Anatomic studies show that the common hepatic branch (CHB) of the vagus contains afferent fibers that innervate sites outside the hepatoportal region, primarily in the gastrointestinal tract. In the current experiments on the anesthetized rat, the source of signals from the CHB was determined by recording CHB neurophysiological responses before and after transection of the gastroduodenal branch (GDB) of the CHB. Serotonin [5-hydroxytryptamine (5-HT)] and CCK-8 were used as probes to stimulate the CHB. Most of the CHB afferent fibers were 5-HT sensitive (56%), and 35% of these were also sensitive to CCK-8. Portal vein vs. jugular vein infusion of 5-HT and CCK-8 and GDB transection showed that 5-HT- and CCK-sensitive fibers innervate the hepatoportal region and areas outside the hepatic hilus (e.g., the gastrointestinal tract). Suppression of basal nerve activity by a 5-HT3 receptor antagonist (Y-25130) suggests that ∼50% of CHB afferent fibers contain 5-HT3 receptors, but none of these fibers appears to be in the hepatoportal region because only in rats with an intact GDB did Y-25130 reduce nerve activity. In summary, these data are in close agreement with anatomic observations on the distribution of the CHB fibers and indicate that neurophysiological studies of the CHB must be carefully evaluated given the prominent role of nonhepatoportal afferent signals recorded from the CHB.
for many years it was commonly thought that the common hepatic branch (CHB) of the vagus was solely a “hepatic” nerve. However, anatomic work showed that the CHB innervates mostly the gastrointestinal (GI) tract, and secondarily the hepatoportal region (3). Although there have been many studies on the behavioral effects of CHB ablation and neurophysiological recording experiments on the CHB (e.g., Refs. 1, 9, 11, 23), little attention has been directed toward separating the roles of hepatic from GI afferent fibers in these studies.
The CHB of the vagus is in a unique position to provide sensory input to the brain from organs involved in nutrition. Ablation of the CHB alters food intake and blocks food aversion produced by an amino acid deficiency (9, 11). CHB afferent fibers are sensitive to carbohydrates, fats, amino acids, hormones, and a large number of signaling factors infused into the portal vein (19–23, 29, 30). Although the responses to portal vein infusion of neurochemicals and nutrients have been attributed to stimulation of hepatic afferent fibers, it is quite plausible that these data represent the activation of GI afferent fibers because chemicals infused into the portal vein can stimulate organ sites outside the liver after entering the general circulation. None of the prior experiments addressed the potential role of nonhepatoportal CHB afferent fibers in the neurophysiological responses of the CHB (19–23, 29, 30).
In the current experiments on the rat, the source of signals from the CHB was assessed by recording neurophysiological responses from the CHB before and after transection of the gastroduodenal branch (GDB) of the CHB. The GDB is a subbranch of the CHB that contains fibers that primarily innervate the upper region of the duodenum and to a lesser extent the stomach and pancreas (3). Recently enhanced neurophysiology techniques were used to record from the CHB (16). Multiunit records, as well as single-unit data, are reported to provide a much larger insight into the population of afferent fibers in the CHB. Two methods were used to assess the contribution of the hepatoportal and GI components to these recordings: GDB ablation and comparisons of nerve activation produced by portal vein vs. jugular vein infusion. Serotonin (5-HT) and cholecystokinin (CCK-8) were used as probes to activate vagal afferent fibers due to their ubiquitous involvement in vagal afferent signaling (e.g., Refs. 15, 28, 31).
MATERIALS AND METHODS
Surgery, Anesthesia, and Chemicals
Experiments were conducted on male CD rats (350–450 g; Charles River, Kingston, NY). Surgery, anesthesia, nerve recording, data acquisition, and data analysis were performed as described previously (16). Rats were anesthetized with pentobarbital sodium (50 mg/kg ip) followed by a continuous infusion of sodium methohexital (Brevital; 25 μl/h) into the left femoral vein. Animals were paralyzed with pancuronium bromide (0.4 mg/h iv) and artificially respirated (95% O2-5% CO2) at 50 cycles/min using constant-volume ventilation. Body temperature was kept at 37°C by a regulated heating pad. Heart rate, blood pressure (right carotid artery), and expired CO2 were continuously monitored to ensure the physiological stability of the preparations. A jugular vein catheter (placed at 1 cm from the heart in the left jugular vein) and in some preparations a portal vein catheter were implanted to test the effects of 5-HT and CCK-8 on vagal afferent activity. The tip of the portal vein catheter was advanced through the ileocecal vein to a point (∼6 cm) at which it would not move further, and then the catheter was withdrawn 4 cm. This technique of portal vein catheter placement was used to ensure that nerve fibers innervating the portal vein and liver would have adequate access to infusates.
