Stress-induced red nucleus attenuation induces anxiety-like behavior and lymph node CCL5 secretion

Abstract

Previous studies have speculated that brain activity directly controls immune responses in lymphoid organs. However, the upstream brain regions that control lymphoid organs and how they interface with lymphoid organs to produce stress-induced anxiety-like behavior remain elusive. Using stressed human participants and rat models, we show that CCL5 levels are increased in stressed individuals compared to controls. Stress-inducible CCL5 is mainly produced from cervical lymph nodes (CLN). Retrograde tracing from CLN identifies glutamatergic neurons in the red nucleus (RN), the activities of which are tightly correlated with CCL5 levels and anxiety-like behavior in male rats. Ablation or chemogenetic inhibition of RN glutamatergic neurons increases anxiety levels and CCL5 expression in the serum and CLNs, whereas pharmacogenetic activation of these neurons reduces anxiety levels and CCL5 synthesis after restraint stress exposure. Chemogenetic inhibition of the projection from primary motor cortex to RN elicits anxiety-like behavior and CCL5 synthesis. This brain-lymph node axis provides insights into lymph node tissue as a stress-responsive endocrine organ.

Fig. 1. Mental health outcomes and immune mediators of stressed individuals.

a Flowchart of the stress period and blood collection of stressed individuals. b PHQ9 scores. c GAD7 scores. d PCL-5 scores (one-way ANOVA, F_3,165 = 3.689, *P = 0.0190). e SRQ-20 scores (one-way ANOVA, F_3,165 = 3.234, **P = 0.0013, *P = 0.0404) of stressed individuals. (n = 41 for control, n = 47 for stress 1 m, n = 40 for stress 6 m, n = 41 for stress 12 m). f–o Plasma levels of MCP-1, CCL5 (two-sided unpaired t test, **P = 0.0022, t = 3.167, df = 79; n = 38 for control, n = 43 for stress), IL-1β, IL-2, IL-4, IL-6, TGF-α, TNF-α, BDNF and cortisol (two-sided unpaired t test, **P = 0.0020, t = 3.273, df = 79; n = 38 for control, n = 43 for stress) in stressed individuals 1 month after leaving the stressful environment. p, q, Plasma CCL5 (n = 38 for control, n = 43 for stress 1 m, n = 39 for stress 6 m, n = 33 for stress 12 m) and cortisol levels (n = 39 for control, n = 43 for stress 1 m, n = 41 for stress 6 m, n = 41 for stress 12 m) in stressed individuals 6 months later and 1 year later (one-way ANOVA, F_3,165 = 10.800, **P = 0.0012, ***P < 0.0001. ctrl control individuals without stressful experience, stress 1 m: stressed individuals 1 month after leaving the stressful environment, stress 6 m: stressed individuals 6 months after leaving the stressful environment, aid 12 m: stressed individuals 12 months after leaving the stressful environment. Data are presented as the mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001. Source data are provided as a Source Data file.

Fig. 2. Acute restraint stress led to anxiety-like behavior and a sharp increase in CCL5 in rats.

