for more information:ali.ma@wecistanche.com

Discussion

The molecular underpinnings of engram formation have been an area of great interest. Here we take advantage of the TRAP mouse model to elucidate the dynamics of chromatin accessibility, 3-D genomic architecture, and gene expression across the lifespan of an engram cell. First, we directly show that memory encoding has a vast and long-lasting effect on chromatin accessibility. Once this chromatin state is established, subsequent events such as memory consolidation and recall appear to have a minor effect on the chromatin landscape. It is possible that the chosen time point for recall may not fully capture the chromatin events of reactivated cells and subsequent chromatin changes might be associated with memory reconsolidation or extinction8. Importantly, in line with the previous publications12, 13,23, our data suggest that memory formation is largely an enhancer-driven phenomenon.

Click to Cistanche UK for memory

Second, our study provides the first comprehensive landscape of 3D-genome architecture during different phases of memory formation, as we showed re-localization of large chromatin segments and specific promoter–enhancer interactions dynamic within these compartments that enable fine-tuning of different transcriptional programs. Moreover, we show that promoters of down-regulated genes interact more frequently with enhancers that harbor transcriptional repressors and vice versa.

Last, our work provides the first evidence for a functional priming event, that is characterized by an initial increase in enhancer accessibility during encoding, without expected transcriptional changes. Our analysis revealed that with reactivation, engram neurons utilize a subset of denovopromoter-enhancer interactions, where primed enhancers are brought in contact with their respective promoters to up-regulate genes involved in mRNA transport and local protein translation in synaptic compartments. These changes corresponded to expected morphological and functional changes. Collectively, it appears that engram cells are marked at the epigenetic level and the interplay between chromatin accessibility, 3D-genome architecture, and promoter-enhancer interactions describe a well-coordinated system that leads to a delayed transcriptional surge during engram reactivation.

As with any work, there were several limitations to the study. The use of the Tamoxifen-dependent TRAP mice allows for temporal resolution but opens a labeling window of about 12 hours which could result in non-specific tagging of neurons that are not part of the CFC experience. Moreover, as previously suggested37, the Arc-tagged neurons might be heterogeneous and include different ensembles that regulate both memory discrimination and generalization. In addition, any time point before 1.5–2 hr (early), will not be sufficient to capture transcriptional and translational events caused by DNA recombination, i.e. the expression of Cre recombinase and the final expression of eYFP gene5. Hence, we believe that many early-transient clusters of DEGs were missed, as it was shown previously that many IEGs return to baseline levels after 2h (except Arc46). Lastly, although our study focused on the hippocampus, structural and synaptic plasticity have been linked to long-term storage of information in other brain regions, such as the prefrontal cortex, amygdala2. Future studies should determine if these epigenetic alterations may represent a global critical process involved in the long-term formation and retention of memories or if it is unique to the hippocampal circuits.

Methods

Mice and behavior

As previously described6, ArcCreERT2(+) mice were bred with R26R-STOP-floxed- yellow fluorescent protein (eYFP) homozygous line. ~140 subsequent male offspring were genotyped using the Jackson Laboratory genotyping protocol. Mice were housed four to five per cage in a 12 hr (06:00–18:00) light-dark colony room at 22°C with 40–60% humidity. Food and water were provided ad libitum. Behavioral testing was performed during the light phase. 1h before the behavioral test, mice received one dose of 4-Hydroxytamoxifen (4- OHT, H6278, Sigma-Aldrich), which opens a ~12 hr labeling window5. In order to minimize the nonspecific eYFP label, one day prior to the behavioral task (e.g. CFC), mice were single housed in a 24h dark/darkroom. In addition, to reduce anxiety levels, a tube was added to the cage for shelter and nesting. On the next day, mice were injected with TAM 1h before the CFC. Following the behavioral task, mice were either euthanized after 1.5–2 h, and brains were collected. or placed back into the darkroom for the following 48 hr. After mice were taken out of the dark, they were returned to normal 12 hr (06:00–18:00) light-dark cycle, in the same room. 3 days later (total of 5 days after the CFC), half of the mice cohort were euthanized and brain were collected, while the other half returned to the testing room, where they were exposed to the fear-inducing cue (tone and context). All mouse work was approved by the Committee for Animal Care of the Division of Comparative Medicine at the Massachusetts Institute of Technology. Cistanche can improve memory.

