THE ANATOMICAL RECORD PART A 280A:827– 835 (2004)

Anatomy and Function of Sensory

Hepatic Nerves
HANS-RUDOLF BERTHOUD*
Neurobiology of Nutrition Laboratory, Pennington Biomedical Research Center,
Louisiana State University, Baton Rouge, Louisiana

ABSTRACT
Vagal and spinal afferent innervation of the portal hepatic area has not
been studied as thoroughly as the innervation of other important organs. It
is generally agreed that unlike noradrenergic sympathetic efferent nerve
fibers, sensory nerve fibers of either vagal or dorsal root/spinal origin do not
directly innervate hepatocytes, but are restricted to the stroma surrounding
triades of hepatic vasculature and bile ducts, and to extrahepatic portions of
the portal vein and bile ducts. For vagal afferent innervation, retrograde
and anterograde tracing studies in the rat have clearly shown that only a
minor portion of the common hepatic branch innervates the liver area, while
the major portion descends in the gastroduodenal branch toward duode-
num, pancreas, and pylorus. Hepatic paraganglia, bile ducts, and portal
vein receive the densest vagal afferent innervation. Calretinin may be a
relatively specific marker for vagal afferent innervation of the portal-he-
patic space. Calcitonin gene-related peptide (CGRP) is a specific marker for
dorsal root afferents, and CGRP-immunoreactive fibers are mainly present
near the intrahepatic vascular bundles and bile ducts, and in the same
extrahepatic compartments that contain vagal afferents. Because of the
specific anatomical organization of hepatic nerves, selective hepatic dener-
vation, whether selective for the vagal or sympathetic division, or for effer-
ents and afferents, is nearly impossible. Great caution is therefore neces-
sary when interpreting functional outcomes of so-called specific hepatic
denervation studies. © 2004 Wiley-Liss, Inc.

Key words: liver; portal vein; bile ducts; vagus; sympathetic in-
nervation; parasympathetic innervation; dorsal
root afferents; cholinergic innervation, calcitonin
gene-related peptide; premotor areas; central auto-
nomic control

Like every important organ, the liver communicates
with the rest of the body through both humoral and neural
signals. The brain is potentially “interested” in a great
deal of sensory information related to the metabolic, im-
mune, and other functions of the liver (Langhans, 1996;
Friedman, 1997; Goehler et al., 2000) and in turn can
modulate these functions through hormones and neural
pathways (Tracey, 2002; Obici and Rossetti, 2003). Neural
communication between the liver and the central nervous
system is bidirectional, with afferent (sensory) and effer-
ent (motor) nerve fibers of both vagal parasympathetic
and spinal sympathetic origin. Neural communication be-
tween the liver and the enteric nervous system is also
indicated by functional studies (Stümpel et al., 2000) but
has not yet been anatomically identified.
Basic knowledge about the wiring of hepatic sensory
nerve fibers is important for the design of experiments
involving neural ablation and other manipulations, and
for the possible outcome of liver transplantations. Another
Grant sponsor: the National Institutes of Health; Grant num-
ber: DK47348 and DK57242.
*Correspondence to: Hans-Rudolf Berthoud, Pennington Bio-
medical Research Center, 6400 Perkins Road, Baton Rouge, LA
70808. Fax: 225-763-0260. E-mail: berthohr@pbrc.edu
Received 28 June 2004; Accepted 28 June 2004
DOI 10.1002/ar.a.20088
Published online 23 August 2004 in Wiley InterScience
(www.interscience.wiley.com).

© 2004 WILEY-LISS, INC.

828 BERTHOUD
major interest is the identity of the adequate stimulus and
the signal transduction cascade in the periphery (the
liver), as well as the central nervous pathways that trans-
form such sensory signals into autonomic, endocrine, and
behavioral adjustments or actions.
GENERAL ANATOMY AND MAJOR ACCESS
ROUTES OF PORTAL-HEPATIC NERVES
According to classical autonomic physiology, parasym-
pathetic innervation of the upper abdominal organs is
accomplished by the vagus nerve. Afferent neurons are
located in the nodose ganglia, with axonal processes to the
periphery and central processes to the nucleus of the sol-
itary tract and other areas of the dorsal vagal complex.
These neurons contain a variety of transmitters, peptides,
and receptors, with glutamate generally released at their
central terminals. Vagal preganglionic efferent or motor
neurons with upper abdominal projections are located in
the dorsal motor nucleus of the vagus. Although most of
these neurons are cholinergic, up to 10% have a cat-
echolaminergic phenotype (Gwyn et al., 1985; Armstrong
et al., 1990; Guo et al., 2001; Hayakawa et al., 2004). The
general branching patterns of the two subdiaphragmatic
vagal trunks and their central efferent and afferent rep-
resentations have been described in detail (Shapiro and
Miselis, 1985; Norgren and Smith, 1988; Berthoud et al.,
1991). These vagal preganglionic neurons terminate on
classically called postganglionic neurons located within
the innervated organs. For the gut, these ganglia are
located in the enteric plexuses (Berthoud, 1995, 1996), and
for the pancreas, in the interlobular pancreatic ganglia
(Berthoud et al., 2001), but with respect to the liver, the
exact location of these parasympathetic ganglia is still not
known (Berthoud and Neuhuber, 1998). No intrahepatic
ganglia have been found in the rat. Although many of
these postganglionic neurons are cholinergic, others con-
tain nitric oxide, vasoactive intestinal peptide, gastrin-
releasing peptide, and possibly other transmitters
(Berthoud, 1996; Berthoud et al., 2001).
On the sympathetic side, the upper abdominal organs
are innervated by preganglionic neurons located in the
intermediolateral column of the thoracic spinal cord, with
postganglionic neurons in the prevertebral ganglia such
as the celiac, superior mesenteric, and perhaps adrenal
ganglia. Generally, the preganglionic neurons are cholin-
ergic and the postganglionic neurons are noradrenergic
and those innervating blood vessels also express neu-
ropeptide Y (NPY). The dorsal root sensory fibers, which
are strictly speaking not part of the sympathetic nervous
system, are the equivalent of the bipolar vagal afferents,
with cell bodies in the dorsal root ganglia (DRG), and
axonal processes in the periphery and spinal cord. Figure
1 schematically depicts vagal and spinal afferent innerva-
tion of the liver and associated organs, including its cen-
tral representations. The organization of vagal and spinal
efferent innervation of this area has been described else-
where (Berthoud and Neuhuber, 1998).
