Galanin is localized in sympathetic neurons of the dog liver

Transcription

1 Galanin is localized in sympathetic neurons of the dog liver THOMAS O. MUNDINGER, 1,3 C. BRUCE VERCHERE, 1,3 DENIS G. BASKIN, 1,2,3 MICHAEL R. BOYLE, 1,3 STEPHAN KOWALYK, 1,3 AND GERALD J. TABORSKY, JR. 1,3 1 Departments of Medicine and 2 Biological Structure, University of Washington, Seattle 98195; and 3 Division of Endocrinology and Metabolism, Veterans Affairs Puget Sound Health Care System, Seattle, Washington Mundinger, Thomas O., C. Bruce Verchere, Denis G. Baskin, Michael R. Boyle, Stephan Kowalyk, and Gerald J. Taborsky, Jr. Galanin is localized in sympathetic neurons of the dog liver. Am. J. Physiol. 273 (Endocrinol. Metab. 36): E1194 E1202, Stimulation of canine hepatic nerves releases the neuropeptide galanin from the liver; therefore, galanin may be a sympathetic neurotransmitter in the dog liver. To test this hypothesis, we used immunocytochemistry to determine if galanin is localized in hepatic sympathetic nerves and we used hepatic sympathetic denervation to verify such localization. Liver sections from dogs were immunostained for both galanin and the sympathetic enzyme marker tyrosine hydroxylase (TH). Galanin-like immunoreactivity (GALIR) was colocalized with TH in many axons of nerve trunks as well as individual nerve fibers located both in the stroma of hepatic blood vessels and in the liver parenchyma. Neither galanin- nor TH-positive cell bodies were observed. Intraportal 6-hydroxydopamine (6-OHDA) infusion, a treatment that selectively destroys hepatic adrenergic nerve terminals, abolished the GALIR staining in parenchymal neurons but only moderately diminished the GALIR staining in the nerve fibers around blood vessels. To confirm that 6-OHDA pretreatment proportionally depleted galanin and norepinephrine in the liver, we measured both the liver content and the hepatic nerve-stimulated spillover of galanin and norepinephrine from the liver. Pretreatment with 6-OHDA reduced the content and spillover of both galanin and norepinephrine by 90%. Together, these results indicate that galanin in dog liver is primarily colocalized with norepinephrine in sympathetic nerves and may therefore function as a hepatic sympathetic neurotransmitter. norepinephrine; hepatic glucose production; hepatic blood flow THE 29-AMINO ACID neuropeptide, galanin, is found in both central and peripheral neurons of several species (4, 20, 26, 30). In the dog, galanin is found in intrinsic nerves of the gut (14) and in postganglionic sympathetic nerves of the pancreas (8). Galanin-like immunoreactivity (GALIR) (1) and galanin mrna (29) have been found in the majority of cell bodies of dog celiac ganglia, which supply postganglionic sympathetic nerves not only to the pancreas but also to the stomach, duodenum, and liver. Stimulation of hepatic sympathetic nerves projecting from the celiac ganglia releases galanin from the liver (17); therefore, galanin may be a sympathetic neurotransmitter in the dog liver like it is in the pancreas (9). Five criteria can be used to classify a peptide as a sympathetic neurotransmitter or neuromodulator (2). First, the peptide is localized in postganglionic sympathetic nerves within the organ. Second, the peptide is released from the organ during sympathetic nerve stimulation. Third, the biological effect of nerve stimulation is not completely blocked by neutralizing classical neurotransmitter (i.e., norepinephrine) action. Fourth, specific peptide antagonism or immunoneutralization abolishes the effect of sympathetic nerve stimulation that persists during adrenergic blockade. Fifth, the local infusion of exogenous peptide is sympathomimetic. In the case of canine hepatic galanin, only the second criterion has been clearly demonstrated (17). However, preliminary data relevant to the fifth criterion suggest that hepatic arterial infusions of galanin potentiates norepinephrine s stimulation of hepatic glucose production (HGP) (27). The goal of the present study was to determine whether or not canine hepatic galanin is localized in sympathetic nerves, thus addressing the first criterion above. To demonstrate that galanin is colocalized with norepinephrine in hepatic sympathetic nerves, we stained liver sections of control dogs for both galanin and tyrosine hydroxylase (TH), an enzyme marker of sympathetic nerves. We also looked for a parallel decrease of GALIR and TH staining in dogs whose livers were sympathectomized by a prior intraportal infusion of 6-hydroxydopamine (6-OHDA). To verify that hepatic galanin is localized in sympathetic nerves, we decreased both the liver content and the hepatic nerve-stimulated release of norepinephrine with 6- OHDA pretreatment and tested for parallel galanin reductions. In addition to measuring sympathetic neurotransmitter release during hepatic nerve stimulation (HNS), we compared the nerve-stimulated changes in HGP and hepatic arterial blood flow (HABF) in control vs. 6- OHDA-pretreated dogs. 6-OHDA pretreatment markedly decreased the HGP response to 8-Hz nerve stimulation yet only modestly reduced the HABF response. These observations, coupled with the staining data, are consistent with greater sympathetic innervation of the vasculature than the hepatocyte in the dog liver. To investigate the physiological implications of such differential innervation, we electrically stimulated the hepatic nerve at low frequency (1 Hz) in control dogs and demonstrated its ability to elicit an HABF response in the absence of an HGP response. MATERIALS AND METHODS Animals, surgical procedures, and 6-OHDA pretreatment. For acute, terminal studies, adult male dogs (28 35 kg) of mixed breed were fasted overnight and surgically prepared as E /97 $5.00 Copyright 1997 the American Physiological Society

