Heightened lateral habenula activity during stress produces brainwide and behavioral substrates of susceptibility

Abstract

Some individuals are susceptible to chronic stress, and others are more resilient. While many brain regions implicated in learning are dysregulated after stress, little is known about whether and how neural teaching signals during stress differ between susceptible and resilient individuals. Here, we seek to determine if activity in the lateral habenula (LHb), which encodes a negative teaching signal, differs between susceptible and resilient mice during stress to produce different outcomes. After (but not before) chronic social defeat stress, the LHb is active when susceptible mice are in proximity of the aggressor strain. During stress, activity is higher in susceptible mice during aggressor interactions, and activation biases mice toward susceptibility. This manipulation generates a persistent and widespread increase in the balance of subcortical vs. cortical activity in susceptible mice. Taken together, our results indicate that heightened activity in the LHb during stress produces lasting brainwide and behavioral substrates of susceptibility.

Keywords: chronic social defeat stress; lateral habenula; learning.

Figure 1.. In susceptible mice, CSDS produced strain-specific aversion and changes in anxiety-like behavior and mobility.

A. Schematic of Chronic Social Defeat Stress (CSDS). B. Schematic of Social Interaction (SI) test. Social zone: 8cm additional radius from the perimeter of the cup containing the social target. C. Left: time spent near the aggressor strain in SI test pre-CSDS. Right: time spent near the aggressor strain in SI test post-CSDS. Dashed line indicates the cutoff for binary categorization of Susceptible/Resilient (based on one standard deviation below the control mean). Control vs stressed post-CSDS: t = 3.5505, p = 6.1699e-04. D. Time spent in the social zone (SI time) before vs after CSDS when the social target was of the self (BL6) strain (left) or other (AKR) strain (right). Self strain SI time in susceptible pre-CSDS vs post-CSDS: t = −6.1047, p = 1.33e-05. Self strain SI time in resilient pre-CSDS vs post-CSDS: t = −2.9147, p = 0.0393. Other strain SI time in susceptible pre-CSDS vs post-CSDS: t = −3.1374, p =0.0393. E. Top: schematic of homecage assay. Bottom: percent of time spent investigating (sniffing and pursuing) social target in freely-moving assay when the social target was a juvenile of the aggressor or self strain (control N = 22, susceptible N = 26, resilient N = 20). Control vs susceptible for aggressor social target: t = 3.2324, p = 0.0023. Susceptible vs resilient for aggressor social target: t = −2.1898, p = 0.0339. Control vs susceptible for self-strain social target: t = −2.1285, p = 0.0387. Susceptible vs resilient for self-strain social target: t = 2.0549, p = 0.0460. F. Relationship between time spent investigating an aggressor strain juvenile in the homecage assay and SI time after CSDS: R = 0.3965, p = 0.0064. G. Top: Schematic of elevated plus maze (EPM). Bottom: percent of time spent in open arms of EPM (control N = 14, susceptible N = 20, resilient N = 31). Control vs susceptible: t = 4.5352, p = 7.6295e-05. Susceptible vs resilient: t = −2.1630, p = 0.0354. H. Relationship between time spent in open arms of the EPM and SI time: R = 0.3100, p = 0.0269. I. Top: schematic of novelty suppressed feeding assay (NSF). Bottom: Latency to feed during NSF (control N = 14, susceptible N = 19, resilient N = 31). Control vs susceptible: t = −2.7623, p = 0.0096. J. Relationship between latency to feed in NSF and SI time: R = −0.4584, p = 0.0008. K. Top: Schematic of chamber exploration assay. Bottom: Percent of time immobile (speed <1cm/s) during chamber exploration (control N = 30, susceptible N = 40, resilient N = 46). Control vs susceptible: t = −3.7733, p = 3.4041e-04. Susceptible vs resilient: t = −2.5642, p = 0.0102. Control vs resilient: t = −2.7313, p = 0.0079. L. Relationship between time spent immobile during chamber exploration and SI time: R = −0.3119, p =0.0035. M. Top: Schematic of open field test (OFT). Bottom: percent of time spent in inner zone of OFT (control N = 14, susceptible N = 20, resilient N = 31). Control vs susceptible: t = 3.5396, p = 0.0013. Control vs resilient: t = 2.6244, p = 0.0102. N. Relationship between time spent in inner zone of OFT and SI time. p-value in C is from an unpaired 2-sided t-test. p-values in D are from paired 2-sided t-tests (with Bonferroni correction for three groups and two strains). p-values in E, G, I, K, M are from unpaired 2-sided t-tests following 1-way ANOVA. p-values in F, H, J, L are from Pearson’s correlations. Shaded areas in F, H, J, L, N represent 95% confidence interval for linear fit. *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001. See Table S1 for statistics details.