The major and possibly only neural access route to the
liver is through the porta hepatis or liver hilus, along the
hepatic artery, portal vein, and bile ducts. An important
issue is the species specificity of this innervation. The
following review is mainly based on literature on rats,
with the occasional comparison with other mammals and
humans.
Several experimental approaches can be used to char-
acterize hepatic innervation anatomically and function-
ally by specific classes of neurons: retrograde and antero-
grade tracing, immunohistochemistry, electrophys-
iological recording, and selective ablation, each with its
strengths and limitations.
VAGAL AFFERENT INNERVATION
Retrograde Tracing Studies
Retrograde tracing studies in rats using injections of the
tracer into various hepatic compartments and analyzing
labeled neurons in the nodose (sensory vagal) ganglia led
to the conclusion that many of the vagal afferent fibers
traveling in the common hepatic branch do not terminate
in the liver, and that among those terminating in the liver,
very few penetrate into the liver lobe parenchyma (Magni
and Carobi, 1983). Injections of horseradish peroxidase
directly into the hepatic branch resulted in the largest
number of retrogradely labeled neurons in the left nodose
ganglion. Considerably fewer neurons were labeled follow-
ing injections into the wall of extrahepatic bile ducts and
the hilus area, and virtually no neurons were labeled
following multiple injections into the hepatic parenchyma
(Magni and Carobi, 1983). The study also demonstrated
that although much fewer, some labeled neurons were
found in the right nodose ganglion after injections into the
liver hilus, suggesting that not all vagal afferent fibers
innervating the liver run through the common hepatic
branch (Magni and Carobi, 1983). In another study, label-
ing was reported in the nodose ganglia following injection
of the retrograde tracer True Blue into the tissue sur-
rounding hepatic artery and portal vein (Barja and Ma-
thison, 1984).
Anterograde Tracing Studies
The anatomical method of choice for investigating vagal
afferent nerve fibers and terminals in the liver is antero-
grade tracing from the nodose ganglia, because there is no
unique neurochemical marker for vagal afferent fibers.
Anterograde tracing with the orange fluorescent carbocya-
nine dye DiI was able to reveal the exact location and
structure of vagal afferent terminals in the various tissue
compartments and essentially confirmed the conclusions
based on retrograde tracing (Berthoud et al., 1992). How-
ever, this method has limitations, too, in that it is difficult
to label vagal afferents in their entirety and there is no
easy verification for completeness.
When tissue blocks containing the hepatic arterial tree
were inspected in whole mounts, it became clear that most
of the DiI-labeled fibers running in the common hepatic
branch did not enter the liver hilus, but turned toward the
bifurcation of the hepatic and gastroduodenal arteries to
follow the latter toward pylorus, duodenum, and pancreas
(Fig. 2A) (Berthoud et al., 1992). This descending gas-
troduodenal component of the common hepatic branch has
been well documented anatomically and functionally
(Stavney et al., 1963; Tiscornia et al., 1976; Matsuo and
Seki, 1978; Berthoud et al., 1991; Phillips et al., 1997).
These observations reinforce the conclusions drawn from
the retrograde studies above (Magni and Carobi, 1983),
that vagal afferent fibers traveling in the common hepatic
branch predominantly innervate tissues other than the
liver and immediately adjacent to the liver.
The highest density of DiI-labeled terminal structures
was found in the wall of bile ducts as they emerge from the
 SENSORY HEPATIC NERVES
Fig. 1. Schematic diagram showing anatomical organization of vagal
and spinal afferent innervation of the rat liver and associated organs and
their central representations. Vagal afferent innervation is shown in blue
solid lines, spinal afferent innervation in red broken lines. Relative abun-
dance of innervation is indicated by thickness of axons and diameter of
dots representing sensory terminals in the various compartments. Dis-
tribution of vagal afferent fibers is based on retrograde (tracer injections
in the liver and associated organs) and anterograde (tracer injected into
the nodose ganglia) tracing. The vagal afferent innervation is predomi-
nantly through the left nodose ganglion and vagus and common hepatic
branch, with a minor contribution through the right nodose ganglion and
vagus/celiac branch. The major portion of the common hepatic branch
829
does not serve the liver, portal vein, and bile ducts, but innervates
duodenum, pancreas, pylorus, and distal gastric antrum. Distribution of
spinal/dorsal root afferents is based on retrograde tracing and CGRP
immunohistochemistry. Spinal sensory innervation is accomplished
through dorsal root afferents at thoracic spinal levels T7 through T13.
Note the very sparse or absent innervation of the hepatic parenchyma by
sensory nerve terminals in the rat. Central representation of vagal affer-
ent input is partly based on electrophysiological recording and c-Fos
studies. Central representation of dorsal root afferents has not been
specifically determined for hepatic innervation and is depicted as gen-
eralized for visceral spinal input.
830 BERTHOUD
Fig. 2. Vagal afferent innervation of the portal hepatic space as
revealed by DiI-tracing from the nodose ganglia in the rat. A: Montage of
low-magnification images from whole-mounted tissue block showing
two major fascicles of the intensely DiI fluorescent (bright white) com-
mon hepatic branch with associated paraganglia (PG) coursing from the
ventral vagal trunk on the esophagus toward the hepatic artery proper
(AHP) along the hepatoesophageal artery (AHE). The major fascicle
becomes the gastroduodenal branch, which bypasses the liver hilus and
descends toward the gastroduodenal artery. The smaller fascicle, the
hepatic branch proper, turns toward the liver hilus, giving off smaller
hepatic lobes, with only the occasional fiber penetrating
deeper into the liver along the triades of hepatic artery,
portal vein, and bile ducts (Fig. 2B and C). No evidence
was found for direct intraparenchymal innervation in this
study. Locally dense accumulations of terminal-like fibers
were also found in the adventitia of the portal vein as it
divides and enters the liver proper (Fig. 2D) (Berthoud et
al., 1992). In addition, many sensory terminals were found
in the paraganglia nestled between and within the fasci-
cles of the common hepatic branch as it travels along the
hepatoesophageal artery from the ventral subdiaphrag-
matic vagal trunk on the esophagus toward the hepatic
artery proper (Prechtl and Powley, 1987; Berthoud et al.,
1992; Berthoud and Patterson, 1996).
