Cross-linking of a Growth Hormone Releasing Factor-binding Protein in Anterior Pituitary Cells*

Growth hormone-releasing factor (GRF) stimulates the release of growth hormone from the anterior pituitary and is related to the peptides of the glucagon/ secretin family. Although the mechanism of action of this hormone has been studied in considerable detail, little is known concerning the GRF receptor itself. We have attempted to label the GRF receptor by chemically coupling the '"I-GRF analog [His', NleZ7]-hGRF(1-32)-NH2 (GRFa) (where Nle is norleucine) to plated rat anterior pituitary cells with the protein cross-linker disuccinimidyl suberate (DSS) (0.1 mM). Verification of biological activity of the "'1-GRFa was confirmed prior to the cross-linking experiments using the re- verse hemolytic plaque assay. Whole cell extracts prepared from the cross-linked cells were subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis followed by autoradiography of the dried gels. Four bands of 72,50,30, and 26 kDa were detected in autoradiograms from cells exposed to the labeled analog for 20 min (22 "C) followed by exposure to DSS for 2 min. The 72-kDa band was interpreted to be bovine serum albumin, which was used as a carrier in initial studies. The 50- and 30-kDa bands were very faint and probably represent nonspecific binding sites since they were unchanged in the presence of excess unla- beled GRFa.

Growth hormone-releasing factor (GRF) stimulates the release of growth hormone from the anterior pituitary and is related to the peptides of the glucagon/ secretin family. Although the mechanism of action of this hormone has been studied in considerable detail, little is known concerning the GRF receptor itself. We have attempted to label the GRF receptor by chemically coupling the '"I-GRF analog [His', NleZ7]-hGRF(1-32)-NH2 (GRFa) (where Nle is norleucine) to plated rat anterior pituitary cells with the protein cross-linker disuccinimidyl suberate (DSS) (0.1 mM). Verification of biological activity of the "'1-GRFa was confirmed prior to the cross-linking experiments using the reverse hemolytic plaque assay. Whole cell extracts prepared from the cross-linked cells were subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis followed by autoradiography of the dried gels. Four bands of 72,50,30, and 26 kDa were detected in autoradiograms from cells exposed to the labeled analog for 20 min (22 "C) followed by exposure to DSS for 2 min. The 72-kDa band was interpreted to be bovine serum albumin, which was used as a carrier in initial studies. The 50-and 30-kDa bands were very faint and probably represent nonspecific binding sites since they were unchanged in the presence of excess unlabeled GRFa. The 26-kDa band was diminished in a concentration-dependent manner by unlabeled rat GRF, GRFa, and to a lesser extent by vasoactive intestinal peptide (VIP). It is unlikely, however, that GRFa was acting at a VIP receptor since the labeled analog did not induce prolactin secretion (VIP is a prolactin secretagogue). GRFa also increased cellular CAMP to levels similar to GRF and greater than VIP. Autoradiographs from gels run under nonreducing conditions revealed the 26-kDa band as the major species, indicating that, if a polymeric form of this binding protein exists, it does not involve disulfide linkages. Thus, the best candidate for the putative GRF receptor is the 26-kDa band. We have further demonstrated that the higher concentrations of DSS used previously (5 mM) result in diffuse autoradiograms with multiple bands, *This work was supported by Grant AM 32632 and Research Career Development Award lKL4NS00601 from the National Institutes of Health. 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.
suggesting that caution should be exercised when interpreting cross-linking data under these conditions. The hypothalamic neuropeptide GRF' is a member of the glucagon/secretin/vIP family and is the primary stimulus for GH release from the somatotroph of the anterior pituitary (AP) (1)(2)(3). The human-derived peptide exists primarily in a 44-and 40-residue form and both possess similar biological activity (3)(4)(5). Mechanistically, GRF induces GH secretion in a CAMP-and calcium-dependent manner (1,6,7) and the regulation of GRF receptor-adenylate cyclase coupling is thought to involve GTP-binding regulatory proteins (8,9). However, the chemical nature of the GRF receptor remains unclear. Initial characerization of GRF binding to the cell surface has shown that binding sites exist in bovine (10) and rat (11) AP cells with Kd values of 3 nM and 41 PM, respectively. Furthermore, GRF binding to rat AP cells is influenced by glucocorticoids (11) and can undergo down regulation (12).
Other studies indicate that GRF can act at VIP receptors in intestinal epithelium (13,14) and pancreas (15), while VIP is known to be a weak GRF agonist in the AP (16). The concentrations of each hormone required for these cross-receptor interactions, however, exceed physiologic levels and are undoubtedly related to the homology of these peptides (14).
Recently, the molecular mass of the putative GRF receptor was reported to be 75 kDa based on chemical cross-linking of '251-hGRF(1-40)-hydroxide (hGRF) to rat AP cells (17). In the present study, we have cross-linked a biologically active, radioiodinated GRF analog (GRFa) to cultured rat AP cells.
The results of our work indicate a lower molecular weight binding site and suggest that the 75-kDa protein is BSA. This discrepancy and the difficulties involved in interpreting crosslinking data are considered.

