Growing body of evidence supports the notion that the microbiota serves as a source of sleep-promoting signals (Brown et al., 1990; Millican et al., 2018; Ogawa et al., 2020, Szentirmai et al., 2019; Szentirmai et al., 2021). We hypothesized that LPS, a component of gram-negative bacterial cell wall, could function as one such sleep-inducing signal. Our main finding is that LPS induces increased sleep and elevated body temperature when injected into the portal circulation. The portal circulation drains blood from organs housing the intestinal microbiota, including the stomach, small intestine, and large intestine. Microbial molecules, such as LPS, enter the portal circulation, thus the internal environment of the host, by the process called translocation. The free translocation of microbial molecules is prevented by the intestinal barrier, nevertheless, significant amounts of these molecules appear in the portal blood under normal healthy conditions (Guerville and Boudry, 2016). The translocation of LPS from the intestinal lumen to the portal blood is facilitated by transport mechanisms associated with the LPS receptor TLR4 (Cetin et al., 2004). LPS is naturally present in both portal and extraportal systemic blood in healthy humans, rats, and mice, rendering it a physiological plasma component of prokaryotic origin (Nolan, 1981; Wiedermann et al., 1999; Goto et al., 1994). Everyday stimuli further facilitate translocation, consequently elevating circulating levels of LPS. For example, a single high-fat meal consisting of three slices of toast spread with 50 g butter elevates plasma LPS levels by 50% in humans (Erridge et al., 2007). Additionally, common occurrences such as consumption of relatively low amounts of alcohol (Ferrier et al., 2006; Sturm et al., 2021), acute or chronic sleep loss (Wang et al., 2021; Li et al., 2024) and mild psychological stress (Vanuytsel et al., 2013; Santos et al., 2000) all contribute to increased circulating levels of LPS.
Given the pervasive presence of LPS in circulatory systems and the observation that systemic administration of high, proinflammatory doses of LPS induces sleep in various species, including humans (Lancel et al., 1995; Kapas et al., 1998; Opp and Toth, 1998; Pollmacher et al., 1993), we postulated that LPS translocated into the portal circulation might serve as a signal from the intestinal microbiota to initiate sleep. If proven valid, we anticipated that mimicking increased LPS translocation into the portal circulation would enhance sleep. Our findings support this hypothesis, revealing that even a minimal dose of 1 µg/kg LPS injected into the portal vein induces increases in NREMS and elevated body temperature.
Multiple lines of evidence suggest that the effects of intraportally injected LPS are not systemic but rather arise from the activation of LPS receptors located in the portal circulation or within the liver. First, our previous work has identified the existence of a sleep-inducing sensory mechanism in the hepatoportal region, sensitive to microbial molecules such as the short-chain fatty acid butyrate and LTA, a cell wall component of gram-positive bacteria (Szentirmai et al., 2019; Szentirmai et al., 2021). Second, systemic administration of LPS at a dose 20 times higher than the lowest effective intraportal dose failed to impact sleep and did not elicit any biologically meaningful effect on body temperature. Moreover, systemic administration of the lowest effective dose of LPS following intraportal administration failed to induce changes in LPS levels in the extraportal systemic circulation. This observation is consistent with the concept that the liver functions as an effective LPS sink. Notably, 1 g liver tissue has the capacity to clear intraportally administered LPS at the rate of 1.5 µg/h during a single passage (Yamaguchi et al., 1982). Considering that the liver weight of a 400 g rat is approximately 14 g (Webster et al., 1947), the liver can remove 21 µg of LPS in an hour or 0.35 µg in one minute. In our experiment, we injected 1 µg/kg (equivalent to approximately 0.4 µg/animal LPS) over a period of two minutes. These theoretical considerations robustly support the notion that low doses of LPS, when directly injected into the portal vein, undergo significant clearance by the liver and do not result in overflow into the hepatic vein.
Several potential cellular targets for LPS exist in the hepatoportal region. The LPS receptor, TLR4, is expressed by various liver cells, including hepatic macrophages (Kupffer cells), sinusoidal endothelial cells, stellate cells, and hepatocytes (Nakamoto and Kanai, 2014). Additionally, the liver and the portal vein wall receive innervation from vagal afferents, which also express TLR4 receptors (Berthoud et al., 1992; Hosoi et al., 2005; Kunda et al., 2014). Given that the sleep-promoting and febrile responses to LPS manifest after a latency of 2–4 hours, it is unlikely that these effects result from the direct activation of sensory nerves by LPS. Instead, it suggests that the effects involve the activation of a slower biochemical machinery leading to the production of other molecular mediators in the liver. Among the liver cell types, Kupffer cells are the most extensively studied targets for LPS. In response to LPS, Kupffer cells produce molecules known to affect sleep, including proinflammatory cytokines such as interleukin (IL)-6 and tumor necrosis factor-α (TNFα), along with PGE2, (Nakamoto and Kanai, 2014; Bowers et al., 1985; Peters et al., 1990).