Direct innervation of hepatocytes could have been ex-
pected in the rat on the basis of functional evidence for
various hepatic sensors in this species. It may be that
intralobular parenchymal innervation in other species
such as the guinea pig, primates, and humans, is not
limited to sympathetic fibers, but also includes vagal sen-
sory fibers. However, a similar anterograde tracing study
has not yet been done in these species.
Immunohistochemistry
Since there is no generally specific neurochemical
marker known to be expressed in vagal afferent fibers, it is
fascicles innervating portal vein, bile ducts, and the different liver lobes.
B: Cross-section of major intrahepatic artery (AHP) showing several
DiI-labeled (bright white) profiles of vagal afferent fibers running in the
surrounding connective tissue. C: Tissue section from hepatic hilus area
showing DiI-labeled vagal afferent fiber terminals near the border be-
tween connective tissue surrounding a major hepatic artery (AHP) and
the hepatic parenchyma (HP). Note that no vagal afferent fibers are
penetrating into the parenchyma in the rat. D: Whole-mounted portal
vein showing vagal afferent fiber terminal (bright white) running in the
fibrous adventitia.
difficult from such experiments to draw conclusions about
vagal afferent identity. However, there may be reasonably
predictive markers for vagal afferents in specific regions
and specific terminal structures. The calcium binding pro-
tein calretinin is a fairly good marker for vagal afferent
innervation of the esophagus (Dutsch et al., 1998), and
both P2X2/X3 (Wang and Neuhuber, 2003) and vesicular
glutamate transporter (Raab and Neuhuber, 2003) can be
used as selective markers for esophageal vagal intragan-
glionic laminar endings (IGLEs). The fact that calretinin-
immunoreactive fiber distribution and terminal structure
in the liver, portal vein, bile ducts, and paraganglia show
many similarities to DiI-traced fibers, together with the
absence of calretinin in dorsal root afferents, suggests that
calretinin might be a relatively specific marker for vagal
afferents in this area (Berthoud and Neuhuber, 1996).
SPINAL AFFERENT INNERVATION FROM
DORSAL ROOT GANGLIA
Since the last reviews on this topic appeared, not much
progress has been made on a more comprehensive identi-
fication of dorsal root sensory neurons innervating the
liver and associated organs. Clearly, calcitonin gene-re-
lated peptide (CGRP) is an excellent marker for these
afferents, and their distribution in the various liver com-
partments and associated organs has been shown exten-
 SENSORY HEPATIC NERVES
sively and in many different species, including humans.
However, many questions regarding target-specific map-
ping and phenotyping of neurons in the dorsal root gan-
glia, as well as their projections to the spinal cord, remain
unanswered. In other words, we do not know how dorsal
root sensory neurons projecting to the liver are different
from those projecting to the gastrointestinal tract or pan-
creas. These questions can only be answered in future
experiments combining retrograde and anterograde trac-
ing techniques with electrophysiological recording.
Retrograde Tracing Studies
Our knowledge is based exclusively on a series of retro-
grade tracing experiments with inspection of the DRGs
but not the spinal cord (Magni and Carobi, 1983). Large
multiple injections of HRP into the parenchyma surround-
ing the hilus yielded relatively few labeled neurons pri-
marily in the left DRGs of Th7 to Th11. Large injections
into the intrahepatic bile ducts yielded about twice as
many neurons with a similar distribution and peak label-
ing occurring at the level of T8 to T10. These data are
consistent with the paucity of CGRP-IR fibers in the pa-
renchyma of rat liver. They suggest that the central ter-
minals and the second-order sensory neurons in the spinal
cord are coextensive with DRG afferents innervating
stomach and pancreas between thoracic levels T7 to T13.
However, neither the central axon terminals of DRG af-
ferents innervating the liver, portal vein, and biliary tree,
nor the second-order neuronal perikarya in the spinal
cord, have been directly identified by tracing, electrophys-
iological, or Fos-induction methods. It is interesting to
note that only about 10% of all afferent fibers in the
thoracic dorsal roots are of visceral origin (some 6,000 –
7,000 fibers in the cat greater splanchnic nerve); the other
90% are somatic afferents. However, more than 75% of all
neurons in the thoracic spinal cord are viscerosomatic,
suggesting a large degree of convergence of visceral and
somatic input to second-order spinal afferent neurons
(Cervero and Foreman, 1990).
Contrary to the classical perception that dorsal root
afferents simply course through the prevertebral ganglia,
it has been demonstrated that some give off short axon
collaterals within these ganglia that impinge on sympa-
thetic postganglionic neurons (Matthews and Cuello,
1982; Aldskogius et al., 1986). Although the functional
implication of such collaterals is not well understood, they
would permit the propagation of reflexes between differ-
ent viscera via the shortest possible pathway.
CGRP Immunohistochemistry
Since no anterograde tracing from the DRGs has been
carried out to date, the description of the peripheral in-
nervation pattern of spinal afferents is based exclusively
on CGRP and SP immunocytochemistry. Because a large
proportion of DRG neurons but only a few nodose ganglion
neurons contain CGRP, it is likely that CGRP-immunore-
active terminals in the liver truly represent DRG affer-
ents.
In the rat, CGRP-immunoreactive nerve fibers and ter-
minals show a similar gross distribution as the vagal
afferents do, with a notable absence within the hepatic
parenchyma. In the rat, dense networks of CGRP-positive
fibers are found in the fibromuscular layer of the biliary
tree, surrounding the portal vein, and in the stromal in-
831
terlobular compartment of mainly the portal areas of the
liver (Sasaki et al., 1986; Goehler and Sternini, 1996;
Akiyoshi et al., 1998). In the guinea pig, no CGRP-immu-
noreactive fibers were found in the parenchyma (Goehler
et al., 1988; Akiyoshi et al., 1998). In human liver, CGRP-
immunoreactive fibers were relatively sparse and limited
to the portal tracts (Akiyoshi et al., 1998; Stoyanova and
Gulubova, 2000).
IMPLICATIONS FOR SELECTIVE HEPATIC
DENERVATION STUDIES
Selective elimination of specific types of nerve fibers is
an important tool for functional studies. Ideally, one
wants to eliminate vagal or spinal innervation selectively,
and among these, eliminate afferents and efferents selec-
tively.