2 GALANIN IN HEPATIC SYMPATHETIC NERVES E1195 described in detail elsewhere (17). Briefly, anesthesia was induced with the ultra short-acting barbiturate thiamylal sodium (Surital, Parke Davis, Morris Plains, NJ; 30 mg/kg iv) and maintained with halothane (0.8%) in 100% oxygen administered by positive pressure ventilation from a calibrated vaporizer. Adequate levels of anesthesia during surgery and experimentation were verified by maintenance of normal blood pressure and heart rate and by an absence of a pedal and an eye reflex. A midline laparotomy was performed to allow placement of sampling catheters and blood flow probes (Transonic Systems, Ithaca, NY). Blood sampling catheters were placed in the femoral artery, portal vein, and hepatic vein, the latter being a Swan-Ganz catheter (Arrow International, Reading, PA) introduced in the femoral vein and fluoroscopically guided into a hepatic vein. The catheter tip was advanced to a point where hepatic venous blood from several branches flowed freely by the uninflated tip, and blood from the inferior vena cava could not flow in a retrograde fashion around the inflated tip. An infusion catheter was placed in the femoral vein for saline administration. The nerve sheath surrounding the common hepatic artery, the anterior hepatic plexus, was dissected free from the vessel midway between the celiac ganglia and the branching of the gastroduodenal artery. The anterior hepatic plexus was ligated and cut, and a stimulation electrode was placed around the proximal end of the anterior hepatic plexus at a distance of one half inch from the cut. Transection of the anterior hepatic plexus ensured that no retrograde, afferent stimulation resulted from HNS. Ultrasonic blood flow probes were placed around the common hepatic artery at the point where the nerve sheath had been stripped from the artery and around the portal vein as it enters the liver. After the completion of the surgery, a 60-min stabilization period preceded basal determinations. Ten days before HNS studies, animals undergoing hepatic sympathectomy received an intraportal infusion of 6-OHDA (Sigma, St. Louis, MO), a treatment that markedly reduces the number of adrenergic nerve terminals in the liver (3). Because extraction of catecholamines by the liver is avid (17), intraportal 6-OHDA was expected to selectively denervate the liver. The adrenal medulla, peripheral cholinergic neurons, and central neurons are unaffected by intraportal 6-OHDA infusion (13). To access the portal vein, a midline laparotomy was performed under aseptic conditions, and a splenic vein was catheterized for intraportal infusion of 6-OHDA (125 µg kg 1 min 1 20 min), as described recently (21). Fifteen minutes before the intraportal infusion of 6- OHDA, the -adrenergic antagonist esmolol (500 µg/kg plus 300 µg kg 1 min 1 iv) was infused to reduce the potentially fatal arrhythmias induced by large amounts of norepinephrine which are released from hepatic noradrenergic nerve terminals immediately after 6-OHDA treatment. Heart rate and mean arterial blood pressure were monitored and kept in acceptable ranges (heart rate 180 beats/min, mean arterial blood pressure 160 mmhg) by further - and -adrenergic blockade (phentolamine, 1.5 mg iv bolus/injection; propranolol, 1.0 mg iv bolus/injection, respectively) as needed. Femoral arterial and portal venous catheters were removed, wounds were sutured, anesthesia was discontinued, and dogs were allowed to recover from surgery only when blood pressure and heart rate were stable in the absence of adrenergic blockade. All animals included in these studies were certified as healthy by the Veterinary Medical Officer of the Veterans Affairs Puget Sound Health Care System (VAPSHCS) and exhibited normal white blood cell counts, hematocrit, temperature, food intake, urination, and defecation before acute, terminal nerve stimulation studies. All research involving animals was conducted in an American Association for Accreditation of Laboratory Animal Care-accredited facility. All protocols were designed to ensure appropriate ethical treatment of the animals and were approved by the Institutional Animal Care and Use Committee of the VAPSHCS. HNS protocol, blood sampling, and assays. Two separate HNSs were performed on each of four control and six 6-OHDApretreated dogs. HNS was performed by electrically stimulating the sheath surrounding the hepatic artery using a model S-44 stimulator (Grass Instruments, Quincy, MA). HNS was performed during hexamethonium (1 mg/kg plus 0.7 µg kg 1 min 1 iv) and atropine (0.25 mg/kg plus 0.4 µg kg 1 min 1 iv) treatment to block the effect of any stimulated parasympathetic nerves that may run in the hepatic arterial sheath. HNS parameters were as follows: frequency 1, 4, or 8 Hz; current 10 ma; pulse duration 1 ms. Blood samples were taken from the femoral artery, portal vein, and hepatic vein, and both hepatic arterial and portal venous blood flows were monitored at each sampling time: 10, 5, and 0 min before; 2.5, 5, and 10 min during; and 15 and 25 min after HNS. The second HNS was performed 45 min after the first HNS. Blood samples drawn for measurement of plasma GALIR concentration were placed in tubes containing several proteolytic enzyme inhibitors (7). Blood samples for norepinephrine analysis were drawn on a mixture (50 µl/2.5 ml blood) of ethylene glycol-bis( -aminoethyl ether)-n,n,n,n -tetraacetic acid (0.09 mg/ml) and