Hippocampal contextualization of social rewards in mice

Abstract

Acquiring and exploiting memories of rewarding experiences is critical for survival. The spatial environment in which a rewarding stimulus is encountered regulates memory retrieval. The ventral hippocampus (vH) has been implicated in contextual memories involving rewarding stimuli such as food, social cues or drugs. Yet, the neuronal representations and circuits underlying contextual memories of socially rewarding stimuli are poorly understood. Here, using in vivo electrophysiological recordings, in vivo one-photon calcium imaging, and optogenetics during a social reward contextual conditioning paradigm in male mice, we show that vH neurons discriminate between contexts with neutral or acquired social reward value. The formation of context-discriminating vH neurons following learning was contingent upon the presence of unconditioned stimuli. Moreover, vH neurons showed distinct contextual representations during the retrieval of social reward compared to fear contextual memories. Finally, optogenetic inhibition of locus coeruleus (LC) projections in the vH selectively disrupted social reward contextual memory by impairing vH contextual representations. Collectively, our findings reveal that the vH integrates contextual and social reward information, with memory encoding of these representations supported by input from the LC.

Fig. 1. The vH discriminates contexts following social reward contextual conditioning.

a Schematic and image of vH tetrode microdrives. b Experimental design of social reward (SR) CPP task with representative trajectories. c Context discrimination score per session. Repeated measures (RM) one-way ANOVA F_(2.8,61.8) = 18.55, P = 2.1 × 10⁻⁸. Pre-Train vs. Train-1 P = 7.9 × 10⁻⁷, Train-2 P = 2.9 × 10⁻⁷, Train-3 P = 1.9 × 10⁻⁶, Pre- vs. Post-Train P = 5.3 × 10⁻⁶. One-sample t-test vs. zero, Pre-Train t₍₂₂₎ = 5.19 P = 1.7 × 10⁻⁵, Train-1 t₍₂₂₎ = 4.1 P = 0.0006, Train-2 t₍₂₂₎ = 5.22 P = 1.5 × 10⁻⁵, Train-3 t₍₂₂₎ = 2.8 P = 0.0104, Post-Train t₍₂₂₎ = 1.7 P = 0.103. d Firing rate of two SR-context neurons at Post-Train. e Normalized firing rate of context neurons at Pre- and Post-Train. Two-way RM-ANOVA, SR-context neurons: session × context, F_(1,73) = 22.52, P = 0.00001; SR-context Pre- vs. Post-Train P = 1.64 × 10⁻¹¹; Post-Train SR- vs. neutral-context P = 4.07 × 10⁻⁸, n = 74 neurons. One-sample t-test, Pre-Train SR-context t₍₇₃₎ = 2.34 P = 0.022; Post-train SR-context t₍₇₃₎ = 4.45 P = 0.00003. Two-way RM-ANOVA, neutral-context neurons: context, F_(1,46) = 19.78 P = 0.000055; Post-Train SR- vs. neutral-context P = 5.11 × 10⁻⁶; neutral-context Pre- vs. Post-Train P = 0.0072, n = 47 neurons. f Normalized firing rate of SR- and neutral-context neurons upon entry into context-B or -A at Pre-Train: context-B, −2 to −1s, neutral-context neurons, t₍₄₅₎ = 2.11 P = 0.041; SR-context neurons, t₍₆₇₎ = 2.46 P = 0.017, and during (g) Post-Train: three-way Mixed-ANOVA, pre-entry group × context × time, F_{(3.7, 448.5)} = 3.94 P = 0.005; post-entry group x context: F_(1,120) = 12.65 P = 0.0005. One-sample t-test, pre-entry into neutral-context, neutral-context neurons: −2 to −1 s, t₍₄₇₎ = 3.01 P = 0.004; −1 to 0 s, t₍₄₇₎ = 2.19 P = 0.034; pre-entry into SR-context, SR-context neurons: −2 to −1 s, t₍₇₃₎ = 0.89 P = 0.38; −1 to 0 s, t₍₇₃₎ = 2.39 P = 0.02. h Correlation between context discrimination and proportion of SR-context neurons during Post-Train. Pearson’s r (2-tailed), r = 0.4 P = 0.0141. All multiple-comparisons were Bonferroni-corrected 2-tailed t-tests. Data are from n = 23 mice. Bar charts are presented as mean ± SEM.