5-HT (10 μg; Sigma), CCK-8 (100 pmol; Bachem), and saline (0.15 M) were infused into the jugular vein and portal vein in volumes of 0.5 ml over 30 s. A specific 5-HT3 antagonist, Y-25130 (Tocris) (12, 13), was infused into the jugular vein for 30 s (0.8 mg/0.8 ml). In two preparations the duration of the effect of Y-25130 on vagal afferent activity was assessed by repeated testing of the response to 5-HT infusion. Saline was used to flush the portal and jugular vein catheters between infusions of 5-HT, CCK-8, and Y-25130. There was at least a 5-min interval between infusions. Animals were euthanized at the conclusion of an experiment by jugular vein infusion of pentobarbital sodium (60 mg). All experiments conformed to established standards of the National Institutes of Health and the Monell Center’s Institutional Animal Care and Use Committee.
Nerve Recording from the CHB of the Vagus
An incision from the top of the xiphoid process to the lower abdomen was made to expose the viscera. The edges of the wound were retracted and elevated to create a large ovoid cavity. After retraction of the stomach caudally and displacement of the liver to right of the animal, the body cavity was filled with warm (37°C) mineral oil, and the electrode platform was lowered into place. The ventral trunk of the vagus was cut centrally ∼5 mm rostral to origin of the CHB and pinned to the recording platform containing four recording electrodes. To eliminate neural signals from other vagal branches, the ventral gastric and accessory celiac branches of the vagus were also transected below the bifurcation of the CHB from the ventral trunk of the vagus.
The nerve was pinned into position across a reference electrode. One to four small nerve filaments (a single nerve filament contains many axons from different neurons) were teased away from the trunk and wrapped around the recording electrodes using fine forceps. Nerve signals were amplified 5–20 K using AC amplifiers (Grass P511; Astromed) with low-frequency (100 Hz) and high-frequency (3 kHz) cutoffs. Neural activity, body temperature, heart rate, blood pressure, and expired CO2 data were saved to computer hard disk using data-acquisition hardware and software (DataWave Tech, Longmont, CO). Nerve signals were digitized at 32 kHz per electrode channel.
In preparations in which the GDB was transected, two loose sutures were placed around the connective tissue between the hepatic hilus and the duodenum. The GDB was lesioned by tightening the sutures and completely transecting the tissue between the sutures with scissors. Confirmation that this connective contained the GDB was confirmed by blunt probing of the serosal surface of the duodenum and stomach. In animals with intact connectives, mechanical probing of the GI tract elicited a robust short-latency increase in CHB responding; however, this mechanoafferent responding was absent in all animals with transected connectives.
Test of Blood Flow After GDB Ablation
The GDB and gastroduodenal artery travel in parallel, and it was therefore necessary to cut the artery along with the nerve. Because cutting the artery might limit the flow of systemically infused chemicals to the intestine and prevent agents from reaching afferent neuron endings in that tissue, this could potentially confound the interpretation of experiments involving the nerve cut. To address this issue, 1 ml of methylene blue dye (0.5%) was infused into the jugular vein over 1 min. Three minutes after the infusion, the duodenum (a 2-cm section distal to the pyloric sphincter) and the liver were excised, weighed, and stored frozen. Each tissue sample was diluted with distilled water (1 g tissue per 4 ml water) and homogenized. The homogenate was diluted further (1 ml homogenate per 9 ml water) and vortexed. This mixture was centrifuged at 10,000 rpm, and the supernatant was assayed in a spectrophotometer at λ = 607 nm (wavelength of maximal absorption for methylene blue).
Spike waveforms were extracted from the continuous stream of data using amplitude threshold levels (CP Analysis; DataWave). When spikes crossed the threshold, 3 ms (1 ms before and 2 ms after the peak) of the waveforms were extracted.
