Autophagy in DG engrams mediates Rac1-dependent forgetting during aging via microglia-mediated synapse elimination

Engrams are considered to be substrates for memory storage, and the functional dysregulation of the engrams leads to cognition impairment. However, the pathological changes of the engrams leading to forgetting, which typically involves a failure in memory retrieval, remains unclear. Here we found that the expression of autophagy protein 7 (Atg7) in dentate gyrus (DG) engrams was dramatically increased in aged mice, leading to the activation of surrounding microglia and impair retrieval of conditioned fear memory. Using transcriptomic and fluorescence in situ hybridization analyses, we demonstrated Toll-like receptor (TLR) pathway were upregulated in DG microglia by overexpressing ATG7 in DG engrams. TLR2/4 in the microglia mediates the excessive synapse elimination and impaired retrieval of fear memory induced by ATG7-depedent autophagy in DG engrams. The expression of Rac1, a Rho-GTPases which mediates active forgetting, was upregulated in aged engrams. Optogenetic activation of Rac1 in DG engrams promoted the expression of ATG7 and autophagy in the engrams, the activation of microglia, and thus impaired the retrieval of fear memory. Interference of the Atg7 expression in the engram and microglia activation prevented the impairment of fear memory retrieval induced by activation of Rac1 in DG engrams. Together, our findings revealed autophagy-dependent remodeling of DG engrams by microglia as a novel interference mechanism of memory retrieval. Introduction Memory, the ability to store learned information, is an essential property of the brain. A given memory is encoded by sparsely distributed engrams which are a population of neurons that become highly activated during learning (Holtmaat and Caroni, 2016). The regulation of engrams in the brain has been viewed as an exclusive prerogative of the neuronal system. In addition to neurons, however, the brain comprises other types of cells, including glia and vasculature. Recent findings support these microglia, the major immune cells in the brain, are closely associated with the neural remodeling from rodents to human in an experience-dependent manner (Cangalaya et al., 2020; Cheadle et al., 2020a; Cong et al., 2020; Wu et al., 2015). Microglia were involved in both neurogenesis-related and aging-dependent memory destabilization (Barrientos et al., 2010; Spittau, 2017; Wang et al., 2020a). C1q-dependent complement pathway is actively involved in synapse elimination by microglia in adult hippocampus (Wang et al., 2020a). High resolution imaging shows that microglia engulf retinal ganglion cell inputs during the period of activity-dependent synaptic pruning during the early stage of postnatal development (Schafer et al., 2012). The cellular mechanisms that underlie synaptic remodeling mediated by microglia is crucial for figuring out the possible pathogenesis of psychiatric disorders due to the impaired synaptic refinement, such as Alzheimer disease (Hammond et al., 2019; Rajendran and Paolicelli, 2018), schizophrenia (Osimo et al., 2019; Sellgren et al., 2019) and autism (Heavner and Smith, 2020). Previous researches have shown that microglia-neuron interaction is necessary for neuronal remodeling and memory precision. Microglia become morphologically and functionally activated by some trigger molecules that believed to be derived from the neurons. Growing evidence demonstrate that the binding between CX3C chemokine receptor 1 (CX3CR1) expressed on microglia and CX3CL1 secreted from neurons attributes to the neuron-microglia communication. Impaired CX3CL1/ CX3CR1 signaling is involved in the pathogenesis of Alzheimer's disease (Fuhrmann et al., 2010). Recently, it is demonstrated that Toll-like receptors (TLRs) are activated by pathogen-associated molecular patterns (PAMPS), and play an important role in host defense response in the CNS (Fiebich et al., 2018; Kawasaki and Kawai, 2014). The activation of TLRs in microglia have been observed in both Alzheimer's and Parkinson’s patients (Fiebich et al., 2018). TLRs-mediated signaling pathways in microglia lead to the translocation of transcription factors, such as NF-B, thereby activating the transcription of several genes including interleukins (ILs) and tumor necrosis factors (TNFs) which mediated synaptic pruning and functional plasticity in neurons (Lewitus et al., 2016; Liao et al., 2016; Werneburg et al., 2017). Growing evidence shows that cell debris or signaling molecules are transferred from neurons to glia to survey neuronal state (Kulkarni et al., 2018; Stowell et al., 2019). Autophagy is a complex self-degradative system that removes damaged cellular debris and aggregated proteins to maintain brain homeostasis and promote neuronal survival (Glick et al., 2010). Autophagy maintains neuronal homeostasis through degrading and recycling the damaged cellular components (Kulkarni et al., 2018). Numerous researches have shown that autophagy is involved in microglia activation in an unconventional secretion pathway (Kulkarni et al., 2018). Recent reports found that autophagy mediated unconventional secretion of cytosolic organelles and a series of cytosolic proteins (Gonzalez et al., 2020). Previous researches shown autophagy is upregulated in the hippocampus when mice were exposed to contextual fear conditioning (CFC), Morris water maze tasks and novel object recognition (Glatigny et al., 2019). However, whether autophagy in engrams affects memory retrieval through activating microglia in an unconventional secretion way remains unclear. In this study, we found that autophagy protein 7 (Atg7) mRNA in the dorsal dentate gyrus (DG) engrams which encodes fear memory, was increased in aged mice, leading to the impairment of fear memory retrieval through TLR2/4-depedent activation of microglia. In addition, Rac1-mediated forgetting signal promoted autophagy in DG engrams. These signaling crosstalk between engrams and microglia lead to synapse elimination of engrams and memory destabilization, which might be a potential interference mechanism of memory retrieval in aging mice. Results The expression of ATG7 is increased in dorsal DG engrams in aged-mice, impairs the retrieval of contextual fear memory, and activates DG microglia. Impaired memory maintenance and increased autophagy of the neuron has been found in Alzheimer’s disease (AD) and aging animal model (Bordi et al., 2016). To assess the autophagy in dorsal DG engram, we used a doxycycline (Dox)-dependent robust activity marking (RAM) system to label the engrams(Sorensen et al., 2016). Aged (24-month-old) mice and young (3-month-old) mice were infected with AAV-RAM−EGFP in dorsal dentate gyrus (DG) to label the engrams during fear conditioning (Fig 1A). 