Stress-activated brain-gut circuits disrupt intestinal barrier integrity and social behaviour

Chronic stress underlies the etiology of both major depressive disorder (MDD) and irritable bowel syndrome (IBS), two highly prevalent and debilitating conditions with high rates of co-morbidity. However, it is not fully understood how the brain and gut bi-directionally communicate during stress to impact intestinal homeostasis and stress-relevant behaviours. Using the chronic social defeat stress (CSDS) model, we find that stressed mice display greater intestinal permeability and circulating levels of the endotoxin lipopolysaccharide (LPS) compared to unstressed control (CON) mice. Interestingly, the microbiota in the colon also exhibit elevated LPS biosynthesis gene expression following CSDS. Additionally, CSDS triggers an increase in pro-inflammatory colonic IFNγ+ Th1 cells and a decrease in IL4+ Th2 cells compared to CON mice, and this gut inflammation contributes to stress-induced intestinal barrier permeability and social avoidance behaviour. We next investigated the role of enteric neurons and identified that noradrenergic dopamine beta-hydroxylase (DBH)+ neurons in the colon are activated by CSDS, and that their ablation protects against gut pathophysiology and disturbances in social behaviour. Retrograde tracing from the colon identified a population of corticotropin-releasing hormone-expressing (CRH+) neurons in the paraventricular nucleus of the hypothalamus (PVH) that innervate the colon and are activated by stress. Chemogenetically activating these PVH CRH+ neurons is sufficient to induce gut inflammation, barrier permeability, and social avoidance behaviour, while inhibiting these cells prevents these effects following exposure to CSDS. Thus, we define a stress-activated brain-to-gut circuit that confers colonic inflammation, leading to impaired intestinal barrier function, and consequent behavioural deficits.


Main Text
Chronic psychosocial stress is a major risk factor for neuropsychiatric disorders including major depressive disorder (MDD), as well as functional gastrointestinal disorders such as irritable bowel syndrome (IBS), two of the most prevalent and debilitating illnesses that show high rates of comorbidity 1 .Potentially underlying this co-morbidity is the recent nding that chronic stress elicits lowgrade in ammation, which is associated with the severity of both MDD and IBS symptoms 3,4 .Importantly, both of these conditions are multifactorial disorders with limited treatment options; thus, there is a need to identify biological changes associated with their pathogenesis.Moreover, increasing evidence suggests that the gut-brain axis, or connections between the central (CNS) and enteric nervous systems (ENS), contributes to the etiology of MDD and IBS 5,6 .However, the mechanisms by which psychological states such as chronic stress in uence gut pathophysiology, including in ammation and permeability, remain poorly understood.Further, the impact of gut in ammation on gut-to-brain signaling and stress-relevant behaviours is largely unexplored.
The intestinal barrier has gained attention for its role in in ammatory diseases, as it regulates the translocation of pro-in ammatory antigens and microbial compounds from the gut lumen into circulation, and is home to 70% of the body's immune cells 7 .An increase in circulating bacterial toxins -or endotoxaemia -is hypothesized to initiate and propagate systemic in ammation by activating innate immune responses during chronic in ammatory conditions 8 .Importantly, there is evidence that the intestinal barrier becomes compromised in patients with MDD or IBS [9][10][11] .However, the mechanisms by which this occurs and the consequences on neuropsychiatric symptoms including behaviour remain largely unknown.In this study, we use a mouse model of chronic psychosocial stress to uncover a pathway through which the CNS, ENS, and immune system communicate to trigger gut in ammation, barrier permeability, and social avoidance behaviour.

