Purification of the neurotensin receptor from bovine brain.

The neurotensin receptor protein, solubilized with digitonin/asolectin from bovine cerebral cortex membranes, was purified to apparent homogeneity by affinity chromatography using immobilized neurotensin. The product exhibits saturable and specific binding of [3,11-tyrosyl-3,5-3H]neurotensin with an apparent affinity (Kd = 5.5 nM) comparable to that measured in intact membranes and crude soluble extracts. The affinity-purified material, after reduction with 100 mM dithiothreitol, in denaturing gel electrophoresis showed a single polypeptide of Mr 72,000. Under nonreducing conditions the apparent Mr, however, was 50,000, suggesting the presence of intramolecular disulfide bonds. The purified neurotensin receptor was judged to be homogeneous, in that (i) only a single polypeptide was detectable; and (ii) the overall purification was 30,000-50,000-fold, giving a specific neurotensin-binding activity close to the theoretical maximum.

The neurotensin receptor protein, solubilized with digitonin/asolectin from bovine cerebral cortex membranes, was purified to apparent homogeneity by affinity chromatography using immobilized neurotensin. The product exhibits saturable and specific binding of [3,1 l-tyro~yl-3,5-~H]neurotensin with an apparent affinity (K,, = 5.5 nM) comparable to that measured in intact membranes and crude soluble extracts. The affinity-purified material, after reduction with 100 m M dithiothreitol, in denaturing gel electrophoresis showed a single polypeptide of M, 72,000. Under nonreducing conditions the apparent M,, however, was 50,000, suggesting the presence of intramolecular disulfide bonds. The purified neurotensin receptor was judged to be homogenous, in that (i) only a single polypeptide was detectable; and (ii) the overall purification was 30,000-50,000-fold, giving a specific neurotensin-binding activity close to the theoretical maximum.
Neurotensin, a tridecapeptide, was first isolated from bovine hypothalamus (1) and is found in the brain (2), gastrointestinal tract (2), and spinal cord (3). Neurotensin is a putative neurotransmitter/neuromodulator in the central nervous system, and in the spinal cord it is potentially important in pain pathways. Radiolabeled neurotensin or various neurotensin analogs have been used to identify specific neurotensin-binding sites in brain homogenates of a number of species including rat (4-7), guinea pig (8), mouse (9), and human (10,11). These have been attributed to neurotensin receptors existing in the neuronal membrane. However, these have not been characterized in the isolated state, and even crude soluble extracts which retain this activity have not hitherto been described. Studies at the membrane level, using photoaffinity labeling or cross-linking with neurotensin derivatives (12), have suggested that the neurotensin receptor contains two proteins with apparent molecular weights of 49,000 and 51,000. This has led to the proposal (12) that the neurotensin receptor in rat exists as a noncovalently linked heterodimer.
We have employed conditions in which highly active neurotensin binding sites are solubilized in high yield, using the nonionic detergent digitonin. The full purification of this * 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  receptor by affinity chromatography and the characterization of its binding parameters are described.

