Forniceal deep brain stimulation rescues hippocampal memory in Rett syndrome mice

Abstract

Deep brain stimulation (DBS) has improved the prospects for many individuals with diseases affecting motor control, and recently it has shown promise for improving cognitive function as well. Several studies in individuals with Alzheimer disease and in amnesic rats have demonstrated that DBS targeted to the fimbria-fornix, the region that appears to regulate hippocampal activity, can mitigate defects in hippocampus-dependent memory. Despite these promising results, DBS has not been tested for its ability to improve cognition in any childhood intellectual disability disorder. Such disorders are a pressing concern: they affect as much as 3% of the population and involve hundreds of different genes. We proposed that stimulating the neural circuits that underlie learning and memory might provide a more promising route to treating these otherwise intractable disorders than seeking to adjust levels of one molecule at a time. We therefore studied the effects of forniceal DBS in a well-characterized mouse model of Rett syndrome (RTT), which is a leading cause of intellectual disability in females. Caused by mutations that impair the function of MeCP2 (ref. 6), RTT appears by the second year of life in humans, causing profound impairment in cognitive, motor and social skills, along with an array of neurological features. RTT mice, which reproduce the broad phenotype of this disorder, also show clear deficits in hippocampus-dependent learning and memory and hippocampal synaptic plasticity. Here we show that forniceal DBS in RTT mice rescues contextual fear memory as well as spatial learning and memory. In parallel, forniceal DBS restores in vivo hippocampal long-term potentiation and hippocampal neurogenesis. These results indicate that forniceal DBS might mitigate cognitive dysfunction in RTT.

Extended Data Figure 1. Time line of forniceal DBS tests in RTT and WT mice

Extended Data Figure 2. Fear memory in RTT mice and WT control animals

All mice were trained with tone-foot shock pairings on day 0. Memory retention was tested 3h, 1d, 3d, and 7d after training. a, b, Impaired fear memory in RTT mice (n = 20) compared to WT littermates (n = 20). These animals were implanted with electrodes but without DBS or sham treatment. A significant main effect of genotype was observed (two-way repeated measures ANOVA followed by Tukey post hoc: context, F_1,38 = 15.32, P < 0.001; cue, F_1,38 = 20.70, P < 0.001). *P < 0.05; **P < 0.01; ***P < 0.001 vs. WT. c, d, Cued fear memory in RTT mice (n = 20) and WT littermates (n = 20) that were implanted with electrodes but without DBS or sham treatment. During the retention test, freezing in the tone phase (T) was significantly more than in the no tone phase (NT) across all the test time points in both WT (a) and RTT mice (b). e-h, Retrieval of cue fear memory in DBS or sham treated RTT and WT mice. During the cued memory test, all four groups of animals actively responded to the tone presentation (WT-sham, n = 21; WT-DBS, n = 21; RTT-sham, n = 14; RTT-DBS, n = 17). There was a significant increase of freezing time in the tone phase (T) compared to the no-tone phase (NT) at each of the test time points over all the groups. *P < 0.05, **P < 0.01, ***P < 0.001 (paired t-test). All data are presented as mean ± s.e.m.

Extended Data Figure 3. Increased handling, but not forniceal DBS, increased locomotor activity and decreased the anxiety level in RTT and WT mice

a, There was no difference among the four DBS-treated groups in the total distance traveled in the open field test (WT-sham, n = 20; WT-DBS, n = 20; RTT-sham, n = 17; RTT-DBS, n = 18; genotype, F_1,71 = 1.13, P = 0.292; treatment, F_1,71 = 0.13, P = 0.724; genotype × treatment, F_1,71 = 0.063, P = 0.803). RTT and WT mice that received DBS/sham treatment traveled longer distances than RTT (n = 20) and WT (n = 20) animals that were implanted with electrodes but did not experience DBS/sham procedures, respectively. b, During the open field test, there was no difference in the center/total distance ratio among the four DBS groups (genotype, F_1,71 = 1.22, P = 0.273; treatment, F_1,71 = 0.0079, P = 0.93; genotype × treatment, F_1,71 = 0.081, P = 0.777). Both RTT and WT mice that received DBS/sham treatment traveled more in the center area compared to implanted RTT and WT animals without DBS/sham procedures. c, In the light-dark test there was no difference in the amount of time spent in the light compartment among the 4 chronically treated groups (n = 12 per group; two-way ANOVA: genotype, F_1,44 = 1.83, P = 0.183; treatment, F_1,44 = 0.057, P = 0.813; genotype × treatment, F_1,44 = 0.33, P = 0.567). Both RTT and WT mice that received DBS/sham treatment spent more time in the light compartment than implanted RTT (n = 15) and WT (n = 14) animals without DBS/sham procedures. *P < 0.05,0 **P < 0.01, ***P < 0.001 (two-tailed t-test). All data are presented as mean ± s.e.m.