Chronic social stress promotes intestinal permeability and endotoxaemia
Using the chronic social defeat stress (CSDS) model (Fig. 1a), which induces social avoidance behaviour in both male (Fig. 1b, c) and female experimental mice (Extended Data Fig. 1a, b), we rst tested if animals exposed to chronic psychosocial stress display increased intestinal permeability.After orally gavaging mice with 4 kDa FITC-dextran, which does not normally cross an intact intestinal epithelium, we assessed FITC uorescence in plasma 1 or 4 h later, corresponding to gastrointestinal transit time from the stomach to the small or large intestine, respectively 12 .At both time points, male and female mice that underwent CSDS showed signi cantly greater gut permeability compared to unstressed control (CON) mice, with plasma FITC concentrations correlating with social avoidance behaviour (Fig. 1d-g, Extended Data Fig. 1c, d).Importantly, when gut permeability was measured before the 10 day CSDS protocol, there were no pre-existing differences in permeability before stress exposure that predicted consequent social avoidance behaviour.In fact, the majority of mice that underwent CSDS displayed increased gut permeability after the 10 day paradigm (Extended Data Fig. 1e-h), indicating that a leaky gut is indeed a consequence of chronic stress.Additionally, several claudins, which form tight junctions between epithelial cells, such as Cldn4, Cldn5, Cldn8, and Cldn15, were transcriptionally downregulated throughout the intestine following CSDS compared to CON mice (Fig. 1h).Further, there was a marked reduction in goblet cell number, but not cell size, in the colons of mice after CSDS (Extended Data Fig. 1i-k), suggesting a loss of mucous, which normally protects the intestinal epithelium against pathogenic invasion 13 .These results demonstrate that chronic stress compromises intestinal barrier integrity.
As a potential consequence of a damaged intestinal barrier, we analysed circulating levels of lipopolysaccharide (LPS), a component of the gram-negative bacterial outer membrane that elicits proin ammatory host responses upon recognition by its receptor toll-like receptor 4 (TLR4) 14 .After CSDS, stressed mice had signi cantly higher plasma LPS levels than CON mice, with circulating LPS concentrations negatively correlating with social interaction behaviour (Fig. 1i, j).Of note, we found no correlation between the degree of wounding acquired during CSDS and plasma LPS levels (Extended Data Fig. 1l); we therefore speculated that stress-induced endotoxaemia originates from gut translocation rather than exogenous infection.We next performed functional metagenomic analysis of colon luminal contents from stressed and CON mice.First, we identi ed that CSDS-exposed mice had a distinct microbiome compared to CON mice, characterized by a shift in beta diversity and a reduced Firmicutes/Bacteroidetes ratio (Extended Data Fig. 2a-e).Next, while the majority of differentially expressed microbial genes regulated by stress were downregulated in colon contents following CSDS, genes involved in the LPS biosynthesis pathway were signi cantly enriched (Fig. 1k, l, Extended Data Fig. 2f, Supplementary Table 1), suggesting that CSDS-induced endotoxaemia may be a consequence of both increased barrier leakiness and LPS production by gut bacteria.Hypothesizing that gut permeability and endotoxaemia could initiate systemic in ammation contributing to social avoidance behaviour, we depleted TLR4 from haematopoietic cells via bone marrow transplantation to restrict gene depletion to peripheral immune cells, then subjected these chimeric mice to CSDS (Extended Data Fig. 3a-d).
Interestingly, we found that mice lacking haematopoietic TLR4 (Tlr4 -/-→WT) were partially protected against social avoidance triggered by CSDS, compared to mice transplanted with wild-type bone marrow (WT→WT) (Fig. 1m, n).Taken together, these results indicate that chronic stress disrupts the intestinal barrier, allowing LPS translocation into circulation, which impacts social behaviour.

Colonic in ammation contributes to stress-evoked gut permeability
We next investigated how stress could impact intestinal permeability, hypothesizing that proin ammatory polarization of enteric immune cells breaks down the gut barrier, as seen in conditions such as in ammatory bowel disease (IBD), obesity, and intestinal ischaemia 15,16 .Focusing on the colon, where most gut bacteria are found, we isolated lamina propria lymphocytes from the colons of both male and female mice under CSDS and CON conditions, and assessed speci c T cell subpopulations by ow cytometry.In both sexes, while there were no differences in total T cell frequencies between CON and CSDS mice, there were signi cantly greater numbers of CD4 + helper T cells, and notably, increased proin ammatory IFNγ + Th1 cells in mice exposed to CSDS, while IL4 + Th2 cells were reduced (Fig. 2a-h, Extended Data Fig. 4a-d).Moreover, we did not observe increases in IL17A + Th17 cells or innate immune cells often associated with colon in ammation, such as monocytes, neutrophils, and dendritic cells (Extended Data Fig. 4e-m).CSDS therefore triggers Th1-mediated colonic in ammation.