Cistanche can improve memory

Contextual fear conditioning

Contextual fear conditioning (CFC) apparatus and the acquisition program were from TSE- Systems. Mice were introduced into the CFC chamber, following 3 minutes of initial habituation 3 kilo Hz tone with 80 DB intensity was presented for 30 seconds and was co- terminated with 0.7 mA foot shock for 1 sec. 70% isopropanol was used between experiments to clean the cages. 5 days later mice were introduced into the original conditioned context paired with the tone and freezing levels were measured for 3 min (TSE software default parameters used to monitor freezing behaviors), to asses contextual memory. Context B was a modified context A. The stainless-steel bars were covered with a white plastic insert, the outer walls of the chambers were covered with shapes of duct tape to change the appearance of the chamber. The chamber was scented with vanilla, the room lighting was much brighter and the mice were transported to the apparatus in a different cart from context A.

Drugs

Recombination in ArcCreERT2 × R26R-STOP-floxed-EYFP transgenic mice was induced using 4-Hydroxytamoxifen (4-OHT, H6278, Sigma-Aldrich). 4-OHT was dissolved in 2.5ml of EtOH /100% and rocked on a shaker for 30 min. Next, 5ml of corn oil was added and rocked on a shaker for 1h. Finally, tubes were placed for 30min at 60°C to allow the full evaporation of EtOH. Single-injection of 50 mg/kg was injected intraperitoneally (i.p.).

Immunohistochemistry (IHC)

Mice were transcardially perfused with ice-cold 25 ml of phosphate-buffered saline (PBS) followed by 40 ml of4% paraformaldehyde in PBS. Brains were removed and post-fixed in 4% PFA overnight at 4°C and transferred to PBS until sectioning. Brains were mounted on a vibratome stage (Leica VT1000S) using superglue and sectioned with a slice thickness of 40 μm. Slices were washed with PBS and blocked using 1 % Bovine Serum Albumin (BSA) prepared in PBS containing 0.1% Triton-X100 (PBST) for 2 h at room temperature. Blocking buffer was aspirated out and the slices were incubated with the appropriate primary antibody (Supplementary Table 11) overnight at 4°C on a shaker. Slices then were washed three times 15 minutes each with the blocking buffer, and then were incubated with the Alexa Fluor 488, 594, or 647 conjugated secondary antibodies for 2 h at room temperature. Following three washes 15 minutes each with blocking buffer and one final wash with PBS for 10minues, slices were mounted with fluromount-G (Electron microscopic Sciences). Cistanche can improve memory.

In Situ Hybridization (RNAscope®)

Samples were prepared based on the manufacturer’s instructions for RNAscope® multiplex fluorescent v2 assay (Advanced cell diagnostics, USA), with few modifications and combined with the Immunohistochemistry (IHC) protocol. First, samples were stained for primary (Arc and GFP) and secondary (Alexa Fluor 488 and 594, respectively) antibodies (as described in the IHC protocol above). Brain slices were embedded on a glass slide and

dehydrated for 30 min in RT. Next, a barrier was drawn around each section with the Immedge™ hydrophobic pen and samples were treated with RNAscope® Hydrogen Peroxide for 10 min at RT (no target retrieval and protease treatment were applied). Based on the manufacturer’s instructions, sections were subjected to probe hybridization (RNAscope® Probe- Mm-Gria1, 26241, Advanced cell diagnostics), amplification, and HRP-C1 signal development using the TSA® Plus Cyanine 5 channel. Cistanche can improve memory.

Imaging analysis

Images were acquired using an LSM 710 or LSM 880 confocal microscopes (Zeiss) with 10×, 20×, 40×, or 63× objectives at identical settings for all conditions. Images were quantified using ImageJ 1.42q or Imarisx64 8.1.2 (Bitplane, Zurich, Switzerland). For each experimental condition, five coronal sections from at least five animals were used for quantification. Cistanche can improve memory.

