Bidirectional coordination of actions and habits by TrkB in mice

Abstract

Specific corticostriatal structures and circuits are important for flexibly shifting between goal-oriented versus habitual behaviors. For example, the orbitofrontal cortex and dorsomedial striatum are critical for goal-directed action, while the dorsolateral striatum supports habits. To determine the role of neurotrophin signaling, we overexpressed a truncated, inactive form of tropomyosin receptor kinase B [also called tyrosine receptor kinase B (TrkB)], the high-affinity receptor for Brain-derived Neurotrophic Factor, in the orbitofrontal cortex, dorsomedial striatum and dorsolateral striatum. Overexpression of truncated TrkB interfered with phosphorylation of full-length TrkB and ERK42/44, as expected. In the orbitofrontal cortex and dorsomedial striatum, truncated trkB overexpression also occluded the ability of mice to select actions based on the likelihood that they would be reinforced. Meanwhile, in the dorsolateral striatum, truncated trkB blocked the development of habits. Thus, corticostriatal TrkB-mediated plasticity appears necessary for balancing actions and habits.

Figure 1

TrkB.t1 overexpression in the oPFC impedes goal-directed action selection. (a) Behavioral testing approach: Mice were trained to nose poke on two ports for food reinforcers. Then, one response was reinforced approximately 50% of the time (‘Non-degraded’), while the probability of reinforcement associated with the other response was greatly decreased (‘Degraded’), given that pellets were delivered non-contingently. Inhibiting responding in this condition is considered goal-directed, while insensitivity to non-contingent pellet delivery is considered habitual. (b) Experimental timeline: Mice were infused with viral vectors, then behaviorally tested. (c) Viral vector constructs (from ref.) are shown. (d) A lentivirus expressing TrkB.t1, GFP, or a half-and-half mixture of both was infused bilaterally into the oPFC. Representative viral vector spread is represented on images from the Mouse Brain Library. White represents the maximal spread and black the smallest. “VLO” refers to the ventrolateral oPFC. (e) Quantitative immunostaining revealed that full-titer lenti-TrkB.t1 infusions generated greater HA immunofluorescence than a half-and-half mixture of lenti-GFP and lenti-TrkB.t1 (“Half-TrkB.t1”) (GFP: n = 4; TrkB.t1: n = 5). Inset: Representative HA immunofluorescence. (f) Mice were trained to respond for food reinforcers. Full-titer TrkB.t1 overexpression reduced response rates (n = 6 mice/group). (g) Further, mice with full-titer lenti-TrkB.t1 were insensitive to action-outcome contingencies, failing to reduce responding when responding was not reinforced. (h) The same data were normalized to response rates generated on the last day of training, such that 0 reflects no change. Response rates increased in the ‘Non-degraded’ condition across groups. Meanwhile, full-titer TrkB.t1 overexpression interfered with response inhibition, such that these mice maintained high levels of responding even when responding was not reinforced (‘Degraded’ condition). (i) With additional exposure to noncontingent pellet delivery, full-titer TrkB.t1 mice were ultimately able to inhibit a nonreinforced response. Bars and symbols represent means + SEMs, *p < 0.05. Behavioral findings are concordant with independent unpublished pilot investigations and post-mortem experiments were conducted at least twice.

Figure 2

Validation of the TrkB.t1-overexpressing virus. (a) Virus-infected oPFC tissue was dissected by tissue punch and immunoblotted for TrkB.T1, revealing elevated TrkB.T1 protein in mice bearing the TrkB.t1-overexpressing virus, as expected (GFP: n = 4; TrkB.t1: n = 5). (b) Phospho-TrkB was also diminished (GFP: n = 12; TrkB.t1: n = 16, representing 2 independent cohorts; applies also to all following panels). (c) Full-length TrkB was unaffected. (d) Phospho-ERK42/44 was reduced, while (e) total ERK42/44 was unchanged. (f) Similarly, we identified a trend for reduced phospho-Akt and (g) no changes in total Akt. (h) Mature BDNF and (i) the pro-form were not significantly affected. (j) Finally, the astrocytic marker GFAP was reduced and (k) the synaptic marker PSD95 was not affected. Representative, unadjusted lanes from the same individual gels are shown with their corresponding loading controls. Molecular weights of each protein are indicated either in, or directly adjacent to, the protein name. Bars represent means + SEMs, *p < 0.05, ^#p = 0.06. Every gel was run at least twice.

Figure 3

TrkB.t1 overexpression in the striatum bidirectionally regulates actions and habits. (a) Lenti-TrkB.t1 or GFP was infused into the DMS or DLS, then sensitivity to action-outcome contingency was tested. (b) Viral vector infusions are represented, with white representing the largest infusion and black the smallest. (c) We detected no group differences during food-reinforced instrumental conditioning. (d) Overexpression of TrkB.t1 in the DMS, however, caused a bias towards inflexible habits, indicated by insensitivity to action-outcome contingencies. (e) Additional nose poke training induced habits in control mice, but overexpression of TrkB.t1 in the DLS blocked these habits from forming. n = 8, 6, 7 for GFP, DMS and DLS, respectively. Bars and symbols represent means + SEMs, *p < 0.05. Results are concordant with independent unpublished pilot investigations.