Fig. 2. The vH represents emotional contexts in a valence-specific manner.

a Behavioural protocol and representative mouse trajectories during the habituation, fear conditioning and test sessions for the contextual fear memory task. b Freezing levels during habituation and test sessions. Two-way RM ANOVA between groups, context × session F_(1,23) = 20.20, P = 0.0002. Context-A vs. context-B during test P = 4.2 × 10⁻⁶, and habituation vs. test at context-A P = 5.7 × 10⁻⁹, n = 24 mice. c Example firing rate traces of fear context neurons at the habituation and test sessions in context-A and -B. d Comparison of neuron proportions at habituation and test sessions. Chi-squared test between sessions, P = 0.0235, n = 24 mice. e Averaged normalized firing rates during habituation and test sessions in context-A and -B for fear (left) and neutral context neurons (right). Two-way RM ANOVA, fear context neurons: session × context F_(1,113) = 53, P = 1 × 10⁻¹¹. Context-A vs. Context-B at the Test session P = 1 × 10⁻¹⁶, n = 114 neurons, n = 24 mice. Neutral context neurons: session × context F_(1,62) = 15.41, P = 0.00022. Context-A vs. Context-B at the Test session, P = 4.8 × 10⁻¹³ n = 64 neurons, n = 24 mice. f Overlap of neurons with preferential activity in the social reward context and fear context. Hypergeometric test 2-sided, OR = 1.1, P = 0.273; Fisher exact test 2-sided P = 0.509; n = 154 neurons in total, n = 23 mice. All multiple-comparisons were Bonferroni-corrected 2-tailed t-tests. Data are presented as the mean ± SEM.

Fig. 3. Emergence of social reward context neurons is contingent on associative learning.

a Experimental design of Mock and Social Reward CPP with representative mouse trajectories. b Discrimination scores during training. Two-way RM-ANOVA, task effect F_(1,8) = 4.9, P = 0.0577. One-sample t-test 2-tailed, SR-CPP Train-2 t₍₈₎ = 3.07 P = 0.0153; Train-3 t₍₈₎ = 2.86 P = 0.0212; all other comparisons non-significant, n = 9 mice. c Discrimination scores during Pre- and Post-Train Tests in the context that was non-preferred at Pre-train for Mock- and SR-CPP. Two-way RM-ANOVA, task effect F_(1,8) = 7.23 P = 0.0275, session × task F_(1,8) = 9.07 P = 0.0168. SR-CPP Pre- vs. Post-Train P = 0.007. One-sample t-test 2-tailed, Mock-CPP Pre t₍₈₎ = 3.78 P = 0.0054; Post t₍₈₎ = 1.31, P = 0.228; SR-CPP Pre t₍₈₎ = 3.28, P = 0.0112; Post t₍₈₎ = 1.54 P = 0.16, n = 9 mice. d Representative firing rate traces of two SR-context neurons. e Normalized firing rate of social reward context neurons during the Post-train Test. Two-way RM-ANOVA, context F_(1,22) = 8.351 P = 0.0085; task F_(1,22) = 10.08 P = 0.0044. SR-CPP comparison of SR vs. neutral context P = 0.0055. One-sample t-test 2-tailed, neutral-context-C t₍₂₂₎ = 1.6 P = 0.12, neutral-context-D t₍₂₂₎ = 2.89 P = 0.0085; neutral-context-B t₍₂₂₎ = 1.5, P = 0.16; SR-context-A t₍₂₂₎ = P = 0.0013; n = 23 neurons, 9 mice. f Normalized firing rate of neutral context neurons during Post-train. Two-way RM-ANOVA, context F_(1,35) = 6.62 P = 0.0145; task F_(1,35) = 7.4 P = 0.0101; context vs. task F_(1,35) = 7.13 P = 0.0114. SR CPP SR vs. neutral context, P = 0.0049. One-sample t-test 2-tailed, neutral-context-C t₍₃₅₎ = 0.92 P = 0.364; neutral-context-D t₍₃₅₎ = 1.15 P = 0.257; neutral-context-B t₍₃₅₎ = 4.27 P = 0.0001; SR-CtxA t₍₃₅₎ = 0.79 P = 0.437; n = 36 neurons, 9 mice. All multiple-comparisons were Bonferroni-corrected 2-tailed t-tests. Data are presented as the mean ± SEM.

Fig. 4. Ventral hippocampal neurons reorganize their context activity upon social reward contextual learning.

a Percentages of context neurons at Pre- and Post-Train. Chi-squared test, χ²₍₂₎ = 11.26 P = 0.0036, n = 23 mice. b Contingency matrix of neuron percentages and counts per context category at Pre- and Post-Train. χ²₍₄₎ = 9.75, P = 0.0448, n = 23 mice. c Schematic and example image of optogenetic strategy. d Schematic of SR-CPP task with laser illumination in SR-context during training. (C-F) n = 6 EYFP, n = 7 eArch3.0. e Behavioural discrimination score during training. Two-way Mixed-ANOVA, group effect F_(1,6) = 7.617 P = 0.0329; f Discrimination score during Pre- and Post-Train. Two-way Mixed-ANOVA, group × session F_(1,4) = 40.98, P = 0.0031. Bonferroni-corrected comparisons, Post-Train EYFP vs. eArch3.0, P = 0.0048. g Schematic of calcium imaging strategy. Inset, processed 1-photon Ca²⁺ signals in vH pyramidal neurons. (G-N) tdTomato: n = 120 neurons, 2 mice; eNpHR3.0: n = 273 neurons, 3 mice. h Representative Ca²⁺ traces of SR-context neurons during SR-CPP at Post-Train. i Percentages of context neurons during Pre- and Post-train. tdTomato: χ²₍₂₎ = 6.78 P = 0.0337; eNpHR3.0: χ²₍₂₎ = 0.94 P = 0.625. j Contingency table of neurons in each context category at Pre-Train and percentage at Post-Train. Numbers indicate percentages of neurons. Non-preferring at Pre-train: χ²₍₂₎ = 8.15 P = 0.017; SR-preferring at Pre-train: χ²₍₂₎ = 2.67 P = 0.263; neutral-preferring at Pre-train: χ²₍₂₎ = 0.92 P = 0.631. k Z-scored Ca²⁺ activity of SR-context neurons upon entry into SR-context at Post-Train. Two-way Mixed-ANOVA, pre-entry group × time F_(4.6,644.8) = 8.27 P = 3.97 × 10⁻⁷; post-entry group F_(1,141) = 49.52 P = 7.87 × 10⁻¹¹. l Z-scored Ca²⁺ activity of neutral-context neurons upon entry into the neutral context at Post-Train. Two-way Mixed-ANOVA, pre-entry group × time: F_(3.95,438.7) = 3.48 P = 0.008; post-entry group: F_(1,111) = 16.19 P = 0.00011. Data are presented as the mean ± SEM.

