Social conflict resolution regulated by two dorsal habenular subregions in zebrafish

See allHide authors and affiliations

Science  01 Apr 2016:
Vol. 352, Issue 6281, pp. 87-90
DOI: 10.1126/science.aac9508

How to win a fish fight

When to cease aggression and escape is an important decision that fighting animals must make. Chou et al. characterized the role of two nuclei in a brain area of the zebrafish called the dorsal habenula (dHb) during social aggression (see the Perspective by Desban and Wyart). Silencing the lateral dHb reduced the likelihood of winning a fight, whereas silencing the medial dHb increased the likelihood of winning. Thus, these two nuclei antagonistically control the threshold for surrender.

Science, this issue p. 87; see also p. 42


When animals encounter conflict they initiate and escalate aggression to establish and maintain a social hierarchy. The neural mechanisms by which animals resolve fighting behaviors to determine such social hierarchies remain unknown. We identified two subregions of the dorsal habenula (dHb) in zebrafish that antagonistically regulate the outcome of conflict. The losing experience reduced neural transmission in the lateral subregion of dHb (dHbL)–dorsal/intermediate interpeduncular nucleus (d/iIPN) circuit. Silencing of the dHbL or medial subregion of dHb (dHbM) caused a stronger predisposition to lose or win a fight, respectively. These results demonstrate that the dHbL and dHbM comprise a dual control system for conflict resolution of social aggression.

Aggression is an evolutionarily conserved behavior critical for animal survival (1, 2). When conflict is unavoidable, animals use aggression to establish a social hierarchy that determines how to share limited resources (2). Most animal conflicts aim at establishing a social hierarchy rather than causing lethal damage to opponents (14), which achieves the best cost-benefit for the group. However, the biological mechanisms governing the resolution of social conflict remain largely unknown. To address this question, we isolated adult male zebrafish for 24 hours and then put them together to fight in pairs (Fig. 1A) (5). Dyadic male zebrafish fights proceed in a stereotypic manner, starting with each animal exhibiting display behaviors, followed by circling and biting attacks, and ending when one fish shows fleeing behavior indicating surrender (Fig. 1B and movie S1) (5).

Fig. 1 Schematic illustration of fighting behavioral task and the Hb-IPN circuit.

(A) Schematic illustration of the fighting behavioral task. (B) Schematic illustration of dyadic fighting behaviors in male zebrafish. (C) Schematic illustration showing a dorsal oblique view of the axon projections from the dHbL to the d/iIPN (red), the dHbM to the i/vIPN (green), and the vHb to the MR (blue) in the adult zebrafish brain. OB, olfactory bulb; PP, para-pineal organ.

To investigate the neural circuits underlying the regulation of social conflict, we focused on the dorsal habenula–interpeduncular nucleus (dHb-IPN) pathway. We previously demonstrated that the lateral subregion of the zebrafish dHb (dHbL) sends axons to the dorsal IPN (dIPN) and the intermediate IPN (iIPN) (Fig. 1C), and efferent axons from the d/iIPN pass through the dorsal raphe to reach the dorsal tegmental area (DTA) (6, 7) containing a putative region corresponding to the mammalian periaqueductal gray (PAG) (Fig. 2A). Because the PAG regulates fight, flight, and freezing behaviors (8), we wondered if the dHbL-d/iIPN pathway signals information critical for fight and flight behaviors during aggressive conflicts. We performed calcium imaging of acute brain slices to visualize neural activity after electrical stimulation of the Hb (Fig. 2A and fig. S1). In both naïve and winner fish, we found an intense activity spot in the dIPN and scattered spots in the DTA region (Fig. 2, B and C, and movies S2 and S3), reflecting activation of the dHbL-d/iIPN-DTA pathway. In contrast, in loser fish we observed intense activity mainly in the ventral IPN (vIPN) and the median raphe (MR) (Fig. 2D and movie S4). In the dIPN, the peak of fluorescence intensity in winners was similar to that in naïve fish but was significantly reduced in losers (Fig. 2E). In the vIPN, we found a higher fluorescence intensity in loser fish than in winner and naïve fish, although it did not reach statistical significance (Fig. 2E).

