Connections of the Mouse Orbitofrontal Cortex and Regulation of Goal-Directed Action Selection by Brain-Derived Neurotrophic Factor

Abstract

Background: Distinguishing between actions that are more likely or less likely to be rewarded is a critical aspect of goal-directed decision making. However, neuroanatomic and molecular mechanisms are not fully understood.

Methods: We used anterograde tracing, viral-mediated gene silencing, functional disconnection strategies, pharmacologic rescue, and designer receptors exclusively activated by designer drugs (DREADDs) to determine the anatomic and functional connectivity between the orbitofrontal cortex (OFC) and the amygdala in mice. In particular, we knocked down brain-derived neurotrophic factor (Bdnf) bilaterally in the OFC or generated an OFC-amygdala "disconnection" by pairing unilateral OFC Bdnf knockdown with lesions of the contralateral amygdala. We characterized decision-making strategies using a task in which mice selected actions based on the likelihood that they would be reinforced. Additionally, we assessed the effects of DREADD-mediated OFC inhibition on the consolidation of action-outcome conditioning.

Results: As in other species, the OFC projects to the basolateral amygdala and dorsal striatum in mice. Bilateral Bdnf knockdown within the ventrolateral OFC and unilateral Bdnf knockdown accompanied by lesions of the contralateral amygdala impede goal-directed response selection, implicating BDNF-expressing OFC projection neurons in selecting actions based on their consequences. The tyrosine receptor kinase B agonist 7,8-dihydroxyflavone rescues action selection and increases dendritic spine density on excitatory neurons in the OFC. Rho-kinase inhibition also rescues goal-directed response strategies, linking neural remodeling with outcome-based decision making. Finally, DREADD-mediated OFC inhibition weakens new action-outcome memory.

Conclusions: Activity-dependent and BDNF-dependent neuroplasticity within the OFC coordinate outcome-based decision making through interactions with the amygdala. These interactions break reward-seeking habits, a putative factor in multiple psychopathologies.

Keywords: Action; Amygdala; Habit; Orbital; Outcome; Striatum.

Fig. 1. The VLO innervates the dorsal striatum and BLA

(a) A representative BDA infusion into the VLO and a drawing of the targeted area are shown (distance from bregma and estimated regional boundaries based on (16)). (b) The VLO innervates the dorsomedial and central striatum along the rostrocaudal axis. (c) Coronal amygdala sections from (16) correspond to magnified depictions shown in (d). (d) Infusions of BDA into the VLO reveal innervation of the anterior BLA and the lateral wall of the posterior BLA, along with light innervation of the intercalated cell masses and moderate innervation of the dorsal endopiriform nucleus. Abbreviations: AIP-posterior agranular insular cortex; AIV-ventral agranular insular cortex; BLA-basolateral amygdala; Cl-claustrum; CPu-caudate putamen; CeA-central amygdala; DEn-dorsal endopiriform nucleus; I-intercalated masses; ic-internal capsule; opt-optic tract.

Fig. 2. The DLO/AI innervates the lateral and ventral striatum and BLA

(a) A representative infusion of BDA into the DLO/AI and a rendering of the targeted area are shown. (b) BDA infusions into the DLO/AI illuminate heavy innervation of the ventral and lateral striatum, and reveal innervation of the posterior AI that is maintained along the rostrocaudal axis. (c) Coronal amygdala sections from (16) correspond to the magnified depictions shown in (d). (d) The DLO/AI sends heavy projections to the anterior BLA. Projections are lighter in the posterior BLA, and preferentially terminate along the lateral wall. Innervation is also noted in the ventral BLA, as well as the posterior AI and the PRh. Abbreviations not defined in Fig. 1: AID-dorsal agranular insular cortex; BLV-basolateral amygdala, ventral part; LH-lateral hypothalamus; M1-primary motor cortex; PRh-perirhinal cortex; VP-ventral pallidum.

Fig. 3. Photomicrographs show representative BDA staining and demonstrate innervation of the striatum and PRh and BLA by the DLO/AI

(a) Projections from the DLO/AI to the striatum are bihemispheric, but labeling is heaviest in the hemisphere ipsilateral to the infusion site, and (b) heavy labeling is noted in the posterior caudate. (c) Representative images of DLO/AI innervation of the BLA, posterior AI, and PRh (rostral to caudal); note avoidance the dorsal endopiriform nucleus. Green outline-BLA; blue outline-DEn; yellow outline-posterior AI; red outline-PRh. Inset: Corresponding images from (16), with regions outlined in black.