Extended Data Figure 4. Forniceal DBS did not alter the pain threshold, motor function, or social behavior in RTT or WT mice

a, There was no group difference in foot shock threshold intensities to evoke flinch, vocalization, or jumping (WT-sham, n = 14; WT-DBS, n = 14; RTT-sham, n = 11; RTT-DBS, n = 12; two-way ANOVA, no significant main effect of genotype, treatment, or genotype × treatment interaction, P > 0.05). b, In a rotarod test (n = 12 mice per group), latency to fall increased over trials but there was no difference among the 4 groups (two-way repeated measures ANOVA: group, F_3,44 = 1.68, P = 0.184; trial, F_7,308 = 34.26, P < 0.001; group × trial interaction, F_21,308 = 1.22, P = 0.230). c, RTT mice showed decreased latency to fall in the wire hang test compared to WT animals, but there was no difference between DBS and sham treated groups for either RTT or WT mice (n = 12 per group; two-way ANOVA: genotype, F_1,44 = 10.41, P = 0.002; treatment, F_1,44 = 0.33, P = 0.566; genotype × treatment interaction, F_1,44 = 0.75, P = 0.392). d, RTT mice showed a decreased latency to fall in the dowel test compared to WT animals, but there was no difference between DBS and sham treated groups for either genotype (n = 12 per group; genotype, F_1,44 = 23.63, P < 0.001; treatment, F_1,44 = 0.0018, P = 0.966; genotype × treatment interaction, F_1,44 = 0.83, P = 0.367). e, f, In the three chamber test, all 4 groups of animals (n = 12 per group) showed a clear preference for the partner mice compared to the object (e). Two-way ANOVA revealed a significant genotype main effect of the interaction time with the partner mice (F_1,44 = 4.56, P = 0.038), indicating altered social behavior in RTT mice (P = 0.063, RTT-sham vs. WT-sham, Tukey post hoc). However, DBS did not change the interaction time with the partners (treatment, F_1,44 = 0.28, P = 0.597; genotype × treatment interaction, F_1,44 = 0.31, P = 0.579) or the object (treatment, F_1,44 = 2.64, P = 0.111; genotype × treatment interaction, F_1,44 = 0.015, P = 0.905) (f). **P < 0.01, ***P < 0.001 (Tukey post hoc in c, d; paired t-test in e). All data are presented as mean ± s.e.m.

Extended Data Figure 5. Forniceal DBS did not alter the body weight, visual or sensorimotor skills in RTT or WT mice

a, All four groups (n = 12 mice per group) showed changes in body weight over time. Two-way repeated measure ANOVA revealed a significant main effect of group (F_3,44 = 6.73, P < 0.001) and age (F_4,176 = 89.32, P < 0.001). Tukey post hoc showed that RTT-sham mice were significantly heavier than WT-sham mice (P = 0.015), but there was no difference in body weight between WT-sham and WT-DBS (P = 0.861) or between RTT-sham and RTT-DBS (P = 0.099) mice. b, Comparison of body weight at the age of 23 weeks among the four groups (two-way ANOVA: genotype, F_1,44 = 10.06, P = 0.003; treatment: F_1,44 = 1.93, P = 0.172). c-e, Swimming test in the water maze task with a flagged platform (n = 18 mice per group). RTT-sham mice did not have different escape latencies than WT-sham controls (c, two-way repeated measures ANOVA: genotype, F_1,34 = 1.73, P = 0.197; genotype × treatment interaction, F_1,34 = 0.133, P = 0.718). DBS did not change the escape latencies in either WT controls (d, treatment, F_1,34 = 0.44, P = 0.513; treatment × day interaction, F_1,34 = 1.24, P = 0.273) or RTT mice (e, treatment, F_1,34 = 2.36, P = 0.134; treatment × day interaction, F_1,34 = 0.41, P = 0.524). *P < 0.05; n.s., not significant. (Tukey post hoc). All data are presented as mean ± s.e.m.

