Effects of VNS on spatial memory after cerebral I/R injury
Rats were trained on the Morris water maze task for 5 d before surgery. On days 7 and 14 after surgery, animals were tested on the task to determine escape latencies (time to find the platform which was always on the same location at any time), path length, and swimming speed. On the swimming trajectory plots, the white box represents the starting place in the water tank, the red line represents the swim path, and the platform is positioned at the end of the red line (Fig. 2a). During the training trial, escape latencies and path length gradually decreased. For example, on the first day of training (day-5), the average escape latencies of the Sham, MCAO/R, and VNS+MCAO/R groups were 36.8, 37.7, and 37.6 s, respectively. However, on the fifth day of training, the escape latencies had decreased to 8.6, 6.4, and 8.4 s respectively (unpaired two-tailed t test: t(76) = 19.04, p < 0.0001 for Sham group; t(35) = 15.47, p < 0.0001 for MCAO/R group; t(44) = 12.31, p < 0.0001). Group differences at any point during the training were not significant. However, the escape latencies for the Sham group on days 7 and 14 after surgery were not significantly different from those of the later stages of training, whereas in the MCAO/R group, the post-surgery escape latencies increased to 19.7 and 16.3 s, respectively. In the VNS treatment group, the escape latencies were 10.9 and 6.6 s (escape latency was shorter on day 14 than on day 7), both of which were significantly lower than those in the MCAO/R group, but not significantly different from those in the Sham group (Fig. 1b) [two way ANOVA: F (2494) = 1.68, p = 0.1881, Bonferroni post hoc test: Sham vs. MCAO/R, p < 0.001 (day 7)]. The path lengths exhibited a similar trend in all three groups, decreasing gradually from an initial 500 cm to 200 cm at the end of training, without significant group differences. While changes in post-surgery path lengths exhibited a similar trend to those observed for escape latencies, path lengths were markedly longer on days 7 and 14 in the MCAO/R group (408.7 and 340.7 cm, respectively) than in the Sham group (156.1 and 151.4 cm, respectively). The path lengths were markedly shorter in the VNS treatment group (205.1 and 122.9 cm, p < 0.05) than in the MCAO/R group, but they were not significantly different from those of the Sham group [Two way ANOVA: F (2488) = 4.11, p = 0.017. Bonferroni post hoc tests: Sham vs. MCAO/R, p < 0.001 (day 7), p < 0.05 (day 14); MCAO/R vs. MCAO/R+VNS, p < 0.05 (days 7, 14) (Fig. 2c). Compared to pre-surgery results, there were no significant changes in swimming speed among any group [Two-way ANOVA, F (2, 150) = 6.97, p = 0.0013] (Fig. 2d). Thus, VNS effectively reduces spatial memory impairment after cerebral I/R-related injury in rats.
Effects of VNS on impaired fear memory after cerebral I/R
Changes in fear memory were examined using the shuttle box avoidance task on days from 5 to 16 after surgery. The number of electric shocks and avoidance responses were recorded on each testing day. The avoidance CR rate, mean shock duration, and latency to avoidance were calculated. As shown in Fig. 2, there was no initial difference in avoidance CR rates between the Sham, MCAO/R, and MCAO/R+VNS groups. For example, on post-surgery day 6, the avoidance CR rates were 20.0, 12.3, and 20.0 %, respectively. With increased training over time, the avoidance CR rates increased and reached 65.8 and 65.5 % at day 16 for the Sham and MCAO/R+VNS groups, respectively, whereas it remained low at 10.6 % in the MCAO/R group [Two-way ANOVA: F (2465) = 71.01, p < 0.0001. Bonferroni post hoc tests: sham vs. MCAO/R, p < 0.05 (days 11–13), p < 0.01 (day 14), p < 0.001 (days 15, 16); MCAO/R vs. MCAO/R+VNS, p < 0.05 (days 10–11), p < 0.001 (days 12–16)] (Fig. 3a). The mean shock duration was negatively associated with the avoidance CR rate for all groups. The mean shock duration had reduced in the Sham group from the initial 18.9 to 6.5 s, whereas in the MCAO/R group, the mean shock duration was markedly