Fig. 2 Losing experience reduced transmission in the dHbL-d/iIPN pathway.

(A) Schematic illustration of the stimulation site in the sagittal brain section. Tel, telencephalon; TeO, tectum opticum; IL, inferior lobe of hypothalamus; DR, dorsal raphe. (B to D) Calcium imaging of brain slices from a naïve (B), winner (C), and loser (D) Tg(narp:GAL4VP16);Tg(UAS:DsRed) double Tg fish. (E) Average peak calcium signal in the dIPN (naïve fish, 1.58 ± 0.55, n = 5 fish; winner, 0.68 ± 0.15, n = 9 fish; loser, 0.20 ± 0.08, n = 8 fish; P = 0.0132, Kruskal-Wallis test with Dunn’s multiple comparison test) and the vIPN (naïve fish, 0.84 ± 0.16; winner, 0.91 ± 0.16; loser, 1.99 ± 0.78; P = 0.2034, Kruskal-Wallis test). (F) Schematic illustration of electrophysiology recording in vivo. (G) Extracellular responses evoked by electrical stimulation to the left Hb in the IPN. Five traces were averaged for each condition. The arrowheads indicate the timing of stimulation. (H) Comparison of the peak LFP amplitude recorded from the dIPN (naïve fish, 0.78 ± 0.02 mV, n = 6 fish; winner, 0.85 ± 0.05 mV, n = 8 fish; loser, 0.65 ± 0.04 mV, n = 8 fish; P = 0.0066, Kruskal-Wallis test with Dunn’s multiple comparison test) and the vIPN (naïve fish, 0.35 ± 0.05 mV; winner, 0.39 ± 0.03 mV; loser, 0.33 ± 0.03 mV; P = 0.5685, Kruskal-Wallis test). The black circles indicate individual values, and the red squares indicate the average for each group. Data from a pair undertaking fighting behavior are linked by a black line. (I and J) Tract tracing by Neurobiotin. The efferents from the dIPN terminated in the DTA (I). The inset in (I) is a close-up view of the boxed area in (I) showing the efferents with boutons (arrowheads). Retrogradely labeled neurons were observed in the DTA (arrows). (J) The efferents from the vIPN terminated in the MR. The inset in (J) is a close-up view of the boxed area in (J) showing the efferents with boutons (arrowheads). Retrogradely labeled neurons were found in the MR (arrows). Scale bars, 100 μm. Error bars in (E) and (H), mean ± SEM. *P < 0.05; ns, not significant.

To further investigate neural activity of the dHbL-d/iIPN and dHbM-i/vIPN circuits in winner and loser states, we performed in vivo electrophysiological recordings of local field potential (LFP) in the IPN (Fig. 2F). In the dIPN, the LFP amplitude evoked by electrical stimulation in the left Hb was similar between winners and naïve fish but was significantly reduced in losers (Fig. 2, G and H). In contrast, naïve fish, winners, and losers showed similar LFP levels in the vIPN (Fig. 2, G and H) (9). Reduction in the transmission of the dHbL-d/iIPN circuit thus reflects a loser state, consistent with our calcium imaging data. The peak time latency and half decay time were similar among all conditions (fig. S2). Moreover, tract tracing results showed that the dIPN neurons coursed dorsally and caudally to extend to the DTA (Fig. 2I), whereas the vIPN neurons projected to the MR (Fig. 2J), which is consistent with our previous data (6) and supports our calcium imaging results.