3 days later after engrams labeling, mice were exposure to context A (the fear conditioned context), and c-Fos immunostaining was performed to assess the reactivation of EGFP engrams. The total number of EGFP engrams in DG and the proportion of c-Fos engrams (c-Fos EGFP) were decreased in aged mice, compared with that in the young mice (Fig 1B), indicating the reactivation of the engrams during the memory retrieval were decreased in aged mice. We then assess the level of autophagy in DG engrams by performing single molecule fluorescence in situ hybridization (sm FISH) against Atg7 (Autophagy Related Protein 7) and Egfp mRNA. The Atg7 mRNA abundance in Egfp engrams were significantly increased in aged mice, compared with the young mice (Fig 1C-D). Autophagy could be specifically manipulated by overexpression or downregulation of ATG7 (Hui et al., 2019; Kim et al., 2017). To assess the role of autophagy in DG engram in memory retrieval, mice were infected with AAV-RAM-Cre, AAV-DIO-Atg7-shRNA-EGFP (or DIO-Scramble-shRNA) in dorsal DG to specifically knock down Atg7 in the engrams (Fig 1E). The expression of Atg7 mRNA was dramatic decreased in Atg7-shRNA expressing (EGFP) engrams (Fig 1F). Knocking down ATG7 in DG engrams increased the freezing level in the conditioning context, indicating the enhanced retrieval of contextual fear memory (Fig 1G-H). Instead, mice were infected with AAV-RAM-Cre, AAV-DIO-Atg7-Flag (or DIO-EYFP) in dorsal DG to specifically overexpress ATG7 in the engrams. Immunostaining of DG showed a dramatic increase of phagocytic marker CD68 puncta within Iba-1 microglia in mice overexpressing ATG7, compared with the EYFP group, indicating that ATG7 in DG engrams might increase the activation of the microglia (Fig 1K). Colony-stimulating factor 1 receptor (CSF1R) signaling is crucial for microglial survival (Spangenberg et al., 2016). Mice were treated with diet containing Pexidartinib (PLX3397) (CSF1R inhibitor) (Wang et al., 2020a) to deplete microglia. Significantly elimination of the microglia in the DG were observed 3 days after PLX3397 administration (Fig 1L-M). Overexpression of ATG7 in DG engrams impaired the retrieval of contexture fear memory in mice, while PLX3397 pretreatment prevented ATG7-induced impairment of memory retrieval (Fig 1N-O). These results demonstrate that ATG7 in DG engrams activates microglia and impairs the retrieval of contexture fear memory. TLR2/4 in microglia mediates ATG7-induced spine elimination in DG engrams To determine mechanism of microglia activation by autophagy in DG engrams, the mice were infected with AAV-RAM-Cre, AAV-DIO-Atg7-Flag or AAV-DIO-EYFP in dorsal DG to overexpress ATG7 in the DG engrams. The dorsal DG was dissected 3 days after engrams labeling, and microglia were enriched using magnetic-activated cell sorting (MACS) (Fig 2A). Fluorescence-activated cell sorting analysis were performed to assess the purity of the enriched microglia (Fig 2B). RNA-Seq and Gene Ontology (GO) of biological process categories with enrichment showed that when overexpression of ATG7 within DG engrams, the upregulated genes in DG microglia were enriched in immune response pathways, including T cell activation and cytokine-mediated signaling pathway (Fig 2C-D). Gene Set Enrichment Analysis (GSEA) revealed that TLR signaling pathway was upregulated in microglia when engrams overexpressed ATG7, indicating the positive correlation between ATG7 overexpression within engrams and TLR signaling upregulation within microglia (Fig 2E). When analyzed the genes involved in immune response, the expression of Toll-like receptor 2 (Tlr2), the most abundantly TLR family member expressed in microglia is increased (Fig 2F). It has been well-documented that TLR2 contributed to the activation of microglia (Liao et al., 2016; Nie et al., 2018; Stirling et al., 2014). The activation of microglia through Toll-like receptor 2/4 (TLR2/4) initiates inflammatory signaling, contributes to the release of Interleukin-1 alpha (IL-1α) and tumor necrosis factor alpha (TNFα), which mediates synaptic pruning (Nie et al., 2018). Consistently, GSEA showed the enrichment of genes involved TNF signal pathway and cytokine-cytokine receptor interaction in microglia when engrams overexpressed ATG7 (Fig 2G), suggesting the ATG7-dependent autophagy in the engrams promote immune response in microglia. The expression Tlr2 and Tlr4 in microglia were determined by smFISH (Fig 2H). CX3C Motif Chemokine Receptor 1 (CX3CR1) were used to label the microglia. The expression of Tlr2/4 mRNA in Cx3cr1 microglia of DG were upregulated by overexpression of ATG7 in the DG engrams, compared with the EYFP group (Fig 2I-K). To explore whether TLR2/4 within microglia were involved in the impaired memory retrieval induced by autophagy within the engrams., we downregulated TLR2/4 in DG microglia with the lentivirus expressing Tlr2/4 microRNA or control microRNA in a Cre-dependent manner (Fenno et al., 2014b). Cx3cr1-CreER mice were infected with the LVDIO-Tlr4miR-Tlr2miR -mCherry or LVDIO-NegmiR-mCherry in the DG, in which Cre recombinase are selectively expressed in microglia(Nie et al., 2018) when treated with tamoxifen, and thus specifically expressed Tlr2/4miR or NegmiR in microglia. Tlr2/4miR-mCherry significantly downregulate the expression of Tlr2 and Tlr4 mRNA in DG microglia isolated by MACS (Fig 3A). To determine whether upregulation of ATG7 in DG engrams interferes with the retrieval of fear memory through Tlr2/4 within microglia, Cx3cr1-CreER mice were infected with AAV-RAM-Flp, AAV-RAM-Frt-Atg7-Flag, and LV-DIO-Tlr2/4miR-mCherry (or NegmiR-mCherry) in dorsal DG (Fig 3B), in which Cre-loxp and Flp-FRT (Fenno et al., 2014a) system were combined to genetic control engrams and microglia separately. After labeling and overexpressing ATG7 in the engrams, TAM was administered to induce the expression of Tlr2/4miR in the microglia of DG (Fig 3C). Overexpression of ATG7 in DG engram impaired the retrieval of contexture fear memory, whereas downregulation of TLR2/4 by lentivirus in the DG microglia impeded the impairment of fear memory retrieval (Fig 3D-E). Neuronal activity is communicated to microglia and may alter microglia–neuron interactions. To assess how TLR2/4 is involved in the interaction between engrams and microglia, Cx3cr1-CreER mice were injected with AAV-RAM-Flp, AAV-RAM-Frt-Atg7-Flag, AAV-RAM-mScarlet-I-post-eGRASP, and either LV-DIO-NegmiR-EGFP or LV-DIO-Tlr2/4miR-EGFP in dorsal DG (Fig 3F). The dendrite of the engrams was labeled with mScarlet-I-post-eGRASP (Choi et al., 2018). Compared with the engrams overexpressing EYFP, the contact area between microglia and the dendrites of the engrams overexpressing ATG7 were increased, and the spine density of these engrams was decreased (Fig 3G-J). Downregulation of TLR2/4 in microglia decreased the microglia-dendrite contact area and resumed the spine density (Fig 3G-J). These data indicate that microglia contact with the engrams and promote the elimination of the spine in a TLR2/4-dependent manner. Forgetting signal Rac1 in DG engrams impairs the retrieval of fear memory and activates autophagy within the engrams The activity of Ras-related C3 botulinum toxin substrate 1 (Rac1), which regulates active forgetting in both fly and mice, is aberrantly elevated in the hippocampal tissues of AD patients and AD animal models (Wu et al., 2019). An age-dependent Rac1 activity-based memory loss was observed in an AD fly model (Wu et al., 2019). Sm FISH showed a significant increased mRNA level of Rac1 within the DG engram of the aged (24-month-old) mice, indicating the age-dependent upregulation of Rac1 (Fig 4A-B). To investigate whether Rac1 activity in DG engrams impairs the retrieval of contexture fear memory, WT mice were infected with AAV-RAM -Cre mixed with either AAV-DIO -C450A–mVenus or AAV-DIO-Pa Rac1-mVenus in bilateral dorsal DG to express photoactivatable Rac1 (Pa Rac1) or light-insensitive mutant (C450A) in the engrams. 