Immune activation state modulates infant engram expression across development

Abstract

Infantile amnesia is possibly the most ubiquitous form of memory loss in mammals. We investigated how memories are stored in the brain throughout development by integrating engram labeling technology with mouse models of infantile amnesia. Here, we found a phenomenon in which male offspring in maternal immune activation models of autism spectrum disorder do not experience infantile amnesia. Maternal immune activation altered engram ensemble size and dendritic spine plasticity. We rescued the same apparently forgotten infantile memories in neurotypical mice by optogenetically reactivating dentate gyrus engram cells labeled during complex experiences in infancy. Furthermore, we permanently reinstated lost infantile memories by artificially updating the memory engram, demonstrating that infantile amnesia is a reversible process. Our findings suggest not only that infantile amnesia is due to a reversible retrieval deficit in engram expression but also that immune activation during development modulates innate, and reversible, forgetting switches that determine whether infantile amnesia will occur.

Fig. 1.. MIA male offspring do not demonstrate infantile amnesia in a contextual fear paradigm.

(A) Developmental trajectory of infant mice. (B) Behavioral schedule. The black lightning symbol represents foot shocks. Mice are trained using contextual fear conditioning (CFC) in context A and tested for recall 1 or 8 days later. S, shock; NS, no shock. (C) Memory recall in context A. Adult (P63) C57BL/6 J male mice (n = 8) froze significantly more in context A than no shock controls (n = 8) both 1 and 8 days after training. Infant (P17) C57BL/6 J male mice (n = 9) froze significantly more than no shock controls during recall 1 day after training. No significant difference in freezing between groups (n = 9) when tested for recall 8 days after training. (D) Representative diagram of maternal immune activation (MIA) in pregnant dams. The syringe indicates poly(I:C) (PIC) injection at E12.5. (E) Social preference index (% time spent exploring social stimulus out of total object investigation time) of MIA adult C57BL/6 J male offspring (n = 10). (F) Marble-burying behavior of MIA adult C57BL/6 J male offspring (n = 10). The marble burying index is plotted on the y axis. (G) Memory recall in male C57BL/6 J MIA offspring in context A after CFC at P17 (n = 10). *P < 0.05, **P < 0.01, and ***P < 0.001 were calculated by (C) two-way analysis of variance (ANOVA) or (E to G) nested ANOVA with Bonferroni post hoc tests. Data are presented as means ± SEM. IL-6, interleukin-6; IL-17a, interleukin-17a; T_H17, T helper 17.

Fig. 2.. Natural reactivation of engram cells for an infant-encoded memory.

(A) Engram labeling method in Ai32-FosTRAP mice. (B) Whole-brain engram labeling of an infant engram. Scale bar, 400 μm. (C) Dentate gyrus (DG) engram cells (ChR2-EYFP⁺ green) in Ai32-FosTRAP mice. Scale bar, 100 μm. (D) Behavioral schedule for engram labeling of the home cage (HC), a context (Cxt), or contextual fear memory (CFC) in P17 Ai32-FosTRAP mice. The black lightning symbol represents foot shocks. The syringe symbol represents 4-hydroxytamoxifen (4-OHT) injection 2 hours after training. (E and F) ChR2-EYFP⁺ cell counts in the DG (N = 4 to 6, n = 4) and amygdala (AMG) (N = 4, n = 4). (G) Example images of DG ChR2-EYFP⁺ (green) and c-Fos⁺ (red) cell counts after recall at P20 and P63. IA, infantile amnesia. Scale bars, 20 μm. (H and K) Behavioral schedule. (I and J) Engram reactivation (c-Fos⁺ChR2-EYFP⁺/ChR2-EYFP⁺) in the DG (N = 4 to 6, n = 4) and AMG (N = 4, n = 4) after recall or no recall (home cage) at P20 and P63. (L) ChR2-EYFP⁺ cell counts in the DG in MIA male offspring (N = 5 to 6, n = 4). (M) Scatter plot of the relationship between the number of EYFP⁺/DAPI (4′,6-diamidino-2-phenylindole) cells and freezing behavior (%) (N = 8, n = 4). (N) c-Fos⁺ cell counts in the DG (N = 4 to 5, n = 4). (O) c-Fos⁺ cell counts in non-engram cells the DG (N = 10, n = 4). (P) c-Fos⁺ChR2-EYFP⁺/ChR2-EYFP⁺ cell counts in the DG (N = 4 to 5, n = 4). (Q) Engram cell dendritic spines in the DG of MIA male offspring after recall at P25 (N = 4, n = 10). Scale bars, 40 μm. (R) Spine density per 10 μm. (S) Average dendritic spine head diameter (in micrometers). (T) Average dendritic spine head volume (in square micrometers). *P < 0.05, **P < 0.01, and ***P < 0.001 calculated by (E, L, and N to P) nested Student’s t test, (R to T) Student’s t test, (M) Pearson’s correlation, or (F, I, and J) nested ANOVA with Bonferroni post hoc tests. Data are presented as ±SEM.