To determine the contribution of this in ammation to stress-induced intestinal permeability and social behaviour de cits, we used mice lacking integrin β7 (ITGβ7), a transmembrane protein necessary for gut homing of immune cells 17 .After con rming that the majority of enteric leukocytes express ITGβ7, and that Itgb7 -/-mice had reduced total leukocytes, B cells, and T cells in the colon (Extended Data Fig. 5a-e) compared to wild-type mice, we put Itgb7 -/-and littermate wild-type control mice through CSDS, and found that ITGβ7-de ciency ameliorates stress-induced colon in ammation (Fig. 2i, j, Extended Data Fig. 5f).In addition, Itgb7 -/-mice were partially resistant to gut permeability, and social avoidance triggered by CSDS (Fig. 2k-m), indicating that intestinal immune cells play a role in these stress-linked phenotypes.

Noradrenergic enteric neurons in uence gut in ammation and barrier permeability
As chronic stress heightens sympathetic tone from the autonomic nervous system (ANS), in conjunction with the colon receiving dense noradrenergic innervation from neurons in the myenteric and submucosal plexus [18][19][20] , we examined if noradrenergic enteric neurons are involved in gut in ammation and barrier permeability during CSDS.We rst assessed if noradrenergic neurons in the myenteric and submucosal plexus become activated by stress.Using iDISCO+ whole-tissue clearing and staining of colon samples from CON and CSDS mice, we found that in the CSDS group, enteric neurons expressing dopamine betahydroxylase (DBH), a marker for noradrenergic neurons, show signi cantly elevated levels of cFos, a molecular marker of neuronal activity (Fig. 3a, b).To understand if these neurons contribute to stressinduced gut pathophysiology, we injected saporin (SAP) conjugated to anti-DBH antibodies throughout the wall of the colon, to effectively ablate DBH + neurons selectively in the colon (Fig. 3c, d).Upon exposure to CSDS, we observed that anti-DBH-SAP blunted the stress-evoked increase in Th1 cells in the colon compared to mice that received control infusions of IgG-SAP (Fig. 3e).Moreover, colonic DBH + cell depletion dampened CSDS-induced gut permeability (Fig. 3f) and social avoidance behaviour (Fig. 3g, h).These experiments demonstrate that noradrenergic enteric neurons become activated by psychosocial stress and regulate colonic in ammation and barrier permeability.
Stress-induced gut pathophysiology is regulated by the CNS Mechanistically, we aimed to identify brain regions that become activated by stress that can convey signals to the ENS to in uence intestinal in ammation and barrier function.To accomplish this, we rst injected pseudorabies virus carrying RFP (PRV-RFP) into the colon as a polysynaptic retrograde tracer, then performed whole-brain clearing with iDISCO+ and light sheet imaging to label CNS regions that project to the colon (Fig. 4a, b).In parallel, we assessed neuronal activity by staining cFos in whole brains following CSDS (Fig. 4c), and generated a comprehensive map of stress-activated gut-innervating brain regions (Fig. 4d, Supplementary Table 2).We identi ed the paraventricular nucleus of the hypothalamus (PVH) as a brain region with both strong connectivity to the colon and signi cantly increased cFos expression following CSDS.Given that the PVH is implicated in regulating immunological processes in peripheral tissues, such as the spleen and bone marrow, as well as homeostatic gastrointestinal functions including motility 21,22 ; we chose to investigate this region further.Our next objective was to characterise neuronal subpopulations in the PVH that innervate the colon.We hypothesized that corticotropin-releasing hormone (CRH)-expressing neurons could be involved, as CRH knockout ameliorates in ammation during experimental colitis, and intracerebroventricular CRH administration can mimic stress-induced colonic dysmotility 23,24 .To assess if PVH CRH + neurons project to the gut, we delivered Cre-dependent adeno-associated virus expressing EYFP (AAV-DIO-EYFP) into the PVH of Crh Cre mice, then three weeks later, injected PRV-RFP into the colons of the same mice (Fig. 4e).Co-localization of EYFP and RFP revealed that approximately 50% of all gut-innervating PVH neurons express CRH (Fig. 4f, g), con rming that these neurons are structurally connected to the colon.