glutathione (0.06 mg/ml), and those for glucose measurement were drawn on EDTA. All blood samples were kept on ice until centrifugation (3,000 revolutions/min, 20 min, 2 C). The centrifuged plasma was decanted and frozen at 20 C until assayed. Unextracted plasma was assayed for galanin by radioimmunoassay (RIA) (10) with a non-cooh-terminally directed antibody raised against porcine galanin. Because the species variable portion of the galanin molecule is limited to the COOH terminus (15, 24), this assay detects GALIR from all species yet examined. Synthetic porcine galanin was used for assay standard, for canine galanin dilutes in parallel with porcine standards (17). Plasma norepinephrine was measured in duplicate with a sensitive and specific radioenzymatic assay (23). The intraand interassay coefficients of variation in this lab are 6% and 12%, respectively. Plasma glucose was measured by the glucose oxidase method (Beckman, Brea, CA). Data analysis. To assess the magnitude of galanin and norepinephrine release from the liver during nerve stimulation, the hepatic spillovers of galanin (HGALSO) and norepinephrine (HNESO) were calculated before and during HNS, as previously described in detail (17) and recently validated (21). Briefly, neurotransmitter spillover was calculated as the total rate of neurotransmitter exiting the liver minus that portion that enters the liver by the hepatic artery and portal vein and escapes hepatic extraction. The fractional extraction of norepinephrine across the liver was previously determined in this lab to be 90% (17). To calculate this hepatic norepinephrine extraction, tritiated norepinephrine (NEN, Boston, MA; 1.5 µci/min iv) was infused, and after a stabilization period, blood samples were drawn from the femoral artery, portal vein, and hepatic vein. Blood flows in the common hepatic artery and portal vein were also measured. Tritiated norepinephrine was extracted from plasma with alumina, radioactivity was counted, and hepatic extraction of tritiated norepinephrine was calculated by an arteriovenous difference equation (17). To monitor the vascular and metabolic responses of the liver to nerve stimulation, we measured blood flow in the common hepatic artery and calculated HGP before and

3 E1196 GALANIN IN HEPATIC SYMPATHETIC NERVES during HNS. HGP was calculated by the method of Myers et al. (22), which incorporates plasma glucose concentrations and blood flows in the common hepatic artery, portal vein, and hepatic vein as well as conversion factors to convert plasma glucose concentrations to blood glucose concentrations. All data are expressed as means SE. Statistical comparison of responses to HNS vs. basal were made using a single-tailed paired t-test. Comparison of neurotransmitter content between liver sections of control vs. 6-OHDA-treated dogs was made using a two-sample t-test. Correlation of the HGALSO and HNESO responses to HNS was performed using the method of least squares. Immediately after the completion of the HNS studies, dogs were euthanized by an overdose of anesthesia and liver sections were immediately excised for hepatic staining and extraction. Immunocytochemistry. The basic procedure for immunocytochemistry has been previously described (11, 28). Pieces of liver (1-cm cubes) were placed in 4% paraformaldehyde in 0.1 M phosphate buffer (ph 7.2) for 2 h, transferred to phosphate buffer overnight, dehydrated, and embedded in paraffin. Sections (5 µm) were incubated overnight (4 C) with antibodies raised in rabbits against porcine galanin (1:4,000) followed by a 1-h, room temperature incubation with fluorescein isothiocyanate (FITC)-labeled goat anti-rabbit immunoglobulin G. The sections were then incubated overnight with monoclonal antibody to TH followed by a 1:20 dilution of Cy3-labeled anti-rabbit polyclonal antiserums for 1 h. Sections were examined with a fluorescence microscope equipped with filters for Cy3 and FITC microscopy. By switching filters, galanin and TH stainings were viewed in the same liver section. Acid extraction and chromatography. To assess hepatic galanin and norepinephrine content, excised liver sections were immediately snap-frozen on dry ice and stored at 70 C. Tissue (1 2 g) was homogenized and boiled in 1 N acetic acid (10 ml/g tissue, 10 min). The homogenate was then centrifuged twice (10,000 revolutions/min, 20 min), and the supernatant was dried and reconstituted in 2 ml of galanin RIA buffer. The reconstituted extract was stored at 20 C until assayed. To characterize the molecular size of GALIR in canine liver, we subjected liver extracts to gel-exclusion chromatography. Liver extracts (1.5 ml) were applied to a 1 50-cm Sephadex G-50 column and eluted with galanin RIA buffer. Fifty 1-ml fractions were collected and run in the galanin RIA. Synthetic porcine galanin (Bachem, Torrance, CA) and blue dextran were used as galanin-sized and void volumesized markers, respectively. RESULTS Hepatic galanin localization. To determine if hepatic galanin is localized exclusively in hepatic sympathetic nerves, we dual-immunostained sections of peripheral dog liver (n 5) for the presence of both GALIR and TH. GALIR-positive staining was observed in nerve trunks (Fig. 1B) and in individual fibers both surround- Fig. 1. Liver sections showing nerve trunks near the portal hilus (A, B) and individual fibers surrounding blood vessels (C, D) or coursing through parenchyma (E, F) immunostained for sympathetic enzyme marker, tyrosine hydroxylase (A, C, E) and galanin (B, D, F). Dual immunostaining for galanin and tyrosine hydroxylase was performed on same liver sections.

