Adult-born neurons immature during learning are necessary for remote memory reconsolidation in rats

Abstract

Memory reconsolidation, the process by which memories are again stabilized after being reactivated, has strengthened the idea that memory stabilization is a highly plastic process. To date, the molecular and cellular bases of reconsolidation have been extensively investigated particularly within the hippocampus. However, the role of adult neurogenesis in memory reconsolidation is unclear. Here, we combined functional imaging, retroviral and chemogenetic approaches in rats to tag and manipulate different populations of rat adult-born neurons. We find that both mature and immature adult-born neurons are activated by remote memory retrieval. However, only specific silencing of the adult-born neurons immature during learning impairs remote memory retrieval-induced reconsolidation. Hence, our findings show that adult-born neurons immature during learning are required for the maintenance and update of remote memory reconsolidation.

Fig. 1. Blocking protein synthesis after spatial memory reactivation impairs both remote memory reconsolidation and adult-born neurons activation.

a Experimental protocol: 2-month-old rats were injected with BrdU 1 week before MWM training. Rats were trained for 6 days and memory was reactivated 4 weeks later. Immediately after reactivation rats were injected (icv) with anisomycin (Ani-R, n = 7) or with aCsf (aCsf-R, n = 10). A group of rats received anisomycin without the reactivation session (Ani, n = 4). Memory was tested 2 days later and rats were sacrificed 90 min after the test. b Latency to find the platform during training and to first cross the position of the platform during the reactivation and test trials. Memory performances of Ani-R rats were impaired compared to those of aCsf-R rats during the test (Tukey’s test: ^###p = 0.0007) and compared to their performances at the reactivation trial (Tukey’s test: ****p < 0.0001). c Latency to cross the position of the platform at the test. Latency was higher for the Ani-R rats compared to that of aCsf-R and Ani rats (Tukey’s test: *p < 0.05). d Zif268 expression in BrdU-IR cells. Percentage of expression was higher in the aCsf-R group compared to that of control home cage (HC) rats (n = 5) and to that of Ani-R rats (Tukey’s test: *p < 0.05, **p < 0.01) but not different than that of Ani rats. e Confocal illustration showing BrdU-IR cells (red) coexpressing the cellular activation factor Zif268-IR (green). Bar scale 10 µm. All data shown are mean ± s.e.m. For statistical details, see Table S1.

Fig. 2. The effect of reconsolidation blockade on retrieval-induced activation is specific to adult-born neurons and not to developmentally-generated cells.

a Experimental protocol: 7-days-old rat pups were injected with CldU and later with Idu at the age of 2-month old. Six weeks later, they were trained for 6 days in the MWM and memory was reactivated 4 weeks later. Immediately after reactivation rats are injected (icv) with anisomycin (Ani-R, n = 12) or with aCsf (aCsf-R, n = 12). A group of rats received anisomycin without the reactivation session (Ani, n = 12). Memory was tested 2 days later and rats were sacrificed 90 min after the test. b Zif268 expression in IdU-IR cells. Percentage of expression was higher in the aCsf-R group compared to that of control home cage (HC) rats (n = 7) and to that of Ani-R rats (Tukey’s test *p < 0.05, **p < 0.01, ***p < 0.001) but not different than that of Ani rats. c Confocal illustration showing IdU-IR cells (red) coexpressing the cellular activation factor Zif268-IR (green). Bar scale 10 µm. d Zif268 expression in CldU-IR cells. Percentage of expression was similar among the groups. e Confocal illustration showing CldU-IR cells (red) coexpressing the cellular activation factor Zif268-IR (green). Bar scale 10 µm. All data shown are mean ± s.e.m. For statistical details, see Table S1.