Figure 2.. After but not before CSDS, aggressor strain-specific responses in the LHb of susceptible mice in the SI test.

A. Left: Location of fiber photometry recordings from cell bodies in the Lateral Habenula (LHb) Middle: GCaMP (AAV5-CaMKII-GCaMP6f or AAV5-syn-jGCaMP7f) expression in LHb cell bodies (green) and DAPI (blue). Right: A confocal image of LHb neurons showing nuclear exclusion of GCaMP. B. Responses in an example mouse to individual visits to the social zone of the aggressor (Figure 1B). C. LHb signal aligned to entry of aggressor (SW) social zone during SI test. Each row is mean response in one mouse, with mice sorted by SI time, from susceptible (green, N = 10) to resilient (purple, N = 10). D. LHb signal aligned to entry of aggressor social zone during pre-CSDS SI test, averaged across individuals in resilient and susceptible groups (mean ± s.e.m. plotted). E. Same as D for post-CSDS SI test. F. Average from 1s to 2s post onset of entry to the social zone in D and E for susceptible (N = 10), resilient (N = 10), and control mice (N = 10). Susceptible pre-CSDS vs post-CSDS: t = −3.7842, p = 0.0389. G. Same as F for self-strain (BL6) social zone entry. H. Same as F for other strain (AKR) social zone entry. I. Correlation between the magnitude of the fluorescence response during the SI test with the aggressor strain post-CSDS and avoidance level in mice from fiber photometry experiments. r = −0.7291, p = 0.0003. J. Neural response as a function of visit number across mice for susceptible (left) and resilient (right) mice pre-CSDS vs post-CSDS. K. Left: Cellular resolution calcium imaging schematic. Right: Example histology with GRIN lens placement above LHb and AAV9-syn-FLEX-GCaMP7f expression (green) and DAPI (blue). L. Left: Example FOV from microendoscope. Right: Same FOV, with identified neurons outlined. M. Example traces of colored neurons from L. N. LHb signal aligned to entry of aggressor social zone during SI test. Each row is a neuron, sorted from susceptible (green, n = 131 neurons, N = 4 mice) to resilient (purple, n = 75 neurons, N = 5 mice). O. LHb signal aligned to entry into aggressor social zone during pre-CSDS SI test, averaged across neurons in resilient and susceptible groups (mean ± s.e.m. plotted). P. Same as O post-CSDS. Q. Average from 1s to 2s post entry into aggressor social zone in O and P plotted for susceptible and resilient groups. Susceptible pre-CSDS vs post-CSDS: t = −5.8304, p <.0001. R. Same as Q for self-strain (BL6) social zone entry. S. Same as Q for other strain (AKR) social zone entry. T. Distribution of responses in resilient and susceptible mice during aggressor strain proximity in the SI test pre-CSDS. U. Same as T post-CSDS. Susceptible vs resilient post-CSDS: k = 0.4123, p = 9.5142e-08. V. Proportion of cells that were significantly responding during aggressor proximity during the SI test after defeat in susceptible (left) and resilient (right) mice (see Figure S2D–G and Methods). W. In significantly activated cells, orrelation between the magnitude of the fluorescence response during aggressor strain proximity in the SI test post-CSDS and avoidance level: r = −0.7732, p = 0.0414. X. Same as W for significantly inhibited cells. Y. Spontaneous event rates in susceptible (green, n = 165 neurons, N = 6 mice) and resilient (purple, n = 71 neurons, N = 5 mice) during a 5 min test in a neutral chamber pre-CSDS. Z. Same as Y post-CSDS. Susceptible vs resilient t = −3.0187, p = 0.0028. p-value in F is from a paired 2-sided t-test (with Bonferroni correction for three groups and three strains). p-value in Q is from an unpaired 2-sided t-test (with Bonferroni correction for two groups and three strains). p-value in U is from a Kolmogorov-Smirnov test. p-values in I, W-X are from Pearson’s correlations. p-values in J, Z are from 2-sided t-tests. Shaded areas in I-J, W-X represent 95% confidence interval for linear fit. *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001. See Table S1 for statistics details.