Data were processed using principal component analysis (Off-line Sorter; Plexon). The approximate centers of clusters (putative single unit scores) were selected manually, and the k-means method (10) was used to automatically select data for inclusion into clusters. Cluster assignments were checked using interspike interval histograms and cross-correlation analysis. A cluster was determined to be single-unit activity when 1) it revealed a refractory period of not less than 3 ms, and 2) no significant cross-correlations were found with any other data cluster (a significant correlation with another cluster might indicate that the spikes from the 2 clusters were really from the same nerve fiber). Spike count histograms and 95% confidence intervals were constructed for 400-s bins of data (100 s before infusion and 300 s after the start of infusion of 5-HT and CCK-8) (NeuroExplorer; NEX Technologies, Littleton, MA). A single unit was determined to be sensitive to 5-HT or CCK-8 if its change in activity exceeded the 95% confidence interval.
In multiunit recordings, spikes were computed for 5-s bins using NeuroExplorer software. Data from different animals were then grouped to compute averages and SEs of the mean. Data were analyzed using ANOVA, and planned comparisons between means were conducted using the least significance difference test (Statistica; StatSoft). Planned comparisons were conducted by comparing the first mean in the analysis (time = −50 s from treatment) to all other means. A criterion of P < 0.05 was used to indicate statistical significance.
CHB Single Units Sensitive to 5-HT and CCK-8
Fifty single units were isolated from 28 CHB nerve filaments in 20 animals. Jugular vein infusion of 5-HT and CCK-8 evoked short-latency responses from the CHB (see Fig. 1A). Fifty-six percent (28 of 50) of the single units were activated by 5-HT treatment, and 35% of these were also stimulated by CCK-8 (10 of 28; e.g., see unit 1 in Fig. 1B). Only a few fibers were sensitive to CCK-8 but not 5-HT (14%, 7 of 50; e.g., see unit 2 in Fig. 1B), and a significant number was not affected by either 5-HT or CCK-8 treatment (30%, 15 of 50; see Fig. 1C).
Site of Action for 5-HT and CCK-8 Activation of the CHB
Portal vein vs. jugular vein comparisons.
Nine nerve filaments from four animals contributed to the multiunit analysis of CHB activity during saline, 5-HT, and CCK-8 infusion into the portal vein and jugular vein. Each animal was tested in all six conditions (site of infusion by infused agent; 2 × 3). Saline infusion elicited no effects on CHB activity, but 5-HT and CCK-8 produced pronounced activation of the CHB by 10–25 s after infusion (see Fig. 2) [a significant 3-way interaction between agent (saline, 5-HT, and CCK-8), site of infusion (portal vs. jugular vein), and time (30 5-s bins; −50 to 100 s); F(58,464) = 2.5, P < 0.0001, ANOVA]. The effects of 5-HT and CCK-8 were greater when infused into the jugular vein than when administered into the portal vein. CCK-8 treatment also caused a longer activation of the CHB when infused into the jugular vein than it did when delivered by portal vein infusion.
Sixteen nerve filaments from 12 animals contributed to the single-unit analysis. Twenty-eight single units were analyzed (15 = 5-HT sensitive, 8 = CCK-8 sensitive, and 5 = 5-HT and CCK-8 sensitive). There were small differences between portal vein and jugular vein infusion in the latency for activation of single-unit activity (see Fig. 3A). To more closely assess differences between portal and jugular vein infusion, recordings for portal vein infusions were subtracted from the jugular vein infusion records; these records are referred to as “transformed records.” Nineteen of the 28 units showed a short latency (10–25 s) peak in the transformed records after infusion of 5-HT or CCK-8, which indicates that jugular vein infusion yielded greater amplitude or was shorter in latency than portal vein infusion (since this suggests that the recorded nerve fiber is stimulated at a site that is outside the liver, perhaps in the GI tract, these CHB units are referred to as GI 5-HT- or CCK-8-sensitive units). In contrast, only three units (of 28) showed a short latency valley after infusion of 5-HT or CCK-8, which indicated that portal vein infusion yielded greater amplitude or was shorter in latency than jugular vein infusion (since this suggests that a recorded nerve fiber is stimulated at a site within the liver or portal vein, these units are referred to as hepatoportal 5-HT- or CCK-8-sensitive units). Representative samples of these transformed records are shown in Fig. 3B. The samples in Fig. 3B were selected from the 5-HT- or CCK-8-sensitive units but not units sensitive to both 5-HT and CCK-8. Units sensitive to both 5-HT and CCK-8 showed only short-latency peaks, i.e., GI 5-HT- or CCK-8-sensitive units, in the transformed records. There were no transformed records of CCK-8-only sensitive fibers that showed a short-latency valley, i.e., hepatoportal CCK-8-sensitive units.
