HCN-channel dendritic targeting requires bipartite interaction with TRIP8b and regulates antidepressant-like behavioral effects

Abstract

Major depressive disorder (MDD) is a prevalent psychiatric condition with limited therapeutic options beyond monoaminergic therapies. Although effective in some individuals, many patients fail to respond adequately to existing treatments, and new pharmacologic targets are needed. Hyperpolarization-activated cyclic nucleotide-gated (HCN) channels regulate excitability in neurons, and blocking HCN channel function has been proposed as a novel antidepressant strategy. However, systemic blockade of HCN channels produces cardiac effects that limit this approach. Knockout (KO) of the brain-specific HCN-channel auxiliary subunit tetratricopeptide repeat-containing Rab8b-interacting protein (TRIP8b) also produces antidepressant-like behavioral effects and suggests that inhibiting TRIP8b function could produce antidepressant-like effects without affecting the heart. We examined the structural basis of TRIP8b-mediated HCN-channel trafficking and its relationship with antidepressant-like behavior using a viral rescue approach in TRIP8b KO mice. We found that restoring TRIP8b to the hippocampus was sufficient to reverse the impaired HCN-channel trafficking and antidepressant-like behavioral effects caused by TRIP8b KO. Moreover, we found that hippocampal expression of a mutated version of TRIP8b further impaired HCN-channel trafficking and increased the antidepressant-like behavioral phenotype of TRIP8b KO mice. Thus, modulating the TRIP8b-HCN interaction bidirectionally influences channel trafficking and antidepressant-like behavior. Overall, our work suggests that small-molecule inhibitors of the interaction between TRIP8b and HCN should produce antidepressant-like behaviors and could represent a new paradigm for the treatment of MDD.

Figure 1. Viral expression of TRIP8b is sufficient to rescue I_h

A.) Schematic of TRIP8b interacting with a single HCN pore-forming subunit. The N-terminal interaction occurs between the CNBD of HCN and an acidic stretch of amino acids in TRIP8b, schematized by a red shape. The C-terminal tail interaction occurs between the TPR domains of TRIP8b in gray and the ‘SNL’ tripeptide of HCN1, 2, and 4. The variably spliced N terminus of TRIP8b is represented by a green rectangle. B.) Schematic showing virus injection site. Coronal brain image generated using the Allen Mouse Brain Atlas. C.) Whole-cell somatic recordings from eGFP-positive CA1 pyramidal neurons four weeks after bilateral viral injection into TRIP8b KO mice. Representative traces are shown during hyperpolarizing current injection. D.) Quantification of sag ratio. For reference, WT mice injected with AAV-eGFP are also shown (WT: 1.23±0.02, n=8; AAV-eGFP: 1.07±0.01, n=9; AAV-TRIP8b: 1.27±0.04, n=8; (F(2,22)=11.64, p<0.001). E.) Quantification of I_h amplitude from somatic voltage-clamp recordings from TRIP8b KO animals injected with the indicated viral construct (AAV-eGFP: 0.24±0.05pA/pF, n=9, AAV-TRIP8b: 0.68±0.10pA/pF, n=7). AAV-eGFP injected wild type mice are included for reference (0.67±0.11 pA, n=8; F(2,21)=7.92, p<0.01). F.) Representative voltage-clamp traces in response to hyperpolarizing voltage steps. All error bars represent ± s.e.m and are described above as mean ± s.e.m. *p<0.05 on Tukey’s test following one way ANOVA.

Figure 2. Dendritic targeting of HCN channels is rescued by viral delivery of TRIP8b

A.) Low power composite image of TRIP8b KO animals unilaterally injected with AAV-TRIP8b. Uninjected hemisphere (left) shows absence of TRIP8b (red, top panel) and weak hippocampal HCN gradients (green, middle and lower panels). Injected hemisphere (right) shows restoration of the distal dendritic enrichment of TRIP8b, HCN1, and HCN2. SO: Stratum oriens, SP: Stratum pyramidale, SR: Stratum radiatum, SLM: Stratum lacunosum moleculare. Scale bar represents 200 µm. B.) Quantification of HCN1 channel expression after viral rescue. A value of ‘1’ represents no change in staining of HCN1 relative to the uninjected hemisphere. Asterisk denotes comparison of HCN1 staining in the SLM (AAV-eGFP: 0.98±0.01 n=6, AAV-TRIP8b= 1.52±0.05 n=4, t=23.81, p<0.001). C/D) TRIP8b KO mice bilaterally injected with AAV-TRIP8b showed more immobility time on forced swim test (FST) and tail suspension test (TST) relative to AAV-eGFP injected controls. Injection of AAV-TRIP8b increased the immobility time on TST (AAV-eGFP = 132±20.9 sec, AAV-TRIP8b 166.16±26.5 sec, t=2.47, p<0.05) and FST (AAV-eGFP = 77.3±11.2 sec, AAV-TRIP8b = 142.8±20.1 sec, t=6.96, p<0.05), indicating a reversal of the increase in antidepressant-like behavioral effects of TRIP8b KO mice (n=6,6). For comparison, uninjected wild type mice and uninjected TRIP8b KO mice are also shown to highlight the increase in antidepressant-like behavior for both TST (WT = 191.8±17.6 sec, TRIP8b KO=134.8±23.4 sec, t=4.47, p<0.05) and FST (WT=142.4±21.8 sec, TRIP8b KO=85.6±9.93 sec, t=5.72, p<0.05) (n=6,5). Note that the comparison between AAV-eGFP and AAV-TRIP8b is distinct from the comparison between wild type and TRIP8b KO mice because the AAV-eGFP/AAV-TRIP8b were both subject to bilateral viral injections while the other groups were not. *P<0.05 two tail unpaired T test. All error bars represent ± s.e.m. and are described above as mean ± s.e.m.

