Hypothalamic neurons that mirror aggression

Abstract

Social interactions require awareness and understanding of the behavior of others. Mirror neurons, cells representing an action by self and others, have been proposed to be integral to the cognitive substrates that enable such awareness and understanding. Mirror neurons of the primate neocortex represent skilled motor tasks, but it is unclear if they are critical for the actions they embody, enable social behaviors, or exist in non-cortical regions. We demonstrate that the activity of individual VMHvl^PR neurons in the mouse hypothalamus represents aggression performed by self and others. We used a genetically encoded mirror-TRAP strategy to functionally interrogate these aggression-mirroring neurons. We find that their activity is essential for fighting and that forced activation of these cells triggers aggressive displays by mice, even toward their mirror image. Together, we have discovered a mirroring center in an evolutionarily ancient region that provides a subcortical cognitive substrate essential for a social behavior.

Keywords: FosTRAP; TRAP2; VMHvl; aggression; cognition; emotion; fiber photometry; miniscope; mirror neurons; social behavior; social cognition; tail rattle; ventromedial hypothalamus.

Figure 1:. VMHvl^PR neurons exhibit aggression-mirroring.

A. Strategy to express GCaMP6s in VMHvl^PR neurons. GCaMP6s expression in VMHvl^PR neurons shown in coronal section (middle panel), with dashed white lines outlining VMH and, more laterally on the right, VMHvl, and dashed orange line outlining the fiber optic tract (FT) dorsally. GCaMP6s+ cells express Esr1 (right panel). B. Schematic of fiber photometry setup in freely moving mice. C-D. Aggressor VMHvl^PR neurons are activated during attack (C) and tail-rattling (D). Dashed vertical line in the peri-event time plot (PETP) shows onset of behavioral display in all Figures. Heatmap above PETPs here and other Figures, unless otherwise specified, shows normalized fluorescence signal from a single representative experimental male during individual sequential epochs of that particular behavior. F_n represents fractional change in fluorescence from baseline fluorescence preceding the event. Experimental male is shaded gray in schematic panels of all Figures. E-F. Observer VMHvl^PR neurons show mirroring activity when witnessing attacks (E) or tail-rattles (F) by demonstrator males. G. Aggressor VMHvl^PR neurons show higher peak activation during attacks than observer VMHvl^PR neurons. H. No difference in peak activation between aggressor and observer VMHvl^PR neurons during tail-rattling. I-K. Aggressor VMHvl^PR neurons are activated during investigation (chemoinvestigation of non-anogenital regions) (I), sniffing (anogenital chemoinvestigation) (J), and grooming (K). L-N. No discernible activation of observer VMHvl^PR neurons during investigation (L), sniffing (M), and grooming (N). Mean (dark trace) ± SEM (lighter shading) of z-scored activity shown in PETPs of all Figures. Mean ± SEM, bar graphs. Base., baseline fluorescence signal; Peak, peak amplitude of fluorescence signal. Scale bars, 100 μm (A, middle) and 20 μm (A, right). n = 14 PR^Cre males. ns = not significant, * p < 0.05, *** p < 0.001, **** p < 0.0001. See also Figure S1.

Figure 2:. Aggression-mirroring by VMHvl^PR neurons requires visual input.

A-C. Imaging observer VMHvl^PR neurons of Trpc2 null males witnessing aggression. (A) Schematic of behavioral paradigm. (B-C) Activation of observer VMHvl^PR neurons during attacks (B) and tail-rattling (C). D-F. Imaging observer VMHvl^PR neurons witnessing aggression under infrared illumination. (D) Schematic of behavioral paradigm. (E-F) No discernible activation of observer VMHvl^PR neurons during attacks (E) or tail-rattling (F). G-I. Imaging observer VMHvl^PR neurons of socially naive males witnessing aggression. (G) Schematic of behavioral paradigm. (H-I) Activation of observer VMHvl^PR neurons during attacks (H) and tail-rattling (I). Mean ± SEM. n = 9 PR^Cre;Trpc2^−/− (B-C), 6 PR^Cre (E-F), 7 PR^Cre (H-I) males. * p < 0.05, ** p < 0.01, **** p < 0.0001. See also Figure S2.

