As a result of neurons transmitting their activity by generating electrical spikes, the chemical substance that facilitates or suppresses vagal nerve spike activity is a candidate interoception mediator. There are several endogenous chemical substances that facilitate or suppress the spontaneous firing of this nerve, such as those listed in the following subsections (see also Table 2.1). It is important to keep in mind that the studies described in the following subsections are conducted using the common hepatic branch of the vagus nerve. This nerve bundle is composed not only of nerves that innervate the hepatoportal region but also nerves that innervate outside this region, mainly the gastrointestinal region (Horn and Friedman 2004), as mentioned earlier. Therefore, the possibility that activity may originate in fibers other than those from hepatoportal regions cannot be completely excluded.

A likely indicator of brain–liver interaction is when metabolism in the liver is altered (Fig. 2.2A). For example, liver-specific disruption of peroxisome proliferator-activated receptor γ (PPARγ) in leptin-deficient obese mice prevents hepatic steatosis, but increases peripheral adiposity to aggravate diabetic phenotypes in the mouse (Matsusue et al. 2005). In contrast, tissue-specific overexpression of PPARγ in the liver induces severe hepatic steatosis, but peripheral adiposity is rather reduced as a result of enhanced lipolysis (Uno et al. 2006). A putative mechanism underlying these homeostatic changes between the liver and adipose tissues is participation of the autonomic nervous system. Namely, information is transmitted from the liver to the brain via neuronal and humoral factors, and then the brain controls adipose tissues using efferent sympathetic nerves. Indeed, the increase of lipolysis in brown adipose tissue (BAT) by overexpression of PPARγ in the liver is abolished by hepatic branch vagotomy (Uno et al. 2006). Furthermore, high-fat feeding induced hepatic glucokinase upregulation, and hepatic glucokinase gene overexpression dose-dependently decreased adaptive thermogenesis by downregulating thermogenesis-related genes in the BAT (Tsukita et al. 2012). This inter-tissue (liver-to-BAT) system is mediated by the afferent vagus from the liver to the brain and sympathetic efferents from the brain to the BAT (Tsukita et al. 2012). The liver–brain–adipose neural axis is found to have an important role in switching the fuel source from glycogen to triglycerides under prolonged fasting conditions (Izumida et al. 2013).

Another likely indicator of brain–liver interaction is the appearance of chronic liver diseases such as hepatitis and cirrhosis (Fig. 2.2B). Chronic inflammation of the liver is a disorder that is frequently associated with sickness behavior (fatigue, malaise, loss of appetite, loss of social interest, and so on) and alteration in mood state (D’Mello et al. 2009; Nguyen et al. 2007; Swain 2006; Swain and Maric 1995; Capuron and Miller 2011; Isik et al. 2007; Jones et al. 2006). In these disorders, significant pathological CNS tissue damage is basically absent (Forton et al. 2002, 2008; Pollak et al. 2003). A mechanism that explains CNS symptoms in periphery-originated diseases satisfactorily is that incorrect information about the state of peripheral organs is transmitted to the brain. It is thought that both neuronal and humoral factors are responsible for this and that vagus nerves are main players in neuronal factors (Goehler et al. 2000). In fact, the common hepatic branch of the vagus nerve can respond to proinflammatory cytokine IL-1β (as mentioned above) (Niijima 1996; Ek et al. 1998). The vagus nerve can be shown to express cytokine receptors such as the IL-1 receptor (Ek et al. 1998). In addition, endotoxin lipopolysaccharide (LPS), a gram-negative bacterial cell wall component, is a potent inducer of TNFα, IL-1β, and IL-6 (Dantzer et al. 2008; Capuron and Miller 2011; Lebbar et al. 1986; Vallieres and Rivest 1997; Laye et al. 2000). It is known that plasma endotoxin is elevated in patients with chronic liver disease (Yamamoto et al. 1994; Swain 2006). Therefore, cytokines released peripherally in response to bacterial endotoxin are thought to be important in communication between the brain and the liver through the vagus nerve. Consistently, in LPS-induced peritoneal inflammation, the NTS to which vagus afferents are projected, and subsequent regions to which NTS neurons project abundantly – the PBN and the hypothalamic paraventricular nucleus – are positive when assessed with expression of immediate early gene c-fos (Wan et al. 1993; Goehler et al. 2000). The intraperitoneal injection of IL-1β not only induces c-fos expression in the brain but also alters social exploratory behavior in the rat (Bluthe et al. 1996). It is known that changes in both c-fos expression and behavior can be reduced by subdiaphragmatic vagotomy (Bluthe et al. 1996).