Figure 3.. During CSDS, elevated LHb activity during attack and other proximal behaviors.

A. Left: Behavioral setup. Right: Example video frame with tracked key points. B. Left: Features calculated from key points. Right: Time series of all features used in behavior quantification. C. Smoothed histogram of t-SNE from features, with clusters numbered by increasing distance between mice (N = 35). D. Mean LHb GCaMP signal across t-SNE behavior space in fiber photometry mice (N = 21). E. Average proximity within each t-SNE cluster. F. For each cluster, mean LHb GCaMP signal plotted against mean centroid distance between mice (R = −0.8215, p = 5.3E-5, N = 17 clusters). G. Top: Density of random forest classified investigation within t-SNE space. Bottom: Example frame of being investigated. Stressed mouse: yellow dot; aggressor mouse: red dot. H. Top: Same as G for attack. Bottom: Example frame of attack. I. Same as G for fighting. Bottom: Example frame of fighting. J. Same as G for fleeing. Bottom: Example frame of fleeing. K. Neural activity in LHb time-locked to being investigated (mean ± s.e.m. plotted). L. Same as K for attack. M. Same as K for fighting. N. Same as K for fleeing. p-value in F is from a Pearson’s correlation. Shaded area in F represents 95% confidence interval for linear fit. *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001. See Table S1 for statistics details.

Figure 4.. From the 1st day of defeat, higher LHb activity in susceptible mice during proximal behaviors and in resilient mice when vigilant.

A. Mean LHb GCaMP dF/F across t-SNE behavior space in susceptible mice (across all 10 days of defeat; N = 11). B. Same as A for resilient mice (N = 10). C. Difference between susceptible and resilient LHb GCaMP dF/F (difference between A and B). D. Being attacked onset-aligned LHb responses during defeat averaged across individuals in resilient and susceptible groups (mean ± s.e.m. plotted). Grey region indicates +/−0.25s surrounding the maxima. E. Fighting back onset-aligned LHb dF/F during defeat. F. Fleeing onset-aligned LHb dF/F during defeat. G. Vigilance onset-aligned LHb DF/F during defeat. H. Average LHb GCaMP dF/F to attack onset from susceptible and resilient groups across defeat (mean ± s.e.m across mice; averaging across labeled gray region (+/−0.25s maxima) in D). Onset activity by SI time, day, and their interaction: main effect of SI time, Z = −2.428, p = 0.015; main effect of day, Z = 0.537, p = 0.591. Interaction, Z = 2.088, p = 0.037. I. Same as H for fighting onset. Onset activity by SI time: main effect of SI time, Z = −2.089, p = 0.037. J. Same as H for fleeing onset. Onset activity by SI time, day, and their interaction: main effect of SI time, Z = −1.607, p = 0.108; main effect of day, Z = −0.539, p = 0.590. Interaction, Z = 2.520, p = 0.012. K. Same as H for vigilance onset. Onset activity by SI time: main effect of SI time, Z = 2.185, p = 0.019. L. Average LHb GCaMP responses to attack onset (labeled gray region in D) on day 1 plotted against SI time for each mouse (N = 21 mice): R = −0.5783, p = 0.0060. M. Same as L for fighting onset: R = −0.4806, p = 0.0274. N. Same as L for fleeing onset: R = −0.5807, p = 0.0057. O. Same as L for vigilance onset: R = 0.4954, p = 0.0224. p-values in H-K are from two-sided GEE. p-values in L-O are from Pearson’s correlations. Shaded areas in L-O represent 95% confidence interval for linear fit. *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001. See Tables S1, S3 for statistics details.

