In guinea pig liver microsomes its apparent molecular mass as well as its highly characteristic reversible ["Hlemopamil binding properties (e.g. high affinity for Zn2+ and Na') are indistinguishable from its properties in various other tissues investigated ( 5 ) . Guinea pig liver microsomes were therefore selected as a starting material for purification. From the specific binding activity (about 30 pmoVmg of membrane protein (11)) we estimated that EBP represents less than 0.07% of the total membrane protein, requiring about 1500-fold purification (to a maximal specific activity of 45 nmoumg protein assuming a 1:1 stoichiometry of drug binding). 70-90% (range, n > 8) of the binding activity was extracted from the membranes after solubilization in 1% (w/v) digitonin. As binding activity time dependently became associated with insoluble protein fractions, low binding recoveries were obtained from most resins to which crude binding activity was adsorbed. This problem was minimized by passing the solubilized material through DEAE-Sepharose under conditions (pH = 9.1, 0.5% (wh) digitonin, 5 m M NaCl) where most of the contaminating protein adsorbed to the resin but about 40% of the binding activity were recovered in the flow-through. No further activity could be eluted from the column with NaCl up to 1 M, indicating that the low recovery was most likely due to protein aggregation. After the DEAE step no further tendency of the binding activity to aggregate was observed and thus facilitated further purification with reasonable recoveries (see Table I). Despite the lack of enrichment of specific binding activity DEAE-Sepharose chromatography allowed removal of several prominent contaminating proteins as revealed by silver staining (Fig. lA 1. The flow-through of the DEAE column was then adsorbed to hydroxylapatite and eluted with 0.9-1.5 M YH,PO,. In the eluate binding, activity was reversibly inhibited by &HYPO, but was recovered after removal of salt by overnight dialysis. Major contaminating proteins (42, 32, and 18 kDa) were then removed by passing the eluate through SP-Sepharose and green-agarose resins (Fig.  lA). The resulting material was enriched about 26-fold with respect to reversible ["H]emopamil binding activity as compared to the solubilized membranes (Table I) and contained a prominent 22-kDa polypeptide (Fig. lA). From reversible binding experiments it was known that the EBP binds divalent cations (including Cu" and Zn2+) with high affinity ( 5 ) . To remove the remaining contaminating protein we exploited this property by further chromatography of the partially purified material on a Cu2+-loaded chelating Sepharose. As expected EBP was more tightly bound to this resin than other proteins and could be eluted a t high purity (Fig. lA, lane 7 ) with 15 m M imidazol. The purified fraction contained 6% of the starting binding activity and less than 0.02% of the total protein indicating an at least 270-fold overall purification from the solubilized material. Obviously this represents a lower estimate because we did not attempt to correct the specific activity for irreversible losses of [3H]emopamil binding activity during the 72-h purification process.
To obtain a more precise estimate of the molecular mass of EBP by SDS-PAGE, the purified protein was subjected to Ferguson analysis as illustrated in Fig. 1B. From these experiments an apparent molecular mass of 27.2 kDa was calculated (Fig. lC), indicating an underestimation of its molecular mass when analyzed at a single polyacrylamide concentration. The migration in SDS-PAGE was not affected by the previous reduction of disulfide bonds (not shown).
To confirm the identity of the purified binding activity with the previously characterized microsomal [3H]emopamil binding site the pharmacological properties of the Cu2+-chelating Sepharose eluate were studied in more detail. Saturation analysis with [3H]emopamil revealed a dissociation constant of 4.4 f 1.5 n~ (n = 3) and a B,, of 8.5 3.2 nmoVmg of protein (n = 3). Binding activity was stable upon storage in 15% (v/v) glycerol a t 5 "C (half-life 4-6 weeks, range, n = 2) but rapidly deteriorated at 37 "C (half-life 6 * 1 min, n = 5).
The binding inhibition profile of reversible [3Hlemopamil interaction by a variety of drugs and cations was essentially unchanged after purification as compared to the membranebound state (Table 11). From all substances tested only the potency of NaCl and the antiischemic drug ifenprodil decreased upon solubilization (not shown) and remained low in the purified preparation (Table 11).
