Glucagon Gene Expression in Vertebrate Brain*

of Immunocytochemistry-immunohistochemistry was performed on formalin-fixed, paraffin-embedded tissue using the avidin-biotin peroxidase complex technique (21). Aprimary antiserum directed against synthetic GLP-I (22) and a primary antiserum directed against glucagon (Dako, Santa Barbara, CA) were used at dilutions of 1:200. The duration of exposure to primary antiserum was 24 h at 4 "C. The reaction product was visualized by detection of peroxidase activity using a solution of 3,3'-diaminobenzidine tetrahydrochloride and hydrogen peroxide. Preabsorption of primary antiserum against GLP-I with synthetic GLP-I (1-37) (a kind gift of Dr. J. F. Habener, Massachusetts General Hospital, Boston) eliminated staining at con-centrations of 6 pglml.


script identical to that produced in pancreas and
intestine gives rise to proglucagon-related peptides in the brain.
The gene encoding preproglucagon is expressed in the A cells of the pancreatic islets and the neuroendocrine L cells of the intestine. Several studies have reported glucagon immunoreactivity in other tissues, including thymus, thyroid, and adrenal gland (1). Reports of glucagon biosynthesis in tissues other than pancreas and intestine, however, have not been widely accepted. For example, initial reports of a glucagon-like peptide with hyperglycemic activity in salivary gland extracts (2, 3) were subsequently discounted due to the presence of tracer degrading activity in the salivary gland extracts (4). A growing body of evidence suggests, however, that the brain may be a potential site of glucagon biosynthesis. Neurons in the retina, hypothalamus, and medulla oblongata have been identified which stain positive for glucagon and glucagon-like peptide I immunoreactivity (5)(6)(7)(8). Immunoreactive glucagon and glucagon-like peptide I of various molecular sizes have also been detected in different regions of canine and rat brain (9)(10)(11)(12). Functional evidence of a biological role * This work was supported by grants from the Medical Research Council of Canada and the National Cancer Institute. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisemnt" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequencefs) reported in this paper has been submitted to the GenBankTM/EMBL Data Bank with accession numberfs) 504040. for proglucagon-derived peptides in the nervous system stems from the observations that intracerebral injections of glucagon produces dosage-dependent hyperglycemia, and both glucagon and GLP-I' receptors have recently been identified in rat brain (12)(13)(14). Recent studies have demonstrated that the glucagon-like peptides activate adenylate cyclase in brain tissue and membrane preparations (14)(15), providing additional evidence for the importance of a brain-derived glucagon system.
Taken together, these data suggest that glucagon and the glucagon-like peptides may function as neuropeptides in selected regions of the nervous system. However, whether these peptides are actually synthesized in the brain or simply taken up from the circulation remains uncertain. To determine if the biosynthesis of glucagon and the glucagon-like peptides occurs in the central nervous system, we sought evidence for the expression of the glucagon or a glucagon-related gene in different regions of the brain. We find that the glucagon gene is expressed in neurons in both the hypothalamus and brainstem and that glucagon gene expression in the brain gives rise to an mRNA transcript that is identical in sequence to that found in pancreas and intestine.

EXPERIMENTAL PROCEDURES
Materials-Restriction enzymes, T4 DNA ligase, DNA polymerase, and alkaline phosphatase were from Pharmacia LKB Biotechnology Inc. [cP~'P]ATP (>800 Ci/mmol) was from ICN Radiochemicals. [a-%]ATP (600 Ci/mmol) was from Amersham Corp. Nitrocellulose membranes were from Schleicher and Schuell. All chemicals were from Sigma or Fisher. Sprague-Dawley rats were obtained from Charles River, Canada. The human neonatal brainstem cDNA library (16) was a kind gift from M. Jaye, Meloy Laboratories, Springfield, VA. The rat glucagon and somatostatin cDNA probes were a gift of Dr. J. Habener, Massachusetts General Hospital. The rat cholecystokinin cDNA was obtained from Dr. J. Dixon, Purdue University.
RNA Analysis-RNA was isolated from tissues as described previously (17). Polyadenylated RNA was prepared by two cycles of oligo(dT)-cellulose chromatography (18). RNA was size-fractionated through a 1.3% agarose-formaldehyde gel, ethidium-stained to assess the integrity and migration of the RNA, and transferred to a nylon membrane. The RNA was fixed on the membrane by UV irradiation, following which prehybridization was performed overnight in 1 X Denhardt's, 4 X SSC, 200 pg/ml salmon sperm DNA, 40% deionized formamide, 0.014 M Tris, pH 7.4. cDNA probes for rat glucagon, somatostatin, and cholecystokinin were labeled by the random priming technique (19) to a specific activity of 5 X 1 0 ' cpmlpg. Hybridization was performed in the same solution with 1 X IO6 cpm/ml of 32P-labeled cDNA probes for 24 h at 42 "C. Final washing conditions were 0.1 X SSC, 0.1% sodium dodecyl sulfate at 65 "C. 1 X SSC is 0.15 M NaCl, 0.3 M sodium citrate. Autoradiography was carried out using Kodak X-Omat film at -70 "C.
Isolation of cDNA Clones-A human neonatal brainstem Xgtll cDNA library was plated at a density of 5 X lo' plaques per plate, and duplicate filters were denatured, neutralized, and baked at 80 "C for 2 h. Filters were prehybridized and hybridized in 1 M NaCl, 1% sodium dodecyl sulfate at 65 "C. Positive clones were purified by replating at lower dilutions, and cDNA inserts were excised from phage DNA by cutting with the restriction enzyme EcoRI. The cDNA inserts were subcloned into the Bluescript plasmid (Stratagene) and sequenced as described previously (20). All ligations, transformations, and plasmid and phage preparations were carried out by standard techniques under Medical Research Council and National Cancer Institute of Canada guidelines.
The abbreviations used are: GLP-I, glucagon-like peptide I; bp, base pair.
Immunocytochemistry-immunohistochemistry was performed on formalin-fixed, paraffin-embedded tissue using the avidin-biotin peroxidase complex technique (21). Aprimary antiserum directed against synthetic GLP-I (22) and a primary antiserum directed against glucagon (Dako, Santa Barbara, CA) were used at dilutions of 1:200. The duration of exposure to primary antiserum was 24 h a t 4 "C. The reaction product was visualized by detection of peroxidase activity using a solution of 3,3'-diaminobenzidine tetrahydrochloride and hydrogen peroxide. Preabsorption of primary antiserum against GLP-I with synthetic GLP-I (1-37) (a kind gift of Dr. J. F. Habener, Massachusetts General Hospital, Boston) eliminated staining at concentrations of 6 pglml.