Extended Data Figure 6. Effect of forniceal DBS on hippocampal electrophysiological signatures

a, Representative traces of local field potentials recorded in the dentate gyrus 1 day before and 3 weeks after DBS/sham treatment. There were no electrographic seizure spikes in any of the four groups of mice after DBS/sham treatment. Scale bars: 10 s, 1 mV. b, Input-output (I/O) curves of the evoked responses of the perforant path recorded in the dentate gyrus in DBS/sham treated mice. For each of the four groups, I/O curves were generated 1 day before and 3 weeks after forniceal DBS. All data points were normalized to the maximum value of the population spike amplitude before DBS/sham and the abscissa represents the 7 increments used in each mouse. The I/O relationship was not altered by DBS in WT-sham (n = 5, F_1,4 = 0.062, P = 0.818), WT-DBS (n = 4, F_1,3 = 0.036, P = 0.861), or RTT-sham (n = 5, F_1,4 = 0.018, P = 0.901). DBS reduced the amplitude of the evoked population spikes from the baseline test in RTT-DBS mice (n = 5, F_1,4 = 6.73, P = 0.060). *P < 0.05 (Tukey post hoc). All data are presented as mean ± s.e.m.

Extended Data Figure 7. Unilateral forniceal DBS induces neuronal activity and stimulates neurogenesis bilaterally in the dentate gyrus

a, Representative images showing that the expression of c-fos gene was increased following forniceal DBS in WT and RTT mice compared to their sham controls, respectively (percentage of ipsilateral c-fos positive cells over the dentate granule cells: WT-sham, 0.26 ± 0.04%; WT-DBS, 34.52 ± 4.62%; RTT-sham, 0.30 ± 0.05%; RTT-DBS, 32.55 ± 3.74%). b, Representative images showing that there were more BrdU⁺ (green), DCX⁺ (red), and merged (yellow) cells in the dentate gyrus in forniceal DBS-treated WT and RTT mice than in their respective sham controls. Scale bar, 100 μm. Con, contralateral; Ips, ipsilateral.

Extended Data Figure 8. The cholinergic antagonist atropine did not alter forniceal DBS-induced enhancement of fear memory

a, Placement of guide cannula and recording electrode into the dorsal hippocampus. b, Hippocampal infusion of 1.0 μg atropine did not change the amplitudes of the evoked potentials of the FFx recorded in the dentate gyrus in both RTT and WT mice. There was no difference of the population spike amplitudes before or after atropine infusion in both RTT mice (n = 5; one-way ANOVA, F_9,36 = 0.69, P = 0.715) and WT controls (n = 3; F_9,18 = 0.99, P = 0.485). c, Representative hippocampal EEG traces before and after vehicle (V) or atropine (A) infusion. Scale bars: 0.5 s, 0.2 mV. d, RTT mice (n = 17) showed less spontaneous hippocampal theta activity than WT animals (n = 20) (**P < 0.01, two-tailed t-test). e, Hippocampal infusion of atropine, but not vehicle, reduced hippocampal theta oscillation in both RTT and WT mice compared to their pre-infusion baselines (WT-V, n = 9; WT-A, n = 11; RTT-V, n = 8; RTT-A, n = 9; *P < 0.05, two-tailed paired t-test; n.s., not significant). f, Hippocampal microinfusion of atropine before fear conditioning training did not alter fear memory in forniceal DBS treated RTT mice or WT controls. Mice in all 4 groups (WT-V, n = 10; WT-A, n = 11; RTT-V, n = 12; RTT-A, n = 13) experienced two weeks of forniceal DBS that was finished three weeks before fear conditioning training. Atropine or vehicle was bilaterally infused into the dorsal hippocampus before training. Memory retention was tested 24 h after training. Two-way ANOVA revealed a significant main effect of genotype (F_1,42 = 10.27, P = 0.003), but there was no difference between atropine- and vehicle-treated mice (treatment, F_1,42 = 0.34, P = 0.562; genotype × treatment interaction, F_1,42 = 0.069, P = 0.794). Atropine did not change cued fear memory, either: two-way ANOVA revealed no difference between genotypes (F_1,42 = 2.99, P = 0.091) or between atropine- and vehicle-treated mice (treatment, F_1,42 = 0.046, P = 0.831; genotype × treatment interaction, F_1,42 = 0.154, P = 0.697). *P < 0.05; n.s., not significant (Tukey post hoc). g, Intrahippocampal atropine infusion alone did not change the basal level of freezing in the contextual test environment in either WT or RTT mice. There was no difference between vehicle- (n = 9) or atropine-treated (n = 6) mice (P > 0.05, two-tailed t-test). h, Schematic representation of the dorsal hippocampus at seven rostral-caudal planes (according to Paxinos and Franklin, 2001) for the microinfusion cites in DBS treated experiments. The numbers on the left represent the posterior coordinate from the bregma. All data are presented as mean ± s.e.m.