To investigate how the dHbL-d/iIPN and dHbM-i/vIPN pathways participate in fighting behaviors, we perturbed these neural circuits in adult zebrafish. To silence the dHbL-d/iIPN pathway, we used a double transgenic line Tg(narp:GALVP16);Tg[UAS:tetanus neurotoxin (TeNT)] in which neurotransmission from the dHbL to the d/iIPN is selectively inhibited (6). We confirmed that TeNT is specifically expressed in the dHbL, by checking its expression in the Tg(narp:GALVP16);Tg(UAS:TeNT);Tg(brn3a-hsp70:GFP) zebrafish (Fig. 3A), and effectively blocks neurotransmission (fig. S3). To inhibit the dHbM-i/vIPN pathway, we first generated a transgenic line Tg(gpr151:GALVP16);Tg(UAS:GFP-TeNT) and then confirmed the gpr151:GALVP16-induced green fluorescent protein (GFP) expression in the dHb (Fig. 3B-1). Next, we crossed the Tg(brn3a-hsp70:GFP-Cre) with a glutamatergic neuron-specific line, Tg(vglut2a:loxP-DsRed-loxP-GFP) (10), in which glutamatergic neurons in the dHbM and ventral Hb (vHb) express GFP (Fig. 3B-2). Finally, we generated a triple transgenic line Tg(gpr151:GAL4VP16);Tg(brn3a-hsp70:GFP-Cre);Tg(UAS:loxP-DsRed-loxP-GFP-TeNT) in which GFP-TeNT was induced intersectionally only in the dHbM, and their axons could be observed in the vIPN (Fig. 3B-3 and fig. S4). These two Tg lines showed normal growth at developmental and adult stages, and the Hb-IPN circuit and structures throughout the whole brain were not affected (fig. S4) (6).

Fig. 3 Inactivation of the dHbL-d/iIPN and the dHbM-i/vIPN circuits differentially affects fight outcome.

(A) In situ hybridization of TeNT (red) and immunofluorescent staining of GFP (green; for detection of brn3a-hsp70:GFP) in coronal sections of adult Tg(narp:GALVP16);Tg(UAS:TeNT);Tg(brn3a-hsp70:GFP) fish brain. (B) Immunofluorescent staining of GFP (green; for detection of GFP-TeNT) and DsRed (red) in the coronal sections of adult Tg(gpr151:GALVP16);Tg(UAS:GFP-TeNT) (B-1), Tg(brn3a-hsp70:GFP-Cre);Tg(vglut2a:loxP-DsRed-loxP-GFP) (B-2), and Tg(gpr151:GALVP16);Tg(brn3a-hsp70:GFP-Cre);Tg(UAS:loxP-DsRed-loxP-GFP-TeNT) (B-3) fish brain. (C) The winning rate of the dHbL-silenced Tg against the WT sibling was significantly lower than the chance rate (0.5) (n = 20 fights; P = 0.02069, binomial test), and the rate of the dHbM-silenced Tg against the WT sibling was significantly higher than the chance rate (0.5) (n = 27 fights; P = 0.02612, binomial test). (D to F) Comparison of fighting duration (D) (dHbL WT sibling versus sibling, 250 ± 48.6 s, n = 15 fights; dHbL-silenced Tg versus WT sibling, 387 ± 54.9 s, n = 18 fights; P = 0.0555; dHbM WT sibling versus sibling, 962.7 ± 10.8 s, n = 16 fights; dHbM-silenced Tg versus WT sibling, 266.8 ± 45.2 s, n = 24 fights; P = 0.0008, Mann-Whitney test), latency (E) (dHbL WT sibling versus sibling, 283.6 ± 55.4 s; dHbL-silenced Tg versus WT sibling, 278.6 ± 42.3 s; P = 0.9431; dHbM WT sibling versus sibling, 158.6 ± 39.6 s; dHbM-silenced Tg versus WT sibling, 67.3 ± 10.4 s; P = 0.012, unpaired t test), and biting number (F) (dHbL WT sibling versus sibling, 130.7 ± 18.7; dHbL-silenced Tg versus WT sibling, 234.4 ± 46.1; P = 0.0638; between dHbM WT sibling versus sibling, 23.1 ± 4.1; and dHbM-silenced Tg versus WT sibling, 114.9 ± 24.1; P = 0.0039, unpaired t test). (G to I) The fighting result and initially dominant rate of WT (G), the dHbL-silenced Tg (H), and the dHbM-silenced Tg (I) fish when they fought against their WT siblings. (J) Schematic illustration of behavioral task for testing experience effects in zebrafish fight. (K and L) The fighting result and initially dominant rate of the dHbL WT sibling winners (K) and the dHbL-silenced winners (L) when they fought against naïve WT siblings in the second fight. Scale bars, 50 μm. Bars in (D), (E), and (F) represent the average of each group. *P < 0.05; **P < 0.01; ***P < 0.001; ns, not significant.