EXPERIMENTAL PROCEDURES
Preparation of Iodinated Peptides-GRFa, hGRF (kindly provided by Drs. J. Rivier and W. Vale, The Salk Institute, La Jolla, CA) and VIP (Peninsula Laboratories) were radioiodinated by the chloramine-T method (18) and purified by reverse phase HPLC on an Ultrasphere ODS column. Labeled compounds were diluted in phosphate-buffered saline and used in cross-linking experiments at a concentration of approximately 1 nM (100 pCi/pg).
Preparation of Cultured Anterior Pituitary Cells-Cells were obtained from freshly dissected AP of male Sprague-Dawley rats (180-220 g, body weight, Hilltop Lab Animals, Inc.) and dispersed according to our established procedure (1,8,19). Dispersed cells were seeded into 24-well plates (Primaria, Falcon) and maintained with RPMI 1640 medium supplemented with 7.5% horse serum, 2.5% fetal bovine serum (all from GIBCO), and antibiotics.
GRFa-Cell Cross-linking-Three days after plating, cells were washed 3 times (1 h each) with bicarbonate-and serum-free RPMI 1640 medium containing 25 mM Hepes (Sigma) and antibiotics. After the third wash, radioiodinated peptide with or without unlabeled hormone was added to each well. Binding was allowed to occur for 20 min, at which time the cross-linker DSS (Pierce Chemical Co.) was added. The cross-linking reaction was run for 2-20 min at 22 "C and was stopped by the addition of 40 mM glycine. After removing the ' The abbreviations used are: GRF, growth hormone releasing factor; rGRF and hGRF, rat and human growth hormone releasing factor, respectively. GRFa, [His', Nle2']-GRF(1-32)-NHz; Nle, norleucine; BSA, bovine serum albumin; AP, anterior pituitary; DSS, disuccinimidyl suberate; Hepes, N-2-hydroxyethylpiperazine-N'-2ethanesulfonic acid; HPLC, high pressure liquid chromatography.

A GRF Binding Protein in Anterior Pituitary
medium from each well, the cells were solubilized in Nonidet P-40 (Sigma) and Laemmli sample buffer as described by Wood and O'Dorisio (20). Electrophoresis and Autoradiography-Solubilized cell extracts were applied to 10% Laemmli gels (21) and run at 20 mA/gel. Gels were then fix-stained and dried, followed by autoradiography for 4-7 days with Kodak XAR-5 film in the presence of a Du Pont Cronex intensifier screen. Developed autoradiograms were analyzed on a Quickscan Jr. densitometer (Helena Laboratories) at 525 nm.
Reverse Hemolytic Plaque Assay-Plated cells were prepared for the hemolytic plaque assay essentially as described by Anderson et al. (22) using the method of Frawley and Neill (23) with modifications (24). Briefly, normal rat AP cells were dispersed with trypsin (0.1%, Cooper Biomedical) and adjusted to a concentration of 300,000 cells/ ml. These cells were mixed with 1 ml of protein-A-coupled ox erythrocytes (Colorado Serum) in Dulbeco's minimal essential medium (GIBCO) containing 0.1% BSA and 20 mM HEPES and loaded directly into a Cunningham chamber. After a 1-h attachment period, the chamber was washed with RPMI 1640 containing 10% heatinactivated horse serum and antibiotics and then the cells were incubated for 20 h at 37 "C (95% 02, 5% C02). Labeled or unlabeled hormone was then added with a 1:150 dilution of GH antibody h at 37 "C. The cells were then incubated with Dulbecco's minimal essential medium containing 0.1% BSA and guinea pig complement (1:50, Colorado Serum) for 15 min at 37 "C. Cells were fixed in 1% glutaraldehyde for 5 min at 4 "C, rinsed, stained, and air-dried. The circumference of a plaque was defined as the first line formed by a complete string of intact erythrocytes, taken back to a common background density (22). Fifty or more randomly chosen plaque areas were determined for each slide using a Zeiss Videoplan system (22). Cellular Cyclic AMP and Protein Levels-Cyclic AMP was extracted from the plated AP cells with 0.1 N HC1 and measured by radioimmunoassay (25). Proteins were measured by the method of Lowry et al. (26) using BSA (Sigma) as a standard.