Selective Vagotomies
Although often intended and referred to, true selective
hepatic vagotomy is nearly impossible. Cutting the com-
mon hepatic vagal branch not only denervates the liver,
but also parts of the distal stomach, pylorus, duodenum,
and pancreas, because a large fascicle of this branch by-
passes the liver hilus and descends along the gastroduo-
denal artery (Figs. 1 and 2A) (Berthoud et al., 1992). In
addition, common hepatic branch vagotomy does not elim-
inate those vagal afferent (and possibly efferent) fibers
reaching the liver via the right subdiaphragmatic trunk,
celiac branch, and common hepatic artery (Fig. 1). Also,
hepatic branch vagotomy not only eliminates vagal fibers
since it was estimated that up to 30% of the nerve fibers
running in this branch are of nonvagal origin (Prechtl and
Powley, 1990). Because, at least in the rat, it is nearly
impossible to transect selectively only those vagal fibers
entering the liver hilus (in the hepatic branch proper), the
most reasonable approach is to compare functional
changes after either common hepatic or gastroduodenal
branch transection (la Fleur et al., 2003; Horn and Fried-
man, 2004). Based on a functional study in rats, Latour
and Lautt (2002) concluded that all of the parasympa-
thetic nerves that regulate insulin sensitivity pass
through the cervical vagus, (common) hepatic branch, and
finally through the anterior hepatic plexus along the com-
mon hepatic artery. However, without anatomical verifi-
cation, it is not possible to conclude what type(s) of nerve
fibers were destroyed by this denervation procedure. Al-
though most of the vagal fibers (efferent and afferent)
innervating the liver proper might be destroyed, there are
most likely also vagal fibers running in the gastroduode-
nal branch and nonvagal fibers affected.
Common hepatic vagal branch ablation can still provide
useful data. It proved very useful for studying sympa-
thetic (efferent) innervation of the rat liver by means of
transneuronal viral tracing with pseudorabies virus (Buijs
et al., 2003). In order to see microscopically the various
fascicles of the common hepatic branch and other small
branches running in the fascia between stomach and liver,
the myelin-specific dye toluidine blue was injected peri-
neurally near its exit from the left subdiaphragmatic
trunk. Pseudorabies virus was then injected into the liver
to delineate sympathetic preganglionic and premotor neu-
rons throughout the neuraxis (Buijs et al., 2003). For this
type of experiment, it does not matter that vagal sensory
innervation to the gastrointestinal tract and pancreas was
also eliminated.
832 BERTHOUD
If selective hepatic vagotomy is difficult, selective he-
patic vagal deafferentation or deefferentation is clearly
impossible. Two basic methods are available for selective
deafferentation. The first method is based on the clear
separation of vagal afferent and efferent rootlets as they
exit the medulla (Norgren and Smith, 1994); the second
method makes use of capsaicin binding to vanilloid recep-
tors expressed by most sensory nerve fibers and inducing
apoptosis. If used locally on nerve branches only contain-
ing vagal fibers, capsaicin can selectively destroy vagal
afferents, leaving vagal efferents intact. Thus, only if the
vagal branches innervating the liver hilus could be iso-
lated and selectively treated with capsaicin would selec-
tive hepatic vagal deafferentation be achieved. However,
this challenging procedure has not yet been attempted,
and it would still not address the problem that some vagal
afferents are capsaicin-insensitive.
The inability to do selective hepatic vagotomy, selective
vagal hepatic deafferentation, and selective vagal hepatic
deefferentation should be borne in mind when interpret-
ing functional studies. Behavioral changes such as food
intake following hepatic branch cuts do not implicate va-
gal fibers supplying the liver more than they implicate
vagal fibers supplying duodenum and pancreas by the
gastroduodenal branch. Furthermore, changes of specific
physiological hepatic parameters (such as enzyme activi-
ties) following hepatic branch vagotomies do not necessar-
ily result from cutting fibers to the liver, but could be the
result of, or an interaction with, changes that occur up-
stream in the simultaneously denervated duodenum or
pancreas.
By far the most anatomical work on vagal liver inner-
vation has been done in rats. However, a lot of functional
studies have traditionally been carried out in dogs. In the
dog, anterior and posterior plexuses can be easily distin-
guished and surgically transected (Lautt, 1980). It ap-
pears that the anterior plexus attached mainly to the
common hepatic artery contains all the parasympathetic
vagal fibers, and the dorsal plexus attached mainly to the
portal vein contains all the sympathetic fibers. However,
as mentioned above, without anatomical verification, the
exact identity of transected fibers remains unknown.
Selective Sympathectomies and Spinal
Deafferentation
Selective transection of hepatic innervation from the
spinal cord is as difficult and impure as selective hepatic
vagotomy, and selective hepatic sympathectomy (spinal
deefferentation) as well as spinal deafferentation is even
more difficult and has not been achieved in any study.
Using surgical methods, the difficulty is not to damage
any vagal component when stripping the hepatic artery/
portal vein plexus. The way vagal afferents are distrib-
uted in most of the nerve bundles making up the periar-
terial and periportal plexuses (Berthoud et al., 1992),
there will always be some inadvertent damage to them, at
least in the rat. Although local perineural capsaicin treat-
ment has been successfully used to ablate sensory fibers
selectively (Jancso and Such, 1983), applied to the hepatic
vascular bundle, it will ablate both vagal and spinal affer-
ents; and applied to the celiac ganglion (Takahashi and
Owyang, 1999), it will ablate not only dorsal root afferents
to the liver, but also to other organs. These problems
should be kept in mind when interpreting physiological
and behavioral functional studies.
However, there is one report in which selective elimina-
tion of sympathetic innervation to the liver without com-
promising vagal innervation was achieved. To eliminate
sympathetic innervation to the liver, the nerve bundles
running along the hepatic artery proper were visualized
using the myelin-specific dye toluidine blue and removed;
in addition, the hepatic artery was ligated and transected
(Buijs et al., 2003). It was then shown that vagal pregan-
glionic neurons in the dorsal motor nucleus, but not sym-
pathetic preganglionic neurons in the spinal cord, were
retrogradely labeled following pseudorabies virus injec-
tion into the left liver lobe. As for the reciprocal experi-
ment with common hepatic branch vagotomy, it does not
matter for this type of tracing experiment that vagal and
sympathetic innervation of neighboring organs is also
compromised by this procedure.
CENTRAL PATHWAYS OF SENSORY
HEPATIC NERVES
Vagal Afferents
Generally, most of the central processes of the bipolar
vagal afferent neurons from the abdomen terminate in the
ipsilateral nucleus of the solitary tract (NTS) and fewer in
the area postrema, dorsal motor nucleus, and the con-
tralateral NTS (Kalia and Mesulam, 1980; Shapiro and
Miselis, 1985; Norgren and Smith, 1988).