Properties of the Purified EBP-The purified polypeptide was specifically photoaffinity labeled with [3H]azidopamil (Fig. 2 A ) . Specific photoaffinity labeling was protected by unlabeled drugs and cations with the same potency as expected from the inhibition of reversible [3Hlemopamil binding (complete block by 0.15 p~ (+)-emopamil and 50 p~ ZnCl,, half-maximal block by 1 1.1~ (+verapamil and incomplete block by 150 m~ NaC1, Fig. 2 A ) . Together with the fact that the photolabeled band strictly comigrated with the silver-stained polypeptide in SDS gels these experiments indicate that the isolated polypeptide forms the Zn2+-sensitive high affinity PAA binding site.
We were unable to obtain evidence for other co-purifying polypeptides >5 kDa (the minimal molecular mass of a polypeptide resolved in our gel system) in gels stained with silver ( Fig. lA) or Coomassie Blue (not shown). In some gels an additional 40-kDa band was detected, which could be identified as a dimer of the purified protein by photoaffinity labeling (Fig. 2 A ) . Assuming that not all proteins are visualized by the above staining techniques, independent proof for the absence of other polypeptides in the purified preparation was obtained by either negative staining with Cu" (20) (n = 3, not shown) or using highly sensitive detection of biotinylated protein by Western blotting (n = 3, not shown). Irrespective of the analysis system used, the apparent purity of the purified preparation was >95%.
We therefore conclude that the EBP is sufficient to form the Zd'-sensitive [3H]emopamil binding subunit and is not associ-ated covalently or noncovalently with any other protein species.
Partial Amino Acid Sequencing of the Purified EBP-TO determine whether the EBP shares sequence homology with any other known protein we obtained partial amino acid sequence information. The purified protein was separated by preparative SDS-PAGE, electroeluted after negative staining with CuCl,, and directly subjected to gas-phase sequencing. A single NH,terminal amino acid sequence was obtained and 29 of the first 33 residues could be determined in three separate experiments (peptide A in Table 111). An alanine rather than a methionine residue was identified in the first cycle suggesting that 1 or more residues have been proteolytically removed post-translationally or upon protein purification. The amino acid sequence is unique and shows no significant homology to known protein sequences in available amino acid or nucleic acid data bases (Swissprot, EMBUGenBank). To obtain internal sequence information the purified protein was digested with TPCK-trypsin (see "Experimental Procedures"), proteolytic fragments were isolated by reverse-phase high performance liquid chromatography and subjected to gas-phase sequencing. Amino acid sequence was obtained from five individual peaks: two corresponded to the NH,-terminal amino acid sequence (not shown), two peaks gave other single sequences (peptides B and C, Table  III), and one peak contained two peptide sequences running at about the same level (peptide D E , Table 111). Like the NH, terminus, the internal amino acid sequence obtained for peptide B revealed also no homology to known protein sequences.
To unequivocally demonstrate that the sequenced protein corresponds to the EBP we raised polyclonal antibodies against a synthetic peptide corresponding to the first 25 NH,-terminal amino acid residues (EBP,,,). As shown in Fig. 2B, 25 pl of the anti-EBP,_,, antiserum specifically immunoprecipitated the [3Hlazidopamil-photolabeled EBP from solubilized membranes. Immunoprecipitation was specific because it was completely blocked in the presence of 1 p~ antigenic peptide and was not observed with preimmune serum. In Western blots affinity purified antibodies specifically recognized EBP in membranes (Fig. 3A, lane 1 ) as well as purified preparations (not shown). Taken together these results clearly demonstrate that the NH,terminal sequence information was derived from EBP.