a Schematic timeline and behavioral paradigm in rats, RS restraint stress. b No difference between control rats and RS rats in distance moved in the open field test (t test, P = 0.5039, n = 9). c The distance moved in the central zone was significantly lower in RS rats than in control rats in the open field test (t test, P = 0.0042, n = 9). d The time spent in the central zone in the open field test was significantly lower in RS rats than in control rats (t test, P = 0.0046, n = 9). e There was a lower open arm ratio in the elevated plus maze test in RS rats than in control rats (t test, P = 0.0002, n = 8 for control and 9 for RS). f The fold change in CCL5 in tissues from stressed and control rats (two-way ANOVA, interaction F_9,94 = 17.83, P < 0.0001, effect of tissues F_9,94 = 17.66, P < 0.0001, effect of restraint stress F_1,94 = 24.86, P = 0.0969, n = 4). g Schematic timeline and behavioral paradigm in rats in which the CLNs were removed. h There was no difference in distance moved in the open field test between sham rats and rats in which the CLNs were removed (two-tailed unpaired t test, t = 1.406, df = 11, P = 0.1902, n = 6). i Rats in which the CLNs were removed spent significantly more time in the central zone in the open field test than sham rats (two-tailed unpaired t test, t = 2.391, df = 11, P = 0.0358, n = 7 for control and 6 for LN removal). j There was a trend toward a higher open arm ratio in rats in which the CLNs were removed than in sham rats (two-tailed unpaired t test, t = 1.682, df = 11, P = 0.1208, n = 6). k Serum CCL5 levels were also measured by ELISA at different times in these two groups. Restraint stress led to a sharp increase in CCL5 expression in both sham rats and rats in which the CLNs were removed. However, the CCL5 levels in sham rats were significantly higher than those in rats in which the CLNs were removed (two-way ANOVA, interaction F_4,20 = 4.44, P = 0.0099, effect of time F_4,20 = 21.02, P < 0.0001, effect of lymphadenectomy F_1,20 = 191.9, P < 0.0001; n = 4 per group). l, m Flow cytometry analysis of the frequency and absolute numbers of different T-cell populations in the CLNs of control and restraint stress-treated rats (two-way ANOVA, interaction F_2,18 = 0.4520, P = 0.6434, effect of cell types F_2,18 = 47.76, P < 0.0001, effect of stress F_1,18 = 9.021, P = 0.0076, n = 4 per group; two-way ANOVA, interaction F_2,18 = 0.5738, P = 0.5734, effect of cell types F_2,18 = 72.87, P < 0.0001, effect of stress F_1,18 = 11.80, P = 0.0030, n = 4 per group). n, o Flow cytometry analysis of CD8 memory cells in the CLNs of control and restraint stress-treated rats (two-way ANOVA, interaction F_1,12 = 0.1071, P = 0.7491, effect of cell types F_1,12 = 0.9506, P = 0.3488, effect of stress F_1,12 = 19.732, P = 0.0009, n = 4 per group). CD8 + CD45RC-: memory CD8 T cells; CD8 + CD45RC + : Naïve CD8 T cells. p Heatmap showing the differentially expressed genes (DEGs) in the cervical lymph nodes of control and RS-treated rats. The DEGs were identified with a fold change of RS/CTRL > 2.0 or <0.5. q Volcano map showing the number of DEGs in cervical lymph nodes in control and RS rats. RS led to the upregulation of 2077 genes and the downregulation of 2620 genes in CLNs. r Functional annotation of DEGs in the adult rat cervical lymph nodes. s Heatmap showing the expression of DEGs encoding translation initiation factors and ribosome genes. Data are presented as the mean ± SEM. The significance of differences in all two-group comparisons was determined by a two-tailed unpaired t test. *P < 0.05, **P < 0.01, ***P < 0.001. Source data are provided as a Source Data file.

Fig. 3. Identification of a cluster of RN glutamatergic neurons upstream of CLNs.

a Schematic timeline and paradigm in rats PRV CLN injections. We first injected pseudorabies virus 152 (PRV), encoding mRFP, directly into the CLNs of adult rats to retrogradely and transsynaptically label upstream neurons. We observed a cluster of mRFP+ neurons bilaterally in specific regions, including the RN, n = 3. b Identification of a cluster of RN glutamatergic neurons upstream of CLNs. c 89.7% of cells projecting to the CLNs were vGlut1⁺ cells (two-tailed unpaired t test, P < 0.0001, scale bar: 100 µm, n = 3). d Current steps of increasing amplitude applied during whole-cell recordings from RN glutamatergic cells in control (black) and RS (red) rats. Scale bar, 300 mV and 100 ms. e Depolarizing current steps evoked fewer action potentials in neurons recorded from RS animals than control animals (two-way ANOVA, interaction F_9,150 = 0.1149, P > 0.99, effect of current steps F_9,150 = 16.63, P < 0.001, effect of restraint stress F_1,150 = 6.761, P = 0.0100; control: n = 7, RS: n = 10 cells for 3 rats). f–i Glutamatergic cells exhibited different intrinsic properties in control and RS rats. f Rheobase, t = 2.235, P = 0.0436, n = 7 for control and 8 for RS. g Saturating I, t = 2.184, P = 0.0453, n = 7 for control and 10 for RS. h Voltage threshold, t = 2.127, P = 0.0504, n = 7–10 per group. i Membrane potential, t = 2.630, P = 0.0189, n = 7 for control and 10 for RS. Data are presented as the mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001. Source data are provided as a Source Data file.