A specific termination zone for hepatic vagal afferents
has been claimed to include the gelatinosus subnucleus of
the NTS (Rogers et al., 1980; Fox and Powley, 1985);
however, because in these studies the tracer was injected
into the common hepatic branch, many of the identified
central axon terminals may belong to afferents projecting
to the gastrointestinal tract via the gastroduodenal vagal
branch (Berthoud et al., 1992; Phillips et al., 1997). The
same criticism can be applied to electrophysiologically
identified NTS neurons in experiments with electrical
stimulation of the common hepatic vagal branch (Adachi,
1984).
In other experiments, activity of NTS neurons was as-
sessed by extracellular recording following infusions of
glucose or hypertonic saline into the portal vein (Kobashi
and Adachi, 1986). By combining peripheral osmotic acti-
vation of (presumably vagal) primary afferents with anti-
dromic stimulation of NTS projection neurons by placing
electrodes into various brain regions, Kobashi and Adachi
(1986) were able to identify the parabrachial nuclei, the
ventrolateral medulla, and the hypothalamus as recipi-
ents of portal hepatic osmosensory information. Using the
induction of the immediate-early gene protein product Fos
as an indicator of neuronal activation, Morita et al. (1997)
found activated neurons in the NTS, the area postrema,
and the paraventricular and supraoptic hypothalamic nu-
clei. However, the population of second-order NTS neu-
rons receiving input from other specific vagal hepatic and
portal sensors has not been comprehensively identified.
The general pattern of dissemination of vagal sensory
information from the NTS is characterized by local in-
tramedullary projections to integrative reticular areas
and various motor nuclei, and by at least three major
ascending pathways (Blessing et al., 1991). There are di-
rect projections from mainly catecholaminergic NTS neu-
rons to the paraventricular nucleus of the hypothalamus
(Fig. 1). Another pathway consists of major projections to
various visceral forebrain areas via the parabrachial nu-
 SENSORY HEPATIC NERVES
clear complex in the pons, and a third pathway to hypo-
thalamic and other forebrain areas involves a relay via the
catecholaminergic A1/C1 ventral medullary area.
Whether all of these pathways are important for process-
ing of specific portal-hepatic signals remains to be deter-
mined.
Spinal Afferents
Ascending projections from the dorsal horn of the spinal
cord cross to the contralateral side and then travel
through the spinothalamic, spinoreticular, and spinomes-
encephalic tracts to mainly the lateral thalamus, giganto-
cellular tegmental field of the brain stem reticular forma-
tion, and to the central gray and parabrachial nuclei in the
midbrain and pons, respectively (Fig. 1) (Cervero and
Foreman, 1990). Some of these fibers also reach the NTS
via the spinosolitary tract (Menetrey and Basbaum, 1987).
It is thought that most of these pathways are concerned
with the conscious sensation and discriminative percep-
tion of visceral pain and the organization of motivational-
affective behavior. Pathways involved in homeostatic con-
trol mechanisms such as water and nutrient balance may
primarily use the spinosolitary and spinomesencephalic
tracts, with the NTS, parabrachial nucleus, and PVN as
central integrative structures.
FUNCTIONS OF PORTAL-HEPATIC
SENSORY NERVES
Control of Energy Metabolism and Food Intake
After the pioneering work of Russek (1981), a role for
the sensory functions of the liver in metabolic and food
intake control was first systematically proposed by
Sawchenko and Friedman (1979). Substantial progress
has been made since then on the role of the liver and its
innervation in food intake, metabolic control, and the de-
velopment of obesity (Scharrer, 1999; Langhans, 2003).
A glucose sensor located not in the liver itself, but in the
portal vein, appears to play a key role in glucose ho-
meostasis and insulin sensitivity. Such a sensor could be
hypothesized based on sensitivity to glucose of vagal af-
ferent fibers running in the common hepatic branch that
persisted after complete removal of the liver (Niijima,
1983) and was subsequently anatomically identified by
anterograde DiI tracing from the nodose ganglia
(Berthoud et al., 1992). More recently, its dependence on
the glucose transporter GLUT-2 (Burcelin et al., 2000) and
the gastrointestinal hormone glucagon-like peptide-1
(GLP-1) (Burcelin et al., 2001) and its presumptive phys-
iological role have been identified (Hevener et al., 1997,
2000, 2001; Cherrington, 1999; Burcelin and Thorens,
2001).
Glucose sensors are also suspected to be located in the
liver itself, as the portal signal appears to be compared to
a reference signal accessible through the hepatic artery
(Cherrington, 1999), but identification of its underlying
neural substrate remains elusive. The intrahepatic glu-
cose sensor may be part of a more general sensor of he-
patic energy status (Friedman, 1997; Scharrer, 1999).
This sensor seems to signal through vagal afferents and is
responsible for increased food intake induced by inhibition
of fatty acid oxidation (Anil and Forbes, 1988; Lutz et al.,
1996, 1997; Horn et al., 2001). Transduction may involve
increases in hepatocyte sodium and calcium concentration
and membrane potential (Scharrer et al., 1997; Boutellier
833
et al., 1999; Friedman et al., 2003; Rawson et al., 2003), as
originally suggested by Russek (1981).
Defensive and Protective Functions
A generally protective role for vagal afferents had been
suggested on the basis of phylogenetic development of the
vagal-paraventricular system (Andrews and Lawes,
1992). A role for the vagus nerve, particularly the hepatic
vagus, has now been demonstrated for immune defense.
Attention has been drawn to the subdiaphragmatic vagus
as a pathway signaling inflammatory events from the liver
to the brain (Goehler et al., 2000), and in turn controlling
in real time the inflammatory process by vagal efferents
(Tracey, 2002). Evidence comes from experiments where
intraperitoneal injections of bacterial endotoxins such as
lipopolysaccharide (LPS) or inflammatory cytokines such
as interleukin-1␤ or tumor necrosis factor (TNF) were
used to induce CNS-mediated illness reactions, e.g., fever,
food avoidance, reduced exploratory behavior, and hyper-
algesia. These reactions could be blocked by subdiaphrag-
matic vagotomy, and in some cases with selective transec-
tion of the common hepatic vagal branch (Watkins et al.,
1995; Bluthe et al., 1996). Peripheral injection of lipopoly-
saccharide or IL-1␤ produced prostaglandin-dependent
activation of vagal afferents as measured by increased
electrical discharge (Ek et al., 1998; Bucinskaite et al.,
2000), c-Fos expression in nodose ganglia (Goehler et al.,
1998), and glutamate release in the nucleus of the solitary
tract (Mascarucci et al., 1998). Furthermore, reversible
inactivation of the dorsal vagal complex blocked LPS-
induced Fos expression in central autonomic nuclei and
social withdrawal (Marvel et al., 2004). In turn, vagal
efferents via release of acetylcholine acting through ␣-7-
nicotinic cholinergic receptors on resident macrophages in
various abdominal organs including the liver suppress
further release of TNF-␣ (Borovikova et al., 2000a, b;
Bernik et al., 2002; Tracey, 2002; Wang et al., 2003).