Oligomeric Structure of EBP-Previous hydrodynamic studies of the EBP in crude digitonin extracts indicated that it must form larger heteroor homo-oligomeric complexes (5). As shown above we were unable to demonstrate the association of purified EBP with other proteins. However, EBP could be loosely associated with another subunit in the crude digitonin extract After photolysis (5) samples were dialyzed against 0.05% (w/v) SDS overnight a t 4 "C. Lyophilized samples were resuspended in sample buffer containing 10 mM N-ethylmaleimide and electrophoresed on a 13% (w/v) polyacrylamide SDS gel. Coomassie-stained gels were prepared for fluorography (exposure time 15 days) as described (5). One of three experiments giving almost identical results is shown. The arrow indicates the migration of EBP monomer and dimer, respectively. B, specific immunoprecipitation of the (-)-[:'Hlazidopamil photoaffnity labeled EBP with anti-EBP,_,,. Microsomal membranes were photolabeled with (-)-["Hlazidopamil (ligand concentration, 21 nhl; protein concentration, 0.5 mg/ml). Photolabeled microsomes were solubilized in 4% (w/v) SDS for 30 min. Solubilized protein was diluted 40-fold in RIA buffer and insoluble material was removed by centrifugation (15 min, 12,500 x g). 0.25 ml of solubilized membranes were then added to 15 pl of protein A-Sepharose coupled with 25 p1 of preimmune (lane 2 ) or anti-EBP,-,, immunoglobulins in the absence (lane 3 ) or presence (lane 4) of 1 p~ antigenic peptide. After incubation for 12 h a t 4 "C the protein A-Sepharose was washed five times with 1.25 ml of RIA buffer and then with 1.25 ml of TBS. 10 pl of sample buffer were added to a final concentration of496 (w/v) SDS, 10% glycerol, 62.5 mM Tris-HCI, pH 6.8 (22 "C), 10 mhf N-ethylmaleimide. 15 pg of photolabeled microsomes were separated in parallel (lane l ) on a 13% (w/v) polyacrylamide SDS gel. The fluorogram after 15 days of exposure is shown. that is disrupted by high ionic strength buffers or alkaline pH employed upon purification. This is unlikely because indistinguishable sedimentation coefficients were measured in sucrose density gradients for EBP in crude digitonin extracts ( s~, , ,~~, = 12.5 * 0.1 S, n = 3) and in purified fractions ( s~, , ,~ = 12.5 * 0.2 S, n = 3). In both cases (results not shown) the EBP immunoreactivity strictly comigrated with the [3Hlemopamil binding activity. Instead the relatively large s~, ,~ value of 12.5 S must be explained by the formation of larger homo-oligomeric complexes. To investigate this possibility we carried out cross-linking experiments employing various homo-and heterobifunctional cross-linking reagents. The formation of covalently crosslinked EBP oligomers in membranes (Fig. 3A) and purified preparations (results not shown) was then monitored by immu-TAIW 111

Amino acid sequences of purified EBP
Amino acid sequences are given using the one-letter abbreviation. Microsequencing was performed as described under "Experimental Procedures." Peptide A: from 330 pmol of EBP the amino-terminal acid sequence was obtained up to the 33rd residue. Alanine in positions 1 and 28 were read a t levels of 113 and 43 pmol, respectively. Separate runs confirmed the amino acid sequence from residues 1 to 11 ( n = 2) and 1-21 ( n = 1). Peptide D E : two sequences were obtained in parallel a t similar levels.  (Fig. 3A). The dimer possessed a molecular mass of 39 2 2 kDa ( n = 5) in SDS gels. Sucrose density gradient centrifugation of glutaraldehyde cross-linked microsomes revealed indistinguishable sedimentation coeficients for the dimer (12.4 2 0.2 S, n = 3) and the monomer ( s~, ,~, , = 12.5 f 0.1 S, n = 3, see above). The dimer was still able to bind ['Hlemopamil with high affinity as 50% of the binding activity were detectable despite complete dimerization (see Fig. 3B 1. Moreover the dimer could be specifically photoaffinity labeled with ["Hlazidopamil (not shown). This provides strong evidence that a homodimer represents the functional form of EBP. DISCUSSION We have recently characterized EBP as an intrinsic membrane protein in various tissues that binds several verapamillike Ca2+ antagonists with affinities comparable to the a, subunit of L-type Ca2+ channels (11). However, this binding site is not associated with L-type Ca2+ channels. It is located in the endoplasmic reticulum membrane (5) and clearly differs in its pharmacological binding profile from the Ca2+ channel. It has high affinity not only for the antiischemic PAA emopamil and the PAA photoaffinity ligand azidopamil but also binds a variety of structurally unrelated compounds, like amiodarone, opipramol, ifenprodil, trifluoperazine, and chlorpromazine with low nanomolar dissociation constants (Table I1 (5)). EBP could therefore mediate effects of these drugs that cannot be attributed to their already known pharmacological targets, like Ltype Ca2+ channels (for PAAs), neurotransmitter transporters (for opipramol), D,-receptors (for trifluoperazine and chlorpromazine), or glutamate receptors (for ifenprodil). Moreover tissue-protective effects have been reported for each of the above drugs (22)(23)(24)(25). Therefore the possibility that EBP could play a pathophysiological role in ischemic or toxic cell death cannot be ruled out.