Figure 1. Forniceal DBS restores contextual fear memory in RTT mice

a, b, Photomicrographs illustrating DBS electrode placement (arrowheads) in the FFx (a) and the recording electrode in the DG (b). cc: corpus callosum; LV: lateral ventricle; D3V: dorsal 3^rd ventricle; DG: dentate gyrus. c, Representative evoked potential trace of the FFx pathway recorded in the dentate. d, Forniceal DBS enhanced contextual fear memory in both WT and RTT mice (WT-DBS, n = 21; WT-sham, n = 21; RTT-DBS, n = 17; RTT-sham, n = 14). There were significant main effects on freezing time among the 4 groups (two-way repeated measures ANOVA: group, F_3,69 = 5.67, P = 0.002; day, F_3,180 = 6.44, P < 0.001; group × day interaction, F_9,180 = 2.15, P = 0.027). Within-genotype analysis revealed a significant DBS effect in both WT (F_1,40 = 8.50, P = 0.006) and RTT mice (F_1,29 = 6.44, P = 0.016). DBS in RTT mice restores contextual fear memory to WT level (RTT-DBS vs. WT-sham: group, F_1,36 = 2.76, P = 0.105). Comparison of contextual fear memory on d1 among the 4 groups revealed a significant main effect (two-way ANOVA followed by Tukey post hoc: genotype, F_1,69 = 8.39, P = 0.005; treatment, F_1,69 = 11.41, P = 0.001). e, Cued fear memory of mice tested in d. There was no difference in cued fear memory between groups over any time point (main effect: group, F_3,69 = 0.88, P = 0.456; day, F_3,180 = 1.65, P = 0.179; group × day interaction, F_9,180 = 0.89, P = 0.538) or on d1 (genotype, F_1,69 = 0.64, P = 0.428; treatment, F_1,69 = 0.11, P = 0.741). *P < 0.05. n.s., not significant. Data presented as mean ± s.e.m. Scattergrams show individual values.

Figure 2. Forniceal DBS rescues spatial learning and memory in RTT mice

In the water maze task, all mice were trained with a hidden platform for 9 days followed by a probe test without the platform 24 h after the last training. There were significant main effects of escape latencies among the 4 groups (n = 18 mice per group) during acquisition training (two-way repeated measures ANOVA: group, F_3,68 = 20.74, P < 0.001; day, F_8,544 = 19.72, P < 0.001). a, RTT-sham mice showed increased escape latencies during training (genotype, F_1,34 = 35.30, P < 0.001; day, F_8,272 = 7.06, P < 0.001), but decreased time in target quadrant (P < 0.01) and fewer platform area crossings during the probe test (P < 0.001) than WT-sham controls. b, Forniceal DBS decreased escape latencies during training in WT-DBS mice compared to WT-sham controls (treatment, F_1,34 = 5.94, P = 0.020; day, F_8,272 = 17.10, P < 0.001; treatment × day, F_8,272 = 2.19, P = 0.028). There was no difference in time spent in the target quadrant (P > 0.05) or in the number of platform area crossings (P > 0.05) between DBS and sham groups during the probe test. c, The RTT-DBS mice showed shorter escape latencies during training (treatment, F_1,34 = 10.31, P = 0.003; day, F_8,272 = 6.13, P < 0.001) but more time in the target quadrant (P < 0.05) and platform area crossings (P < 0.05) during the probe test than RTT-sham controls. d, There was no difference between RTT-DBS mice and WT-sham controls in escape latencies during training (group, F_1,34 = 2.91, P = 0.097; group × day interaction, F_8,272 = 0.80, P = 0.606), time in the target quadrant (P > 0.05), or number of crossings of the platform area (P > 0.05) during the probe test. *P < 0.05, **P < 0.01, ***P < 0.001. n.s., not significant (Tukey post hoc in acquisition; two-tailed unpaired t-test between groups and paired t-test within group in probe). Data presented as mean ± s.e.m. Scattergrams show individual values.