Six nerve filaments from four animals contributed to the multiunit analysis of CHB activity during 5-HT and CCK-8 jugular vein infusion before and after cutting the GDB. CHB responses to 5-HT and CCK-8 were greatly diminished after cutting the GDB (see Fig. 4) [a significant 3-way interaction between nerve cut (cut/no cut), agent (5-HT or CCK-8) and time (30 5-s bins; −50 to 100 s); F(29,145) = 3.1, P < 0.0001, ANOVA]. CCK-8 produced a longer duration effect on CHB activation when the GDB was intact (see Fig. 4).
Intestinal blood flow after cutting the gastroduodenal artery.
The three groups of animals consisted of control animals with no dye infusion (normal control; n = 4), control animals with dye infusion (n = 5), and GDB-ablated animals with dye infusion (n = 5). To eliminate background absorbance, the mean absorbance measures for the normal control group, not infused with dye (intestine = 8.9% and liver = 6.6% absorbance), were subtracted from the absorbance values for animals with dye infusion. There were no significant differences between control and GDB-ablated animals in the amount of dye detected in the duodenum (control = 16.3 ± 10.9% and cut = 12.9 ± 7.1% absorbance; 1-tailed t-test, P = 0.4) or liver (control = 6.9 ± 1.8% and cut = 3.6 ± 0.7% absorbance; 1-tailed t-test, P = 0.06). Although it was not possible to excise the diffuse pancreatic tissue, dye was observed in the pancreas in both control and GDB-ablated animals.
Role of 5-HT3 Receptors in Basal and Evoked CHB Activity
Sixteen nerve filaments from 10 animals were used for this analysis. All nerve filaments were tested with jugular vein infusion of 5-HT before and after jugular vein infusion of the 5-HT3 receptor antagonist Y-25130; however, only eight of these nerve filaments (from 5 animals) were also tested with jugular vein infusion of CCK-8. There was a significant interaction between antagonist (Y-25130 or saline), time (30 5-s bins, −50 to 100 s), and agent (5-HT or CCK-8) [F(29,203) = 5.1, P < 0.0001, 3-way interaction effect]. CHB responses to 5-HT were greatly diminished after infusion of Y-25130; however, responses to CCK-8 were unchanged by the 5-HT3 antagonist (see Fig. 5). In two nerve filaments from two animals the current dose of Y-25130 resulted in a suppression of 5-HT responses for >1 h, which indicates a long duration of action of Y-25130 on 5-HT3 receptors. Note that in Fig. 5, CHB responses to 5-HT and CCK were tested within 15 min of infusion with Y-25130.
Contribution of GDB 5-HT3 Receptors to CHB Activity
To assess the effects of 5-HT3 antagonism on CHB activity without input from the GDB, 14 CHB nerve filaments were examined in five animals. Y-25130 was administered 5–10 min after cutting the GDB. GDB ablation produced a significant reduction in CHB activity [F(29,377) = 6.6, P < 0.0001, 1-way repeated-measures ANOVA]. Y-25130 had no effect of CHB activity after GDB nerve cut (P > 0.05, 1-way ANOVA) (see Fig. 6).
The current studies indicate that most of the neural activity recorded from the CHB is not derived from nerve fibers innervating the liver or portal vein. Cutting the GDB led to a 65–80% (see Figs. 4 and 6) decrease in basal response rate from the CHB. It is most probable, based on anatomic studies of the CHB, that this GDB input is specifically from the first segment of the duodenum (3, 24). The GDB component of the CHB appears to consist mainly of nerve fibers with 5-HT3 receptors. The 5-HT3 receptor antagonist Y-25130 resulted in a large decrease in basal and 5-HT-evoked CHB responding (see Fig. 5), and Y-25130 did not produce a change in CHB activity after transection of the GDB (see Fig. 6). Not all of the 5-HT activation of the CHB is due to 5-HT3 receptors because Y-25130 produced only an ∼80% reduction in the multiunit response to 5-HT treatment (see Fig. 5). It is possible that 5-HT also stimulates CHB neurons with 5-HT4 receptors, which have been observed in the vagal system (5, 6). CCK-8 activation of the CHB was not affected by pretreatment with Y-25130 (see Fig. 5).