Figure 3:. Individual aggressor and observer VMHvl^PR neurons are activated during attacks and tail-rattles.

A. There could be complete, partial, or no overlap between aggressor and observer VMHvl^PR neurons. B. Strategy to express GCaMP6s in VMHvl^PR neurons (left panel). Coronal section through the VMH with GRIN lens track (GT) visible dorsal to the GCaMP6s+ cells in the VMHvl (right panel). Dashed white lines outline VMH and VMHvl and dashed orange lines outline the GT. C. Setup for miniscope imaging of VMHvl^PR neurons in freely moving mice. Inset shows raw GCaMP6s fluorescence overlaid with contours of segmented cells that were significantly active during behavioral testing. D-G. Individual aggressor or observer VMHvl^PR neurons are activated, inhibited, or silent during attacks (D,F) or tail-rattles (E,G). Raster of mean activity of individual neurons from 5 males shown as heatmap. Cells are ordered starting with most activated on top and most inhibited at the bottom, and row numbers do not correspond to the same cells across panels. Peaks and troughs of GCaMP6s fluorescence of neurons classified as activated or inhibited corresponded to z-scores >1.5 and <−1.5, respectively, in these and subsequent panels of Figs. 3, S3, 4, S4. H-K. Population dynamics of activated aggressor and observer VMHvl^PR neurons, with inset pie-charts showing percent of activated cells for attacks (H,J) and tail-rattles (I,K). L. No difference in peak amplitude of activity of activated VMHvl^PR neurons between aggressors and observers for either attacks or tail-rattles. Mean ± SEM. n = 152 aggressor and 88 observer VMHvl^PR neurons from 5 PR^Cre males. Scale bar = 200 μm. See also Figure S3.

Figure 4:. Individual VMHvl^PR neurons are co-activated during aggressor and observer paradigms.

A-D. Co-activation of individual neurons during attacks in aggressor and observer paradigms. Segmented cells during representative imaging sessions of a male tested as aggressor or observer (A). Overlapping sets of VMHvl^PR neurons are co-activated in participants and witnesses (B), with distinct activation dynamics at a few time-points (C), but comparable peak amplitude of, or net, activation (D). E-H. Co-activation of individual neurons during tail-rattles in aggressor and observer paradigms. Segmented cells during representative imaging sessions of a male tested as aggressor or observer (E). Overlapping sets of VMHvl^PR neurons are co-activated in participants and witnesses during tail-rattles (F), with comparable activation dynamics (G) and peak amplitude of, or net, activation (H). I-L. Co-activation of individual neurons during attacks and observation of tail-rattles. Segmented cells during representative imaging sessions of a male tested as aggressor or observer (I). Overlapping sets of VMHvl^PR neurons are co-activated in participants and witnesses (J), with distinct activation dynamics at a few time-points (K), but comparable peak amplitude of, or net, activation (L). M-P. Co-activation of individual neurons during tail-rattles and observation of attacks. Segmented cells during representative imaging sessions of a male tested as aggressor or observer (M). Overlapping sets of VMHvl^PR neurons are co-activated in participants and witnesses (N), with distinct activation dynamics at a few time-points (O), but comparable peak amplitude of, or net, activation (P). Mean ± SEM. AUC, area under the curve. n = 5 PR^Cre males. #, significant difference at multiple time-points. See also Figure S4 and Table S1.

Figure 5:. An aggression mirror-TRAP strategy indelibly tags observer VMHvl neurons.