3 days after engrams labeling, the light pulse (150 ms; 1 Hz; 1 hr) was delivered into dorsal DG to opto-activate Rac1 before the retrieval of contexture fear memory (Fig 4C-D). c-Fos immunostaining was performed 1.5 hrs after the memory retrieval. The percentage of the c-Fos component was decreased in the engrams overexpressing Pa-Rac1, compared with the C450A group, suggesting optical activation of Rac1 in DG engrams decreased the reactivation of engrams during memory retrieval (Fig 4D-F). Additionally, the retrieval of contexture fear memory was significantly impaired 1 hr after optical activation of Pa-Rac1 in DG engram (Fig 4H), and this effect was lasted up to 24 hrs after optical activation of Pa-Rac1 (Fig 4I), indicating the Rac1 activation induced a long-term effect on memory retrieval. Instead, optical inhibition of Rac1 T17N within the engrams increased the freezing level in context A, indicating the enhanced retrieval of contexture fear memory (Fig 4K). Nevertheless, the enhancement was not observed 24 hrs after Rac1 inhibition (Fig 4L), indicates that the activation of Rac1 in DG engrams might be an autonomously process that contributes to the forgetting of fear memory. Previous researches showed that Rac1 induces autophagy through activating JNK signal pathway, which enhance the expression of multiple ATG genes including Atg5 and Atg7 via a Foxo-dependent transcription pathway (Zhou et al., 2015). To assess whether Rac1 interferes with fear memory retrieval via promoting autophagy in DG engram, a Cre-dependent autophagy flux reporter system AAV-DIO-RFP-GFP-LC3 were conducted (Castillo et al., 2013) based on different pH stability of EGFP and RFP fluorescent proteins (Fig 4M-N). 1 hr after optical activation of Rac1 within the DG engrams, the autophagy signal was evaluated by analyzing the ratio of RFP and GFP fluorescent intensity. There was a substantial increase ratio of the RFP/Yellow puncta in the DG engrams (Fig 4O). These data suggest that age-dependent upregulation of Rac1 signaling in DG engram facilitates autophagy flux within the engrams and impairs fear memory retrieval. Activation of Rac1 within DG engram upregulates the expression of ATG7 and activates microglia via TLR2/4 Accumulating evidence suggest Rac1 mediates reversible forgetting. Activation of Rac1 within CA1 engrams affects memory maintenance (Lv et al., 2019). To assess the correlation of Rac1 dependent memory loss with autophagy within engrams, mice were infected with AAV-RAM–Cre, AAV-DIO -C450A –mVenus or AAV-DIO-Pa Rac1-mVenus in bilateral dorsal DG (Fig 5A). Optical activation of Rac1 in the DG engrams significantly increased the mRNA level of Atg7 in the DG engrams, compared with C450A control group (Fig 5B-C). CD68 and Iba1 immunostaining after optical stimulation were performed, and a significant increased intensity of CD68 puncta in the Iba1 microglia was observed in Pa Rac1 group (Fig 5D-E). Furthermore, smFISH following optical stimulation of Rac1 within DG engrams showed a significant increased Tlr2/4 mRNA level within microglia around the fiber tip in Pa Rac1, compare with the C450A group (Fig 5F-H). To assess the causal link of Tlr2/4 in microglia in Rac1-induced microglia activation, Cx3cr1-CreER mice were infected with AAV-RAM-Flp, AAV-Frt-Pa Rac1 (or Frt-C450A), and LV-DIO-Tlr2/4miR-mCherry (or -DIO-NegmiR-mCherry) in bilateral dorsal DG. After engrams labeling, the mice were injected with TAM to downregulate TLR2/4 in microglia in a Cre-dependent manner. The light pulse was delivered 3 days after labeling, and the immunohistochemistry was performed 1 hr after optical stimulation (Fig 5I). The colocalization of Iba1 with mCherry indicated the specific expression of Tlr2/4miR-mCherry in DG microglia (Fig 5J). Optical activation of Rac1 in the DG engrams significantly increased the density of CD68 puncta in the microglia surrounding the tips of the optical fibers, while knocking down TLR2/4 within microglia abolished the upregulation of CD68 puncta (Fig 5K-L). These data indicate TLR2/4 in microglia is required for microglia activation induced by Rac1-activation within the engrams. ATG7 within the engrams and TLR2/4 within the microglia mediates the impaired retrieval of fear memory induced by Rac1 activation in DG engrams. To investigate the involvement of ATG7 within the DG engrams and DG microglia in the impaired retrieval of contexture fear memory induced by Rac1 activation, WT mice were infected with AAV-RAM -Cre, AAV-DIO-Atg7-shRNA-EGFP (or scramble shRNA), and AAV-DIO-Pa Rac1-mVenus (or C450A) bilaterally in dorsal DG. Optical activation of Pa Rac1 within DG engrams led to decreased freezing level in the fear conditioned context, indicating the impaired retrieval of contexture fear memory (Fig 6A-C), whereas downregulation of ATG7 in the DG engrams impeded the decreased freezing level in the fear conditioned context (Fig 6A-C), suggesting ATG7 within the engrams mediates the impaired the retrieval of fear memory induced by Rac1 activation. Administration of microglia inhibitor PLX3397 attenuated the decreased freezing level induced by optical activation of Rac1 within DG engrams (Fig 6D-F). No differences were detected in the open field or elevated plus maze tests, suggesting locomotor activity or anxiety level in mice were not affected by Rac1 activation or microglia depletion (Fig 6G-H). Furthermore, in Cx3cr1-CreER mice, knocking down TLR2/4 within DG microglia prevented the impaired retrieval of contexture fear memory induced by the optical activation of Rac1 within the DG engrams (Fig 6I-K), without affecting locomotor activity or anxiety level in mice (Fig 6L-M). Collectively, our observations suggest that the activation of Rac1within DG engram impaired the retrieval of contexture fear memory via ATG7 within the DG engrams, as well as TLR2/4-dependent microglia activation in DG.