4 GALANIN IN HEPATIC SYMPATHETIC NERVES E1197 Fig. 2. Liver sections showing blood vessels (A, B) and parenchyma (C, D) from control (A, C) and 6- hydroxydopamine (6-OHDA Rx)-treated (B, D) dogs immunostained for galanin. Galanin-positive nerve fibers (arrows) are frequently observed in stromal tissue surrounding blood vessels in liver from both control (A) and 6-OHDA-treated dogs (B) as well as in liver parenchyma of control dogs (C). Liver parenchyma of 6-OHDA-treated dogs (D) did not have appreciable galanin-positive staining. In all panels, bar 40 µm. ing blood vessels (Fig. 1D) and coursing through liver parenchyma (Fig. 1F). Similar localization was observed for TH (Fig. 1, A, C, and E). Indeed, most of the GALIR staining was colocalized with TH. Neither GALIR-positive nor TH-positive staining was observed in intrahepatic neuronal cell bodies or other hepatic cell bodies. To verify that hepatic galanin is localized in noradrenergic nerves, we stained liver sections of dogs that were previously subjected to local, hepatic chemical sympathectomy induced by intraportal 6-OHDA infusion (n 5). The parenchymal staining of GALIR fibers that was observed in control liver (Fig. 2C) was virtually absent in liver sections from 6-OHDA-treated dogs (Fig. 2D). Surprisingly, GALIR staining of nerves surrounding blood vessels (Fig. 2A) was still prominent after 6- OHDA treatment (Fig. 2B). Hepatic galanin content. To quantitate the amount of hepatic galanin localized to noradrenergic nerves, we measured the molar content of hepatic GALIR and norepinephrine in liver sections of control and 6-OHDAtreated dogs. The GALIR content of liver sections from control animal was pmol/g tissue (n 5, Fig. 3A), whereas that of animals pretreated with 6-OHDA was pmol/g tissue (n 5). The norepinephrine content of liver sections from control animals was 3, pmol/g tissue (Fig. 3B), whereas that of animals pretreated with 6-OHDA was pmol/g tissue. Thus 6-OHDA pretreatment decreased the hepatic content of GALIR by 87% and decreased the hepatic content of norepinephrine by 94%. To verify that the decrement of hepatic GALIR content after 6-OHDA pretreatment was due to a reduction in the amount of a galanin-sized molecule, acid extracts of peripheral liver tissue of control and 6-OHDA-pretreated dogs were subjected to size-exclusion gel chromatography. Figure 4 depicts the average gel filtration profiles of control (n 5) and 6-OHDApretreated (n 5) liver extracts. The majority of GALIR present in liver extracts of control animals coeluted with synthetic porcine galanin, and the amount of GALIR eluting at the porcine galanin marker in 6-OHDA-pretreated animals was reduced by 77% compared with controls. Therefore, GALIR appears to be predominantly due to a galanin-sized peptide, and 6-OHDA treatment markedly reduces the liver content of this peptide. Hepatic galanin release during nerve stimulation. To demonstrate that 6-OHDA pretreatment depleted hepatic GALIR from a releasable neuronal pool, we examined the effect of 6-OHDA pretreatment on the hepatic spillover of GALIR and norepinephrine during 8-Hz HNS. The calculated HGALSO in control dogs (n 8) increased from a baseline of 0 3 pmol/min to an average of pmol/min ( pmol/min, P 0.01, Fig. 5A) between 5 and 10 min of HNS, whereas HGALSO in 6-OHDA-pretreated dogs (n 12) increased from a baseline of 0 0 pmol/min to only an average of 2 1 pmol/min ( 2 1 pmol/min, P 0.05; Fig. 5A). Thus 6-OHDA pretreatment decreased the HGALSO response to 8-Hz HNS by 97%.