To test whether PVH CRH + neurons can control enteric immunity and barrier function, we expressed the excitatory designer receptor exclusively activated by designer drugs (DREADD) hM3Dq in the PVH of Crh Cre mice.Similar to a previously established method to induce stress responses by chronically activating CRH + cells 25 , we performed chemogenetic activation of PVH CRH + neurons by administering clozapine N-oxide (CNO) once daily for 10 consecutive days, to mimic conditions during the 10 day CSDS paradigm (Fig. 4h, i).Interestingly, independent of stress, chronic PVH CRH + neuron stimulation was su cient to induce low-level colon in ammation and gut barrier permeability, and diminish social interaction behaviour (Fig. 4j-n).Importantly, CNO administration alone did not impact gut physiology or social behaviour in the absence of PVH DREADD expression (Extended Data Fig. 6a-e).As PVH CRH + neurons may in uence peripheral immune responses and behaviour through neuroendocrine mechanisms via the HPA axis 26 , we administered metyrapone to block corticosterone synthesis during chemogenetic excitation of PVH CRH + cells and found no effects on gut barrier permeability (Extended Data Fig. 7 a, b).Moreover, HPA inhibition was found to worsen social avoidance behaviour induced by PVH CRH + neuron activation (Extended Data Fig. 7c, d); thus, central regulation of gut pathophysiology and associated behavioural de cits may be exerted through direct neural connections via the ANS rather than the HPA axis.Lastly, we explored whether silencing PVH CRH + neurons could mitigate CSDS-induced gut and behavioural phenotypes using the inhibitory DREADD hM4Di (Fig. 4o, p).Inhibition of PVH CRH + neurons throughout CSDS partially prevented stress-induced gut in ammation, barrier permeability and social avoidance (Fig. 4q-u).Collectively, we identi ed that CRH + neurons in the PVH are directly linked to the colon and can regulate enteric physiology during stress.

Discussion
Here, we describe a pathway in which psychosocial stress activates CRH + neurons in the PVH, which relay signals to the ENS.Noradrenergic enteric neurons then trigger Th1-mediated colonic in ammation, which contributes to intestinal barrier permeability and endotoxaemia.Consequently, circulating endotoxins are detected by haematopoietic TLR4 to promote social avoidance behaviour.Our ndings are consistent with previous reports showing that activating PVH CRH + neurons can re-capitulate in ammatory leukocyte dynamics in the bone marrow and spleen, similar to what is observed following stress exposure 27,28 , and that depleting CRH receptor type 1 (CRFR1) from the PVH prevents CSDS-induced anxiety-like behaviour 29 .Further, CRH has been implicated in in ammation associated with experimental colitis 23 .Interestingly, a recent study found that chronic restraint stress exacerbates dextran sodium sulphate (DSS)-induced colitis through a mechanism dependent on the HPA axis, enteric glia, and colonic monocytes 30 .Together with our ndings, this suggests that depending on disease context, chronic stress may elicit colonic in ammation through either the peripheral nervous system or HPA axis, and involve enteric neurons and glia, along with colonic T cells and monocytes.In the case of CSDS, our data suggest that PVH CRH + neurons control intestinal permeability and in ammation through enteric neurons via the ANS, but not necessarily through HPA activation or stress hormones.