RESULTS AND DISCUSSION
Chemical cross-linking has been used with considerable success to identify several peptide hormone receptors including insulin (27), glucagon (28,29), VIP (20,30), and prolactin (31). Conditions for cross-linking vary according to the system under study, the nature of the cross-linker, and the specificity and biological activity of the probe.
In this study, we have used GRFa as a probe for the identification of GRF-binding sites. This radioiodinated analog has been reported to bind GRF-specific sites and induce GH secretion from the AP (11). Using the reverse hemolytic plaque assay, we confirmed the biological activity of both unlabeled and radioiodinated GRFa, and determined that this analog did not induce prolactin secretion (Table I). Th' 1s was an important control because GRF has been reported to interact with VIP receptors which can stimulate prolactin release (13-16). Accordingly, as a positive control, we validated that VIP stimulates prolactin release in the hemolytic plaque assay (data not shown). In contrast, radioiodinated hGRF (at a comparable specific activity to GRFa) was virtually inactive as a GH secretogogue (Table I). GRFa also stimulated cellular cyclic AMP levels to an extent similar to that for hGRF and to a much higher degree than comparable concentrations of VIP (Table 11). These data indicate that Y -G R F a is a more suitable probe than "'1-hGRF for the study of the GRF receptor.
When AP cells were incubated with Y -G R F a followed by cross-linking with low (0.1 mM) concentrations of DSS for 2 min, four different molecular weight species were apparent in autoradiograms from sodium dodecyl sulfate-polyacrylamide gels (Fig. 1, lane 4 ) . A 75-kDa band was present only if BSA was included in the cell medium (Fig. 1, lanes 4-6). Indeed, incubation of 1251-GRFa with BSA in the absence of AP cells also yielded an identical 72-to 75-kDa band. Two other bands Hormone secretion in anterior pituitary cells measured by the reverse hemolytic plaque assay AP cells were prepared for the reverse hemolytic plaque assay as described under "Experimental Procedures." Cells were incubated for 1 h in the presence of hormone and antibody followed by a 15-min treatment with guinea pig complement at 37 "C. Plaque area analysis was performed as previously described (22,24). Unless otherwise indicated, each plaque area represents the mean f S.E. of 4-5 slides (n) with at least 50 plaques measured per slide. PRL, prolactin.

Condition
Hormone

I1
Cellular cyclic AMP levels in cultured A P cells exposed to GRFa, hGRF, and VIP AP cells were cultured as described under "Experimental Procedures." Experiments were conducted under conditions identical to those for cross-linking. After incubation of peptides with cells for 20 min, the medium was rapidly removed from each well and cells were acid-extracted for CAMP which was measured by radioimmunoassay (25). Cellular protein was solubilized with 0.2 N NaOH and determined by the method of Lowry et al. (26). Each value represents the mean of six wells f S.E. for a representative experimental series.

Condition
Cyclic AMP (50 and 30 kDa) were also detected in the presence of cells but were very faint. These bands probably represent nonspecific binding sites because they were preserved even when cells were co-incubated with excess, unlabeled hormone ( Fig.   2 A ) .
The fourth band (26 kDa) was the most prominent and decreased in intensity when excess (1 WM) unlabeled GRFa or human GRF was co-incubated with the A P cells (Fig. 1, lanes  1-6). The 26-kDa species also decreased in a concentrationdependent manner in response to unlabeled rat GRF or (to a significantly lesser extent) VIP (Fig. 2, A and B). It is not surprising that higher concentrations of VIP decreased the intensity of the GRFa band since VIP and rat GRF share considerable homology (12 of 32 amino acids) (32). Although VIP stimulates prolactin secretion from the AP, VIP can also act as a weak GRF agonist at concentrations greater than 10 nM (16). This correlates well with the results in Fig. 2B in which a significant decrease in the intensity of the 26-kDa band is only apparent if unlabeled VIP is co-incubated with Iz5I-GRFa at concentrations above 10 nM. Thus, in addition to the data from the reverse hemolytic plaque assay for prolactin, these results indicate that it is unlikely that the 1 2 predominant GRFa binding is to VIP receptors. Somatostatin, which does not act at the GRF receptor (33,34), had no effect on the 26-kDa band (Fig. 2, A and B ) .