5 E1198 GALANIN IN HEPATIC SYMPATHETIC NERVES Fig. 3. Content of liver galanin-like immunoreactivity (GALIR; A)or norepinephrine (NE; B) in control (n 5) and 6-OHDA-pretreated (n 5) dogs. Values are means SE. The concentration of GALIR in hepatic venous plasma, an alternative index of hepatic neurotransmitter release (12), increased in control dogs from a baseline of pm to an average of pm Fig. 4. Gel elution profile of GALIR in liver extracts of control (n 5) and 6-OHDA-pretreated (n 5) dogs. Blue dextran was used for void volume marker (V o ), and synthetic porcine galanin was used for galanin peptide marker (GAL). Values are means SE. Fig. 5. Change of hepatic galanin spillover (HGALSO; A) or hepatic norepinepherine spillover (HNESO; B) during 8-Hz hepatic nerve stimulation (HNS) in control and 6-OHDA-pretreated dogs. Values are means SE. ( pm, P 0.01) between 5 and 10 min of 8-Hz HNS. In contrast, the concentration of GALIR in hepatic venous plasma of 6-OHDA-pretreated dogs increased from a baseline of 37 5 pm to an average of 46 5pM[ 9 9 pm, P not significant (NS)]. 6-OHDA pretreatment thereby decreased the increment of hepatic venous GALIR during HNS by 96%. The HNS-induced increment of GALIR in femoral arterial plasma was less than that seen in the hepatic vein. The concentration of GALIR in femoral arterial plasma increased by pm (P 0.05) from a baseline of pm in control dogs but did not increase ( 1 5 pm, P NS) from a baseline of 38 4pMin 6-OHDA-pretreated dogs. Thus 6-OHDA pretreatment abolished the arterial GALIR increment during HNS. The magnitude of the norepinephrine response to HNS was correlated to that of the GALIR response, and 6-OHDA treatment reduced the magnitude of each by the same percentage. The calculated HNESO in control dogs increased from a baseline of pmol/min to an average of 1, pmol/min ( 1, pmol/min, P 0.01; Fig. 5B and Table 1). In control dogs, there was a significant positive correlation between the HNESO and HGALSO responses to HNS (r 0.95, P , n 8). HNESO in 6-OHDA-pretreated

6 GALANIN IN HEPATIC SYMPATHETIC NERVES E1199 Table 1. Changes in HNESO, HGP, and common HABF during HNS in control and 6-OHDA-pretreated dogs 1-Hz HNS Control 4-Hz HNS Control 8-Hz HNS Control 8-Hz HNS 6-OHDA n HNESO, pmol/min * 1, * HGP, mg kg 1 min * * * HABF, ml/min 22 9* 35 16* 56 6* 35 7* Values are means SE; n, no. of dogs. HNS, hepatic nerve stimulation; 6-OHDA, 6-hydroxydopamine; HNESO, hepatic norepinephrine spillover; HGP, hepatic glucose production; HABF, hepatic arterial blood flow. * Significant response vs. basal. dogs increased from a baseline of pmol/min to an average of pmol/min ( pmol/min, P NS, Fig. 5B and Table 1). Thus 6-OHDA pretreatment produced a decrease of HNESO (96%) that was identical to the decrease of HGALSO (97%). The concentration of norepinephrine in hepatic venous plasma in control dogs increased from a baseline of pg/ml to an average of 1, pg/ml ( 1, pg/ml, P 0.01) between 5 and 10 min of HNS. In contrast, the concentration of norepinephrine in hepatic venous plasma in 6-OHDA-pretreated dogs increased from a baseline of pg/ml to only pg/ml ( pg/ml, P NS). 6-OHDA pretreatment thereby decreased the increment of hepatic venous norepinephrine to HNS by 91%. The increments of femoral arterial norepinephrine in response to HNS were less than those of the hepatic vein in both control and 6-OHDA-pretreated dogs. Arterial plasma norepinephrine increased pg/ml (P 0.005) from a baseline of pg/ml in control dogs and increased pg/ml (P 0.05) from a baseline of 27 8 pg/ml in 6-OHDA-pretreated dogs. 6-OHDA pretreatment thereby decreased the increment of femoral arterial norepinephrine to HNS by 81%. Metabolic and vascular responses during nerve stimulation. To determine if 6-OHDA pretreatment produced an equivalent reduction of the metabolic and vascular responses to HNS, we compared the HGP and HABF responses to HNS in control vs. 6-OHDA-treated dogs. HGP in control dogs increased from a baseline of mg kg 1 min 1 to an average of mg kg 1 min 1 ( mg kg 1 min 1, P 0.025, Fig. 6 and Table 1) between 5 and 10 min of 8-Hz HNS. In contrast, HGP in 6-OHDA-pretreated dogs increased from a baseline of mg kg 1 min 1 to mg kg 1 min 1 ( mg kg 1 min 1, P 0.05) in response to HNS. Thus 6-OHDA pretreatment decreased the HGP response to 8-Hz HNS by 86%. Arterial plasma glucose rose progressively during the 10 min of 8-Hz HNS. Plasma glucose in control dogs rose from a baseline value of 93 3 mg/dl to an average of mg/dl ( 21 6 mg/dl, P 0.005) between 5 and 10 min of HNS, whereas plasma glucose in 6-OHDA-pretreated dogs rose 5 2 mg/dl (P 0.025) from a baseline of 95 2 mg/dl. Thus 6-OHDA pretreatment decreased the arterial glucose response to 8-Hz HNS by 76%. HABF in both control and 6-OHDA-pretreated animals decreased during 8-Hz HNS. HABF in control dogs decreased from a baseline of ml/min to an average of 64 5 ml/min ( 56 6 ml/min, P ; Fig. 5 and Table 1) between 5 and 10 min of 8-Hz HNS. The decrement of HABF in 6-OHDApretreated dogs was 35 7 ml/min (P 0.005; Fig. 7 and Table 1) from a baseline of ml/min. Thus 6-OHDA pretreatment decreased the HABF response to 8-Hz HNS by only 38%, in marked contrast to the % reduction of other parameters measured. Low-frequency nerve stimulation. Because 6-OHDA pretreatment decreased the HGP response more than the HABF response during high-frequency (8-Hz) nerve stimulation, we sought to determine if low-frequency nerve stimulation preferentially stimulated the vasculature over the hepatocyte. To simulate lower levels of neural activation in control dogs, we electrically stimulated the hepatic nerve at 1 and 4 Hz and measured HABF, HGP, and HNESO (see Table 1). One-hertz Fig. 6. Change of hepatic glucose production (HGP) during 8-Hz HNS in control and 6-OHDA-pretreated dogs. Values are means SE. Fig. 7. Change of common hepatic arterial blood flow (HABF) during 8-Hz HNS in control and 6-OHDA-pretreated dogs. Values are means SE.