longer. The mean shock durations for the Sham group on post-surgery days 5 and 16 were 45.0 and 35.5 s, respectively with no significant group differences. In the VNS-treated rats, the mean shock duration was much shorter than that of the MCAO/R group, but was similar to that of the Sham group. Moreover, the 10-s mean shock duration on post-surgery day 16 was significantly shorter than the 22.3 s observed at day 5 [Two-way ANOVA, F (2315) = 165.78, p < 0.0001. Bonferroni post hoc tests: sham vs. MCAO/R, p < 0.01 (days 3–16); MCAO/R vs. MCAO/R+VNS, p < 0.05 (days 3–16)] (Fig. 3b). Avoidance latencies increased in the Sham group from 10.1 s on day 5 to 25.9 s on day 16. However, the avoidance latency of the MCAO/R group remained short at 6.5 and 3.6 s on days 5 and 16, respectively. The avoidance latency in the VNS group was between that of the Sham and MCAO/R groups and increased slowly from 10.0 s on day 5 to 16.8 s on day 16 [Two-way ANOVA, F (2323) = 42.73, p < 0.0001. Bonferroni post hoc tests: sham vs. MCAO/R, p < 0.01 (days 11–13); MCAO/R vs. MCAO/R+VNS, p < 0.05 (days 14, 16)] (Fig. 3c). These results indicate that VNS can effectively improve memory impairment in fear-conditioned animals after I/R-related injury.
The effects of VNS and neurotoxin DSP-4 on NE levels in cortical and hippocampal brain regions
Dopamine beta-hydroxylase (DβH), the enzyme that catalyzes the conversion of dopamine to norepinephrine, is released from sympathetic neurons. Therefore, in order to examine the effect of VNS and DSP-4 on NE levels in the cortical and hippocampal brain regions, we measured the expression of DβH using western blotting. Figure 4 shows that the DβH protein was inhibited by neurotoxin DSP-4.
Damage to catecholaminergic neurons inhibited retention of the VNS-mediated effect on spatial memory
The neurotoxin DSP-4, a chemical agent that damages noradrenergic neurons, was administered intraventricularly 30 min prior to surgery. Training procedures were performed as previously described and swimming trajectories were recorded on the Morris water maze task (Fig. 5a). As shown in Fig. 4, trained rats (day-1) could quickly locate the platform. On post-surgery day 7, the escape latencies of the DSP-4+MCAO/R group and the DSP-4+MCAO/R+VNS group were 640.3 and 416.6 s, respectively, which were significantly slower than the mean escape latency of the DSP-4+Sham group (119.5 s). Escape latencies did not significantly differ between the groups on post-surgery day 14 compared with those on post-surgery day 7 [Two-way ANOVA: F (2,140) = 7.61, p = 0.0007. Bonferroni post hoc tests: DSP-4+Sham vs. DSP-4+MCAO/R, p < 0.001 (days 7, 14); DSP-4+Sham vs. DSP-4+MCAO/R+VNS, p < 0.01 (day 7), p < 0.001 (day 14) (Fig. 5b). The swimming path length of rats in the DSP-4+Sham group was 63.1 cm on day-1 and 119.5 and 90.7 cm on post-surgery days 7 and 14, respectively. The swimming path length of the DSP-4+MCAO/R group increased from 117.8 cm before surgery to 640.3 cm and 410.27 cm on post-surgery days 7 and 14, respectively. The swimming path length of the DSP-4+MCAO/R+VNS group was similar to that of the DSP-4+MCAO/R group, and increased from 97.9 cm before surgery to 416.6 and 460.8 cm on post-surgery days 7 and 14, respectively. The swimming path lengths of the DSP-4+MCAO/R group and the DSP-4+MCAO/R+VNS group on post-surgery days 7 and 14 were not significantly different but were markedly longer than those in the DSP-4+Sham group [Two-way ANOVA: F (2,150) = 9.84, p < 0.0001. Bonferroni post hoc tests: DSP-4+Sham vs. DSP-4+MCAO/R, p < 0.001 (day 7), p < 0.01 (day 14); DSP-4+Sham vs. DSP-4+MCAO/R+VNS, p < 0.01 (day 7), p < 0.001 (day 14)] (Fig. 5c). No changes in swimming speeds occurred between pre- and post-surgery testing [Two-way ANOVA: F (2150) = 6.97, p < 0.0013. Bonferroni post hoc tests: p > 0.05.] (Fig. 5d). These results indicate that the previously observed protective effects of VNS on I/R-induced spatial memory impairment can be reversed by DSP-4, which damages noradrenergic neurons. Thus, VNS may exert its effects by increasing NE release.