Fig. 4. Neuronal activity dysfunction in the RN glutamatergic neurons of control and RS rats.

a Schematic of the fiber photometry setup. Ca²⁺ transients were recorded from RN^GCaMP6 rats. b Schematic of fiber implantation in GCaMP6-expressing RN glutamatergic neurons. Scale bar, 100 µm, n = 6. c GCaMP6 signals from RN glutamatergic neurons aligned to the moment of exploratory behavior of the central zone in the open field. Control: black; RS: red. Gray areas indicate rats in the central zone. d Difference in average fluorescence between control rats and RS-treated rats. The average fluorescence in RS-treated rats was significantly lower than that in control rats (two-tailed unpaired t test, t = 9.082, P < 0.0001, n = 6 rats per group). e, f Heatmap of the RN glutamatergic neuronal response to exploratory behavior in control and RS-treated rats. g, h Averaged calcium signals in response to exploratory behavior in control and RS-treated rats. Black and red lines, mean calcium signal during exploratory behavior; gray area and red area, SEM. i Quantification of the change in calcium signals from exploratory behavior in control and RS-treated rats (two-tailed unpaired t test, t = 3.180, P = 0.0038, n = 6 rats per group). Data are presented as the mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001. Source data are provided as a Source Data file.

Fig. 5. Inhibiting or ablating RN glutamatergic neurons triggered anxiety-like behavior and an increase in CCL5.

a Schematic of AAV-CaMKII-hM4D(Gi)-mCherry injection bilaterally into the RN. Left: confocal image showing CaMKII-hM4D(Gi)-mCherry (red) expression in RN neurons. Right: an enlarged image from the left panel, n = 3. b Current-evoked action potentials in a representative hM4Di-infected neuron recorded before, during, and after CNO perfusion (10 mM) (n = 6 neurons; two-tailed paired t test, t = 10.99, P = 0.0001; t = 14.26, P < 0.0001, respectively). c Inhibiting RN glutamatergic neurons had no effect on the locomotion abilities of adult rats (two-tailed unpaired t test, t = 1.235, P = 0.2357, n = 17 rats for control and 15 rats for hM4Di) but decreased the time spent in the central zone in the OFT (two-tailed unpaired t test, t = 3.686, P = 0.0022, n = 17 rats for control and 15 rats for hM4Di). d Inhibiting RN glutamatergic neurons led to a downward trend in the open arm ratio and frequency (two-tailed unpaired t test, t = 2.579, P = 0.0150, n = 15–17 rats; t = 1.101, P = 0.2797, n = 17 rats for control and 15 rats for hM4Di). e CCL5 levels in serum were increased in hM4Di rats after CNO injection but not in control rats (two-tailed paired t test, t = 2.484, P = 0.0475; t = 0.2531, P = 0.8087, respectively, n = 4). f Schematic of the bilateral coinjection of AAV-CaMKII-Cre-GFP and AAV-EF1α-DIO-taCasp3-TEVP into the RN. Confocal image showing reduced vGlu1⁺ signals ⁽red) in the RN, n = 3. g Ablating RN glutamatergic neurons increased the anxiety level of adult rats, which showed less time spent in the central zone in the OFT than controls (two-tailed unpaired t test, t = 2.783, P = 0.0155, n = 9 rats for control and 6 rats for ablate group). h DIO-taCasp3 rats spent less time in the open arms than controls (two-tailed unpaired t test, t = 3.750, P = 0.0024, n = 9 rats for control and 6 rats for ablate group). i Ablating RN glutamatergic neurons led to an increase in CCL5 levels in serum compared to controls (two-tailed unpaired t test, t = 3.757, P = 0.0094, n = 4 rats). j Schematic of AAV-CaMKII-hM3D(Gq)-mCherry injection bilaterally into the RN. Confocal image showing CaMKII-hM3D(Gq)-mCherry (red) expression in RN neurons. k Current-evoked action potentials in a representative hM3Dq-infected neuron recorded before, during, and after CNO perfusion (10 mM) (n = 6 neurons; two-tailed paired t test, t = 15.34, P < 0.0001; t = 13.22, P < 0.0001, respectively). l Activating RN glutamatergic neurons had no effect on the locomotion abilities of adult rats (two-tailed unpaired t test, t = 1.225, P = 0.2396, n = 8–9 rats) but increased the time spent in the central zone in the OFT (two-tailed unpaired t test, t = 3.022, P = 0.0086, n = 9 rats for control and 8 rats for hM3Dq group). m Activating RN glutamatergic neurons led to an increase in the open arm ratio and frequency in the EPM test (two-tailed unpaired t test, t = 2.432, P = 0.0280, n = 8–9 rats; t = 2.431, P = 0.0281, n = 9 rats for control and 8 rats for hM3Dq group). n CCL5 levels in serum were decreased in hM3Dq stress-treated rats after CNO injection but not in RS rats (two-tailed paired t test, t = 3.461, P = 0.0038; t = 0.3152, P = 0.7573, respectively, n = 4). Data are presented as the mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001. Source data are provided as a Source Data file.

