Cyclin D2-knock-out mice with attenuated dentate gyrus neurogenesis have robust deficits in long-term memory formation

Neurobehavioral studies have produced contradictory findings concerning the function of neurogenesis in the adult dentate gyrus. Previous studies have proved inconsistent across several behavioral endpoints thought to be dependent on dentate neurogenesis, including memory acquisition, short-term and long-term retention of memory, pattern separation, and reversal learning. We hypothesized that the main function of dentate neurogenesis is long-term memory formation because we assumed that a newly formed and integrated neuron would have a long-term impact on the local neural network. We used a cyclin D2-knock-out (cyclin D2−/−) mouse model of endogenously deficient dentate neurogenesis to test this hypothesis. We found that cyclin D2−/− mice had robust and sustained loss of long-term memory in two separate behavioral tasks, Morris water maze (MWM) and touchscreen intermediate pattern separation. Moreover, after adjusting for differences in brain volumes determined by magnetic resonance (MR) imaging, reduced dentate neurogenesis moderately correlated with deficits in memory retention after 24 hours. Importantly, cyclin D2−/− mice did not show deficits in learning acquisition in a touchscreen paradigm of intermediate pattern separation or MWM platform location, indicating intact short-term memory. Further evaluation of cyclin D2−/− mice is necessary to confirm that deficits are specifically linked to dentate gyrus neurogenesis since cyclin D2−/− mice also have a reduced size of the olfactory bulb, hippocampus, cerebellum and cortex besides reduced dentate gyrus neurogenesis.


Supplemental Figure 4
Cyclin D2 -/mice showed decreased swim velocity in acquisition of Morris water maze task. Swim velocity was recorded during water maze training sessions, and cyclin D2 -/mice (N = 27) demonstrated slower swim speeds compared to WT (N = 20). Data shown as mean ± SEM. * indicates p <.05, main effect of genotype in repeated measures two-way ANOVA.

Supplemental Figure 5
Independent cohort replicates long-term spatial memory deficit in cyclin D2 -/mice. A, Cyclin D2 -/mice (N = 11) were able to acquire the location of the platform over the course of 7 training days. WT mice (N = 14), again, performed significantly better in latency to platform. B, Both WT and cyclin D2 -/had intact short-term memory during a probe trial 3 hours after the last training session. C, At 24 hours, cyclin D2 -/showed a deficit in spatial memory of the platform location. WT demonstrated intact memory by spending significantly more time in the target quadrant compared to the other quadrants. D-F, Cyclin D2 -/mice did not show intact long-term memory of the platform location when probed at 1-week, 2-weeks, and 4-weeks, whereas WT do show intact long-term memory of the platform location. Data shown as mean ± SEM. * indicates p < .05 target quadrant versus other quadrants.

Supplemental Figure 6
Cyclin D2 -/mice demonstrate long-term memory deficits in the Morris water maze when slower swim speed is accounted for. To ensure that phenotype observed was not due to reduced swim speed, Morris water maze data was analyzed by looking at the distance travelled within each quadrant. A, Three hours after the last training session, WT and cyclin D2 -/mice both swam around more in the target quadrant compared to the other quadrants, suggesting both groups had intact short-term memory. B-E, The long-term spatial memory deficits cyclin D2 -/mice displayed were reinforced in this analysis; WT travelled more in the target quadrant at the 24-hours, 1-week, 2week and 4-week probes compared to the cyclin D2 -/mice. Data shown as mean ± SEM. * indicates p < .05 target quadrant versus other quadrants.

Supplemental Figure 7
Dentate neurogenesis adjusted for differences in total brain volume moderately correlates with time spent in the target quadrant 24 hours after Morris water maze acquisition learning. A, A weak correlation was found between the number of DCX+ cells normalized to total brain volume and time spent in target quadrant 3 hours after acquisition. B, A moderate correlation (r= 0.33314, F(1,18) = 2.25, p= 0.1512) between DCX+ cell count adjusted for total brain volume and time spent in target quadrant was seen 24 hours after acquisition. C-F, The correlation between DCX+ cell count adjusted for total brain volume was weaker at later time points and across all tested time points. Table 1 Statistical data of Supplemental Figures Supplemental Figure 8 Cyclin D2 -/mice have intact visual discrimination in a non-hippocampal pairwise touchscreen assay. A, Subjects were trained in operant touchscreen chambers and then a non-hippocampal, non-spatial task was employed. WT (N = 9) and cyclin D2 -/mice (N = 10) acquired the task in a similar number of sessions, reaching the criterion of 80% accuracy in two consecutive days in an average of 29 sessions. B, The proportion of individual subjects that completed acquisition at each training day (survival curve) was not significantly different between genotype. Data shown as mean ± SEM.

Group
Metric