Objective To investigate the correlation between hippocampal gap junction connexin Cx36 and cognitive dysfunction in post-stroke depression（ PSD） rats， and to explore the potential pathogenesis of cognitive dysfunction of post-stroke depression. Methods A total of 42 rats were randomly divided into normal group， stroke group， depression group， PSD group， normal saline group， carbenoxolone（ CBX） group and all trans retinoic acid（ ATRA） group， with 6 rats in each group. The depression group was randomly given a chronic unpredictable mild stimulus every day and raised alone. The focal ischemic rat model was induced by injecting endothelin-1 into the motor cortex of stroke rats. PSD group， normal saline group， CBX group and ATRA group were superimposed with depression stimulation on the basis of stroke group to establish PSD model. At the same time of PSD modeling， PSD group did not intervene； In the normal saline group， 1 ml of normal saline was injected intraperitoneally every day； ATRA group received intraperitoneal injection of 1 ml ethanol and phosphate buffer（ 1∶9） with a concentration of 1 mg/ml every day； CBX group was injected intraperitoneally every day according to the standard of 20 mg/kg. On the 28th day after stroke， sucrose preference test， Morris water maze test and shuttle test were used to detect the loss of interest， spatial memory， learning-memory ability of rats to verify whether the PSD cognitive impairment model was successfully induced. RT-PCR experiment was used to detect the expression changes of Cx36 mRNA in rat hippocampus. Western blot and immunofluorescence were used to detect the changes of Cx36 protein expression and average fluorescence intensity in rat hippocampus. Results On the 28th day after operation， the body weight of the rats in the PSD group［ （257.05±4.74）g］ was significantly lower than that of the normal group［ （352.00±7.99）g］ and the stroke group［ （303.95±4.63）g］， and the difference was statistically significant（ P＜ 0.05）. The drinking rate of sugar water in PSD group［ （61.92±3.12）%］ was lower than that in normal group［ （83.40±5.38）%］， stroke group［ （88.03±3.65）%］ and ATRA group［ （71.15±4.55）%］， higher than the CBX group［ （49.62±5.85）%］， and the difference was statistically significant（ P ＜ 0.05）. The Morris water maze latency［ （51.18±4.14）s］ in the PSD group was significantly higher than that in the normal group［ （9.05±2.22）s］， the stroke group ［（9.06±2.25）s］ and the ATRA group［ （32.92±2.29）s］ ， lower than the CBX group［ （91.13±3.27）s］， and the difference was statistically significant（ P＜ 0.05）. The times of active escapes in the shuttle experiment in the PSD group［ （10.67±1.51） times］ was lower than that in the normal group［ （18.00±2.10） times］， the stroke group［ （16.5±1.87） times］ and the ATRA group［ （13.83±1.17） times］， Higher than CBX group［ （7.00±1.26） times］， and the difference was statistically significant（ P ＜ 0.05）. The relative expression of Cx36 mRNA in hippocampus of PSD group（ 0.50±0.03） was lower than that of normal group（ 1.04±0.07）， stroke group （1.07±0.10） and ATRA group（ 0.76±0.11）， but higher than that of CBX group（ 0.20±0.06）， and the difference was statistically significant（ P＜ 0.05）. The relative expression of Cx36 protein in hippocampus of PSD group （0.60±0.07） was lower than that of normal group（ 0.86±0.05）， stroke group（ 0.88±0.03） and ATRA group （0.77±0.07）， but higher than that of CBX group（ 0.56±0.02）， and the difference was statistically significant （P＜0.05）. The average fluorescence intensity of Cx36 protein in hippocampus of PSD group（ 0.60±0.07） was lower than that of normal group（ 1.00±0.03）， stroke group（ 0.97±0.03） and ATRA group（ 0.50±0.02）， but higher than that of CBX group（ 0.56±0.02）， and the difference was statistically significant（ P ＜ 0.05）；The mean fluorescence intensity of Cx36 protein in the hippocampus of the PSD group（ 0.36±0.03） was lower than that of the normal group（ 1.00±0.03）， the stroke group（ 0.97±0.03） and the ATRA group（ 0.50±0.02）， but higher than that of the CBX group（ 0.21±0.03）（P＜0.05）. Conclusions The expression of Cx36 protein in the hippocampus of PSD rats was decreased. CBX intervention could reduce the expression of Cx36 protein and aggravate the cognitive dysfunction of the rats. ATRA intervention could increase the expression of Cx36 protein and the cognitive dysfunction of the rats was slightly improved.