Introduction
Memory, the ability to store learned information, is an essential property of the brain. A given memory is encoded by sparsely distributed engrams which are a population of neurons that become highly activated during learning (Holtmaat and Caroni, 2016). The regulation of engrams in the brain has been viewed as an exclusive prerogative of the neuronal system. In addition to neurons, however, the brain comprises other types of cells, including glia and vasculature. Recent findings support these microglia, the major immune cells in the brain, are closely associated with the neural remodeling from rodents to human in an experience-dependent manner (Cangalaya et al., 2020;Cheadle et al., 2020a;Cong et al., 2020;Wu et al., 2015). Microglia were involved in both neurogenesis-related and aging-dependent memory destabilization (Barrientos et al., 2010;Spittau, 2017;Wang et al., 2020a).
C1q-dependent complement pathway is actively involved in synapse elimination by microglia in adult hippocampus (Wang et al., 2020a). High resolution imaging shows that microglia engulf retinal ganglion cell inputs during the period of activity-dependent synaptic pruning during the early stage of postnatal development (Schafer et al., 2012). The cellular mechanisms that underlie synaptic remodeling mediated by microglia is crucial for figuring out the possible pathogenesis of psychiatric disorders due to the impaired synaptic refinement, such as Alzheimer disease (Hammond et al., 2019;Rajendran and Paolicelli, 2018), schizophrenia (Osimo et al., 2019;Sellgren et al., 2019) and autism (Heavner and Smith, 2020).
Previous researches have shown that microglia-neuron interaction is necessary for neuronal remodeling and memory precision. Microglia become morphologically and functionally activated by some trigger molecules that believed to be derived from the neurons. Growing evidence demonstrate that the binding between CX3C chemokine receptor 1 (CX3CR1) expressed on microglia and CX3CL1 secreted from neurons attributes to the neuron-microglia communication. Impaired CX3CL1/ CX3CR1 signaling is involved in the pathogenesis of Alzheimer's disease (Fuhrmann et al., 2010). Recently, it is demonstrated that Toll-like receptors (TLRs) are activated by pathogen-associated molecular patterns (PAMPS), and play an important role in host defense response in the CNS (Fiebich et al., 2018;Kawasaki and Kawai, 2014). The activation of TLRs in microglia have been observed in both Alzheimer's and Parkinson's patients (Fiebich et al., 2018).
TLRs-mediated signaling pathways in microglia lead to the translocation of transcription factors, such as NF-B, thereby activating the transcription of several genes including interleukins (ILs) and tumor necrosis factors (TNFs) which mediated synaptic pruning and functional plasticity in neurons (Lewitus et al., 2016;Liao et al., 2016;Werneburg et al., 2017).
Growing evidence shows that cell debris or signaling molecules are transferred from neurons to glia to survey neuronal state (Kulkarni et al., 2018;Stowell et al., 2019). Autophagy is a complex self-degradative system that removes damaged cellular debris and aggregated proteins to maintain brain homeostasis and promote neuronal survival (Glick et al., 2010). Autophagy maintains neuronal homeostasis through degrading and recycling the damaged cellular components (Kulkarni et al., 2018). Numerous researches have shown that autophagy is involved in microglia activation in an unconventional secretion pathway (Kulkarni et al., 2018). Recent reports found that autophagy mediated unconventional secretion of cytosolic organelles and a series of cytosolic proteins (Gonzalez et al., 2020). Previous researches shown autophagy is upregulated in the hippocampus when mice were exposed to contextual fear conditioning (CFC), Morris water maze tasks and novel object recognition (Glatigny et al., 2019). However, whether autophagy in engrams affects memory retrieval through activating microglia in an unconventional secretion way remains unclear.
In this study, we found that autophagy protein 7 (Atg7) mRNA in the dorsal dentate gyrus (DG) engrams which encodes fear memory, was increased in aged mice, leading to the impairment of fear memory retrieval through TLR2/4-depedent activation of microglia. In addition, Rac1-mediated forgetting signal promoted autophagy in DG engrams. These signaling crosstalk between engrams and microglia lead to synapse elimination of engrams and memory destabilization, which might be a potential interference mechanism of memory retrieval in aging mice.