Our unpublished preliminary data, demonstrating the complete abolition of the sleep-inducing effects of systemic LPS in cyclooxygenase (COX)-2 knockout mice, prompted further investigation into the involvement of prostaglandins in mediating the effects of intraportally administered LPS. Notably, pretreatment with the COX inhibitor indomethacin significantly suppressed the sleep-inducing actions of LPS, underscoring the role of prostaglandins in mediating these effects. The findings that intraportal injections of PGE2 and the prostaglandin precursor AA also exhibit sleep-promoting effects indirectly support the involvement of hepatic prostaglandin production in LPS-induced sleep. Peripherally produced prostaglandins have been demonstrated to possess sleep-promoting properties. Nicotinic acid, a potent stimulant for prostaglandin synthesis in keratinocytes and Langerhans cells of the skin, exhibits robust sleep-promoting effects. Research has shown that sleep induced by nicotinic acid is abolished by indomethacin (Szentirmai and Kapas, 2019).
The role of prostaglandins in LPS-induced sleep has been previously investigated in mice (Oishi et al., 2015). In that study, somnogenic effects of high doses (~ 100 µg/kg) of systemically administered LPS were not blocked by the COX inhibitor meloxicam, and the effects were not attenuated in mice lacking EP3 receptors in the nervous system or mice with a total body KO of microsomal PGE synthase-1 or the PGD2 receptor type DP. There was a slight reduction in the sleep-inducing effects in mice with a nervous system-specific knockout of the EP4 receptor. It is important, however, to emphasize the differences between our model, where a low dose of LPS was injected intraportally and did not elicit changes in systemic LPS levels, and the above mouse model, where high doses of LPS were injected systemically, likely reaching central and other peripheral targets. Although LPS does not cross the blood-brain barrier (Banks and Robinson, 2010), it can still act on central sites, such as endothelial cells of the cerebral vessels, or neurons and glia at sites where the blood-brain barrier is incomplete and permeable to LPS (Matsumura et al., 1998; Blatteis et al., 1983). When LPS reaches brain target sites, it can induce fever and sleep (Krueger et al., 1986; Zielinski et al., 2017; Steiner and Branco, 2000). Furthermore, the sleep-promoting actions of high doses of systemically administered LPS are also attributed to uncoupling protein-1-dependent thermogenic mechanisms of the brown adipose tissue (Szentirmai and Kapas, 2018). These central and extrahepatic peripheral sleep-promoting effects of LPS may occur independently of the prostaglandin system.
Increased REMS observed after lower doses of LPS, and the REMS suppression following the highest intraportal dose, align with the notion that increasing doses of systemically administered LPS trigger REMS suppression (Szentirmai and Krueger, 2014). The observation that indomethacin pretreatment suppressed both body temperature and REMS, coupled with the finding that high LPS doses also suppress both body temperature and REMS, suggests a potential relationship between reduced REMS and body temperature. This is consistent with the concept that the actual thermoregulatory changes themselves may play a role in sleep regulation (Gilbert et al., 2004).
It has been proposed that hepatic prostaglandin production is a key factor in fever induced by systemic LPS administration (Roth and Blatteis, 2014). Our findings are consistent with this notion. Our results not only demonstrate an increase in sleep but also reveal an elevation in body temperature following the lowest dose of intraportally injected LPS, strongly suggesting a hepatic site of action for LPS-induced fever. The observed 4-h latency to increased body temperature may be attributed to the de novo production of a febrile mediator. The complete abolition of fever by indomethacin underscores the critical role of prostaglandins in this response.
