1 Main text

Ingestion of food contaminated with bacterial toxins induces a series of defensive responses, including nausea, retching, and vomiting [1, 2]. Nausea is an uncomfortable sensation that serves as a teaching signal for conditioned flavor avoidance (CFA) to prevent organisms from later consuming the same poisonous food. Retching and vomiting facilitate the rapid expulsion of ingested toxic food through contractions of the gastrointestinal tract. These defensive responses play an important role in the survival of humans and animals.

Researchers have intensively explored how the brain initiates these toxin-induced defensive responses. Three recent studies suggest that an area called the dorsal vagal complex (DVC) may serve as the brain center for initiating these defensive responses [3,4,5] (Fig. 1). The DVC includes three nuclei, i.e., area postrema (AP), nucleus of the solitary tract (NTS), and dorsal motor nucleus of the vagus (DMV). Ingested toxins (e.g., bacterial toxins or therapeutic drugs) may activate DVC neurons via at least two pathways. The first pathway involves blood-borne toxins directly activating neurons in the AP, which lacks a blood–brain barrier [3, 4]. The second pathway involves a gut-to-brain axis that relays toxin signals from the gastrointestinal tract to NTS neurons [5]. In this commentary, we discuss the results of these studies.

Fig. 1
figure 1

Two pathways to mediate toxin-induced defensive responses. A Schematic diagram showing the ingestion of food contaminated with toxins in mice and the dorsal vagal complex (DVC). B Schematic diagram showing that toxins in the blood or gut may activate brainstem neurons via two neural pathways. In the first pathway, blood-borne toxins may directly activate excitatory and inhibitory neurons in the AP, which lacks a blood–brain barrier. The second pathway involves a gut-to-brain axis that relays toxin signals from the gastrointestinal tract to Tac1 + excitatory NTS neurons. These two pathways may converge onto the PBN to promote nausea. In addition, the Tac1 + NTS neurons may induce retching-like behavior via NTS-rVRG pathway

Full size image

Zhang et al. explored the role of the AP in nausea sensation [3]. Since the previous studies on AP’s role in nausea cannot exclude the influence of nearby neurons, a clear analysis of AP neuron types is lacking. By using single-nucleus RNA sequencing, they identified four principal excitatory neuron types (clusters 1–4) and three principal inhibitory neuron types (clusters 5–7). Clusters 2–4 were of particular interest due to the role of GLP1R agonists in suppressing appetite and evoking nausea [3]. Nausea is indicated by CFA in the lab research. Water-restricted mice are placed in the water-free test arena for 3 days. On day four, two bottles in the arena are filled with either cherry-flavored water or grape flavored water. Mice are injected with saline water alone or conditioned factor, such as clozapine N-oxide (CNO). On day five, both bottles are filled with water. On day six, one bottle contains grape-flavored water and the other one contains cherry-flavored water. The consumptions of each flavored water are recorded and show whether mice display CFA. Chemogenetic activation of clusters 1–4 showed that activation of clusters 2 and 4 induced CFA, suggesting that these two clusters may function in nausea sensation. The researchers then examined whether these two clusters were required for nausea sensation in response to toxins. Ablation of GLP1R + AP neurons (clusters 2, 3, and 4) abolished CFA to exendin-4, lithium chloride, and lipopolysaccharide, and reduced food intake, suggesting a minor role of circuits in mediating appetite; ablation of GFRAL + AP neurons (cluster 4) abolished CFA to lithium chloride and lipopolysaccharide; ablation of neither cluster 2 nor cluster 4 eliminated CFA to exendin-4, likely because the response to exendin-4 involved both clusters.

The researchers mapped the projections of AP neurons and found that excitatory and inhibitory AP neurons displayed different projection patterns. Excitatory neurons projected to many brain areas, including the parabrachial nucleus (PBN), nearby NTS, and autonomic motor nuclei, while inhibitory neurons formed connections with AP neurons themselves and partly with NTS neurons. Clusters 2 and 4 primarily projected to calcitonin gene-related peptide (CGRP) + PBN neurons, previously reported as alarm neurons [3]. These results revealed the possible circuit underlying nausea sensation.

Transcriptome data also indicated that the calcium sensing receptor (CaSR) gene were highly expressed in clusters 2 and 4. Cinacalcet, an agonist of CaSR, caused nausea, while ablation of GLP1R + AP neurons (clusters 2, 3, and 4) abolished CFA to cinacalcet. Therefore, Zhang and colleagues hypothesized that any receptor agonist can induce nausea-associated behaviors as long as the receptor is located in a certain region of the AP.

As AP inhibitory neurons also form connections with local neurons [3], Zhang et al. studied the function of inhibitory AP neurons in nausea and employed channelrhodopsin-2 (ChR2)-assisted circuit mapping and whole-cell recordings to investigate downstream targets of AP inhibitory neurons [4]. Results showed that AP inhibitory neurons were connected to most AP excitatory neurons, as well as some NTS excitatory neurons and other AP inhibitory neurons. Notably, CFA induced by growth differentiation factor 15 (GDF15) and lithium chloride was abolished when AP inhibitory neurons were chemogenetically activated, suggesting that AP inhibitory neurons inhibit the activity of nausea-promoting excitatory neurons.