Results
The expression of ATG7 is increased in dorsal DG engrams in aged-mice, impairs the retrieval of contextual fear memory, and activates DG microglia.

Impaired memory maintenance and increased autophagy of the neuron has been found in
Alzheimer's disease (AD) and aging animal model (Bordi et al., 2016). To assess the autophagy in dorsal DG engram, we used a doxycycline (Dox)-dependent robust activity marking (RAM) system to label the engrams (Sorensen et al., 2016). Aged (24-month-old) mice and young (3-month-old) mice were infected with AAV-RAM−EGFP in dorsal dentate gyrus (DG) to label the engrams during fear conditioning ( Fig 1A). 3 days later after engrams labeling, mice were exposure to context A (the fear conditioned context), and c-Fos immunostaining was performed to assess the reactivation of EGFP + engrams. The total number of EGFP + engrams in DG and the proportion of c-Fos + engrams (c-Fos + EGFP + ) were decreased in aged mice, compared with that in the young mice (Fig 1B), indicating the reactivation of the engrams during the memory retrieval were decreased in aged mice.
We then assess the level of autophagy in DG engrams by performing single molecule fluorescence in situ hybridization (sm FISH) against Atg7 (Autophagy Related Protein 7) and Egfp mRNA. The Atg7 mRNA abundance in Egfp + engrams were significantly increased in aged mice, compared with the young mice (Fig 1C-D). Autophagy could be specifically manipulated by overexpression or downregulation of ATG7 (Hui et al., 2019;Kim et al., 2017). To assess the role of autophagy in DG engram in memory retrieval, mice were infected with AAV-RAM-Cre, AAV-  in dorsal DG to specifically knock down Atg7 in the engrams (Fig 1E).
The expression of Atg7 mRNA was dramatic decreased in Atg7-shRNA expressing (EGFP + ) engrams ( Fig 1F). Knocking down ATG7 in DG engrams increased the freezing level in the conditioning context, indicating the enhanced retrieval of contextual fear memory (Fig 1G-H). Instead, mice were infected with AAV-RAM-Cre, AAV-DIO-  in dorsal DG to specifically overexpress ATG7 in the engrams. Immunostaining of DG showed a dramatic increase of phagocytic marker CD68 puncta within Iba-1 + microglia in mice overexpressing ATG7, compared with the EYFP group, indicating that ATG7 in DG engrams might increase the activation of the microglia ( Fig 1K). Colony-stimulating factor 1 receptor (CSF1R) signaling is crucial for microglial survival (Spangenberg et al., 2016). Mice were treated with diet containing Pexidartinib (PLX3397) (CSF1R inhibitor) (Wang et al., 2020a) to deplete microglia. Significantly elimination of the microglia in the DG were observed 3 days after PLX3397 administration (Fig 1L-M).
Overexpression of ATG7 in DG engrams impaired the retrieval of contexture fear memory in mice, while PLX3397 pretreatment prevented ATG7-induced impairment of memory retrieval (Fig 1N-O).
These results demonstrate that ATG7 in DG engrams activates microglia and impairs the retrieval of contexture fear memory.

TLR2/4 in microglia mediates ATG7-induced spine elimination in DG engrams
To determine mechanism of microglia activation by autophagy in DG engrams, the mice were infected with AAV-RAM-Cre, AAV-DIO-Atg7-Flag or AAV-DIO-EYFP in dorsal DG to overexpress ATG7 in the DG engrams. The dorsal DG was dissected 3 days after engrams labeling, and microglia were enriched using magnetic-activated cell sorting (MACS) (Fig 2A). Fluorescence-activated cell sorting analysis were performed to assess the purity of the enriched microglia ( Fig 2B). RNA-Seq and Gene Ontology (GO) of biological process categories with enrichment showed that when overexpression of ATG7 within DG engrams, the upregulated genes in DG microglia were enriched in immune response pathways, including T cell activation and cytokine-mediated signaling pathway ( Fig 2C-D). Gene Set Enrichment Analysis (GSEA) revealed that TLR signaling pathway was upregulated in microglia when engrams overexpressed ATG7, indicating the positive correlation between ATG7 overexpression within engrams and TLR signaling upregulation within microglia ( Fig   2E). When analyzed the genes involved in immune response, the expression of Toll-like receptor 2 (Tlr2), the most abundantly TLR family member expressed in microglia is increased (Fig 2F). It has been well-documented that TLR2 contributed to the activation of microglia (Liao et al., 2016;Nie et al., 2018;Stirling et al., 2014). The activation of microglia through Toll-like receptor 2/4 (TLR2/4) initiates inflammatory signaling, contributes to the release of Interleukin-1 alpha (IL-1α) and tumor necrosis factor alpha (TNFα), which mediates synaptic pruning (Nie et al., 2018). Consistently, GSEA showed the enrichment of genes involved TNF signal pathway and cytokine-cytokine receptor interaction in microglia when engrams overexpressed ATG7 (Fig 2G), suggesting the ATG7-dependent autophagy in the engrams promote immune response in microglia. The expression Tlr2 and Tlr4 in microglia were determined by smFISH ( Fig 2H). CX3C Motif Chemokine Receptor 1 (CX3CR1) were used to label the microglia. The expression of Tlr2/4 mRNA in Cx3cr1 + microglia of DG were upregulated by overexpression of ATG7 in the DG engrams, compared with the EYFP group ( Fig 2I-K).
To explore whether TLR2/4 within microglia were involved in the impaired memory retrieval induced by autophagy within the engrams., we downregulated TLR2/4 in DG microglia with the lentivirus expressing Tlr2/4 microRNA or control microRNA in a Cre-dependent manner (Fenno et al., 2014b). Cx3cr1-CreER mice were infected with the LV-DIO-Tlr4miR-Tlr2miR -mCherry or LV-DIO-NegmiR-mCherry in the DG, in which Cre recombinase are selectively expressed in microglia (Nie et al., 2018) when treated with tamoxifen, and thus specifically expressed Tlr2/4miR or NegmiR in microglia. Tlr2/4miR-mCherry significantly downregulate the expression of Tlr2 and Tlr4 mRNA in DG microglia isolated by MACS ( Fig 3A). To determine whether upregulation of ATG7 in DG engrams interferes with the retrieval of fear memory through Tlr2/4 within microglia,