Fig. 5. Neurons activated by female odour encode food reward and female conspecifics.

a Schematic of behavioural paradigm for the presentation of unconditioned stimuli (US). b Representative calcium traces of neurons activated by each US in the schematic above. Shaded area indicates periods in stimulus context or foot-shock presentation. c–f Counts of neurons that were activated and not activated by unconditioned stimuli. Significance was tested by 2-sided Fisher’s exact test. c Counts of female odor-active, and non-active neurons between sucrose-active and sucrose non-active neurons: P = 0.0012, and between female-active and female non-active neurons: P = 0.0001. d Counts of female conspecific-active, and non-active neurons between sucrose-active and sucrose non-active neurons: P = 0.488. e Counts of female odor-active, and non-active neurons between fox odor-active and fox odor non-active neurons: P = 0.0002, and between shock-active and shock non-active neurons: P = 0.565. f Counts of fox odor-active, and non-active neurons between shock-active and shock non-active neurons: P = 0.452. g Matrix of Spearman’s rank correlation coefficient ρ (2-tailed) for the response activation to female-odor vs. sucrose: P = 0.0007, female-odor vs. female conspecific: P = 0.0001, female-odor vs. fox-odor: P = 0.0002, female conspecific vs. fox-odor: P = 0.0001. N = 411 neurons, 5 mice.

Fig. 6. Social reward contextual memory involves locus coeruleus projections to the vH.

a Schematic of optogenetic inhibition of LC→vH terminals. (A-E) EYFP: n = 7; eArch3.0: n = 8 mice. b Images of eArch3.0 expression in LC soma and projections in vH. Expression was verified for all 15 mice. c Optogenetic inhibition protocol during SR CPP training. d Behavioural discrimination score at training. Two-way Mixed-ANOVA, no group effect F_(1,7) = 0.17 P = 0.69. e Discrimination score at Pre- and Post-train. Two-way Mixed-ANOVA, group × session F_(1,5) = 11.33 P = 0.020; Bonferroni-corrected 2-tailed t-test, Post-train EYFP vs. eArch3.0 P = 0.0313. f Schematic of eArch3.0 expression in LC neurons, and optic fibre and silicon probe in vH. g Trial-averaged responses of laser-activated and -inhibited neurons in vH of eArch3.0^LC→vH. Dashed lines mark laser onset/offset. h Averaged responses across trial time. i Percentage of neurons activated or inhibited by LC→vH terminal inhibition. Chi-squared test, χ²_{(1, N = 593)} = 5.73 P = 0.0166. (F-I), EYFP^LC→vH n = 311 neurons from 4 mice, eArch3.0^LC→vH n = 282 neurons from 3 mice. j Simultaneous optogenetic inhibition of LC projections and in vivo calcium imaging from vH pyramidal neurons. j–n tdTomato: n = 84 neurons, 2 mice; eNpHR3.0: n = 179 neurons, 2 mice. k Images of eNpHR3.0 expression in LC soma (bottom left), LC→vH projections and GCaMP6f in vH pyramidal neurons (top left, right). Yellow arrows: LC varicosities, white arrows: LC varicosities superimposed on GCaMP6f+ pyramidal neurons. Bottom right, raw and processed images of 1-photon calcium signals in vH pyramidal neurons. Expression was verified for all 4 mice. l Calcium traces of pyramidal neurons during Post-Train. m Proportions of SR-context neurons during Post-Train. Chi-squared test, P = 0.0096. n Z-scored calcium activity upon entry into the SR- or neutral-context during Post-Train. Top, two-way Mixed-ANOVA, pre-entry group × time F_(2.3,163.9) = 3.98 P = 0.016; post-entry group F_(1,71) = 13.65 P = 0.00043. Bottom, two-way Mixed-ANOVA, post-entry group × time F_(1.65,77.46) = 3.82 P = 0.034. Data are presented as the mean ± SEM.