LITERATURE CITED
Adachi A. 1984. Projection of the hepatic vagal nerve in the medulla
oblongata. J Auton Nerv Syst 10:287–293.
Akiyoshi H, Gonda T, Terada T. 1998. A comparative histochemical
and immunohistochemical study of aminergic, cholinergic and pep-
tidergic innervation in rat, hamster, guinea pig, dog and human
livers. Liver 18:352–359.
Aldskogius H, Elfvin LG, Forsman CA. 1986. Primary sensory affer-
ents in the inferior mesenteric ganglion and related nerves of the
guinea pig: an experimental study with anterogradely transported
wheat germ agglutinin-horseradish peroxidase conjugate. J Auton
Nerv Syst 15:179 –190.
Andrews PLR, Lawes INC. 1992. A protective role for vagal afferents:
a hypothesis. In: Ritter SR, Barnes CD, editors. Neuroanatomy and
physiology of abdominal vagal afferents. Boca Raton: CRC Press. p
221–248.
Anil MH, Forbes JM. 1988. The roles of hepatic nerves in the reduc-
tion of food intake as a consequence of intraportal sodium propi-
onate administration in sheep. Q J Exp Physiol 73:539 –546.
Armstrong DM, Manley L, Haycock JW, Hersh LB. 1990. Co-localiza-
tion of choline acetyltransferase and tyrosine hydroxylase within
neurons of the dorsal motor nucleus of the vagus. J Chem Neuro-
anat 3:133–140.
Barja F, Mathison R. 1984. Sensory innervation of the rat portal vein
and the hepatic artery. J Auton Nerv Syst 10:117–125.
Bernik TR, Friedman SG, Ochani M, DiRaimo R, Susarla S, Czura CJ,
Tracey KJ. 2002. Cholinergic antiinflammatory pathway inhibition
of tumor necrosis factor during ischemia reperfusion. J Vasc Surg
36:1231–1236.
834 BERTHOUD
Berthoud HR, Carlson NR, Powley TL. 1991. Topography of efferent
vagal innervation of the rat gastrointestinal tract. Am J Physiol
260:R200 –R207.
Berthoud HR, Kressel M, Neuhuber WL. 1992. An anterograde trac-
ing study of the vagal innervation of rat liver, portal vein and
biliary system. Anat Embryol (Berl) 186:431– 442.
Berthoud HR. 1995. Anatomical demonstration of vagal input to nic-
otinamide acetamide dinucleotide phosphate diaphorase-positive
(nitrergic) neurons in rat fundic stomach. J Comp Neurol 358:428 –
439.
Berthoud HR. 1996. Morphological analysis of vagal input to gastrin
releasing peptide and vasoactive intestinal peptide containing neu-
rons in the rat glandular stomach. J Comp Neurol 370:61–70.
Berthoud HR, Patterson LM. 1996. Innervation of rat abdominal
paraganglia by calretinin-like immunoreactive nerve fibers. Neuro-
sci Lett 210:115–118.
Berthoud HR, Neuhuber WL. 1996. An anatomical analysis of vagal
and spinal afferent innervation of rat liver and associated organs.
In: Shimazu, T, editor. Liver innervation and the neural control of
hepatic function p 31– 42. London: John Libbey.
Berthoud HR, Neuhuber WL. 1998. Peripheral and central functional
neuroanatomy of sensory and motor innervation of the portal-he-
patic axis and biliary system. In: Haussinger DJK, editor. Liver and
nervous system. Lancaster, U.K.: Kluwer. p 17–33.
Berthoud HR, Patterson LM, Zheng H. 2001. Vagal-enteric interface:
vagal activation-induced expression of c-Fos and p-CREB in neu-
rons of the upper gastrointestinal tract and pancreas. Anat Rec
262:29 – 40.
Blessing WW, Li YW, Wesselingh SL. 1991. Transneuronal transport
of herpes simplex virus from the cervical vagus to brain neurons
with axonal inputs to central vagal sensory nuclei in the rat. Neu-
roscience 42:261–274.
Bluthe RM, Michaud B, Kelley KW, Dantzer R. 1996. Vagotomy
blocks behavioural effects of interleukin-1 injected via the intra-
peritoneal route but not via other systemic routes. Neuroreport
7:2823–2827.
Borovikova LV, Ivanova S, Nardi D, Zhang M, Yang H, Ombrellino M,
Tracey KJ. 2000a. Role of vagus nerve signaling in CNI-1493-
mediated suppression of acute inflammation. Auton Neurosci 85:
141–147.
Borovikova LV, Ivanova S, Zhang M, Yang H, Botchkina GI, Watkins
LR, Wang H, Abumrad N, Eaton JW, Tracey KJ. 2000b. Vagus
nerve stimulation attenuates the systemic inflammatory response
to endotoxin. Nature 405:458 – 462.
Boutellier S, Lutz TA, Volkert M, Scharrer E. 1999. 2-mercaptoac-
etate, an inhibitor of fatty acid oxidation, decreases the membrane
potential in rat liver in vivo. Am J Physiol 277:R301–R305.
Bucinskaite V, Kurosawa M, Lundeberg T. 2000. Effect of interleukin-
1beta on subdiaphragmatic vagal efferents in the rat. Auton Neu-
rosci 85:93–97.
Buijs RM, la Fleur SE, Wortel J, Van Heyningen C, Zuiddam L,
Mettenleiter TC, Kalsbeek A, Nagai K, Niijima A. 2003. The supra-
chiasmatic nucleus balances sympathetic and parasympathetic out-
put to peripheral organs through separate preautonomic neurons.
J Comp Neurol 464:36 – 48.