Fig. 3.. Artificial updating of infant-encoded engram cells permanently reinstates a “lost” infantile memory.

(A) Behavioral schedule for optogenetic reactivation of infant DG engram cells in adulthood. (B) Representative image of optogenetic stimulation of DG engram cells in Ai32-FosTRAP–labeled mice. Scale bar, 400 μm. (C) Freezing levels of infant mice (n = 10) during natural memory recall. (D) Memory recall in context C (engram reactivation) with light-off and light-on epochs. (E) Freezing for the two light-off and light-on epochs averaged. (F and I) ChR2-EYFP⁺ cell counts in the DG (N = 4, n = 4) and AMG (N = 5, n = 4) after optogenetic stimulation at P63. (G and J) c-Fos⁺ cell counts in the DG and AMG. (H and K) c-Fos⁺ChR2-EYFP⁺/ChR2-EYFP⁺ cell counts in the DG and AMG. (L) Behavioral schedule for artificial updating of an infant engram. (M) Freezing levels during recall test in context B and context A (n = 10). The experimental light group froze significantly more in context A. (N and O) Histological representation of engram reactivation in the (N) DG and (O) RSC during recall in context A. DAPI⁺ (blue), ChR2-EYFP⁺ (green), c-Fos⁺ (red). Scale bars, 20 μm. (P and Q) DG (N = 6, n = 4) and RSC (N = 4 to 5, n = 4) c-Fos⁺ChR2-EYFP⁺/ChR2-EYFP⁺ cell counts after recall in context A. NL, no light; L, light. *P < 0.05, **P < 0.01, and ***P < 0.001 calculated by (F to K, P, and Q) nested Student’s t test or (C, E, and M) two-way ANOVA with Bonferroni post hoc tests. Data are presented as ±SEM.

Fig. 4.. Infantile amnesia for context, object, and maze memory.

(A and C) Behavioral schedule for novel object recognition. (B) Interaction index (%) of infant C57BL/6 J mice during novel object recognition test 1 or 8 days after acquisition (n = 9). (D) Interaction index of MIA male infant offspring (n = 9 to 10). (E) Behavioral schedule for DG optogenetic reactivation of an infant-encoded engram for an object. (F) Interaction index of adult mice after optogenetic stimulation (n = 9 to 10). (G) Behavioral schedule for the Barnes maze in infant C57BL/6 J mice. (H and I) Time spent in each zone during the test (H) 1 or (I) 8 days after training. (J) Example representations of navigational search strategies. (K and L) Infant and adult search strategies during testing 1 and 8 days after training. (M) Behavioral schedule for the Barnes maze in infant C57BL/6 J male MIA offspring. (N, P, and R) Time spent in each zone by MIA male offspring during test 8 days after training (n = 9 to 11). (O, Q, and S) Heatmap analysis of probe test in (O) Ctrl, (Q) PIC, and (S) IL-17a male MIA offspring. (T) Behavioral schedule for DG optogenetic reactivation of an infant-encoded engram for the Barnes maze. (U and W) Time spent in each zone during the test (n = 8 to 9). (V and X) Heatmap analysis of probe test in (V) no light and (X) light groups. *P < 0.05, **P < 0.01, and ***P < 0.001 calculated by (H, I, U, and W) ANOVA, (D, N, P, and R) nested ANOVA, (B) two-way repeated-measures ANOVA, or (F) two-way ANOVA with Bonferroni post hoc tests. Data are presented as ±SEM.

Fig. 5.. IL-17a is required for MIA effects on infantile amnesia.

(A) Representative diagram of MIA in Il17a KO mice. (B) Behavioral schedule for CFC in Il17a KO male mice. (C) Freezing levels of male Il17a KO infant mice during recall 1 or 8 days after training. (D) Behavioral schedule for postnatal immune activation in C57BL/6 J infant mice. The syringe symbol represents PIC injection at P3, P7, and P14. (E) Behavioral schema for CFC. (F) Freezing levels of infant male mice when tested 8 days after training. (G) Behavioral schedule for IL-17a injection 3 hours before memory encoding (P17) or recall (P25) in MIA C57BL/6 J offspring. (H and I) Freezing levels of infant male mice when tested 8 days after training. *P < 0.05, **P < 0.01, and ***P < 0.001 calculated by (F, H, and I) nested Student’s t test or (C) nested ANOVA with Bonferroni post hoc tests. Data are presented as ±SEM.