7 E1200 GALANIN IN HEPATIC SYMPATHETIC NERVES HNS produced no significant change of either HGP or HNESO, yet produced a significant decrease of HABF. Four-hertz HNS increased HGP and HNESO and further decreased HABF. Therefore, low-frequency (1-Hz) HNS in control dogs produced a disparity between vascular and metabolic responses that was similar to that seen during high-frequency (8-Hz) stimulation in 6-OHDA-treated dogs. DISCUSSION Galanin is released from the dog liver during HNS (17), and preliminary data suggest that galanin potentiates norepinephrine-stimulated increases of HGP (27). Galanin may therefore be a sympathetic neurotransmitter in dog liver that aids in glucose mobilization. The observation presented in this paper that GALIR is colocalized with TH in liver nerves suggests that galanin is expressed by hepatic sympathetic nerves. The colocalization of GALIR and TH in hepatic nerves is consistent with the colocalization of GALIR peptide (1) and galanin mrna (29) with TH in roughly 90% of the cell bodies of the celiac ganglia, the sympathetic ganglia which project the majority of postganglionic nerves to the liver. Colocalization of GALIR with TH in liver nerves is also consistent with the corelease of galanin and norepinephrine from the liver during HNS (see below) (17). The lack of GALIR staining in hepatic neuronal cell bodies or other cells in the liver is evidence that galanin is not expressed by hepatocytes, Kupffer cells, Ito cells, parasympathetic nerves, or sensory nerves. Together these data support the hypothesis that galanin is a sympathetic neuropeptide in the dog liver. To independently confirm that galanin is localized in sympathetic nerves of dog liver, we examined the effect of 6-OHDA on neural GALIR and TH staining. As expected, staining of GALIR and TH in parenchymal nerves was not observed in liver sections of dogs treated with 6-OHDA, but, surprisingly, staining of GALIR nerves around blood vessels was only modestly reduced in comparison to controls. The disproportionate decrease of neural GALIR staining between the parenchyma and the vasculature after 6-OHDA treatment could be due to a more dense innervation of the canine hepatic vasculature compared with the parenchyma, as suggested in other species by previous reviews (25, 31). Equal degrees of parenchymal and vascular denervation by 6-OHDA treatment might then account for the residual staining observed around the vasculature. Alternatively, it is possible that the liver parenchyma is perfused more completely than the vasculature with portal venous blood. Therefore, the portal venous infusion of 6-OHDA could preferentially denervate the parenchyma. Regardless of which explanation is correct, because 6-OHDA selectively destroys noradrenergic nerve terminals (16), the observation of a marked reduction of GALIR in parenchymal nerves strongly supports the hypothesis that galanin is a sympathetic neuropeptide in dog liver. To further confirm that galanin is localized in sympathetic nerves, we examined the effect of 6-OHDA treatment on GALIR content in acid extracts of liver sections. The marked reduction of GALIR and norepinephrine content after 6-OHDA treatment paralleled the marked reduction of GALIR staining in the liver parenchyma. Again, because the neurolytic action of 6-OHDA is specific to noradrenergic nerves, these data again indicate that galanin is colocalized with norepinephrine in hepatic sympathetic nerves. The high correlation between changes in HNESO and HGALSO during HNS suggests that galanin and norepinephrine are coreleased from the same hepatic sympathetic nerve fibers. Furthermore, 6-OHDA pretreatment produced a marked and parallel reduction of HGALSO and HNESO responses to HNS. Interestingly, the molar ratio of galanin to norepinephrine spillover (1:30) in control dogs during HNS was markedly different from the molar ratio of galanin to norepinephrine content (1:750) in control liver tissue. These data suggest either that there are significant nonreleasable stores of hepatic norepinephrine or that there is preferential release of galaninergic storage vesicles during the high-frequency nerve stimulation employed in these experiments. The latter explanation has been proposed for other neuropeptides (18, 19). Nonetheless, the high correlation between the spillover of galanin and norepinephrine in control dogs during HNS and the proportional decrease of releasable galanin and norepinephrine in 6-OHDA-treated dogs suggests that galanin is localized in hepatic sympathetic nerves. The spillover of galanin and norepinephrine from the liver during HNS was accompanied by an increase of HGP and arterial glucose. Although norepinephrine is a known stimulator of HGP (6), preliminary data suggest that coreleased peptides such as galanin may augment the norepinephrine-stimulated HGP response (27). 