MATERIALS AND METHODS
[3H]Neurotensin (40-70 Ci/mmol) was from Du Pont-New England Nuclear and neurotensin was from Cambridge Research Biochemicals. Other materials, as well as methods not specified, were as given previously (13,14). Digitonin (Analar) was from BDH; less soluble impurities were removed by dissolving at 100 "C, cooling at room temperature, and filtering.
Membrane Preparation and Solubilization-Using bovine brain cortex (14), membranes were prepared in the presence of protease inhibitors as previously described for rat brain (13). Frozen membranes were thawed and gently agitated (20 "C for 60 min) at 0.5-1.0 mg of protein/ml in buffer A 10 mM TES'.KOH (pH 7.5), 1 mM EGTA.K+, 2 mM MgSO,, 1 mM benzamidine HCI, 0.02% (w/v) bacitracin, and 0.002% (w/v) soybean trypsin inhibitor. The membranes were then pelleted at 45,000 X g at 4 "C for 30 min and resuspended in buffer A; digitonin and asolectin were added to give a final concentration of 2% (w/v) and 0.06%, respectively. After centrifugation at 120,000 X g at 4 "C for 60 min, the supernatant was stored at -20 "C for use.
Affinity Chromatography-Affi-Gel 10 (2 ml), prewashed with 3 volumes of ice-cold water, was gently agitated (3 h, 4 "C) in 2 ml of 20 mM HEPES (pH 7.4) containing 12 pmol of neurotensin and a trace amount of [3,11-tyr0syl-3,5-~H]neurotensin, followed by a 30min incubation at 20 "C. Unbound ligand was washed away and the remaining active sites on the gel were blocked by reaction with 1 M ethanolamine HC1, pH 8.0 (1 h, 20 "C). The gel was subsequently washed with 10 volumes of 20 mM HEPES (pH 7.4), followed by 5 volumes of 10 mM TES. KOH (pH 7.51, and stored at 4 "C in the latter buffer containing 0.02% sodium azide. The amount of covalently bound neurotensin was estimated, from the trace label incorporated, to be 4-5 pmollml packed gel. The yield of coupling was 30% and did not increase with a longer reaction time. This neurotensin/Affi-Gel 10 was pre-equilibrated with 5 volumes of buffer A containing 10 p~ 1,lO-phenanthroline, 0.1% (w/v) digitonin, and 0.003% (w/v) asolectin. Crude solubilized extract (10 ml; 15-20 mg of protein) was adjusted to be in the same medium and loaded at 20-30 ml/h onto a 2-ml column of the gel. This was then washed with 40 volumes (40-50 ml/h) of the equilibration buffer from which the soybean trypsin inhibitor and bacitracin were omitted and with the asolectin concentration increased to 0.06%. The bound receptor was eluted with 2.5 volumes of the latter buffer containing also 250 mM NaCI, was dialyzed against 5 liters of 10 mM TES. KOH (pH 7.5), 1 mM EGTA.K+, 2 mM MgSO,, 1 mM benzamidine.HC1 and the fractions were assayed immediately for [3,11-tyro~yl-3,5-~H] neurotensin binding. After each preparation the column was washed with 5 volumes of 1% Na+ cholate, 0.5 M NaCI, 0.1 M Tris.HC1 (pH 8.5) and could then be reused.
SDS-PAGE-Protein samples for slab SDS-PAGE (14) were precipitated with chloroform/methanol (16)   The binding parameters of the bovine brain neurotensin receptor The parameters were determined from Scatchard plots of the equilibrium specific binding data (as in Fig. 1). Assay conditions for membranes were 10 mM TES.KOH (pH 7.5), 2 mM M$', plus the protease inhibitors at 20 "C for 1 h. For crude soluble or purified receptor, conditions were as for membranes but with 0.06% asolectin and 0.1% digitonin also present. Values are the mean f S.E. of three independent experiments, except that, for the purified samples, due to the variation in the estimated values of the protein content (see "Materials and Methods"), the B , values are the highest and lowest obtained.

RESULTS AND DISCUSSION
Purification of the Neurotensin Receptor by Affinity Chromatography-Membranes were incubated with 2 mM M$+ (13) and then solubilized in a medium containing the detergent digitonin and the phospholipid mixture asolectin. In these conditions, the high-affinity binding of [3,11-tyrosyl-3,5-3H]neurotensin was maintained in solution (Table I,