Figure 5.. Closed-loop activation of LHb during CSDS produces susceptibility.

A. Left: Location of virus injections and fiber targeting of cell bodies in the Lateral Habenula (LHb). Right: Example histology of virus expression. B. Schematic of attack-triggered stimulation. Each detected attack frame during defeat triggered 5 pulses of 20Hz activation. C. Example of a defeat session with laser light delivery triggered on attack of the closed-loop mouse. Bottom inset is a 10s segment. D. Probability of a laser train, as a function of time relative to attack onset. E. Density of activation in t-SNE space. F. Distribution across sessions of percent of defeat session that mice received laser (mean is 21.34% of the defeat session). G. Difference in SI time between opsin (ChR2 or ChRmine) and control group (YFP). Opsin vs control: t = −2.1263, p = 0.0461. H. Difference in open-arm time in the elevated plus maze between opsin and control group. Opsin vs control: t = −2.2934, p = 0.0328. I. Difference in center time in the open field between opsin and control group. Opsin vs control: t = −2.8231, p = 0.0105. J. Difference in latency to feed in novelty suppressed feeding assay between opsin and control group. K. Difference in SI time between opsin (NpHr) and control group (YFP). L. Difference in open-arm time in the elevated plus maze between opsin (NpHr) and control group. Opsin vs control: t = −3.9140, p = 0.0010. M. Difference in center time in the open field between opsin (NpHr) and control group. N. Difference in latency to feed in novelty suppressed feeding assay between opsin (NpHr) and control group. p-values in G-N are from unpaired 2-sided t-tests. *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001. See Table S1 for detailed statistics.

Figure 6.. Activation of the LHb during defeat produces durable, brainwide changes.

A. Schematic of closed-loop attack-triggered LHb stimulation during CSDS. The Fos dataset includes 10 mice that received LHb stimulation during CSDS and 44 unstimulated control mice. B. Approximately one week after the conclusion of CSDS, mice were placed for 10 min into the cage of a novel aggressor that was restrained under a wire cup. There was no LHb stimulation during this assay. The mice were then sacrificed one hour later for Fos analysis. C. Left: example brainwide Fos imaging data. A 50-plane (100-μm) maximum intensity projection is shown. The insets are shown after background subtraction and filtering. Right: All detected cells overlaid on the Allen CCF for the example section on the left. D. Difference in Fos⁺ cell density across LHb-stimulated resilient (N = 4) and susceptible (N = 6) mice (top), across unstimulated resilient (N = 30) and susceptible (N = 14) mice (middle), and across all LHb-stimulated (N = 10) and all unstimulated (N = 44) mice (bottom). E. Individual brain regions sorted by the estimated contribution of SI time to Fos⁺ cell counts based on GLM coefficients for mice that received LHb stimulation. Significantly different regions are highlighted with red (resilient-activated) and blue (susceptible-activated) boxes. F. Individual brain regions sorted by the estimated contribution of SI time to Fos⁺ cell counts based on GLMM coefficients for unstimulated control mice. G. Individual Allen CCF brain regions sorted by the estimated contribution of LHb stimulation to Fos⁺ cell counts based on GLMM coefficients across all mice. H. Comparison of distributions of LHb stimulation coefficients (from G) across all brain regions in cerebral cortex (n = 61 regions), forebrain nuclei (n = 83 regions), and midbrain/hindbrain (n = 56 regions). I. Correlation between the estimated contribution to Fos⁺ cell counts of SI time in LHb-stimulated mice (from E; y-axis) vs. of LHb stimulation across all mice (from G; x-axis) (n = 200 regions). J. Correlation between the estimated contribution to Fos⁺ cell counts of SI time in unstimulated control mice (from F; y-axis) vs. of LHb stimulation across all mice (from G; x-axis). Significance in E-G is based on GLM or GLMM coefficient estimate z-tests corrected for 10% false discovery rate. Error bars in H represent median ± interquartile range. Shaded areas in I-J represent 95% confidence interval for linear fit. p-values in H are from Kolmogorov-Smirnov tests with Hochberg-Bonferroni correction for multiple comparisons. p-values in I-J are from Pearson correlations. *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001. See Tables S1, S5 for detailed statistics.