Next, we let the dHbL- and the dHbM-silenced fish fight against their wild-type (WT) siblings. The dHbL-silenced fish showed a significantly higher tendency (15 out of 20, 75%) to lose fights (Fig. 3C). We also confirmed that the result does not vary among the groups that were raised separately according to different birth dates (see the supplementary text). This predisposition to losing is not simply due to weaker physical fitness of the Tg fish. The duration, latency, and number of bites (“biting number”) of fights were similar between the fights of WT siblings versus WT siblings and those of the dHbL-silenced fish versus WT siblings (Fig. 3, D to F) [also see the results of a swimming performance test for physical fitness evaluation (fig. S5, movie S5, and materials and methods)]. Moreover, anxiety level, locomotion activity, and aggressiveness level (measured by the novel tank-diving test, open-field test, and mirror-biting test) of the dHbL-silenced fish were also similar to those of WT siblings (figs. S6 and S7A, and materials and methods). In contrast, the dHbM-silenced fish tended to win (19 out of 27, 70%) (Fig. 3C). The latency of a fight was significantly shorter in the fights of the dHbM-silenced fish versus WT siblings than that in those of WT siblings versus WT siblings (Fig. 3E), although the mirror-biting frequency was similar (fig. S7B). Physical fitness, anxiety level, locomotion capability, and the responses to conditioned stimuli after cued fear conditioning were similar between the dHbM-silenced fish and their WT siblings (figs. S5, S6, and S8).

We next analyzed time-lapse changes in biting frequencies during the whole fighting period. In WT fish, individuals that bit more frequently than their opponents during the first one or two minutes of the fight tended to win (15 out of 20, 75%), indicating that initial dominance could be an effective predictor of fighting results (Fig. 3G, fig. S9, and table S1). The results of fights were not caused by individual intrinsic fitness. Indeed, as previously reported, at 1 hour after the first fight, WT zebrafish exhibited “winner effects.” A previous winning experience increases the probability of winning the subsequent contest (5). However, when we identified and separated both winner and loser for 24 hours and let them fight against another naïve fish, we found that the previous winner and loser did not show a higher probability of winning or losing the subsequent contests (figs. S10 and S11 and tables S2 and S3).

The WT siblings of the dHbL- and dHbM-silenced fish also showed a fighting pattern similar to WT fish (fig. S12 and S13 and tables S4 and S5). In the pairs of the dHbL-silenced fish versus WT siblings, all the WT siblings that showed initial dominance won the fight (Fig. 3H, fig. S14, and table S6). In contrast, only 42% of the dHbL-silenced fish that showed initial dominance won (Fig. 3H), even though these Tg fish did not show a lower rate in becoming initially dominant (n = 20 fish; P = 0.8684, binomial test) (Fig. 3H). When the dHbM-silenced fish fought against their WT siblings, they showed a significantly higher tendency to be initially dominant (80%) than their WT siblings (20%) (n = 25 fish; P = 0.002039, binomial test) (Fig. 3I, fig. S15, and table S7). Among 20 Tg fish that showed initial dominance, 15 fish won (Fig. 3I). Quite intriguingly, three out of five (60%) dHbM-silenced fish that showed initial inferiority eventually won (Fig. 3I). These results indicate that the dHbM-silenced fish showed a higher tendency to win than their WT siblings, even when they were initially inferior.