Damage to catecholaminergic neurons inhibits retention of the VNS-mediated effect on fear memory
Rats were treated intraventricularly with DSP-4 30 min before surgery and shuttle boxes were used to assess the number of electric shocks, mean shock duration, and avoidance latencies on post-surgery days 5–16. As shown in Fig. 5, the avoidance CR rate increased gradually with continued training in the DSP-4+Sham group. For example, the avoidance CR rate increased from 26.0 % on post-surgery day 6 to 76.0 % on post-surgery day 16. However, the avoidance CR rates did not improve with training in the DSP-4+MCAO/R group, yielding avoidance CR rates of 18.6 % and 10.0 % at post-surgery days 6 and 16, respectively. Moreover, the avoidance CR rates remained low in the DSP-4+MCAO/R+VNS group, at 15.0 and 21.4 % on post-surgery days 6 and 16, respectively [Two-way ANOVA: F (2324) = 71.01, p < 0.0001. Bonferroni post hoc tests: DSP-4+sham vs. DSP-4+MCAO/R, p < 0.01 (day 8, days 13–14), p < 0.05 (day 16)] (Fig. 6a). For the mean shock duration in the DSP-4+Sham group, the initial rate diminished from 30.1 to 9.9 % on post-surgery day 16. In the DSP-4+MCAO/R group, the mean shock durations were 54.9 and 63.5 % on days 6 and 16, respectively, with no significant differences between pre- and post-training rates. In the DSP-4+MCAO/R+VNS group, the mean shock durations did not differ significantly and were 36.3 and 49.3 % on days 6 and 16, respectively. Although the mean shock duration in the DSP-4+MCAO/R+VNS group was slightly lower than that in the DSP-4+MCAO/R group on post-surgery day 16, it was substantially higher than that in the DSP-4+Sham group [Two-way ANOVA: F (2299) = 61, p < 0.0001. Bonferroni post hoc tests: DSP-4+Sham vs. DSP-4+MCAO/R, p < 0.05 (day 11), p < 0.01 (day 8), p < 0.001 (days 12–16); DSP-4+Sham vs. DSP-4+MCAO/R+VNS, p < 0.05 (days 12, 16), p < 0.01 (day 13)] (Fig. 6b). Avoidance latency in the DSP-4+Sham group gradually increased with continued training, from 10.1 % on day 6 to 21.1 % on day 16. In contrast, no change occurred between pre- and post-training in the DSP-4+MCAO/R group, with avoidance durations of 6.8 and 4.6 % on days 6 and 16, respectively. The avoidance latency for the DSP-4+MCAO/R+VNS group was not significantly different from that of the DSP-4+MCAO/R group, at 4.4 and 8.3 % at days 6 and 16, respectively [Two-way ANOVA: F (2298) = 96.98, p < 0.0001. Bonferroni post hoc tests: DSP-4+Sham vs. DSP-4+MCAO/R, p < 0.01 (day 12), p < 0.001 (days 8, 11, 13, 14, 15); DSP-4+Sham vs. DSP-4+MCAO/R+VNS, p < 0.05 (day 16), p < 0.001 (days 8, 11, 13, 14, 15)] (Fig. 6c). There was no difference between pre- and post- training values for the DSP-4+MCAO/R group (Fig. 6c). These results indicate that the VNS-mediated improvement in the I/R-induced behavioral impairment in fear-conditioned rats can be inhibited by treatment with DSP-4, providing further evidence that NE may mediate the effects of VNS.