Recent single-cell and single-nucleus sequencing datasets have con rmed the presence of neurons expressing DBH and noradrenaline in the colon 19,20 .Our results showing that colonic DBH + neurons become hyperactive during stress suggest that there might be enteric neuron subsets implicated in psychiatric and in ammatory disorder co-morbidity.Although there is evidence that noradrenaline acts on β2-adrenergic receptors (β2AR) on CD4 + T cells to in uence their differentiation and cytokine expression, opposing outcomes have been reported depending on timing of activation.While noradrenaline can suppress IFNγ production by Th1 cells, β2AR stimulation on naïve CD4 + T cells results in greater IFNγ expression upon differentiation into Th1 cells 31 .Thus, we hypothesize that colonic noradrenergic neurons act on naïve CD4 + T cells to provoke gut in ammation during stress.
Moreover, we speculate that Th1-mediated in ammation causes gut permeability as IFNγ is known to directly disrupt intestinal epithelial barrier formation, permitting bacterial translocation 32 .However, it is possible that additional colonic immune cells are involved, as other pro-in ammatory cytokines such as TNFα have been demonstrated to downregulate epithelial tight junction proteins during IBD 33 .While measurements of endotoxaemia during chronic stress and psychiatric disorders have yielded mixed results, faecal samples from patients with MDD or anxiety show enrichment for LPS biosynthesis pathway genes, consistent with our ndings in mice 9,34 .In addition, whole-body TLR4 depletion in mice is protective against stress-induced social avoidance and learned helplessness 35,36 .Our experiments using bone marrow chimeric mice speci cally implicate haematopoietic TLR4 in this process.Notably, peripheral blood cells from patients with MDD or from mice susceptible to CSDS also display heightened sensitivity to TLR4 stimulation [37][38][39] .
The stress-activated brain-gut circuits described in this study represent potentially novel targets for the treatment of stress-related psychiatric disorders, in a tissue that is more accessible for pharmacological intervention than the brain.As antidepressant drugs are predominantly administered orally, it is possible that existing therapeutic options may exert effects through the ENS and intestinal immune system.Notably, antidepressants are effective in treating IBS, although whether the mechanisms of action are local or top-down through the CNS, remain to be fully elucidated 40,41 .It is therefore critical for future work to further dissect how the bi-directional communication between the brain and gut, which involves the immune system, becomes dysregulated under pathological conditions, and to identify new treatment targets to prevent or reverse such effects.
Bone marrow transplantation.To generate chimeric mice lacking TLR4 in haematopoietic cells, 6 weekold male CD45.1 recipient mice were lethally irradiated (11 Gy, in two 5.5 Gy doses, 4 h apart) using an Xrad 320 Irradiator (Precision X-Ray, Madison, CT).Bone marrow cells were then isolated from the femurs and tibia of 10-12 week-old male littermate wild-type or Tlr4 -/-donor mice, passed through a 70 μm cell strainer, and re-suspended at a concentration of 1 x 10 7 cells/mL in PBS.Recipient mice were then anaesthetized with iso urane and a total of 1 x 10 6 donor cells were injected retro-orbitally.Chimeric mice were given neomycin trisulphate (0.2% w/v in drinking water; N1876, Sigma-Aldrich, St. Louis, MO) for 10 days, then allowed to recover for an additional 3 weeks before experimentation.

In vivo interventions and procedures
Chronic social defeat stress.Chronic social defeat stress (CSDS) in male 2 and female 42 mice was performed as previously reported.For CSDS in male mice, 4-6 month-old male CD-1 retired breeder mice were screened for aggressive behaviour for three consecutive days, and non-aggressive mice were excluded.Aggressive CD-1 mice were housed on one side of a perforated Plexiglas partition in a hamster cage (26.7 x 48.3 x 15.2 cm) at least two days before CSDS.Experimental mice were then subjected to an encounter with an aggressive CD-1 mouse for 10 min (5 min for chimeric or denervated mice) per day, then transferred to the opposite side of the partition to allow for sensory but not physical interaction for the remainder of the day.This procedure was repeated for 10 consecutive days with a new aggressor each day.Unstressed control mice were pair-housed across a perforated Plexiglas partition.For CSDS in female mice, aggressors were generated by rst crossing homozygous Esr1 Cre mice with CD-1 mice, then bilaterally injecting Cre-dependent AAV2-hSyn-DIO-hM3D(Gq)-mCherry (44361-AAV2, Addgene, Watertown, MA) into the ventrolateral subdivision of the ventromedial hypothalamus in the heterozygous F1 offspring.Aggressive behaviour was subsequently elicited by intraperitoneally (i.p.) injecting mice with 1.0 mg/kg clozapine-N-oxide (CNO; 4936, Tocris, Bristol, United Kingdom) to activate ERα + cells 30 min before CSDS.Experimental mice were then subjected to a 5 min physical encounter once per day with a new aggressor for each day for 10 consecutive days.Following the nal day of CSDS, experimental mice were single-housed (male) or group-housed (female).All mice were carefully examined for wounding throughout CSDS experiments, with mice with excess wounding according to previously established criteria excluded 43 .