These experiments suggest potential strategies for distinguishing neural responses from the CHB based on site of activation. Jugular vein vs. portal vein infusion of 5-HT and CCK-8 produced neural activation patterns that suggest that nerve fibers innervate the GI tract and the hepatoportal region (see Fig. 3). However, few liver/portal vein units were found using jugular vein and portal vein infusion comparisons. A significant number of single units (11 of 28 units) with no distinct differences between jugular and portal vein infusion was observed. A lack of a difference between jugular and portal vein infusion may indicate that these nerve fibers have processes in both the hepatic region and the GI tract. Nerve fiber counts at different levels of the vagus suggest that many vagal nerve fibers collateralize within the peritoneal cavity (25). To produce greater resolution in neural responses between jugular and portal vein infusion, the present infusion comparison strategy may be enhanced by using lower doses of agents and slower infusion rates.
A single-unit analysis of CHB activity showed that most units were 5-HT sensitive. This is in close agreement with other reports that show 5-HT sensitivity as a prominent feature of vagal signaling (4, 14). The results also support earlier work showing that CCK-8 activates CHB units recorded from the cervical vagus but identified as CHB fibers by electrical stimulation of the CHB (7). In another study jejunal mesenteric afferent fibers were shown to have independent pathways for activation by 5-HT and CCK (15); however, in the current experiments a subgroup of vagal units (10 of 50 CHB single units) showed activation to both 5-HT and CCK-8. This discrepancy may reflect a difference in the afferent fibers that were recorded because much of the jejunum is innervated by the celiac branch of the vagus and not the CHB, and mesenteric nerve bundles also contain spinal and intestinofugal fibers (2, 15). Dual activation of vagal afferent fibers by 5-HT and CCK-8 may explain why odansetron, a 5-HT3 antagonist, attenuates the suppression of feeding behavior produced by CCK treatment (8).
The gastoduodenal artery, a small branch of the celiac artery, was cut along with GDB in ablation experiments, which might have affected the flow of infused chemicals, e.g., 5-HT, to intestinal and hepatic sites. The results of dye infusion experiments show that there was no significant disruption of blood flow to the duodenum and liver after transection of the artery. This indicates that a change in blood circulation does not account for the large effects observed after GDB ablation for induced (5-HT and CCK infusion) and basal CHB nerve activity.
Multi- and single-unit recordings were used in the current study to obtain a more complete record of CHB activity. Single-unit recordings are ideal because they allow for a much finer analysis of neuronal characteristics. However, there are two reasons why it can be important to supplement single-unit data with multiunit data. First, a significant limitation of single-unit studies is that only a few fibers can be recorded due to the arduous process of isolating single fiber activity. For example, the number of single fibers recorded in the vagus literature is characteristically <30 (e.g., Refs. 7, 27). Although high-yield recording techniques were used to isolate 50 single units in the current study, this represents only a small portion of the several thousand fibers in the CHB (2, 25). Multiunit recordings of bundles of axons allow for a much broader sampling of neural activity from the CHB. Second, it is plausible that some of the smaller-diameter fibers that were not isolated in single-unit recordings would be sampled in multiunit experiments, resulting in less sampling bias. It might be argued that multiunit data are difficult to interpret because there can be some variability or drift in the recording over time. This is a common problem to both multi- and single-unit recordings. There is indeed drift in many neurophysiological recordings during the first 10–20 min. However, in the present studies, which were conducted after a delay of 30 min, the recordings were very stable, as noted by the lack of basal spike rate changes. Furthermore, in the present experiments the use of mean group data should effectively control for any remaining variability between recordings.
In summary, the present experiments indicate that care must be taken in conducting and analyzing neurophysiological studies of the CHB because this nerve branch receives multiple inputs from different organ sites, most of which are not from the liver or portal vein. Past work that has emphasized the hepatic component of CHB recordings may need to be reinterpreted in light of the current findings because no attempts were made to separate the contribution of nonhepatic influences in prior studies (7, 17–19, 21, 22, 26, 30). It is recommended that neurophysiological studies on the CHB routinely incorporate GDB ablation and jugular vs. portal vein infusion to distinguish the neural signals of interest.
This work was supported by National Institute of Diabetes and Digestive and Kidney Diseases Grants DK-02894 and DK-36339.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2004 the American Physiological Society