Fig. 4. VLO-selective Bdnf knockdown interferes with goal-directed action selection

(a) A task schematic is shown. Mice are trained to generate two distinct responses. Then, the likelihood that one response will be reinforced is decreased. Preferential engagement of the remaining response during a probe test is interpreted as goal-directed action selection, while engaging both responses equivalently — despite contingency degradation — is codified as habitual behavior. (b) Bdnf was knocked down bilaterally in the VLO. Infusion sites are summarized on images from (60). Black represents the largest viral vector spread, and white the smallest. (c) Inset: Infusions resulted in decreased BDNF expression in homogenized VLO tissue. Mice were trained to nose poke for food reinforcers; Bdnf knockdown reduced response rates, particularly when the response requirement escalated from a fixed ratio 1 to random interval schedule (final 2 sessions). Rates represent total responses on both apertures. (d) Mice with Bdnf knockdown were also unable to select between actions that were more, vs. less, likely to be reinforced (non-degraded vs. degraded) following instrumental contingency degradation; instead, they engaged familiar habit-like response patterns, generating both responses equally. (e) BDNF expression in the downstream amygdala correlated with the degree of impairment following VLO-targeted Bdnf knockdown, with low BDNF associated with robust responding on the ‘degraded’ nose poke aperture. Amygdala BDNF expression was also reduced overall in mice with VLO-targeted Bdnf knockdown (inset). Symbols and bars represent means+SEMs, except in (e) where each symbol represents a single mouse.*p<0.05.

Fig. 5. Functional disconnection of the VLO and amygdala results in habits

(a) Histological representations of unilateral cortical viral vector infusions and amygdala lesions are transposed onto images from (60). Black represents the largest and white the smallest. (b) Mice were trained to nose poke for food reinforcers; response rates did not differ between groups. (c) Mice with asymmetric infusions were, however, insensitive to instrumental contingency degradation. With an additional training session, mice ultimately were able to develop outcome-directed response strategies, indicating that contralateral infusions delayed, but did not block, action-outcome conditioning. Symbols and bars represent means+SEMs,*p<0.05.

Fig. 6. Rescue of goal-directed decision-making and regulation of VLO dendritic spines

(a) Intact mice were extensively trained to respond for food reinforcement. Escalating random interval schedules are indicated. (b) The TrkB agonist, 7,8-DHF, preserved sensitivity to action-outcome contingency degradation, despite extended response training, while control mice developed habitual response strategies as expected. (c) In a subsequent experiment, pretreatment with the TrkB antagonist, ANA-12, blocked this effect. Group means are represented at left, and individual mice are represented at right. Response acquisition curves for these mice are provided in Suppl. Fig. S1. (d) Separate mice were trained to nose poke using a fixed ratio 1 schedule of reinforcement. (e) 7,8-DHF had no effects when mice would be expected to engage in goal-directed decision-making strategies. (f) Additionally, 7,8-DHF had no effects on extinction conditioning. Each arrow represents an injection immediately following the test session. (g) Layer V dendritic spines were imaged and enumerated in these mice. 7,8-DHF increased spine density in the VLO. Representative dendritic branches are adjacent. (h) Next, mice with VLO-targeted Bdnf knockdown were trained to respond for food reinforcers. (i) Bdnf knockdown induced inflexible habit-like responding following instrumental contingency degradation as expected, but 7,8-DHF rescued response selection strategies. Another group of knockdown mice instead received the Rho-kinase inhibitor fasudil, which also rescued goal-directed response strategies. (j) The same data are represented as the percentage of total responses directed towards the intact response-outcome contingency. The pink bar at ~50% indicates that Bdnf knockdown mice responded at chance levels, while 7,8-DHF and fasudil restored selective responding. Bars and symbols represent means+SEMs except in c, right. *p<0.05 as indicated following t-tests, **p<0.04 relative to all other groups. “RI” refers to random interval schedules of reinforcement, and “FR” refers to fixed ratio 1 training used throughout.

Fig. 7. Gi-DREADD stimulation results in habit-like behavioral inflexibility

(a) Mice were infused bilaterally with AAV-CaMKII-GFP or AAV-CaMKII-hM₄D(Gi)-mCitrine. Infusions sites are represented. mCitrine-expressing neurons are shown; the gray box overlaid on the histology image represents the location of these neurons. (b) Response rates did not differ during instrumental response acquisition. (c) When CNO was paired with the degradation of an instrumental contingency, control GFP-expressing mice subsequently preferentially generated the response more likely to be reinforced. Rates are represented in 2×5-min. bins. By contrast, Gi-DREADD-expressing mice initially preferred the ‘non-degraded’ response, but this goal-directed response strategy decayed. (d) The same findings are represented in bar graph form, again indicating that Gi-DREADD stimulation induced non-selective responding. Symbols and bars represent means+SEMs,*p<0.05.