Figure 3. Forniceal DBS rescues hippocampal synaptic plasticity in freely-moving RTT mice

a, Superimposed traces of the perforant path recorded in the DG 5 min before (gray) and 55 min after (black or red) tetani. b, RTT-sham mice (n = 12) showed impaired LTP compared to WT-sham group (n = 11) on day 0 (two-way repeated measures ANOVA: genotype, F_1,21 = 11.34, P = 0.003; time, F_15,315 = 40.51, P < 0.001; genotype × time interaction, F_15,315 = 9.36, P < 0.001), d1 (genotype, F_1,21 = 7.46, P = 0.012; time, F_3,63 = 5.15, P = 0.003), and d2 (genotype, F_1,21 = 6.50, P = 0.019). c-d, Forniceal DBS enhanced LTP in both WT and RTT mice (WT-DBS, n = 12; WT-sham, n = 11; RTT-DBS, n = 13; RTT-sham, n = 12). Two-way repeated measures ANOVA revealed significant main effects of population spike amplitudes among the 4 groups on day 0 (group, F_3,44 = 17.25, P < 0.001; time, F_15,660 = 167.28, P < 0.001; group × time interaction, F_45,660 = 14.50, P < 0.001), d1 (group, F_3,44 = 21.53, P < 0.001; time, F_3,132 = 7.69, P < 0.001), d2 (group, F_3,44 = 16.21, P < 0.001; time, F_3,132 = 8.96, P < 0.001), and d5 (group, F_3,44 = 8.42, P < 0.001; time, F_3,132 = 8.35, P < 0.001). c, Forniceal DBS enhanced LTP in WT controls on day 0 (treatment, F_1,21 = 12.16, P = 0.002; time, F_15,315 = 121.93, P < 0.001; treatment × time interaction, F_15,315 = 10.91, P < 0.001), d1 (treatment, F_1,21 = 18.99, P < 0.001; time, F_3,63 = 4.77, P = 0.005), d2 (treatment, F_1,21 = 11.25, P = 0.003; time, F_3,63 = 3.72, P = 0.016), and d5 (treatment, F_1,21 = 9.44, P = 0.006; time, F_3,63 = 6.73, P < 0.001). d, Forniceal DBS enhanced LTP in RTT mice on day 0 (treatment, F_1,23 = 11.86, P = 0.002; time, F_15,345 = 45.02, P < 0.001; treatment × time interaction, F_15,345 = 10.31, P < 0.001), d1 (treatment, F_1,23 = 14.60, P < 0.001; time, F_3,69 = 2.91, P = 0.041), and d2 (treatment, F_1,23 = 19.45, P < 0.001; time, F_3,69 = 6.09, P < 0.001). e, There was no difference of LTP between RTT-DBS and WT-sham (P > 0.05 for all the test days). Arrow, LTP induction. *P < 0.05, **P < 0.01, ***P < 0.001. n.s., not significant. Data presented as mean ± s.e.m.

Figure 4. Forniceal DBS stimulates hippocampal neurogenesis in WT and RTT mice

One day after completing the two weeks of forniceal DBS/sham treatment, animals were perfused for immunohistochemical detection of BrdU- and DCX-positive cells in the DG. a-d, Representative images ipsilateral to the DBS/sham at low (top; scale bar, 100 μm) and high magnification (bottom; scale bar, 50 μm) showing BrdU⁺ cells (green), DCX⁺ cells (red), and the merge (yellow) from each of the four groups. New neurons were located in the innermost layer of the dentate gyrus. e-g, Summary of immunoreactive cell counting (n = 6 mice per group). Two-way ANOVA revealed a significant main effect on the numbers of BrdU⁺ cells (e; treatment, F_1,20 = 23.49, P < 0.001), DCX⁺ cells (f; genotype, F_1,20 = 5.65, P = 0.028; treatment, F_1,20 = 29.65, P < 0.001), and BrdU⁺/DCX⁺ double staining cells (g; treatment, F_1,20 = 32.99, P < 0.001). Tukey post hoc indicated that RTT-sham mice had fewer BrdU⁺ (e) and DCX⁺ (f) cells than WT-sham controls. Forniceal DBS increased the numbers of BrdU⁺ (e), DCX⁺ (f), and BrdU⁺/DCX⁺ (g) double staining cells in WT-DBS and RTT-DBS mice compared to their respective sham controls. *P < 0.05, **P < 0.01, ***P < 0.001. Scattergrams show individual values.