Quantification of tagged cells—The surfaces module was utilized to detect Arc+ and eYFP+ cells. Double positive cells (co-localization of Arc+/eYFP+) were then counted using the Surface-To-Surface Xtension module. After applying the ‘area’ filter (>6μM) to remove unspecific labeling, quantification of the double-positive cells was carried in the CA1, CA3 and DG. Cistanche can improve memory.

Quantification of Eif and Gria1 proteins/mRNA—The surfaces module was utilized to detect and 3D render neurons axons based on the GFP signal. EiF2a, Eif4E, Eif3E, and Gria1 -positive puncta were then counted using the spots module. Finally, the Spots Close - To -Surface Xtension module was run to find the subset of spots that are closer to the surface objects than the defined 1uM threshold, and exclusion of spots that fall outside this range.

Spines morphology—Spines were classified to three types base on the spine length (L),diameter of the head (Dh) and the diameter of the neck (Dn). Stubby spines – Dn = L, Thin spines – L > Dn and Dh ≥ Dn, Mushroom spines – Dh ≥ 2Dn, Enlarged mushroom spines – Dh ≥ 2Dn.Cistanche can improve memory.

Cistanche can improve memory

Tissue preparation

Hippocampal tissue was extracted and flash frozen. After extraction, samples were immediately homogenized in 0.5 mL ice-cold PBS with protease inhibitors (Pi) (11836170001, Roche). To mitigate the sorting impact as much as possible, in NC-RNAseq and capture-HiC experiments only, the suspension was fixed with 1% (for NC-RNAseq/HiC) or 2% (pc-HiC) paraformaldehyde for 10 min, quenched with Glycine 2.5M for 5 min, and washed twice with 5 mL of PBS. This process preserves the transcriptional and chromatin landscape and minimizes sorting bias. For ATACseq, RNAseq and capture-HiC, the suspension was centrifuged at 1200 g, 4°C, for 5 minutes and the pellet was resuspended in 5 mL NF-1 hypotonic buffer (0.5% Triton X-100/0.1M Sucrose/5mM MgCl2/1mM EDTA/ 10mM Tris-HCl, pH 8.0, Pi). Next, suspension was dounce-homogenized (pestle A) with 30 strokes and the pestle was washed with extra 5mL NF-1 buffer for a combined total of 10 mL. Suspension were collected all in a 15 mL conical tube, filtered with 70 μm mesh filter (08–771-2, Falcon) and incubated on ice for 30 min. Pelleted nuclei (all centrifuge were 4°C, 1,600 × g for 15 min) were resuspended in 10 mL PBS+1%BSA+Pi and incubated on ice for 1 h. Nuclei were spun down and the pellet was resuspended in 1 mL PBST+1%BSA+Pi and incubated with NeuN, Arc and GFP primary antibodies for overnight in 4°C. Unbound antibodies were washed out twice with 5 mL of PBS and finally resuspended in 1 mL PBS+1%BSA+Pi and secondary antibodies for 2–4 h at 4°C. Following two washes, nuclei were resuspended in 0.5 mL (PBS+1%BSA+Pi) and filtered with 40 μm mesh filter for FACS sorting. 4′,6-Diamidine-2′-phenylindole dihydrochloride (DAPI) was added directly to the FACS tubes (1:5000, 10236276001, Sigma-Aldrich). Cistanche can improve memory.

Flow cytometry

For flow cytometry, nuclei were gated first using forward and side scatters pulse area parameters (FSC-A and SSC-A) excluding debris, followed by exclusion of aggregates using pulse width (FSC-W and SSC-W). Then isotype control staining was used to set NEUN+, ARC+, and GFP+ gates. To secure a homogeneous neuronal population, the NeuN+ population was gated before gating populations based on ARC+ and GFP+ fluorescence. We sorted four different populations from three different time points; Approximately 1.5–2 hours after exposure to FS i) NeuN+/GFP+ ii) NeuN+/GFP-. After five days, in the absence of retrieval, we collected iii) NeuN+/GFP+. In a different cohort, mice were re-exposed to the conditioned stimulus. 1.5–2 hours after re-exposure we collected iv) NeuN+/ARC+/GFP+. All nuclei were sorted into 1.5 mL Eppendorf tubes containing; 500 μl PBS+1%BSA+Pi for ATCAseq, 200 μl of digestion buffer (RecoverAll™ Total Nucleic Acid Isolation Kit for FFPE, Cat. AM1975, Invitrogen) for RNAseq, 500 μl fresh ice-cold lysis buffer (10mM Tris pH=8, 10mM NaCl, 0.2% Igepal CA-630, Pi and PCR grade water) for Hi-C/capture-HiC. Data analysis was performed with Flowjo (v10).