Burcelin R, Dolci W, Thorens B. 2000. Glucose sensing by the hepa-
toportal sensor is GLUT2-dependent: in vivo analysis in GLUT2-
null mice. Diabetes 49:1643–1648.
Burcelin R, Thorens B. 2001. Evidence that extrapancreatic GLUT2-
dependent glucose sensors control glucagon secretion. Diabetes 50:
1282–1289.
Burcelin R, Da Costa A, Drucker D, Thorens B. 2001. Glucose com-
petence of the hepatoportal vein sensor requires the presence of an
activated glucagon-like peptide-1 receptor. Diabetes 50:1720 –1728.
Cervero F, Foreman RD. 1990. Sensory innervation of the viscera. In:
Loewy AD, Spyer KM, editors. Central regulation of autonomic
functions. Oxford: Oxford University Press. p 104 –125.
Cherrington AD. 1999. Control of glucose uptake and release by the
liver in vivo. Diabetes 48:1198 –1214.
Dutsch M, Eichhorn U, Worl J, Wank M, Berthoud HR, Neuhuber
WL. 1998. Vagal and spinal afferent innervation of the rat
esophagus: a combined retrograde tracing and immunocytochemi-
cal study with special emphasis on calcium-binding proteins.
J Comp Neurol 398:289 –307.
Ek M, Kurosawa M, Lundeberg T, Ericsson A. 1998. Activation of
vagal afferents after intravenous injection of interleukin-1beta: role
of endogenous prostaglandins. J Neurosci 18:9471–9479.
Fox EA, Powley TL. 1985. Longitudinal columnar organization within
the dorsal motor nucleus represents separate branches of the ab-
dominal vagus. Brain Res 341:269 –282.
Friedman MI. 1997. An energy sensor for control of energy intake.
Proc Nutr Soc 56:41–50.
Friedman MI, Graczyk-Millbrandt G, Ji H, Rawson NE, Osbakken
MD. 2003. 2,5-anhydro-D-mannitol increases hepatocyte sodium:
transduction of a hepatic hunger stimulus? Biochim Biophys Acta
1642:53–58.
Goehler LE, Sternini C, Brecha NC. 1988. Calcitonin gene-related
peptide immunoreactivity in the biliary pathway and liver of the
guinea-pig: distribution and colocalization with substance P. Cell
Tissue Res 253:145–150.
Goehler LE, Sternini C. 1996. Calcitonin gene-related peptide inner-
vation of the rat hepatobiliary system. Peptides 17:209 –217.
Goehler LE, Gaykema RP, Hammack SE, Maier SF, Watkins LR.
1998. Interleukin-1 induces c-Fos immunoreactivity in primary af-
ferent neurons of the vagus nerve. Brain Res 804:306 –310.
Goehler LE, Gaykema RP, Hansen MK, Anderson K, Maier SF,
Watkins LR. 2000. Vagal immune-to-brain communication: a vis-
ceral chemosensory pathway. Auton Neurosci 85:49 –59.
Guo JJ, Browning KN, Rogers RC, Travagli RA. 2001. Catecholamin-
ergic neurons in rat dorsal motor nucleus of vagus project selec-
tively to gastric corpus. Am J Physiol Gastrointest Liver Physiol
280:G361–G367.
Gwyn DG, Ritchie TC, Coulter JD. 1985. The central distribution of
vagal catecholaminergic neurons which project into the abdomen in
the rat. Brain Res 328:139 –144.
Hayakawa T, Takanaga A, Tanaka K, Maeda S, Seki M. 2004. Dis-
tribution and ultrastructure of dopaminergic neurons in the dorsal
motor nucleus of the vagus projecting to the stomach of the rat.
Brain Res 1006:66 –73.
Hevener AL, Bergman RN, Donovan CM. 1997. Novel glucosensor for
hypoglycemic detection localized to the portal vein. Diabetes 46:
1521–1525.
Hevener AL, Bergman RN, Donovan CM. 2000. Portal vein afferents
are critical for the sympathoadrenal response to hypoglycemia.
Diabetes 49:8 –12.
Hevener AL, Bergman RN, Donovan CM. 2001. Hypoglycemic detec-
tion does not occur in the hepatic artery or liver: findings consistent
with a portal vein glucosensor locus. Diabetes 50:399 – 403.
Horn CC, Friedman MI. 2004. Separation of hepatic and gastrointes-
tinal signals from the common “hepatic” branch of the vagus. Am J
Physiol Regul Integ Comp Physiol 287:R120 –R126.
Horn CC, Tordoff MG, Friedman MI. 2001. Role of vagal afferent
innervation in feeding and brain Fos expression produced by met-
abolic inhibitors. Brain Res 919:198 –206.
Jancso G, Such G. 1983. Effects of capsaicin applied perineurally to
the vagus nerve on cardiovascular and respiratory functions in the
cat. J Physiol 341:359 –370.
Kalia M, Mesulam MM. 1980. Brain stem projections of sensory and
motor components of the vagus complex in the cat: II, laryngeal,
tracheobronchial, pulmonary, cardiac, and gastrointestinal
branches. J Comp Neurol 193:467–508.
Kobashi M, Adachi A. 1986. Projection of nucleus tractus solitarius
units influenced by hepatoportal afferent signal to parabrachial
nucleus. J Auton Nerv Syst 16:153–158.
la Fleur SE, Ji H, Manalo SL, Friedman MI, Dallman MF. 2003. The
hepatic vagus mediates fat-induced inhibition of diabetic hyperpha-
gia. Diabetes 52:2321–2330.
Langhans W. 1996. Role of the liver in the metabolic control of eating:
what we know—and what we do not know. Neurosci Biobehav Rev
20:145–153.
Langhans W. 2003. Role of the liver in the control of glucose-lipid
utilization and body weight. Curr Opin Clin Nutr Metab Care
6:449 – 455.
 SENSORY HEPATIC NERVES
Latour MG, Lautt WW. 2002. The hepatic vagus nerve in the control
of insulin sensitivity in the rat. Auton Neurosci 95:125–130.
Lautt WW. 1980. Hepatic nerves: a review of their functions and
effects. Can J Physiol Pharmacol 58:105–123.
Lutz TA, Niijima A, Scharrer E. 1996. Intraportal infusion of 2,5-
anhydro-D-mannitol increases afferent activity in the common he-
patic vagus branch. J Auton Nerv Syst 61:204 –208.
Lutz TA, Diener M, Scharrer E. 1997. Intraportal mercaptoacetate
infusion increases afferent activity in the common hepatic vagus
branch of the rat. Am J Physiol 273:R442–R445.