6-OHDA pretreatment decreased the HGP response to HNS by an amount similar to the reduction of nerve-stimulated galanin and norepinephrine spillover. Thus sympathetic neurotransmitters mediate the HGP response during HNS, but the relative roles of norepinephrine and galanin in glucose mobilization require clarification. Although the HGP response to HNS was markedly reduced after 6-OHDA pretreatment, the HABF response was only modestly reduced. Because the majority of HABF is supplied to the liver, these data suggest that the nerves of the liver that remain after 6-OHDA treatment predominantly supply the vasculature. Alternatively, a sympathetic neurotransmitter that is not colocalized with norepinephrine could modulate a significant portion of the vasoconstrictor response to HNS. However, this latter explanation is highly speculative, and the staining data in this paper support the former explanation, since some GALIR-positive fibers remain near the vasculature after 6-OHDA pretreatment. In addition, GALIR staining of control liver sections show dense innervation of the vasculature with more diffuse innervation of the liver parenchyma, a pattern suggested by the available literature, as summarized by Woods et al. (31) and Sawchenko and Friedman (25) The possibility of differential innervation of hepatic vasculature vs. hepatocytes is also indirectly supported

8 GALANIN IN HEPATIC SYMPATHETIC NERVES E1201 by the effect of low-frequency nerve stimulation on HABF and HGP in control dogs. For example, low levels (1 Hz) of HNS in control dogs produced vasoconstriction without increasing HGP (see Table 1), a pattern similar to that seen in 6-OHDA-pretreated dogs receiving high-frequency (8-Hz) HNS. Thus it appears that the more highly innervated vasculature can respond to degrees of neural activation that are insufficient to stimulate an HGP response from the less densely innervated hepatocytes. These data do not, however, rule out the alternative explanation that vascular endothelium is more sensitive than the hepatocyte to low levels of released norepinephrine. In addition to increasing HGP, HNS produced an increase of arterial GALIR. The arterial increment of GALIR was reduced by 90% by 6-OHDA pretreatment, demonstrating that it was of hepatic origin. Because small increments of arterial galanin can have significant physiological effects (5), this finding raises the possibility that galanin released from hepatic nerves may act as a neurohormone as well as a hepatic neurotransmitter. One possible neuroendocrine effect of hepatic galanin could be to decrease pancreatic insulin secretion (10). This potential neuroendocrine effect may occur during stress, for several stresses activate hepatic sympathetic nerves (21). If so, galanin released from the liver could aid in restraining glucosestimulated insulin release, thereby contributing to stress hyperglycemia. Conclusion. The finding that galanin is colocalized with TH in nerves of dog liver argues that galanin resides in hepatic sympathetic nerves. The finding that 6-OHDA treatment produces a parallel reduction of hepatic galanin and norepinephrine content and spillover supports the hypothesis that galanin is a sympathetic neurotransmitter in the dog liver. In addition, the ability of HNS to significantly increase circulating galanin levels, coupled with the potency of galanin to inhibit insulin secretion (5), suggests a neuroendocrine role for hepatic galanin. Finally, the hepatic vasculature is more responsive than parenchymal hepatocytes to low levels of HNS, suggesting more dense sympathetic innervation of the vascular smooth muscle, a hypothesis consistent with the staining data from control and 6-OHDA-treated dogs. This work was supported by the Medical Research Service of the Department of Veterans Affairs and by National Institute of Diabetes and Digestive and Kidney Diseases Grants DK-12829, DK-17047, and DK Address for reprint requests: T. O. Mundinger, Division of Endocrinology and Metabolism (151), Veterans Affairs Puget Sound Health Care System, 1660 S. Columbian Way, Seattle, WA Received 12 May 1997; accepted in final form 14 August REFERENCES 1. Ahren, B., G. Böttcher, S. Kowalyk, B. E. Dunning, F. Sundler, and G. J. Taborsky, Jr. Galanin is co-localized with noradrenaline and neuropeptide Y in dog pancreas and celiac ganglion. Cell Tissue Res. 261: 49 58, Ahren, B., G. J. Taborsky, Jr., and D. Porte, Jr. Neuropeptidergic versus cholinergic and adrenergic regulation of islet hormone secretion. Diabetologia 29: , Allman, F. D., E. L. Rogers, D. A. Caniano, D. M. Jacobowitz, and M. C. Rogers. Selective chemical hepatic sympathectomy in the dog. Crit. Care Med. 10: , Bishop, A. E., J. M. Polak, F. E. Bauer, N. D. Christofides, F. Carlei, and S. R. Bloom. Occurrence and distribution of a newly discovered peptide, galanin, in the mammalian enteric nervous system. Gut 27: , Boyle, M. R., C. B. Verchere, G. McKnight, S. Mathews, K. Walker, and G. J. Taborsky, Jr. Canine galanin: sequence, expression and pancreatic effects. Regul. Pept. 50: 1 11, Connolly, C. C., K. E. Steiner, R. W. Stevenson, D. W. Neal, P. E. Williams, K. G. M. M. Alberti, and A. D. Cherrington. Regulation of glucose metabolism by norepinephrine in conscious dogs. Am. J. Physiol. 