We have recently shown that a,-binding sites also represent high affinity PAA-binding polypeptides (12). EBP cannot be classified a s a typical SBS due to its low affinity for a ligands like ditolylguanidine and pentazocine (Table 11). However, EBP shares many pharmacological and biochemical properties with SBSs: it possesses a similar molecular mass (about 27 kDa) and sedimentation coefficient in sucrose gradients ( 3 , is also localized in the endoplasmic reticulum membrane (5,26,27), and displays indistinguishable high affinity for a wide variety of drugs (including emopamil itself, a s well a s azidopamil, opipramol, ifenprodil, verapamil, and amiodarone). Therefore it seems justified to classify a receptors and EBP as members of a superfamily of small high affinity drug-binding proteins with ern blots with affinity purified anti-EBP,,, were performed as described under "Experimental Procedures." The arrow indicates the migration of EBP monomer and dimer, respectively. B , comparison of the sedimentation velocity of native and glutaraldehyde cross-linked digitonin-solubilized microsomes. Microsomes were solubilized as described above, cross-linked in the presence of 0.01 M glutaraldehyde a t 4 "C for 5 h, and dialyzed overnight a t 4 "C against 50 m M NaCI, 20 m M YH,PO,, pH 7.5. Microsomes incubated in the absence (W) or presence ( 0 ) of glutaraldehyde were separated on a 5-30% (w/v) sucrose density gradient in O.l%z (w/v) digitonin, 50 mM NaCI, 20 mM Tris-HCI, pH 7.4 (37 "C), 4% (v/v) glycerol a t 210,000 x g for 3.5 h (4 "C). 6-Galactosidase ( I , 15.9 S), catalase (2, 11.3 S), lactate dehydrogenase ( 3 , 7.3 SI, and cytochrome c ( 4 , 1.7 S ) were included as markers. Fractions were collected from the bottom of the gradients and analyzed for specific ['Hlemopamil binding activity and enzyme activity. C, immunoblotting of sucrose density gradient separated glutaraldehyde cross-linked microsomes. 0.012 ml of each sucrose density gradient fraction were subjected to Western blotting and immunostaining with affinity purified anti-unknown function. Pharmacologically highly related binding sites may well fit into this family (like the recently discovered high affinity opipramol-and ifenprodil-binding proteins (28,29)).
To facilitate the further characterization of this family of proteins we purified EBP to apparent homogeneity using five chromatographic steps. The high affinity of the protein for divalent cations resulted in tight binding to Cu'+-chelating Sepharose, which provided efficient purification. Despite the relatively low abundance of this polypeptide in the endoplasmic reticulum membrane, enough material could be obtained for NH,-terminal sequence analysis.
The binding affinities for a variety of drugs including (r ligands (Table 11) was unaffected by solubilization and purification of EBP. A significant decrease in binding afiinity was only found for ifenprodil (70-fold) and sodium (13-fold). This indicates that EBP forms more than one binding domain that participates in high affinity drug and cation interaction.
The absence of other proteins associated with purified fractions clearly demonstrates that EBP alone is able to form the functionally active Zn"-sensitive high affinity PAA binding site. Our cross-linking experiments suggest that EBP most likely exists as a functional homodimer although the formation of larger homo-oligomers cannot be completely excluded.
The lack of association with another protein as well as its unique NH,-terminal amino acid sequence (33 amino acid residues) rules out its identity with any other known protein. This also excludes that EBP is identical to the catalytic subunit of any known drug metabolizing microsomal enzyme as has been speculated for SBSs (17,18).
The antiserum-EBP,_,, raised against the NH,-terminal sequence specifically recognized the EBP in Western blots and immunoprecipitated the ['H1azidopamiLphotolabeled protein.
No other band was specifically immunostained in guinea pig liver microsomes. We have previously demonstrated the presence of the (r, binding site in our preparations, which was also photoaffinity labeled with azidopamil (12). The absence of cross-reactivity of anti-EBP,-,, with the azidopamil photolabeled 27-kDa band indicates that the NH,-terminal sequence of EBP is not conserved in SBSs. However, the possibility of other structural similarities with SBSs cannot yet be ruled out.
Further characterization of EBP will be necessary to clarify the function of this novel microsomal protein and its potential pathophysiological role during ischemia. The results presented here pave the way for the elucidation of the complete primary structure of EBP.