Fig. 3. Effect of silencing immature or mature adult-born neurons on memory retrieval.

a Experimental protocol: 2-month-old rats were injected with Gi-GFP RV (n = 20) or its control GFP RV (n = 14) 1 week before MWM training. Rats were trained for 6 days and memory was reactivated 4 weeks later. Thirty minutes or 1 h before the reactivation test, CNO (1 mg/kg) was injected (i.p.) (Gi-GFP—30 min n = 10, Gi-GFP—1 h n = 10, GFP—30 min n = 7, GFP—1 h n = 7). b The latency to cross the platform was higher in the group of rats that received the Gi-GFP-RV and the CNO 1 h before the reactivation test compared to that of the other groups (Tukey’s test: Gi-GFP—30 min vs Gi-GFP—1 h, *p < 0.05; Gi-GFP—1 h vs GFP—1 h, **p < 0.01). c Experimental protocol: 2-month-old rats were injected with Gi-GFP RV (n = 10) or its control GFP RV (n = 7) 1 week before MWM training. Rats were trained for 6 days and memory was reactivated 2 days later. One hour before reactivation, CNO (1 mg/kg) was injected (i.p.). Rats were killed 90 min after the test. d The latency to cross the platform was similar between groups. e The adult-born neurons were not activated as the percentage of BrdU cells expressing Zif268 in trained-GFP rats (WM, n = 6) was similar to that of home cage animals (HC, n = 5) (two-tailed T-test t₉ = 1.604, p = 0.1433). f The number of BrdU-IR cells was higher in the group of trained rats that received the GFP-RV compared to that of home cage animals (HC) (two-tailed T-test, t₉ = 3.881, p = 0.004). g Illustration of GFP-labeled cells (green) expressing DCX(magenta). Example shown is a representative of a total of >24 sections from 4 rats. Scale bar: 20 µm. All data shown are mean ± s.e.m. For statistical details, see Table S1.

Fig. 4. Silencing during reconsolidation, neurons that were immature at the time of learning, impairs long-term memory persistence.

a Experimental protocol: 2-month-old rats were injected with Gi-GFP RV (n = 5) or its control GFP RV (n = 6) 1 week before MWM training. Rats were trained for 6 days and memory was reactivated 4 weeks later. Thirty minutes before reactivation, CNO (1 mg/kg) was injected (i.p.). Memory was tested 2 days later (Test) and all animals were killed 90 min after the test. b Latency to find the platform during training and to first cross the position of the platform during the reactivation and test trial. Memory performances of Gi-GFP rats were impaired at Test compared to those of GFP RV rats (Tukey’s test: **p < 0.01). c Zif268 expression in BrdU-IR cells. Percentage of expression was higher in the GFP rats compared to that of control home cage (HC, n = 5) rats and to that of Gi-FGP rats (Tukey’s test: ****p < 0.0001). d Number of BrdU-IR cells. The number of BrdU-IR cells was similar between the GFP and Gi-GFP groups and for both higher compared to that of home cage (HC, n = 5) animals (Tukey’s test: *p < 0.05). e Confocal illustration showing BrdU-IR cells (red) coexpressing the cellular activation factor Zif268-IR (green). Bar scale 10 µm. f Zif268 expression in GFP-IR cells. Percentage of expression was higher in the GFP rats compared to that of control HC rats (n = 5) and to that of Gi-FGP rats (Tukey’s test: **p < 0.01, ****p < 0.0001). g Confocal illustration showing GFP-IR cells (green) coexpressing the cellular activation factor Zif268-IR (magenta). Bar scale 10 µm. All data shown are mean ± s.e.m. For statistical details, see Table S1.

Fig. 5. Silencing during reconsolidation, neurons that were immature at the time of learning, impairs long-term memory persistence whereas silencing neurons that were mature at the time of learning had no impact on memory.