Moreover, indomethacin pretreatment not only completely prevented LPS-induced fever but, intriguingly, transformed the response into profound hypothermia. Similar hypothermic responses to LPS have been reported in other models with suppressed prostaglandin production or in mice deficient in IL-6, IL-10, and NF-κB (Kozak et al., 1994; Zhang et al., 2003; Toth and Opp, 2001; Morrow and Opp, 2005; Jhaveri et al., 2006). These collective findings support the notion that LPS has two independent effects on body temperature (Dogan et al., 2002). It promotes fever through the activation of the proinflammatory IL-6/IL-10 – PGE2 – NF-κB axis and, independently, it induces hypothermia. The activation of the proinflammatory arm likely masks the hypothermic effects, but inhibiting the proinflammatory process unveils hypothermia. Furthermore, higher doses of systemic LPS exhibit biphasic effects on body temperature, with fever preceded by an initial hypothermic phase. This dual action of LPS on body temperature may explain the observed biphasic response. Additionally, beyond the "classic" LPS receptor TLR4, LPS also binds to and activates transient receptor potential A1 (TRPA1) receptors on vagal afferents (Meseguer et al., 2014). Transsection of the subdiaphragmatic vagal trunk abolishes the hypothermic effect of high doses of LPS (Kapas et al., 1998), while the activation of vagal TRPA1 receptors by IL-1 or 2-methyl-2-thiazoline induces hypothermia (Matsuo et al., 2021; Silverman et al., 2023). Therefore, it is possible that the hypothermic actions of LPS are mediated through the activation of vagal TRPA1 receptors.
EEG SWA reflects the prevalence of delta waves during NREMS. EEG SWA is influenced by sleep pressure (Borbély et al., 2016), and can also change independently of sleep-wake activity in response to various drugs such as atropine (Bradley, 1968), metabolic alterations (Lewis et al., 1974), or the activity of thermoregulatory mechanisms (Berger et al., 1998; Gaenshirt et al., 1954). The biphasic SWA responses after low-dose intraportal LPS treatment were similar to those observed in response to 100 µg/kg LPS given ip (Kapas et al., 1998). It remains unclear whether the LPS-induced suppression of EEG SWA reflects reduced NREMS pressure or if it represents a sleep-independent effect on cortical neuronal activity. In either scenario, the effect likely originates from the hepatoportal region, as evidenced by the lack of effect after extraportal systemic LPS treatment. Indomethacin abolished the first phase of EEG SWA suppression, suggesting the involvement of prostaglandins. Previously, we reported that subdiaphragmatic vagotomy abolishes the effect of systemic LPS treatment on EEG SWA (Kapas et al., 1998). Furthermore, vagal stimulation has a profound effect on EEG activity (Grastyan et al., 1952; Bonvallet and Sigg, 1958; Chase et al., 1967). These observations are consistent with the notion that the EEG effects of LPS may be mediated, at least in part, by vagus afferents in response to prostaglandin production in the liver.
Indomethacin pretreatment did not completely prevent LPS-induced sleep, suggesting that besides prostaglandins, other mediators may also play a role in the effects of LPS. Kupffer cells, stellate cells, and sinusoidal endothelial cells produce TNFα and IL-6 in response to LPS (Seki et al., 2001; Thirunavukkarasu et al., 2005). Both cytokines have NREMS-promoting actions, potentially contributing to the effects of LPS (Kapas et al., 1992; Szentirmai and Kapas, 2019; Hogan et al., 2003). These mediators may be released into the hepatic vein and reach distant targets through the systemic circulation. Alternatively, they could act locally on afferent neurons, as both the vagus and spinal nerves provide afferent innervation to the liver. Vagal afferents express the PGE2 receptors EP3 and EP4 as well as TNFα p55 receptors (Hermann et al., 2005; Steinberg et al., 2016; Ek et al., 1998). Both PGE2 and TNFα stimulate afferent vagal activity (Birrell et al., 2013; Hermann et al., 2005). It is possible that locally produced PGE2, and possibly other mediators, stimulate vagal afferents to send somnogenic signals to brain core sleep circuits. This aligns with findings indicating that the vagus nerve carries sleep-inducing input to the brain (Armitage et al., 2003; Bonvallet and Sigg, 1958), and the sleep-promoting and fever-inducing actions of systemically administered high doses of LPS are suppressed by vagal nerve transection (Kapas et al., 1998; Opp and Toth, 1998).
In summary, our findings strongly support the idea that LPS, translocated from the intestinal lumen into the portal circulation, plays a role in modulating sleep and body temperature. We propose that, in conjunction with other bacterial metabolites such as butyrate and LTA, and along with other potential microbial molecules, LPS contributes to a complex molecular signaling repertoire targeting the liver to activate hepatoportal sensors. The signals arising from these sensors likely traverse to central sleep circuits, utilizing either neuronal or humoral pathways. This multifaceted molecular interplay underscores the intricate connections between the gut microbiota, microbial molecules and the regulation of sleep-wake cycles.