To investigate the cellular mechanism of this inhibition, Zhang et al. focused on GIPR AP neurons (cluster 6 in their previous study), as glucose-dependent insulinotropic polypeptide (GIP) can suppress GLP1-induced adverse responses [4]. Mapping of GIPR neurons revealed that they mainly projected to local excitatory neurons and some NTS neurons. To investigate the circuit recruited by GIP, the researchers used whole-cell recordings and found that GIPR neurons were directly activated by GIP; GIPR neuronal mapping also suggested that they formed connections with most excitatory AP neurons. They further applied whole-cell recordings to identify the particular receptors of GIPR neurons and found that GIPR neurons selectively inhibited GFRAL neurons (cluster 4).

However, it is unclear how the toxins in the gut induce nausea and vomiting during food poisoning. Recently, Xie et al. explored the gut-to-brain circuit that mediates toxin-induced defensive responses [5]. Unlike previous studies using emesis-competent species as animal models, they developed a mouse-based paradigm to study the gut-to-brain axis and how the brain detects and responds to toxins at the cellular level. They observed unusual mouth-opening actions in mice after intraperitoneal injection of Staphylococcal enterotoxin A (SEA) instead of saline. The abnormal mouth-opening activity showed greater amplitude and longer duration than occasional spontaneous mouth-opening action, resembling retching behavior. Additional physiological evidence, such as electromyogram (EMG) results, further characterized retching-like behavior. The researchers also demonstrated that SEA caused nausea in mice, and the physiology of retching-like behavior and nausea in mice was analogous to vomiting and nausea in emesis-competent species. Thus, the team developed an experimental paradigm to investigate the gut-to-brain axis in mice.

Using the FosTRAP2 strategy, Xie et al. showed that DVC neurons were activated during SEA-induced defensive responses, suggesting a gut-to-brain axis in mice. To identify the cell types controlling defensive responses in the DVC, they employed fluorescent in situ hybridization (FISH) and discovered several neuropeptide markers, including preprotachykinin 1 (Tac1), cholecystokinin (Cck), and neuropeptide Y (Npy). Tac1 + DVC neurons were predominantly located in the NTS. As Tac1 encodes neuropeptides targeting the neurokinin 1 receptor, which plays an important role in vomiting in emetic animals [6], the researchers considered that Tac1 + DVC neurons may function in toxin-induced defensive responses in mice.

Chemogenetic manipulation revealed that only inactivation of Tac1 + DVC neurons significantly impaired SEA-induced retching-like behavior and CFA, highlighting the critical role of Tac1 + DVC neurons in SEA-induced defensive responses. In subsequent experiments, the researchers found that Tac1 + DVC neurons released glutamate and Tac1-encoded neuropeptides, both required for defensive responses to SEA.

The researchers next investigated the compartments involved in the gut-to-brain circuit. Results from retrograde tracing and FISH showed that Tac1 + DVC neurons were mainly innervated by 5-hydroxytryptamine receptor 3A (Htr3a) + ipsilateral vagal sensory neurons. Strikingly, the terminals of the labeled vagal neurons in the gastrointestinal tract were close to 5-HT + enterochromaffin (EC) cells. Thus, the researchers postulated that 5-hydroxy tryptamine (5-HT) released by EC cells may activate vagal neurons, as demonstrated in subsequent experiments. Anterograde tracing revealed that the ipsilateral rostral part of the ventral respiratory group (rVRG) and PBN were the main target regions of Tac1 + DVC neurons. Chemogenetic manipulation of rVRG- and PBN-projecting Tac1 + DVC neurons confirmed that they were required and sufficient for SEA-induced retching-like behavior and CFA, respectively. Taken together, circuits involving EC cell-vagal sensory neuron-DVC-rVRG/PBN were suggested for SEA-induced defensive responses in mice.

Xie and colleagues used a common chemotherapeutic drug (doxorubicin) as a reagent to validate the gut-to-brain axis. Doxorubicin induced both retching-like behavior and CFA in mice. Glutamate and Tac1-encoded neuropeptides released by Tac1 + neurons and 5-HT synthesized by EC cells were required for doxorubicin-induced defensive responses. Furthermore, rVRG- and PBN-projecting Tac1 + DVC neurons were also required for doxorubicin-induced retching-like behavior and CFA, respectively.

The research team also wondered how the gut-to-brain axis is stimulated by toxins and hypothesized that the axis may be recruited indirectly through the immune-neuroendocrine circuit. They tested this hypothesis by deleting critical genes in the immune-neuroendocrine circuit, which resulted in impaired defensive responses to SEA and doxorubicin.

In summary, these three recent studies shed light on the mechanisms underlying toxin-induced defensive responses in mice (Fig. 1). Zhang and his teamfocused on AP neurons and proposed that clusters 2 and 4 mediate nausea sensation and cluster 6 suppresses nausea by forming inhibitory connections with cluster 4. Xie and his teamdeveloped an experimental paradigm to study toxin-induced defensive behavior in mice and proposed a gut-to-brain axis controlling retching-like behavior and CFA, consisting of 5-HT + intestinal EC cells, Htr3a + vagal sensory neurons, Tac1 + DVC neurons, and rVRG/PBN (downstream to upstream). These findings lay a solid foundation for further defining the circuits underlying defensive responses to toxins and elucidating effective interventions during chemotherapy to reduce therapeutic drug-induced side effects. Notably, two pathways that mediated nausea suggested in the studies converge onto PBN. The specificity of two pathways also remains unclear. SEA and doxorubicin apply gut-to-brain axis, while LiCl does not; the study of gut-to-brain axis also do not rule out the possibility that defensive responses to SEA and doxorubicin are mediated by AP. Further studies are needed to better understand the neural circuits and identify effective methods to suppress toxin-induced defensive responses.