a Experimental protocol: 2-month-old rats were injected with Gi-GFP RV (n = 21) or its control GFP RV (n = 10) 1 week before MWM training. Rats were trained for 6 days and memory was reactivated 4 weeks later GFP RV (n = 10) and Gi-GFP-RV (n = 11). Thirty minutes before reactivation, CNO (1 mg/kg) was injected (i.p.). A group of rats Gi-GFP-RV (n = 10) received CNO but was not reactivated (nR). Memory was tested 2 days later (Test) and again 2 weeks later (Test 2). b Latency to find the platform during training and to first cross the position of the platform during the reactivation and test trials. Memory performances of Gi-GFP rats were impaired at Test 2 compared to those of GFP RV rats and compared to their own performances at the reactivation trial (Tukey’s test: **p < 0.01; ^##p < 0.01). c Latency to cross the position of the platform at Test 2. Latency was higher for the Gi-GFP rats compared to that of GFP rats (Tukey’s test: *p < 0.05). d Experimental protocol: 2-month-old rats were injected with Gi-GFP RV (n = 11) or its control GFP RV (n = 6) 6 weeks before MWM training. Rats were trained for 6 days and memory was reactivated 4 weeks later. 30 min before reactivation, rats were injected (i.p.) with 1 mg/kg CNO. Memory was tested 2 days later (Test) and again 2 weeks later (Test 2). e Latency to find the platform during training and to first cross the position of the platform during the reactivation and test trials. Memory performances of Gi-GFP rats and GFP rats are similar. All data shown are mean ± s.e.m. For statistical details, see Table S1.

Fig. 6. Silencing during reconsolidation, neurons that were immature at the time of learning, impairs memory update.

a Experimental protocol: 2-month-old rats were injected with Gi-GFP RV (n = 12) or its control GFP RV (n = 12) 1 week before MWM training. Rats were trained for 6 days and memory was reactivated 4 weeks later. Thirty minutes before reactivation, rats were injected (i.p.) with 1 mg/kg CNO. Memory was tested 2 days later (Test) and again 2 weeks later (Test 2). At the end of each reactivation and test sessions, the Atlantis platform (indicated by ○) raised and rats were put on the platform for 30 s. b, c Cross entries and time spent in the MWM zones (T: Target zone; O: others zones). The reactivation with the Atlantis platform led to an increase in the number of entries and the time spent in the target zone at the test for the GFP control rats at test T and T2 (Tukey’s test: *p < 0.05, **p < 0.01, ***p < 0.001). Atlantis reactivation had no effect on Gi-GFP performances at the tests. GFP rats enter more and spent more time in the target zone during tests than Gi-GFP rats (Tukey’s test: *p < 0.05, **p < 0.01). d Density plot for grouped data: The color level represents the lowest (Min) to the highest (Max) location frequency in pixels. e Gallagher Proximity: Average distance of rats from the former platform during the first 20 s. Efficiency: Percentage of distance crossed in the represented triangular zone. f, g Gallagher Proximity and Efficiency during reactivation and tests. Atlantis reactivation led to an increase in Gallagher and Efficiency performances at the tests for the GFP control rats (Tukey’s test: *p < 0.05, **p < 0.01, ***p < 0.001). Atlantis reactivation had no effect on Gi-GFP performances at the tests. All data shown are mean ± s.e.m. For statistical details, see Table S1.

Fig. 7. Silencing during reconsolidation, neurons that were 6-week old at the time of learning, has no impact on memory updating.

a Experimental protocol: 2-month-old rats were injected with Gi-GFP RV (n = 12) or its control GFP RV (n = 11) 6 weeks before MWM training. Rats were trained for 6 days and memory was reactivated 4 weeks later. Thirty minutes before reactivation, rats were injected (i.p.) with 1 mg/kg CNO. Memory was tested 2 days later (Test) and again 2 weeks later (Test 2). At the end of each reactivation and test sessions, the Atlantis platform (indicated by ○) raised and rats were put on the platform for 30 s. b, c Cross entries and time spent in the MWM zones (T: Target zone; O: others zones). The reactivation with the Atlantis platform led to an increase in the number of entries and the time spent in the target zone at the test for both GFP control rats and Gi-GFP rats (Tukey’s test: **p < 0.01, ***p < 0.001). d Density plot for grouped data: The color level represents the lowest (Min) to the highest (Max) location frequency in pixels. e, f Gallagher Proximity and Efficiency during reactivation and tests. The reactivation with the Atlantis platform led to an increase in Gallagher and Efficiency performances at the test for both GFP control rats and Gi-GFP rats (Tukey’s test: **p < 0.01, ***p < 0.001). All data shown are mean ± s.e.m. For statistical details, see Table S1.