As mentioned above, at 1 hour after the first fight, WT zebrafish exhibited “winner effects.” We investigated whether the same effect also held true in the dHbL-silenced fish (Fig. 3J). We found that 90% of WT siblings of the dHbL-silenced fish that won the first fight also won a second fight against a naïve fish (Fig. 3K). All WT sibling winners of the first fight showed initial dominance in the second fight (Fig. 3K, fig. S16, and table S8). In contrast, only 30% of the dHbL-silenced winners won the second fight (3 out of 10). The winning rate of the dHbL-silenced winners in the second fight was similar to that when the naïve dHbL-silenced fish fought against their WT siblings, as described above (Fig. 3C), indicating that previous winning experiences did not increase the probability for winning the subsequent fight in the dHbL-silenced fish. Only 50% of the dHbL-silenced winners of the first fight showed initial dominance in the second fight, and 60% of these lost the second fight (Fig. 3L, fig. S17, and table S9). The impairment of winner effects in the dHbL-silenced fish was not due to accumulated fatigue during the first fight, because the duration and biting number of the second fight was not reduced compared with the first fight (fig. S18). In contrast, we did not observe a higher winning tendency in the dHbM-silenced loser fish. Most the dHbM-silenced losers and their WT siblings lost in the second fight, as did most of the WT losers (loser effects).

The mechanisms by which the dHbL-d/iIPN and dHbM-i/vIPN circuits facilitate winner and loser behaviors may be due to their projections to the DTA and MR, respectively. The reduction in transmission of the dHbL-d/iIPN pathway with an accompanying reduction in the activity of the DTA (Fig. 2D) that contains a putative homologous brain region of the mammalian PAG might switch fish behaviors from attacking to fleeing. On the other hand, the dHbM-i/vIPN pathway may increase the tendency to lose through inhibitory outputs to repress the activity of the MR that was proposed to reduce anxiety- and depressive-like behaviors by increasing resilience to aversive stimuli (11). As the dHbL and dHbM antagonistically modulate the outcome of fights, it is possible that the interaction between these two regions determines the fighting results. Because we did not find neural connections between the dHbL and the dHbM, neural activity and interaction between the dIPN and the vIPN may play a key role in regulating fight resolution. Therefore, it is possible that the dHb-IPN pathway acts as a binary switchboard to affect fight outcomes. The left and right dHbL and dHbM are known to show characteristic size asymmetry (Fig. 1C) (12). A recent study demonstrated that impairment of epithalamic left-right asymmetry increases anxiety in zebrafish (13). Thus, it is intriguing to examine whether disruption of left-right asymmetry also influences the outcomes of fights. In addition, we found that loss of the ventral Hb (vHb)-raphe transmission impairs proper loser behaviors to avoid further attacks from the winners (see fig. S19 and movies S6 and S7 for details) (14). In conclusion, different parts of the Hb differentially regulate the social conflict—i.e., the two subregions of the dHb regulate the resolution of the social conflict, and the vHb contributes to the maintenance of the loser behaviors.

Supplementary Materials

Materials and Methods

Supplementary Text

Figs. S1 to S19

Tables S1 to S9

Movies S1 to S7

References (1622)

References and Notes

  1. The apparent discrepancy between LFP and calcium imaging of the vIPN may be because the LFP reflects the summation of excitatory postsynaptic potentials and calcium imaging reflects the summation of whole-cell excitation.
  2. The vHb-silenced Tg fish showed lower physical fitness and higher tendency to lose (fig. S19). The result of impairment of proper loser behaviors in the vHb-silenced fish is consistent with our previous conclusion (15) that the vHb activity reflects a negative expectation value and is essential for learning adaptive avoidance behavior.
Acknowledgments: We thank K. Kawakami for providing the Tol2 and pT2MUASTeTxLC constructs, RIKEN Advanced Science Institute’s advanced technology support division for production of experimental chambers, C. Yokoyama for advice in preparing the manuscript, and R. Nakayama for assistance in the identification of transgenic fish line. We thank all members of the Okamoto laboratory for support and advice, the Research Resource Center of RIKEN Brain Science Institute for animal care, and the National BioResource Project of Japan for fish strains. All requests for the materials (strains and plasmids) should be made to H.O. They are made available upon exchange of a materials transfer agreement. This work was supported by a Grant in Aid for Scientific Research (23120008) and Strategic Research Program for Brain Science from the Ministry of Education, Culture, Sports, Science and Technology of Japan and the Core Research for Evolutional Science and Technology, Japan Science and Technology Agency, Japan. The supplementary materials contain additional data.
View Abstract

Stay Connected to Science

Navigate This Article