261 (Endocrinol. Metab. 24): E764 E772, DeHaen, C. S., S. A. Little, J. M. May, and R. H. Williams. Characterization of proinsulin-insulin intermediates in human plasma. J. Clin. Invest. 62: , Dunning, B. E., B. Ahren, R. C. Veith, G. Böttcher, F. Sundler, and G. J. Taborsky, Jr. Galanin: a novel pancreatic neuropeptide. Am. J. Physiol. 251 (Endocrinol. Metab. 14): E127 E133, Dunning, B. E., and G. J. Taborsky, Jr. Galanin sympathetic neurotransmitter in endocrine pancreas? Diabetes 37: , Dunning, B. E., and G. J. Taborsky, Jr. Galanin release during pancreatic nerve stimulation is sufficient to influence islet function. Am. J. Physiol. 256 (Endocrinol. Metab. 19): E191 E198, Francis, B. H., D. G. Baskin, D. R. Saunders, and J. W. Ensink. Distribution of somatostatin-14 and somatostatin-28 gastrointestinal-pancreatic cells of rats and humans. Gastroenterology 99: , Garceau, D., N. Yamaguchi, R. Goyer, and F. Guitard. Correlation between endogenous noradrenaline and glucose released from the liver upon hepatic sympathetic nerve stimulation in anesthetized dogs. Can. J. Physiol. Pharmacol. 62: , Gilman, A. G., L. S. Goodman, T. W. Rall, and F. Murad. The Pharmalogical Basis of Therapeutics (7th ed.). New York: Macmillan, 1985, p Gonda, T., E. E. Daniel, T. J. McDonald, J. E. T. Fox, B. D. Brooks, and M. Oki. Distribution and function of enteric GAL-IR nerves in dogs: comparison with VIP. Am. J. Physiol. 256 (Gastrointest. Liver Physiol. 19): G884 G896, Kaplan, L. M., E. R. Spindel, K. J. Isselbacher, and W. W. Chin. Tissue-specific expression of the rat galanin gene. Proc. Natl. Acad. Sci. USA 85: , Kostrzewa, R. M., and D. M. Jacobowitz. Pharmacological actions of 6-hydroxydopamine. Pharmacol. Rev. 26: , Kowalyk, S., R. Veith, M. Boyle, and G. J. Taborsky, Jr. Liver releases galanin during sympathetic nerve stimulation. Am. J. Physiol. 262 (Endocrinol. Metab. 25): E671 E678, Lundberg, J. M., J. Pernow, and J. S. Lacroix. Neuropeptide Y: sympathetic cotransmitter and modulator? News Physiol. Sci. 4: 13 17, Lundberg, J. M., A. Rudehill, A. Sollevi, E. Theodorsson- Norheim, and B. Hamberger. Frequency- and reserpinedependent chemical coding of sympathetic transmission: differential release of noradrenaline and neuropeptide Y from pig spleen. Neurosci. Lett. 63: , Melander, T., T. Hokfelt, A. Rokaeus, J. Fahrenkrug, K. Tatemoto, and V. Mutt. Distribution of galanin-like immunoreactivity in the gastro-intestinal tract of several mammalian species. Cell Tissue Res. 239: , Mundinger, T. O., and G. J. Taborsky, Jr. Activation of hepatic sympathetic nerves during hypoxic, hypotensive and glucopenic stress. J. Auton. Nerv. Syst. 63: , Myers, S., O. P. McGuinness, D. W. Neal, and A. D. Cherrington. Intraportal glucose delivery alters the relationship between

9 E1202 GALANIN IN HEPATIC SYMPATHETIC NERVES net hepatic glucose uptake and the insulin concentration. J. Clin. Invest. 87: , Peuler, J. D., and G. A. Johnson. Simultaneous single isotope radioenzymatic assay of plasma norepinephrine, epinephrine and dopamine. Life Sci. 21: , Rokaeus, A., and M. Carlquist. Nucleotide sequence analysis of CDNAs encoding a bovine galanin precursor protein in the adrenal medulla and chemical isolation of bovine gut galanin. FEBS Lett. 234: , Sawchenko, P. E., and M. I. Friedman. Sensory functions of the liver a review. Am. J. Physiol. 236 (Regulatory Integrative Comp. Physiol. 5): R5 R20, Skofitsch, G., and D. M. Jacobowitz. Immunohistochemical mapping of galanin-like neurons in the rat central nervous system. Peptides 6: , Taborsky, G. J., Jr., and T. O. Mundinger. Differential actions of hepatic sympathetic neuropeptides. Diabetes 46: Verchere, C. B., S. Kowalyk, D. J. Koerker, D. G. Baskin, and G. J. Taborsky, Jr. Evidence that galanin is a parasympathetic, rather than sympathetic, neurotransmitter in the baboon pancreas. Regul. Pept. 67: , Verchere, C. B., S. Kowalyk, G. H. Shen, M. R. Brown, M. W. Schwartz, D. G. Baskin, and G. J. Taborsky, Jr. Major species variation in the expression of galanin messenger ribonucleic acid in mammalian celiac ganglion. Endocrinology 135: , Walker, L. C., N. E. Rance, D. L. Price, and W. S. I. Young. Galanin mrna in the nucleus basalis of meynert complex of baboons and humans. J. Comp. Neurol. 303: , Woods, S. C., G. J. Taborsky, Jr., and D. J. Porte, Jr. Central nervous system control of nutrient homeostasis. In: Handbook of Physiology. The Nervous System. Intrinsic Regulatory Systems of the Brain. Bethesda, MD: Am. Physiol. Soc., 1986, sect. 1, vol. IV, chapt. 7, p