Cx3cr1-CreER mice were infected with AAV-RAM-Flp, AAV-RAM-Frt-Atg7-Flag, and
LV-  in dorsal DG (Fig 3B), in which Cre-loxp and Flp-FRT (Fenno et al., 2014a) system were combined to genetic control engrams and microglia separately. After labeling and overexpressing ATG7 in the engrams, TAM was administered to induce the expression of Tlr2/4miR in the microglia of DG ( Fig 3C). Overexpression of ATG7 in DG engram impaired the retrieval of contexture fear memory, whereas downregulation of TLR2/4 by lentivirus in the DG microglia impeded the impairment of fear memory retrieval (Fig 3D-E).
Neuronal activity is communicated to microglia and may alter microglia-neuron interactions. To assess how TLR2/4 is involved in the interaction between engrams and microglia, Cx3cr1-CreER AAV-RAM-mScarlet-I-post-eGRASP, and either LV-DIO-NegmiR-EGFP or LV-DIO-Tlr2/4miR-EGFP in dorsal DG ( Fig 3F). The dendrite of the engrams was labeled with mScarlet-I-post-eGRASP (Choi et al., 2018). Compared with the engrams overexpressing EYFP, the contact area between microglia and the dendrites of the engrams overexpressing ATG7 were increased, and the spine density of these engrams was decreased (Fig 3G-J). Downregulation of TLR2/4 in microglia decreased the microglia-dendrite contact area and resumed the spine density ( Fig 3G-J). These data indicate that microglia contact with the engrams and promote the elimination of the spine in a TLR2/4-dependent manner.

Forgetting signal Rac1 in DG engrams impairs the retrieval of fear memory and activates autophagy within the engrams
The activity of Ras-related C3 botulinum toxin substrate 1 (Rac1), which regulates active forgetting in both fly and mice, is aberrantly elevated in the hippocampal tissues of AD patients and AD animal models ). An age-dependent Rac1 activity-based memory loss was observed in an AD fly model . Sm FISH showed a significant increased mRNA level of Rac1 within the DG engram of the aged (24-month-old) mice, indicating the age-dependent upregulation of Rac1 (Fig 4A-B). To investigate whether Rac1 activity in DG engrams impairs the retrieval of contexture fear memory, WT mice were infected with AAV-RAM -Cre mixed with either AAV-DIO -C450A-mVenus or AAV-DIO-Pa Rac1-mVenus in bilateral dorsal DG to express photoactivatable Rac1 (Pa Rac1) or light-insensitive mutant (C450A) in the engrams. 3 days after engrams labeling, the light pulse (150 ms; 1 Hz; 1 hr) was delivered into dorsal DG to opto-activate Rac1 before the retrieval of contexture fear memory (Fig 4C-D). c-Fos immunostaining was performed 1.5 hrs after the memory retrieval. The percentage of the c-Fos + component was decreased in the engrams overexpressing Pa-Rac1, compared with the C450A group, suggesting optical activation of Rac1 in DG engrams decreased the reactivation of engrams during memory retrieval (Fig 4D-F). Additionally, the retrieval of contexture fear memory was significantly impaired 1 hr after optical activation of Pa-Rac1 in DG engram (Fig 4H), and this effect was lasted up to 24 hrs after optical activation of Pa-Rac1 (Fig 4I), indicating the Rac1 activation induced a long-term effect on memory retrieval.
Instead, optical inhibition of Rac1 T17N within the engrams increased the freezing level in context A, indicating the enhanced retrieval of contexture fear memory ( Fig 4K). Nevertheless, the enhancement was not observed 24 hrs after Rac1 inhibition (Fig 4L), indicates that the activation of Rac1 in DG engrams might be an autonomously process that contributes to the forgetting of fear memory.
Previous researches showed that Rac1 induces autophagy through activating JNK signal pathway, which enhance the expression of multiple ATG genes including Atg5 and Atg7 via a Foxo-dependent transcription pathway (Zhou et al., 2015). To assess whether Rac1 interferes with fear memory retrieval via promoting autophagy in DG engram, a Cre-dependent autophagy flux reporter system AAV-DIO-RFP-GFP-LC3 were conducted (Castillo et al., 2013) based on different pH stability of EGFP and RFP fluorescent proteins (Fig 4M-N). 1 hr after optical activation of Rac1 within the DG engrams, the autophagy signal was evaluated by analyzing the ratio of RFP and GFP fluorescent intensity. There was a substantial increase ratio of the RFP/Yellow puncta in the DG engrams ( Fig   4O). These data suggest that age-dependent upregulation of Rac1 signaling in DG engram facilitates autophagy flux within the engrams and impairs fear memory retrieval.