A. Schematic of FosTRAP2 strategy to genetically tag activated, Fos-expressing neurons. B,C. Aggression-activated, FosTRAP2-tagged neurons (Aggression-TRAP, tdTomato+, red) show comparable Fos induction (green) following performance or observation of aggression. Asterisks (B) label tdTomato+ and Fos– cells. For all TRAPing studies in this and subsequent Figures, mice were subjected to two rounds of the behavior being tested and provision of 4OHT to maximize number of neurons expressing the Cre-dependent transgene. D. Schematic of aggression mirror-TRAP strategy to express GCaMP6s in VMHvl neurons for fiber photometry. E,F. Significant activation of aggression mirror-TRAPed, GCaMP6s+ VMHvl neurons when the male is observing attacks (E) or tail-rattles (F). Mean ± SEM. n = 4 (B,C) Fos^iCreERT;Ai14 and 8 (E,F) Fos^iCreERT2 males. * p < 0.05, *** p < 0.001. Scale bar = 50 μm. See also Figure S5.

Figure 6:. Activity of aggression-mirroring VMHvl neurons is essential for territorial aggression.

A. Schematic of aggression mirror-TRAP strategy to express the inhibitory chemogenetic actuator DREADDi in VMHvl neurons. B. Schematic of resident-intruder test of territorial aggression, with the aggression mirror-TRAPed resident male expressing DREADDi in VMHvl cells. C. Rasters of individual resident males showing reduced aggression upon inhibition of aggression-mirroring VMHvl neurons. In this and subsequent Figures, each consecutive pair of vehicle and CNO rasters, starting from the top row, represents behavioral displays of a resident encountering an unfamiliar intruder on different days in his cage. Rasters in this and subsequent Figures are only shown for males displaying behaviors being tested in at least one condition (vehicle or CNO). D-G. Inhibition of aggression-mirroring VMHvl neurons reduces the likelihood that males attack or tail-rattle (D), increases the latency to initiate either of these two behaviors (E), and reduces the number and duration of aggressive events (F-G). Mean ± SEM. n = 10 Fos^iCreERT2 males. * p < 0.05, ** p < 0.01. See also Figure S6.

Figure 7:. Activation of aggression-mirroring VMHvl neurons increases aggressivity.

A. Schematic of aggression mirror-TRAP strategy to express the excitatory chemogenetic actuator DREADDq in VMHvl neurons. B. Schematic of resident-intruder test of territorial aggression, with the aggression mirror-TRAPed resident male expressing DREADDq in VMHvl cells. C. Rasters of individual resident males exhibiting increased aggressivity upon chemogenetic activation of aggression-mirroring VMHvl neurons. D-G. Activation of aggression-mirroring VMHvl neurons did not significantly increase likelihood of initiating attacks or tail-rattles (D), but it reduced the latency to start attacking (E) and the increased the number of attacks inflicted upon the intruder male (F). Other parameters of attack and tail-rattles were unaltered (E-G). H-I. Activation of aggression-mirroring VMHvl neurons increases tail-rattling to a mirror. Schematic of aggression mirror-TRAPed resident male expressing DREADDq in VMHvl neurons with a mirror in his cage (H). Activation of aggression-mirroring VMHvl neurons increased the likelihood of tail-rattling ~3-fold and the number of tail-rattles ~10-fold (I). J. Schematic of working model of how the activity of VMHvl^PR neurons may reflect agonistic states. VMHvl^PR neurons are sensitive to inputs from diverse, potentially inter-related sources of information, and their activity can drive physical acts of aggression (offensive or defensive) or reflect aggression-mirroring. Dashed arrows indicate that input to or output from VMHvl^PR neurons may occur via a multi-synaptic relay, and curved dashed arrow depicts the possibility that there may be cross-talk between VMHvl^PR neurons directly or via local interneurons. Mean ± SEM. n = 12 (B-G) and 6 (H-I) Fos^iCreERT2 males. * p < 0.05, ** p < 0.01. See also Figure S7.