Magni F, Carobi C. 1983. The afferent and preganglionic parasympa-
thetic innervation of the rat liver, demonstrated by the retrograde
transport of horseradish peroxidase. J Auton Nerv Syst 8:237–260.
Marvel FA, Chen CC, Badr N, Gaykema RP, Goehler LE. 2004.
Reversible inactivation of the dorsal vagal complex blocks lipo-
polysaccharide-induced social withdrawal and c-Fos expression in
central autonomic nuclei. Brain Behav Immun 18:123–134.
Mascarucci P, Perego C, Terrazzino S, De Simoni MG. 1998. Gluta-
mate release in the nucleus tractus solitarius induced by peripheral
lipopolysaccharide and interleukin-1 beta. Neuroscience 86:1285–
1290.
Matsuo Y, Seki A. 1978. The coordination of gastrointestinal hor-
mones and the autonomic nerves. Am J Gastroenterol 69:21–50.
Matthews MR, Cuello AC. 1982. Substance P-immunoreactive periph-
eral branches of sensory neurons innervate guinea pig sympathetic
neurons. Proc Natl Acad Sci USA 79:1668 –1672.
Menetrey D, Basbaum AI. 1987. Spinal and trigeminal projections to
the nucleus of the solitary tract: a possible substrate for somatovis-
ceral and viscerovisceral reflex activation. J Comp Neurol 255:439 –
450.
Morita H, Yamashita Y, Nishida Y, Tokuda M, Hatase O, Hosomi H.
1997. Fos induction in rat brain neurons after stimulation of the
hepatoportal Na-sensitive mechanism. Am J Physiol 272:R913–
R923.
Niijima A. 1983. Glucose-sensitive afferent nerve fibers in the liver
and their role in food intake and blood glucose regulation. J Auton
Nerv Syst 9:207–220.
Norgren R, Smith GP. 1988. Central distribution of subdiaphragmatic
vagal branches in the rat. J Comp Neurol 273:207–223.
Norgren R, Smith GP. 1994. A method for selective section of vagal
afferent or efferent axons in the rat. Am J Physiol 267:R1136 –
R1141.
Obici S, Rossetti L. 2003. Minireview: nutrient sensing and the reg-
ulation of insulin action and energy balance. Endocrinology 144:
5172–5178.
Phillips RJ, Baronowsky EA, Powley TL. 1997. Afferent innervation of
gastrointestinal tract smooth muscle by the hepatic branch of the
vagus. J Comp Neurol 384:248 –270.
Prechtl JC, Powley TL. 1987. A light and electron microscopic exam-
ination of the vagal hepatic branch of the rat. Anat Embryol (Berl)
176:115–126.
Prechtl JC, Powley TL. 1990. The fiber composition of the abdominal
vagus of the rat. Anat Embryol (Berl) 181:101–115.
835
Raab M, Neuhuber WL. 2003. Vesicular glutamate transporter 2
immunoreactivity in putative vagal mechanosensor terminals of
mouse and rat esophagus: indication of a local effector function?
Cell Tissue Res 312:141–148.
Rawson NE, Ji H, Friedman MI. 2003. 2,5-anhydro-D-mannitol in-
creases hepatocyte calcium: implications for a hepatic hunger stim-
ulus. Biochim Biophys Acta 1642:59 – 66.
Rogers RC, Kita H, Butcher LL, Novin D. 1980. Afferent projections to
the dorsal motor nucleus of the vagus. Brain Res Bull 5:365–373.
Russek M. 1981. Current status of the hepatostatic theory of food
intake control. Appetite 2:137–143.
Sasaki Y, Hayashi N, Kasahara A, Matsuda H, Fusamoto H, Sato N,
Hillyard CJ, Girgis S, MacIntyre I, Emson PC, et al. 1986. Calcito-
nin gene-related peptide in the hepatic and splanchnic vascular
systems of the rat. Hepatology 6:676 – 681.
Sawchenko PE, Friedman MI. 1979. Sensory functions of the liver: a
review. Am J Physiol 236:R5–R20.
Scharrer E, Rossi R, Sutter DA, Seebacher MC, Boutellier S, Lutz TA.
1997. Hyperpolarization of hepatocytes by 2,5-AM: implications for
hepatic control of food intake. Am J Physiol 272:R874 –R878.
Scharrer E. 1999. Control of food intake by fatty acid oxidation and
ketogenesis. Nutrition 15:704 –714.
Shapiro RE, Miselis RR. 1985. The central organization of the vagus
nerve innervating the stomach of the rat. J Comp Neurol 238:473–
488.
Stavney LS, Kato T, Griffith CA, Nyhus LM, Harkins HN. 1963. A
physiologic study of motility changes following selective gastric
vagotomy. J Surg Res 23:390 –394.
Stoyanova, II, Gulubova MV. 2000. Immunocytochemical study on the
liver innervation in patients with cirrhosis. Acta Histochem 102:
391– 402.
Stümpel F, Scholtka B, Jungermann K. 2000. Stimulation by portal
insulin of intestinal glucose absorption via hepatoenteral nerves
and prostaglandin E2 in the isolated, jointly perfused small intes-
tine and liver of the rat. Ann NY Acad Sci 915:111–116.
Takahashi T, Owyang C. 1999. Mechanism of cholecystokinin-induced
relaxation of the rat stomach. J Auton Nerv Syst 75:123–130.
Tiscornia OM, Martinez JL, Sarles H. 1976. Some aspects of human
and canine macroscopic pancreas innervation. Am J Gastroenterol
66:353–361.
Tracey KJ. 2002. The inflammatory reflex. Nature 420:853– 859.
Wang ZJ, Neuhuber WL. 2003. Intraganglionic laminar endings in
the rat esophagus contain purinergic P2X2 and P2X3 receptor im-
munoreactivity. Anat Embryol (Berl) 207:363–371.
Wang H, Yu M, Ochani M, Amella CA, Tanovic M, Susarla S, Li JH,
Yang H, Ulloa L, Al-Abed Y, Czura CJ, Tracey KJ. 2003. Nicotinic
acetylcholine receptor alpha7 subunit is an essential regulator of
inflammation. Nature 421:384 –388.
Watkins LR, Maier SF, Goehler LE. 1995. Immune activation: the role
of pro-inflammatory cytokines in inflammation, illness responses
and pathological pain states. Pain 63:289 –302.