Activation of Rac1 within DG engram upregulates the expression of ATG7 and activates microglia via TLR2/4
Accumulating evidence suggest Rac1 mediates reversible forgetting. Activation of Rac1 within CA1 engrams affects memory maintenance (Lv et al., 2019). To assess the correlation of Rac1 dependent memory loss with autophagy within engrams, mice were infected with AAV-RAM-Cre, AAV-DIO -C450A -mVenus or AAV-DIO-Pa Rac1-mVenus in bilateral dorsal DG ( Fig 5A). Optical activation of Rac1 in the DG engrams significantly increased the mRNA level of Atg7 in the DG engrams, compared with C450A control group (Fig 5B-C). CD68 and Iba1 immunostaining after optical stimulation were performed, and a significant increased intensity of CD68 puncta in the Iba1 + microglia was observed in Pa Rac1 group (

ATG7 within the engrams and TLR2/4 within the microglia mediates the impaired retrieval of fear memory induced by Rac1 activation in DG engrams.
To investigate the involvement of ATG7 within the DG engrams and DG microglia in the impaired retrieval of contexture fear memory induced by Rac1 activation, WT mice were infected with

Discussion
In this study, we found that the expression of Rac1-and ATG7 in DG engrams were increased fin aged mice. The upregulation of Rac1 and ATG7 activates microglia, promotes spine remodeling, and thus impairs the retrieval of contexture fear memory. Besides, TLR2/4 within microglia is essential for autophagy-dependent crosstalk between engrams-microglia. Those result reveal a novel molecular mechanism underlying the memory destabilization in aged mice that contributed by the interaction between DG microglia and engrams (Fig 7).

Autophagy In Neurodegenerative Diseases
Autophagy is a natural self-degradative process that clean out damaged organelles and protein aggregates in the body to maintain cell homeostasis. The dysfunction of autophagy is believed to be involved in the pathogenesis of Alzheimer's disease (AD), which is a neurodegenerative disorder with deficiency in memory and cognitive functions. Genome-wide analysis showed the transcriptional upregulation of autophagy pathways in AD patients (Lipinski et al., 2010). Other researches demonstrated that autophagy kinase complexes such as BECN1-PIK3C3 were reduced in hippocampus of AD mouse model (Lachance et al., 2019). APP/PS1 mice lacking NF-E2 related factor 2 (Nrf2) had increased accumulation of multivesicular bodies, endosomes and lysosomes in neurons, leading to defective autophagy and increased Alzheimer's disease-like pathology (Joshi et al., 2015). For Parkinson's disease (PD), previously studies have implicated PINK1 and Parkin mutations induced defective mitophagy, thereby, accumulating damaged mitochondria and triggering Parkinson's Disease (PD) (Mancias and Kimmelman, 2016;Sarraf et al., 2013). Here, we found the autophagy in DG engrams which store the learned information is increased in aged mice ( Fig 1D), and enhanced autophagy in DG engram by overexpressing ATG7 in young mice led to impaired retrieval of fear memory ( Fig 1O).

Autophagy signaling in engrams mediates spine elimination induced by microglia activation
Microglia-neuron signaling transduction is necessary for synaptic remodeling by microglia.
Numerous researches have shown that microglia activation plays an important role in pruning of exuberant synaptic connections and maintaining brain homeostasis during development (Kim et al., 2017;Lui et al., 2016;Weinhard et al., 2018). Fn14 within thalamic neurons interacted with TWEAK signals within microglia to engage the number of spines on thalamic neurons (Cheadle et al., 2020b).
Cytokine interleukin-33 (IL-33) can be released from DG neurons in an expression-dependent manner and bind to the IL-33R in microglia, thus promoting microglial engulfment of the extracellular matrix and leading to the increased spine density (Nguyen et al., 2020). Previous researches have shown the induction of autophagy enhanced secretion of the proinflammatory cytokine IL-1β via an unconventional secretion pathway depending on ATG5 (Dupont et al., 2011).
IL-1β was found to drive microglia activation (Todd et al., 2019) and induce NFκB activation in microglia (Ferreira et al., 2010;Krasnow et al., 2017). Accumulative evidence indicates microglia mediates synapses remodeling and neural circuits reorganization in an experience-and learning-dependent manner (Miyamoto et al., 2016;Parkhurst et al., 2013), which is essential for memory precision throughout life. Aberrant microglial activation correlates with synapse loss and cognitive decline in AD mouse models in a complement-dependent process (Hong et al., 2016;Shi et al., 2015). Microglia deficit in synaptic engulfment causes weak synaptic transmission, impaired brain connectivity, which is tightly associated with autism and other neuropsychiatric disorders (Kim et al., 2017;Zhan et al., 2014). Here, we found that autophagy within the DG engrams activated the microglia ( Fig 1K) to contact and eliminate the spine (Fig 3I-J), providing a new mechanism for the dysregulation between neurons and microglia under pathological condition.

Microglial TLR2/4 mediates the crosstalk between the engrams and microglia
Toll-like receptors (TLRs) in microglia acts as a surveil lancer recognizing the pathogen-associated molecular patterns (PAMPs) and endogenous ligands of host to respond to microenvironment changes. Previous evidence suggest α-synuclein released from neurons when suffering stress, injury or other stimulation interacts with TLR2/4 in microglia, leading to microglia activation and engulfment of α-synuclein in turn (Choi et al., 2020;Kim et al., 2013). TLR4 performs a critical role in Aβ-induced NLRP3 inflammasome and inhibition of TLR4/NLRP4 pathway alleviate extracellular Aβ-induced neuroinflammation (Liu et al., 2020). Ubiquitin-specific protease 8 (USP8) attenuates cognition deficiency and motor dysfunction in lipopolysaccharide(LPS)-induced mice models through inhibiting TLR4/MyD88/NF-kB pathway and increasing the expression of anti-inflammatory factor (Zhao et al., 2020). Recent studies have shown crucial roles of TLR4-dependent activation of microglia in neurodegenerative diseases, such as AD and PD (Fernandez-Arjona et al., 2019;Fiebich et al., 2018;Su et al., 2016;Yao et al., 2013). In our study, we found TLR2/4 within DG microglial is necessary for neuron-microglia cross-talk and autophagy-induced spine elimination by microglia (Fig 3G-J), and thus provides a potential therapeutic target for cognitive decline and new insights into the mechanisms of pathogenesis of neuronal degeneration. GSEA analysis indicated the TNF signal pathway and cytokine-cytokine receptor interaction in microglia, which mediates synaptic pruning (Nie et al., 2018), was upregulated in ATG7-overexpressing mice. The causal link between the autophagy dysregulation within the engrams and the release of IL-1α and TNFα by microglia in pathologic conditions such as aging need to be investigate in future.