samples 1 and 2).
For the affinity gel used, neurotensin was immobilized on Affi-Gel 10 beads. Since the N terminus of neurotensin is blocked by a pyroglutamyl residue, the N-hydroxysuccinimide reactive group on the spacer arm of Affi-Gel 10 can be expected to react only via the amide chain of Am5 of neurotensin. This would leave free the 8-13 segment of neurotensin, which is essential for receptor binding (11).
This neurotensin-affinity gel bound 40-70% (range of several experiments) of the receptor activity present in the crude extract applied. The specific neurotensin-binding activity remained tightly bound, even after washing the column extensively. As a result of this tight binding, elution with M neurotensin alone resulted in a low recovery (8%) of bound activity, since most of the receptor was found in the Na' cholate wash of the column after each run. Moreover, the prolonged dialysis required for the removal of the eluting neurotensin resulted in losses of receptor activity. Since the SDS-PAGE protein profiles of the neurotensin-eluted and Na+ cholate-eluted protein were found to be the same, complete recovery of the tightly bound receptor was accomplished instead by elution with 250 mM NaCI. We have observed that NaCl strongly decreases (with ICs0 = 35 mM) the affinity of the bovine neurotensin receptor.' In addition to being able to dialyze the salt out quickly, the transition from high to low salt concentrations in the presence of asolectin and detergent during dialysis may stabilize the receptor activity by increasing the phospholipid to detergent ratio in the micelles.
Properties of the Purified Neurotensin Receptor-After the removal of NaCl by dialysis, the binding of [3,11-tyrosyl-3,5-3H]neurotensin to the purified receptor was specific and appeared to be saturable within the same concentration range *Mills, A., Demoliou-Mason, C. D., and Barnard, E. A. (1988) J. Neurochem, in press. of [3,11-tyro~yl-3,5-~H]neurotensin seen for the crude soluble receptor (Fig. 1); nonspecific binding was reasonably low and even for the highest [3,1l-tyro~yl-3,5-~H]neuroten~in concentration used it was less than 30% of the total binding. Scatchard plots (Fig. 1, inset) identified a single set of sites (in the range up to 30 nM ligand concentration) with an apparent equilibrium dissociation constant, KO, similar to that obtained in crude soluble receptor extracts (Table I). The value obtained was only 1.6-fold higher than that obtained with intact membranes, so that under the conditions used for solubilization and purification the receptor activity is well preserved.
On the basis of the estimated protein content of the purified receptor (see "Materials and Methods"), the -fold purification achieved in all the experiments was in the range of 18,000-36,000 (Tables I and 11), when compared with the neurotensinbinding activity present in intact bovine cortex membranes. Comparison with the activity present in crude soluble extracts gave, in fact, a higher -fold purification, 30,000-51,000 (Tables  I and 11 Table I. Inset, Scatchard plot. All points denote the averages of triplicate determinations. "The activity was determined at 12 nM [3,11-tyro~yl-3,5-~H]neurotensin, in the assay conditions noted in Table I. Not corrected for the higher KD value in solution (see Table I), and therefore an underestimate. e Values are for a single preparation (except where noted). For the range of binding activity found in all preparations, see Table I.
The maximum purification found in the series of preparations, and using the B,, values, was 36,000-fold (see Table I), relative to the membrane specific activity. ues show that sufficient specificity for full purification is introduced at the stage of the binding to the immobilized neurotensin, together with the affinity change of this receptor on raising the salt concentration.
SDS-PAGE of the purified receptor protein, when denatured in the presence of 100 mM dithiothreitol, showed a single protein band of M , 72,000 after either Coomassie Blue staining (not shown) or silver staining (Fig. 2, lane I). Under nonreducing conditions (Fig. 2, lane 2), a single band of M , 50,000 was instead visible, suggesting the presence of intramolecular disulfide bonds. Under these conditions a larger amount of aggregated protein was observed on the top of the gel, suggesting that there is some aggregation of the receptor protein during precipitation for SDS-PAGE. Increases in apparent M , after vigorous dithiothreitol reduction have been found (and interpreted similarly) for several other homopolymeric receptors including the &-adrenergic receptor (18), the interleukin-3 receptor (19), a gonadotropin receptor (20), and the p opioid receptor (21,22).
In a control experiment, in which an excess of free neurotensin was added to the sample prior to the stage of binding to the neurotensin/Affi-Gel 10 column (with all other steps being as before), no protein bands were detected in the final analysis by SDS-PAGE even after overstaining the gel (Fig.   2, lane 3); this preblocking effect confirmed that the M, 72,000 polypeptide (with M , 50,000, unreduced) is specific to the neurotensin receptor. On the basis of the apparent size ( M , 72,000) of the single protein identified by SDS-PAGE, the theoretical specific binding activity (assuming one binding site per receptor molecule) would be 13.9 pmol/g protein; this is reasonably close to the B,, (7.4-12.7 pmol/g protein) 1 2 The protein eluted from the affinity column was analyzed on a 10% polyacrylamide gel, with silver staining. The receptor, =350 ng, was first denatured with SDS a t 100 "C either with 100 mM dithiothreitol (lunes 1 and 3) or without dithiothreitol ( l a n e 2). Lune 3, an equivalent volume of a sample from a control preparation, in which crude extract was preincubated with 10" M neurotensin (1 h, 20 "C) prior to its application to the affinity gel, was similarly analyzed; a11 stages were as for the unblocked preparation of lane I , but the gel was overstained for the detection of any minor bands. The migration positions (kDa) of standard proteins (phosphorylase b, bovine serum albumin, ovalbumin, and soybean trypsin inhibitor) in a parallel track are shown on the left. Coomassie Blue staining of similar gels also showed only the band seen here. obtained from the saturation binding measurements on the purified receptor. Taking into account the difficulty in estimating the low amounts of protein present, these results suggest that the neurotensin receptor has been purified to homogeneity.

66-
Apparent molecular weights, as measured under similar conditions, in the region of 70,000-80,000 daltons have been reported for other single-chain receptor species associated with G-proteins, e.g. the &adrenergic receptors (M, 65,000-70,000) (18), the muscarinic receptor (Mr 80,000) of a number of species (23), and the rat opioid p receptor (M, 66,000-72,000) (21,22,24). The sizes of the two protein subunits of the neurotensin receptor (M, 49,000 and 51,000), previously identified in rat brain membranes by photolabeling and crosslinking studies (12), are in disagreement with the single, larger size observed here. This may be the result of proteolytic degradation or deglycosylation.
In reality, where the true polypeptide M, is known .from cDNA cloning, the true M , values in each case of the above mentioned type tend to be in the range of 50,000-55,000. The rest of the masses stated are due either to attached oligosaccharides or to systematic error in the estimation of the M , of membrane proteins by SDS-PAGE. This knowledge should become available soon for the neurotensin receptor, since the isolation of its subunit permits us to determine some internal protein sequence therein and hence to start its cDNA cloning.