Dentate gyrus activin signaling mediates the antidepressant response

Abstract

Antidepressants that target monoaminergic systems, such as selective serotonin reuptake inhibitors (SSRIs), are widely used to treat neuropsychiatric disorders including major depressive disorder, several anxiety disorders, and obsessive-compulsive disorder. However, these treatments are not ideal because only a subset of patients achieve remission. The reasons why some individuals remit to antidepressant treatments while others do not are unknown. Here, we developed a paradigm to assess antidepressant treatment resistance in mice. Exposure of male C57BL/6J mice to either chronic corticosterone administration or chronic social defeat stress induces maladaptive affective behaviors. Subsequent chronic treatment with the SSRI fluoxetine reverses these maladaptive affective behavioral changes in some, but not all, of the mice, permitting stratification into persistent responders and non-responders to fluoxetine. We found several differences in expression of Activin signaling-related genes between responders and non-responders in the dentate gyrus (DG), a region that is critical for the beneficial behavioral effects of fluoxetine. Enhancement of Activin signaling in the DG converted behavioral non-responders into responders to fluoxetine treatment more effectively than commonly used second-line antidepressant treatments, while inhibition of Activin signaling in the DG converted responders into non-responders. Taken together, these results demonstrate that the behavioral response to fluoxetine can be bidirectionally modified via targeted manipulations of the DG and suggest that molecular- and neural circuit-based modulations of DG may provide a new therapeutic avenue for more effective antidepressant treatments.

Fig. 1. Behavioral responders and non-responders to FLX treatment following CORT administration.

a Timeline of experiment. b Kaplan–Meier survival curve (left) and scatterplot (right) of NSF data showing individual latency to eat values across all four groups. c–e Left panels represent Two-way ANOVA of all treatment groups and right panels represent One-Way ANOVA of CORT+VEH, CORT+FLX responders, and CORT+FLX non-responders for: EPM open arm entries and open arm duration (c), FST immobility (d), and Negative Affect Index (e). f Regression analyses correlating NSF latency to eat with EPM open arm duration (left) and FST immobility (right). g In a separate cohort of CORT+FLX mice, persistence of response was determined by assessing NSF behavior after 3 weeks of FLX (time point 0), and then again 1, 4, and 6 months later. For the NSF survival curve, line shading shows SEM of each group (n = 15–23 per group). Scatterplots, horizontal lines, and bars show group means with error bars indicating SEM (n = 7–16 per group after CORT+FLX mice are divided into responders and non-responders).

Fig. 2. Dentate Gyrus mRNA expression of Activin signaling components correlates with behavioral response to FLX treatment following CORT administration.

a–f Left panels represent Two-way ANOVA of all treatment groups and middle panels represent One-Way ANOVA of CORT+VEH, CORT+FLX responders, and CORT+FLX non-responders for DG mRNA expression of: Activin A (a), the Activin receptors acvr1a (b), acvr1b (c) and acvr1c (d), and the intracellular signaling proteins smad2 (e) and smad3 (f). Right panels show regression analyses correlating NSF latency to eat with DG mRNA expression of: Activin A (a), the Activin receptors acvr1a (b), acvr1b (c) and acvr1c (d), and the intracellular signaling proteins smad2 (e) and smad3 (f). Scatterplots, horizontal lines, and bars show group means with error bars indicating SEM (n = 12–14 per group).

Fig. 3. CSDS exposure leads to behavioral responders and non-responders and altered DG Activin signaling.

a Timeline of experiment and diagram of CSDS and Social Interaction (SIT) paradigms. b Kaplan–Meier survival curve (left) and scatterplot (right) of NSF data showing individual latency to eat values across all four treatment groups. c–e Left panels represent Two-way ANOVA of all treatment groups and right panels represent One-Way ANOVA of CORT+VEH, CORT+FLX responders, and CORT+FLX non-responders for: EPM open arm entries and open arm duration (c), FST immobility (d), and Negative Affect Index (e). f Regression analyses correlating NSF latency to eat with EPM open arm duration (left) and FST immobility (right). (g-l) Left panels represent Two-way ANOVA of all treatment groups and middle panels represent One-Way ANOVA of CORT+VEH, CORT+FLX responders, and CORT+FLX non-responders for DG mRNA expression of: Activin A (g), the Activin receptors acvr1a (h), acvr1b (i), and acvr1c (j), and the intracellular signaling proteins smad2 (k) and smad3 (l). Right panels show regression analyses correlating NSF latency to eat with DG mRNA expression of: Activin A (g), the Activin receptors acvr1a (h), acvr1b (i) and acvr1c (j), and the intracellular signaling proteins smad2 (k) and smad3 (l). For survival curves, line shading shows SEM of each group (n = 12–14 per group). Scatterplots, horizontal lines, and bars show group means with error bars indicating SEM (n = 6–18 per group).

Fig. 4. Chronic Activin A infusions into DG convert FLX non-responders into responders while chronic Inhibin A infusions convert responders into non-responders.

a Timeline of experiment and coordinates of infusions for ventral DG and ventral CA1 infusions into FLX non-responders. b Kaplan–Meier survival curve (left panel) and scatterplot (right panel) of NSF data showing individual latency to eat values across all three FLX non-responder treatment groups: vehicle infusions into DG (VEH_DG), Activin A infusions into DG (ACTIVIN_DG), and Activin A infusions into CA1 (ACTIVIN_CA1). c–e One-Way ANOVA of VEH_DG, ACTIVIN_DG, and ACTIVIN_CA1 for: EPM open arm entries and open arm duration (c), FST immobility (d), and Negative Affect Index (e). f Regression analyses correlating NSF latency to eat with EPM open arm duration (left) and FST immobility (right). g Timeline of experiment and coordinates of infusions for ventral DG and ventral CA1 infusions into FLX responders. h Kaplan–Meier survival curve (left panel) and scatterplot (right panel) of NSF data showing individual latency to eat values across all three FLX non-responder treatment groups: vehicle infusions into DG (VEH_DG), Inhibin A infusions into DG (INHIBIN_DG), and Inhibin A infusions into CA1 (INHIBIN_CA1). One-Way ANOVA of VEH_DG, INHIBIN_DG, and INHIBIN_CA1 for: EPM open arm entries and open arm duration (i), FST immobility (j), and Negative Affect Index (k). l Regression analyses correlating NSF latency to eat with EPM open arm duration (left) and FST immobility (right). For survival curves, line shading shows SEM of each group (n = 12 per group). Scatterplots, horizontal lines, and bars show group means with error bars indicating SEM.

Fig. 5. Activin A infusions into DG are a more effective augmentation therapy than commonly used second-line treatments.

a Timeline of experiment. b Kaplan–Meier survival curve (left) and scatterplot (right) of NSF data showing individual latency to eat values across all six CORT+FLX non-responder treatment groups: CORT+FLX (FLX), CORT+FLX switched to CORT+FLX+Activin A into DG (FLX+ACTIVIN_DG), CORT+FLX switched to CORT+sertraline (SER), CORT+FLX switched to CORT+bupropion (BUP), CORT+FLX switched to CORT+venlafaxine (VEN), and CORT+FLX switched to CORT+FLX+bupropion (CORT+FLX+BUP). c Graphical depiction of proportion of CORT+FLX non-responders converted into responders following different second-line antidepressant treatments.