RESULTS AND DISCUSSION
To determine whether expression of the glucagon or a related gene could be detected in brain tissues, total cellular RNA was prepared from rat cortex, cerebellum, striatum, hippocampus, pons, pituitary, hypothalamus, and brainstem. Northern blot analysis of 30 pg of RNA from each region of the brain showed no hybridizable glucagon mRNA transcripts except for the brainstem, which contained a glucagon mRNA species of -1300 bp after a 7-day exposure. We next prepared poly(A+) RNA from the above tissues and repeated the Northern blot analysis (Fig. 1). Glucagon mRNA transcripts were easily detectable in 5 pg of poly(A+) RNA from adult brainstem. A considerably weaker signal was seen in fetal brainstem, despite loading double the amount of poly(A+) RNA in this lane. No glucagon mRNA transcripts could be convincingly detected in RNA prepared from adult or fetal hypothalamus, due to considerable background in the expected region of the glucagon mRNA transcript. The glucagon mRNA transcript in rat brainstem (-1300 bp) was just slightly larger  To verify the integrity of the hypothalamic and brainstem RNAs and to compare the relative abundance of neuropeptide mRNA transcripts in different regions of the brain, the Northern blot shown in Fig. 1 was rehybridized with cDNA probes for somatostatin and cholecystokinin, two neuropeptide genes expressed at relatively high levels in vertebrate brain (23). These experiments demonstrated abundant yet differing amounts of somatostatin and CCK mRNAs in both adult and fetal hypothalamus, despite the apparent lack of detectable glucagon mRNA transcripts in the same RNA preparations (Fig. 1, B and C). Although the mRNA signals obtained with the somatostatin and CCK probes appeared similar, the relative levels of abundance of these mRNAs in poly(A+) RNA from brainstem and hypothalamus were clearly different. In view of the suggestive, albeit inconclusive, presence of a faint glucagon mRNA transcript in RNA from rat hypothalamus (Fig. l), we attempted to more conclusively demonstrate the presence of glucagon mRNA transcripts in rat hypothalamus. A second Northern blot was run with 2 pg of poly(A') RNA prepared from rat brainstem ( B S ) or hypothalamus (H) and is shown in Fig. 2. A 20-h exposure was sufficient to visualize the relatively abundant glucagon mRNA transcripts in the brainstem preparation, but no definite glucagon mRNA transcript was detected in the RNA prepared from hypothalamus. However, prolonged autoradiographic exposure of the same blot for 12 days resulted in the appearance of a single band in the hypothalamic lane, the same size as the glucagon mRNA transcript detected in the brainstem. Thus, the glucagon gene is also expressed in the hypothalamus, albeit a t levels approximately 100-fold lower than in brainstem. In contrast to the marked regional variation in abundance of glucagon mRNA transcripts, the same blot was rehybridized with a somatostatin cDNA probe, as shown in Fig. 2C. Nearly identical amounts of somatostatin mRNA transcripts were detected in the brainstem and hypothalamic RNA preparations, consistent with the results obtained in Fig. 1B.
To define in greater detail the structure of the preproglucagon mRNA produced in cells of neural origin, we screened cDNA libraries prepared from both hypothalamus and brainstem RNA with a full-length rat pancreatic preproglucagon cDNA. No positive clones were obtained after screening 1 X

IP-11
Ala Ile VI1 GlU GlU Leu Gly Arc Arc  lo6 recombinant phage from a rat hypothalamic cDNA library.