Fig. 6. Inhibiting the M1-RN pathway induced anxiety-like behavior.

a Left: injection of AAV1-hSyn-Cre into the M1 and AAV-CaMKII-DIO-GCaMP6 into the RN. Right: a representative confocal image of GCaMP6⁺ neurons in the RN. b, c Heatmap showing the Ca²⁺ signals evoked by exploratory behavior in the OFT in RN glutamatergic neurons from control and RS-treated rats. d, e Averaged responses of control and RS-treated rats (black and red line, mean calcium signal during exploratory behavior; gray area and red area, SEM). f Quantification of the change in calcium signals in response to exploratory behavior in control and RS-treated rats (two-tailed unpaired t test, t = 2.429, P = 0.0223, n = 6 rats per group). g Left: injection of AAV1-hSyn-Cre into the M1 and AAV-CaMKII-DIO-hM4Di-mCherry into the RN. Right: a representative confocal image of hM4Di⁺ neurons in the RN. h Current-evoked action potentials in a representative hM4Di-infected neuron recorded before, during, and after CNO perfusion (10 mM) (n = 6 neurons; two-tailed paired t test, t = 17.33, P < 0.0001; t = 17.89, P < 0.0001, respectively). i Left: locomotion abilities were not affected by inhibiting M1-RN projections (two-tailed unpaired t test, t = 0.4938, P = 0.6291, n = 8 rats per group). Inhibiting M1-RN projections decreased the time spent in the central zone in DIO-hM4Di rats compared to mCherry control rats (two-tailed unpaired t test, t = 4.072, P = 0.0011, n = 8 rats per group). Right: representative tracks of control and hM4Di rats in the OFT. j Both the open arm ratio and frequency decreased significantly in hM4Di rats compared to mCherry control rats (two-tailed unpaired t test, t = 2.181, 2.649, P = 0.0468, 0.0191, n = 8 rats per group). Right: representative tracks of control and hM4Di rats in the EPM. k CCL5 levels in serum were significantly increased after CNO injection (two-tailed paired t test, t = 5.984, P = 0.0010, n = 4 rats). Data are presented as the mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001. Source data are provided as a Source Data file.