DECLARATION OF INTERESTS
The authors declare no competing financial interests.

Resource availability
Further information and requests for resources and reagents should be directed to and will be fulfilled by corresponding author, lanma@fudan.edu.cn and ffwang@fudan.edu.cn.

Animals
Cx3cr1-CreERT2 mice (RRID:IMSR_JAX:021160) (Parkhurst et al., 2013) were purchased from Jackson Laboratory (Sacramento, CA, USA), and were bred to C57BL/6 J for more than 6 generations. C57BL/6J male mice were obtained from the Shanghai Laboratory Animal Center (CAS, Shanghai, China) and housed in reverse light-dark cycle (lights-off at 8:00 a.m., lights-on at 8:00 p.m.) with free access to food and water. All experiment procedures for animals were performed strictly in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals, and were approved by the Animal Care and Use Committee of Fudan University.

Plasmid construction
The

Surgery
Mice were anesthetized with 2% isoflurane during the surgery and were bilaterally injected with 500 nl of purified AAV (10 12  For optogenetic activation of Rac1 in vivo, optic fibers were connected to 473 nm laser diode via a FC/PC adaptor (Newdoon Inc., Hangzhou, China). The laser diode was adjusted to ~20 mW at the end of the fiber. The light pulse was delivered by light-emitting diode (Newdoon Inc.) at 150 ms, 1 Hz for 1 hr, and the freezing behavior was measured 1 hr after stimulation. Cx3cr1-CreER mice were administered with Tamoxifen (T5648, Sigma-Aldrich) dissolved in ethanol and corn oil by intraperitoneal injection (150 mg/kg) for 3 consecutive days to induce Cre-dependent TLR2/4 miR expression in microglia.

Elevated plus maze (EPM) test
The elevated plus maze apparatus composed of two open arms and two closed arm (34.5 cm length × 6.3 cm width × 19.5 cm height) were placed 75 cm above the floor. Mice were allowed to explore EPM freely for 6 min. Mouse behaviors were recorded with an overhead camera. The apparatuses and testing area were wiped with 75% alcohol in each test interval. Videos were analyzed with

Locomotor Activity Test
Mice were placed in an open field chamber (43.2 cm length × 43.2 cm width × 30.5 cm height, Med-Associates, USA) and allowed to freely explore the space for 30 min. Behavior of mice were monitored with an overhead camera. The overall activity was quantified with TopScan automated detection system (CleverSys, Reston, VA, USA) and the total distance traveled and percentage time spent in the center zone were measured.

Immunofluorescence
Mice were transcardially perfused with saline followed by 4% paraformaldehyde (dissolved in 1 × PBS). Brains were isolated and immersed in 4% PFA at 4 °C overnight for post fixation, and then the brains were transferred to 30% sucrose in PBS for 2 days. Brains were frozen and 40-m coronal brain slices were prepared with a cryostat microtome (Leica CM3050S, Nussloch, Germany). For immunostaining, slices were incubated with blocking buffer (PBS containing with 0.3% Triton X-100 and 10% normal donkey serum) for 1 hr at room temperature (RT) and then incubated with appropriate primary antibodies at 4 °C overnight. The primary antibody used are: anti-c-Fos

RNA sequencing analysis
Quality of reads was evaluated using Fastp (Jiang et al., 2021), all samples passed quality control, and reads were aligned to mm10 (GRCm38 from Ensembl) using Hisat2 (Kim et al., 2019). Mapped reads were counted using featureCounts (Wang et al., 2020b) and DESeq2 package (Niu et al., 2018) was used to perform differential gene expression analysis. Gene ontology term enrichment analysis was performed using clusterProfiler (Yu et al., 2012). Gene set enrichment analysis of KEGG were carried out by gseKEGG (part of clusterProfiler) and the results of GSEA were visualized with gseaplot2 (created by Guangchuang Yu). The data have been submitted to NCBI Gene Expression Omnibus (GEO) under accession number GSE169019.

Quantitative RT-PCR
To downregulate Tlr2/4 in microglia, Cx3cr1-CreER mice were bilaterally injected with LV-DIO -NegmiR-mCherry or LV-DIO-Tlr2/4miR -mCherry in dorsal dentate gyrus. Two weeks later, the mice were intraperitoneally injected with tamoxifen (T5648, Sigma-Aldrich) at a dose of 150 mg/kg for 3 consecutive days. Dorsal DG was isolated and the microglia were purified using magnetic-activated cell sorting and mRNA in microglia was extracted with Trizol. cDNA was obtained from total RNA using PrimeScript RT reagent Kit (RR037A, Takara, Shiga, Japan) and then

Statistical analysis
Data were analysed with SPSS 22 software (IBM, Armonk, NY, USA), and plotted by Graphpad Prism. Our sample sizes were based on our previous research (Jiang et al., 2021;Niu et al., 2018;Wang et al., 2020b). The normality test and homogeneity of variance test were performed by the Shapiro-Wilk test and Levene's test. Tukey's multiple comparisons test were performed after student's t test (Unpaired, two tailed), or one-factor analysis of variance (ANOVA) followed.
Bonferroni's post hoc analysis was performed after repeated-measures (RM) ANOVA. Data that does not fit a normal distribution were analyzed with a nonparametric test. Two-sample Kolmogorov Smirnov test was used for cumulative frequency plot analysis. Statistical significance was represented as *P<0.05; **P<0.01, and ***P<0.001. All data are presented as mean ± SEM.    Scale bar: 50 m. Data are presented as mean ± SEM. *P < 0.05, ***P < 0.001. right:10 m. Data are presented as mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001.  The expression of autophagy-related gene 7 (ATG7) in dorsal DG engrams is increased in aged mouse, leading to the activation of the surrounding microglia and the spine elimination of the engrams in a TLR2/4-dependent manner. Rac1, a mediator of forgetting, is upregulated in aged engrams and its activation promotes ATG7 expression and microglia activation to interfere with the retrieval of contexture fear memory.