G A T G A G A T G A A C A C C A T T C T T G A T A A T C T T G C C G C C A G G G A C T T T A T A AAC TGG TTG ATT CAG ACC AAA ATC ACT GAC AGG
Initial screening of a human neonatal brainstem cDNA library produced a single clone that hybridized strongly under stringent conditions to the rat preproglucagon cDNA probe, in keeping with the low abundance of glucagon mRNA transcripts detected in fetal brainstem (Fig. 1). This cDNA, BS8, was used to rescreen the cDNA library for additional clones.
A total of 3 x lo6 recombinant phage were screened and three additional clones were identified, BS3, BSA, and BSB. These three clones were characterized by restriction mapping and DNA sequencing and were found to be identical. The structure and restriction map of the human glucagon brainstem cDNA is depicted in Fig. 3 human preproglucagon gene and contains the entire coding sequence for a signal peptide, glucagon, glicentin, and glucagon-like peptides I and 11.
Expression of the glucagon gene in pancreas and intestine gives rise to a single glucagon mRNA species which, following translation, is processed to glucagon in the pancreas and glicentin and the GLPs in the intestine (22,(27)(28). To determine if the glucagon mRNA transcripts were translated in brain neurons, we examined human brain tissue for the presence of glucagon and GLP-I immunoreactivity. GLP-Iimmunoreactive neurons were observed in the medulla oblongata, in the region of the dorsal motor nucleus of the vagus (Fig. 4, u and 6). Neurons containing GLP-I immunoreactivity were also detected in the region of the paraventricular nucleus ( Fig. 4, a and b). Neurons containing GLP-I immunoreactivity were also detected in the region of the paraventricular nucleus in the hypothalamus (Fig. 4c). Positive staining was completely abolished by absorption with excess synthetic GLP-I peptide (Fig. 4b), confirming the specificity of immunostaining for authentic peptide. Serial sections of hypothalamus that stained for GLP-I were negative with various antisera to glucagon (not shown). In contrast, brainstem nuclei that contained GLP-I immunoreactivity also stained weakly with antisera to glucagon, suggesting that cell-specific post-translational processing may contribute to the diversity of glucagon gene expression in different regions of the brain.
Previous immunochemical studies of glucagon and glucagon-related peptides in brain tissues (6)(7)(8) suggested that posttranslational processing of proglucagon in the brain may differ from that seen in pancreas or intestine. Recent studies have indicated that neurons in the brainstem differ from those in the hypothalamus in that both stain with antiserum against GLP-I, yet the hypothalamic neurons do not stain with antiserum against pancreatic glucagon. Evidence for the translation of proglucagon mRNA in different regions of the brain derives from multiple previous studies that have demonstrated both glucagon and GLP-I-immunoreactive neurons in brainstem and hypothalamus (4)(5)(6)(7)(8)(9)(10)(11)14). Moreover, gel filtration chromatography in combination with radioimmunoassays have detected multiple immunoreactive species with antisera to both glucagon and GLP-I (9, 11,14). Interestingly, we found that glucagon mRNA transcripts were much more abundant in the brainstem compared with hypothalamus, yet a recent study of GLP-I immunoreactivity in brain tissues found twice as much GLP-I immunoreactivity in hypothalamus compared with brainstem (14). Moreover, the detection of GLP-I-immunoreactive neurons in the hypothalamus with failure to detect mRNA transcripts in these neurons by in situ hybridization using a GLP-I-directed oligonucleotide has been recently described (8). These results suggest that either the proglucagon mRNA is less abundant but either translated or stored more efficiently in the hypothalamus than in brainstem or that glucagon or the glucagon-like peptides may be synthesized in the brainstem and transported along brainstem afferents to nuclei in the hypothalamus. The existence of such brainstem projections has been recently described (29). An alternative explanation that invokes a novel glucagon-related gene that encodes a GLP-I-like peptide is less likely but cannot be excluded. We have failed to detect additional glucagon-related mRNA transcripts by Northern blot analysis after hybridizing at lower stringencies, and Southern blot analyses of rat or human DNA using glucagon cDNA probes at lower stringencies are consistent with the presence of only one glucagon gene (24)(25)(26).* The detection of glucagon mRNA transcripts in hypothalamus and brainstem, concomitant with the isolation of a full-length cDNA clone encoding proglucagon from a brainstem cDNA library, provides strong evidence for the importance of local synthesis of glucagon and the D. J. Drucker and S. Asa, unpublished observations. glucagon-like peptides in the brain. Future studies will attempt to address the target sites of action and physiological relevance of this newly described family of proglucagon-derived neuropeptides.