ChIP-qPCR

Immediately after the following sorting, 10,000 cells were pelleted at 4 °C, 1,600 × g for 15 min, and resuspended in SDS lysis buffer. ChIP was conducted based on the manufacturer’s instructions for ChIP-IT High Sensitivity® (HS) Kit (Catalog No. 53040, Active Motif, USA) using anti- H3K4me1 (Abcam, ab8895, 1:100) and anti-H3K27ac (Abcam, ab4729, 1:100) antibodies. ChIP assays were performed in a Bio-Rad CFX96 connect real-time PCR unit using SsoFast™ EvaGreen® qPCR reagent mix (Bio-Rad, 172–5202). Samples were assayed in triplicates and results were normalized to input. Sequences of primer pairs for ChIP–qPCR assays are listed in Supplementary Table 12.

Cistanche can improve memory

ATAC-seq

libraries preparation—ATAC seq libraries were prepared as previously described in49 with minor modification. 10,000 cells from each group, were pulled from 8–10 different mice and considered as N=1. 3–4 biological replicates (N=3–4) were used for library preparation. Immediately after the following sorting, cells were spun down (4°C, 1,600 × g for 15 min) and resuspended in fresh 50 ul cold lysis buffer (0.15% NP-40 instead of Igpal) and rocked on ice for 10 min. Next, segmentation reaction (Nextera DNA sample prep kit 24 reactions, FC-121–1030, Illumina) was carried out in 25 ul volume (12.5 μl TD, 1.25 μl TDE1, 11.25 μl nuclease-free water), for 30 min at 37c with gentle rocking. Augmented DNA fragments were amplified and barcoded using PCR reaction (1 cycle; 5min – 72c, 30sec – 98c, 10 cycles; 15sec – 98c, 30sec – 60c, 3min – 72c). Following the manufacturer's protocol, amplified DNA was cleaned and purified using 1.8x AMPure beads (AmpureXP beads, Beckman Coulter, A63880). After bioanalyzer QC for library size and distribution, libraries were sequenced on the Illumina Nextseq 550 platform at the MIT BioMicro Center. Cistanche can improve memory。

ATAC-seq analysis—The raw fastq data of 40-bp paired-end sequencing reads were aligned to the mouse mm9 reference genome using Bowtie 2.0 in paired-end mode50. The Samtools collection51 was used to sort, remove PCR duplicates (red up), index BAM files (index), and calculate library statistics (flagstat, see Supplementary Table 13). BAM files were downsampled from properly aligned, paired, and unique reads to 50M. Open chromatin peaks were analyzed using MACS252 software Differentially accessible regions (DARs) were identified by Diffbind v1.16.353 with default parameters and settings to use DESeq2 (cutoff; FDR < 0.01; fold changes > 1.5). The full matrix of normalized counts (RPKM54) was retrieved by the DiffBind package. For a generation of bigwig tracks, pileup signal files (fragment pileup per million reads) in biography format were constructed by using the MACS2 call peak function with the -B –SPMR flags. Then, normalized fold enrichment (FE) over input lambda tracks were generated using the bdgcmp function ofMACS2 with the setting - m FE. Then bedGraph files were converted to bigwig format.Cistanche can improve memory

IGV55 tools were used for the visualization of bigwig files. ChromHMM56 was used to establish a chromatin state model from two independent studies that utilized bulk hippocampus tissue before and after foot shock21,22 and to determine fold enrichment offers over the state model. Motif enrichment analysis57,58 and peaks annotation were performed by HOMER59 tools. All codes are available in a supplementary software file