Brain regions activated by inflammation were elucidated by getting healthy male volunteers (some of whom received typhoid vaccination or saline) to conduct a Stroop task during functional magnetic resonance imaging. Typhoid but not saline injection produced a robust inflammatory response and a significant increase in fatigue, confusion, and impaired concentration in participants. Performance of the Stroop task under inflammation-activated brain conditions involved the participation of the brainstem, thalamus, amygdala, cingulate, and bilateral mid and anterior insula in afferent interoception. Activity changes in these regions can be confirmed to be a result of inflammation (Harrison et al. 2009).

Dantzer et al. (2008) is an excellent review on the relation between inflammation and sickness behavior. The authors show that decompensation of sickness behavior could be a risk factor for mood disorders.

A large part of the information collected by interoception generates the reflex to maintain the body’s homeostasis. The hypothalamus controls a vast array of these physiological processes including sleep/wake cycles, sexual behavior and reproduction, and metabolic control such as thermoregulation, energy intake/expenditure, glucose metabolism, lipid metabolism, and food and water intake. In many cases, sympathetic neurons play a central role in these reflex reactions. There are excellent reviews on the role that sympathetic neurons and the hypothalamus play in homeostatic control (Saper 2002; Kalsbeek et al. 2014). To avoid duplication, I would like to introduce here another brain structure, the insular cortex, as a major ascending destination of interoception.

The insular cortex is part of the cerebral cortex and is located deep in the lateral sulcus in humans. In rodents, it surrounds the rhinal fissure and is between the primary/secondary somatosensory and piriform cortices in the most anterior–posterior levels (Paxinos and Franklin 2001). The insular cortex is normally divided into three subdivisions (Cechetto and Saper 1987): the agranular insular cortex (AI), which surrounds the rhinal fissure; the granular insular cortex (GI), which is located just ventral to the secondary somatosensory cortex; and the dysgranular insular cortex (DI), which is between the AI and GI. Each subdivision is thought to involve sensory information, in particular. For example, the AI is thought to be involved in nociceptive and autonomic information processing (Cechetto and Saper 1987; Jasmin et al. 2003; Burkey et al. 1996, 1999); the GI plays a crucial role in modulating visceral function (Yamamoto et al. 1981, 1984); and the DI participates in gustatory processing (Cechetto and Saper 1987). It is known that ascending visceral afferents, which project from ventroposterolateral (VPL) and ventroposteromedial (VPM) parvicellular (pc) nuclei of the thalamus and from the PBN, are localized in the GI and DI (Allen et al. 1991). On the other hand, integrated limbic afferents from the infralimbic cortex and the mediodorsal nucleus of the thalamus project to the AI (Allen et al. 1991). The convergence of visceral and limbic information may be used by the insular cortex to integrate autonomic response with ongoing behavior and emotion.

Stimulation of the vagus nerve has been clinically utilized in the treatment of epilepsy and depression (Beekwilder and Beems 2010; Bonaz et al. 2013). The first attempt at vagus stimulation using an implanted electrode in humans was reported in 1988 for the treatment of drug-resistant epilepsy (Penry and Dean 1990). In 1997, vagus nerve stimulation was approved by the Food and Drug Administration (FDA) in the United States as an adjunctive treatment for drug-resistant seizures, and then in 2005 for depression. It can be demonstrated that the activation of vagus afferents suppresses seizure in rats (Krahl et al. 1998, 2001). The latter therapy is based on mood improvement in patients with epilepsy receiving vagus nerve stimulation (Groves and Brown 2005), but its effectiveness has not yet been confirmed by meta-analysis because of insufficient data (Martin and Martin-Sanchez 2012). Vagus nerve stimulation is effected by wrapping an electrode around the left vagus nerve in the neck (Reid 1990).

The precise mechanism underlying vagus nerve stimulation in epilepsy and depression is obscure, but it is postulated that vagus nerve stimulation modulates the activity of the following neuronal networks via the NTS (Bonaz et al. 2013): (1) activation of the thalamus and thalamocortical pathway; (2) suppression of the amygdala and hippocampus; and (3) increased norepinephrine and serotonin release in widespread cerebral regions. The locus coeruleus (LC), the main origin of noradrenergic projection, mediates at least some of the effects of vagus nerve stimulation in attenuating seizures because LC lesions suppress the antiepileptic effect of vagus nerve stimulation (Krahl et al. 1998). The LC is directly connected to the NTS. Activity mapping using c-fos in rats shows that positive signals are increased in the amygdala, cingulate, LC, and hypothalamus in response to antiepileptic stimulation of the vagus nerve (Naritoku et al. 1995). In humans, functional neuroimaging studies suggest that the thalamus, cerebellum, orbitofrontal cortex, limbic system, hypothalamus, and medulla exert long-term changes in their activity in response to vagus stimulation (Lomarev et al. 2002; Chae et al. 2003).