Social interaction test.Social interaction (SI) testing was performed as previously described 2 , 24 h after the nal day of CSDS.Following a 1 h habituation period under red light conditions in the behavioural suite, mice were allowed to freely explore a Plexiglas arena (45 x 45 x 45 cm, Nationwide Plastics, Arlington, TX) with an empty wire enclosure on one end for 2.5 min.Next, experimental mice were removed from the arena, and a novel social target mouse was placed into the wire enclosure.
Experimental mice were then returned to the arena for an additional 2.5 min.Locomotor activity was tracked and recorded using a Basler GenICam (acA1300-60, Basler, Ahrensberg, Germany) coupled to Noldus Ethovision version 11.0 (Noldus Information Technology, Leesburg, VA).SI ratio was calculated as the ratio of time spent in the 24 x 14 cm interaction zone around the wire enclosure in the presence of a social target divided by time spent in the absence of a social target.Corner time was calculated as the cumulative time spent in the two 9 x 9 cm corner zones opposite of the wire enclosure in the presence of a social target.FITC-Dextran gut permeability assay.FITC-Dextran with an average molecular weight of 3000 -5000 Da (FD4, Sigma-Aldrich) was prepared at a concentration of 120 mg/mL in mouse drinking water.Mice were fasted for at least 4 h, then orally gavaged with 600 mg/kg FITC-Dextran using 20 G x 38 mm exible gavage needles (FTP-20-38, Instech Laboratories, Plymouth Meeting, PA).At least one mouse per experiment was gavaged with drinking water alone for background subtraction.After 1 or 4 h, trunk blood or blood from the submandibular vein was collected into heparin-coated microcentrifuge tubes (41.1503.150,Sarstedt, Newtown, NC), then centrifuged at 375 x g for 10 min.Plasma was separated into new tubes then diluted 1 in 10 in PBS.Diluted plasma or FITC-Dextran standards in PBS were added to black 96-well plates, and plasma FITC-Dextran concentrations were measured on a SpectraMax Gemini XS uorescence microplate reader (Molecular Devices, San Jose, CA) using an excitation wavelength of 490 nm and emission wavelength of 520 nm.Mean uorescence intensity in plasma from water-gavaged mice was subtracted as background uorescence from all samples, and plasma FITC-Dextran concentrations were calculated by interpolating uorescence values against a standard curve.
Local colonic denervation.Anti-DBH-SAP (IT-03, Advanced Targeting Systems, Carlsbad, CA) or control IgG-SAP (IT-18, Advanced Targeting Systems) were prepared at a nal concentration of 0.7 mg/mL in sterile saline with 0.005% Fast Green FCF dye (F7252, Sigma-Aldrich) for visualization.Seven week-old male C57BL/6J mice were anaesthetized with iso urane, and the surgical site was shaved and sterilized with iodine and alcohol swabs.An approximately 1 cm lower abdominal midline incision was made with sterile surgical scissors, and the caecum and colon were carefully extracted from the peritoneal cavity onto a saline-moistened gauze pad using sterile forceps and cotton swabs.Under a dissection microscope, anti-DBH-SAP or IgG-SAP was injected into the colonic mucosa using a NanoFil Submicroliter Injection System (World Precision Instruments, Sarasota, FL) using 36 G beveled needles (NF36BV-2, World Precision Instruments), connected to a Legato 180 syringe pump (KD Scienti c, Holliston, MA).Approximately 120 nL of anti-DBH-SAP or IgG-SAP was injected at each injection site at a rate of 40 nL/s throughout the length of the colon.The caecum and colon were then returned to the peritoneal cavity, and the abdominal muscles and skin were sutured.Mice were given 10 days to recover post-surgery before experimentation.