A GRF Binding Protein in Anterior Pituitary
The low molecular weight of the GRF-binding protein prompted us to ask if the 26-kDa species is a proteolytic fragment of a larger subunit and if a larger disulfide-linked binding complex exists. No such higher molecular weight bands were detected in autoradiograms when cell extracts were prepared in the presence of protease inhibitors (1 mM phenylmethylsulfonyl fluoride, 25 pg of leupeptin/ml) or if gels were run in the absence of sulfhydryl reducing agents, &mercaptoethanol or dithiothreitol (data not shown).
If a polymeric form of the GRF receptor exists, it is probably through noncovalent interactions.
The difference between the mass of the GRF-binding protein in our study (26 kDa) and the higher mass species (75 kDa) recently reported (17) led us to expose AP cells to the high concentrations of DSS and the long cross-linking time described in this earlier report. Under those conditions (5 mM DSS, 20-min cross-linker incubation), we obQined autoradiograms with a diffuse appearance (Fig. 3) SRIF, somatostatin. appearance of these patterns to the presence of glycoproteins, we think that excess cross-linker and longer incubation are responsible because much more distinct autoradiograms are evident a t 0.1 mM DSS for 2 min (compare Figs. 1 and 2A to Fig. 3). Furthermore, with 5 mM DSS we resolved multiple bands both for 12sI-GRFa and 12sI-VIP (Fig. 3). As with GRFa, 1251-VIP also formed a 72-to 75-kDa complex only in the presence of BSA (Fig. 3, lane 7), confirming the earlier work of Wood and O'Dorisio (20). The multiple bands obtained a t these higher concentrations of DSS are most likely due to nonspecific cross-linking of cellular proteins to the labeled hormone, although it is possible that smaller GRF-binding subunits may have cross-linked to form higher molecular weight complexes. In this regard it is important to note that,  in time course studies for cross-linking at 0.1 mM DSS, we observed an increase in the intensity of the 26-kDa GRFa band up to 30 min; no higher mass species was evident (data not shown). We have also observed that excess unlabeled GRFa can diminish the 75-kDa band under conditions of high cross-linker concentrations (Fig. 3, lane 4 ) and occasionally at 0.1 mM DSS as well (Fig. 1, lane 6 ) . Because albumin is known to bind several circulating hormones, this interaction with GRF is not surprising.
Another possible explanation for the discrepancy in GRFbinding protein size is the use of different GRF probes. The data from the hemolytic plaque assay indicate that '*'I-hGRF (which exposes the 75-kDa binding site (17)) has very little biological activity at a specific activity comparable to '*'I-GRFa (Table I). This may be due to the iodination in the human GRF peptide of Tyr-1 and the possible oxidation of Met-27 which could have compromised the binding and activity of this hormone (11). In contrast, the biological activity of "'I-GRFa was comparable to that of the unlabeled analog, confirming and extending the results of Seifert et al. (11). Finally, it is always possible that the differences in these results are due to conditions which we have not recognized. However, acceptable current criteria for the identification of a specific GRF-binding site in AP cells have been fulfilled. These include the graded decrease in the labeling of the 26-kDa band with unlabeled peptides, and the appropriate rank order of potency (rGRF > GRFa > VIP, no effect by soma-tostatin-14) and biological activity. These results also illustrate the importance of proper conditions and controls for cross-linking, as well as the caution needed to interpret crosslinking data.
In conclusion, we provide evidence for a specific GRFbinding protein in AP cells. The bioactivity and specificity of the GRF analog and the conditions with which we have conducted the cross-linking experiments strongly suggest that this protein is the putative GRF receptor or a subunit thereof. Although the smaller size of this receptor is intriguing, it is not unique since the prolactin receptor is reported to have a mass of only 30 kDa (31). The isolation of this binding protein and subsequent reconstitution into an adenylate cyclase system will ultimately establish whether or not the 26-kDa species is capable itself of activating this second messenger system, thus fulfilling a requirement for a hormone receptor.