Figure 3. Loss of either TRIP8b-HCN binding site blocks viral rescue of I_h

Ai–iii.) Schematic of mutant TRIP8b isoforms used showing predicted results of mutating either binding site. B.) Representative traces from EGFP-positive CA1 pyramidal neurons from each condition during hyperpolarizing current injections. Traces are scaled to ΔV_max to facilitate comparison of sag ratio. C.) Representative voltage-clamp recordings in response to hyperpolarizing voltage steps highlight slowly activating inward current. D.) Quantification of the sag ratio showed no difference by ANOVA (AAV-eGFP: 1.07±0.01, n=9; AAV-N13A: 1.11±0.02, n=7; AAV-Δ58: 1.09±0.01, n=6; F(2,19)=2.095, p>0.05). For comparison the sag ratio for AAV-TRIP8b infected neurons is reproduced from Figure 1. E.) Voltage-clamp measurements of I_h also showed no difference by ANOVA). (AAV-eGFP: 0.24±0.05 pA/pF, n=9; AAV-N13A: 0.282±0.09 pA/pF, n=7; AAV-Δ58: 0.27±0.04 pA/pF, n=5; F(2,19)=0.09, p>0.05). For comparison, the I_h amplitude from AAV-TRIP8b infected neurons is reproduced from Figure 1. All error bars represent ± standard error of the mean and are described above as mean ± s.e.m.

Figure 4. Both TRIP8b-HCN binding sites are required for dendritic targeting of HCN channels

A.) Confocal images demonstrating that unilateral injection of AAV-N13A causes a reduction in HCN1 staining in the SLM. Top panels show staining for HCN1 (red), middle panel shows staining for eGFP (green), and the bottom panel shows a composite image. Images in the left column are from the uninjected (control) hemisphere, while the right column shows the AAV-N13A injected hemisphere. B/C.) Unilateral injection of either AAV-Δ58 or AAV-Δ58/N13A fails to rescue HCN1 distal dendritic enrichment. Display of images is identical to that in A. Scale bar represents 100 µm. D.) Quantification of HCN1 distal enrichment. HCN1 staining intensity in regions of interest from the virally injected hemisphere were scaled by staining in the contralateral (uninjected) hemisphere. For comparison, the AAV-TRIP8b rescue experiment is reproduced from Figure 2. E.) Quantification of HCN1 staining in the SLM layer of CA1 (AAV-eGFP= 0.98±0.01, AAV-Δ58=0.98±0.04, AAV-N13A=0.78±0.04, AAV-Δ58/N13A=0.99±0.07). A one way ANOVA comparing staining intensity in the SLM of the different conditions (AAV-eGFP, AAV-Δ58, AAV-N13A, AAV-Δ58/N13A) was found to be significant (F(3,17)=22.375, p<0.05). Tukey’s post hoc tests found a difference between AAV-N13A and AAV-eGFP (p<0.05), AAV-N13A and AAV-Δ58 (p<0.05), and AAV-N13A and AAV-N13A/Δ58 (p<0.05). A one way ANOVA comparing HCN1 staining in the cell body layer of the different conditions (AAV-eGFP, AAV-N13A, AAV-Δ58, AAV-Δ58/N13A) was not significant (F(3,17)=0.59, p>0.5) hence post hoc tests were not performed. F.) TRIP8b KO mice bilaterally injected with AAV-N13A, but not AAV-Δ58, show less immobility time on TST and FST. A one way ANOVA examining the immobility time on FST found a significant difference (AAV-eGFP=148.42±18.8 sec, AAV-N13A=40.2±6.7 sec, AAV-Δ58=127.4±25.0 sec, F(2,18)=67.34, P<0.05), and follow up Tukey’s tests revealed a difference between AAV-eGFP and AAV-N13A as well as a difference between AAV-N13A and AAV-Δ58. G.) A one way ANOVA examining the immobility time on TST was significantly different (AAV-eGFP=130±12.2 sec, AAV-N13A=44.9±20.1 sec, AAV-Δ58=114±10.0 sec, F(2,17)=62.32, p<0.05) with Tukey’s test showing differences between AAV-eGFP and AAV-N13A and between AAV-Δ58 and AAV-N13A. *p<0.05 on Tukey’s HSD test following one way ANOVA. All error bars represent ± standard error of the mean and are described above as mean ± s.e.m.