Histology.Mice were euthanized with 10% chloral hydrate and transcardially perfused with ice-cold PBS.
Colon tissue was extracted and xed in 10% neutral buffered formalin (HT501128, Sigma-Aldrich) in a 'Swiss roll' con guration for 48 h, then para n-embedded using a Tissue-Tek VIP E300 Tissue Processor (Sakura, Torrance, CA).Formalin-xed, para n-embedded colon tissue was cut into 5 μm sections, and periodic acid-Schiff (PAS) staining was performed using the PAS Reaction Kit (k047, Poly Scienti c R&D Corp., Bay Shore, NY) according to the manufacturer's instructions to label goblet cells.Brie y, depara nized and rehydrated slides were incubated with 0.5% periodic acid for 5 min, Schiff Reagent for 15 min and Harris haematoxylin for 5 min, then quickly dipped in 0.5% acid alcohol followed by 1% lithium carbonate.Sections were dehydrated in 95% ethanol and cleared with xylene, then mounted onto slides.Images were acquired using a Zeiss Axio Imager.M1 microscope (Zeiss).Goblet cell size and number were analysed using ImageJ (National Institutes of Health, Bethesda, MD).
To amplify PRV-RFP signal in brain tissue, mice were euthanized with 10% chloral hydrate then transcardially perfused with ice-cold PBS followed by 4% paraformaldehyde (PFA).Brains were post-xed in 4% PFA for 12 h at 4 °C.Coronal sections were prepared on a vibratome at a thickness of 50 μm.Brain sections were incubated in blocking solution (PBS + 3% normal donkey serum + 0.3% Triton X-100) for 2 h, then incubated with primary antibodies against RFP (1:800, 600-401-379, Rockland Immunochemicals, Pottstown, PA) for 2 h.Samples were then washed three times with PBS and incubated in donkey antirabbit IgG (H+L) secondary antibody, Alexa Fluor 568 (1:1000, A10042, ThermoFisher Scienti c) for 2 h.Sections were washed three times with PBS before staining with DAPI (1 μg/mL) for 20 min.Slides were then mounted onto coverslips with EcoMount.Images were acquired using a Zeiss LSM 780 confocal microscope (Zeiss, White Plains, NY).Flow cytometry.For staining of blood samples, red blood cells were rst lysed using BD Pharm Lyse (555899, BD Biosciences) according to manufacturer's instructions, then washed in FACS buffer.For all samples, cells were blocked using anti-CD16/32 (2.5 μg/mL, BE0307, clone 2.4G2, Bio X Cell, Lebanon, NH), and incubated with Fixable Viability Dye eFluor 780 (1:4000, 65-0865-14, ThermoFisher Scienti c) on ice for 30 min, then washed.Cell surface staining was then performed with speci ed antibodies (Supplementary Table 4) on ice for 30 min.For staining of intracellular cytokines, cells were incubated in xation/permeabilization buffer (554714, BD Biosciences) on ice for 30 min, washed with BD Perm/Wash buffer (554714, BD Biosciences), then stained with speci ed antibodies (Supplementary Table 4) at °C in the dark overnight.For cell counting, CountBright absolute counting beads (C36950, ThermoFisher Scienti c) were used.Cells were washed, then re-suspended in FACS buffer before acquisition on a BD LSRFortessa cell analyzer (BD Biosciences).Data were analysed using FlowJo version 10.6.2 (BD Biosciences).iDISCO+ whole-tissue clearing staining.The iDISCO+ protocol was from http://www.idisco.info,with incubation times indicated for brain/colon tissues, respectively.

Figure 2 Stress
Figure 2