4R Tau Modulates Cocaine-Associated Memory Through Adult Hippocampal Neurogenesis

Background: Drug memory that generally develops with drug-paired contextual stimuli and drug administration is critical for the development, persistence and relapse of drug addiction. Previous studies have suggested that adult hippocampal neurogenesis (AHN) plays a role in cocaine memory formation; however, the underlying mechanism is not fully understood. Methods: Conditioned place preference (CPP), self-administration and locomotor activity were used to investigate the role of Tau in cocaine-associated memory formation. Virus-mediated gene transfer, western blot, immunohistochemistry, ow cytometry analysis, Tau-interacting proteomics, co-immunoprecipitation, and mutation of 4R Tau were performed. Results: Hippocampal expression of Tau was signicantly decreased during the cocaine-associated memory formation. Genetic overexpression of four microtubule-binding repeats Tau (4R Tau) in the hippocampus disrupted cocaine memory by suppressing AHN. Furthermore, 4R Tau directly interacted with phosphoinositide 3-kinase (PI3K)-p85 and impaired its nuclear translocation and PI3K-AKT signaling, processes required for hippocampal neuron proliferation. Conclusions: 4R Tau modulates cocaine memory formation by disrupting AHN, suggesting a novel mechanism underlying cocaine memory formation and provide a new strategy for the treatment of cocaine addiction. a Schematic diagram of re-expression 2N4R Tau and EdU injection in Tau-KO mice with cocaine CPP conditioning. b Re-expression 2N4R Tau in the dDG signicantly attenuated cocaine CPP (n=9 per Endogenous PI3K-p85 was immunoprecipitated with antibodies against PI3K-p85 from tissue lysates

cocaine memory becomes stronger and more resistant to extinction [14]. However, the precise mechanism underlying the role of AHN in cocaine memories formation has not been completely elucidated.
Tau is a microtubule-associated scaffolding protein whose major function is the stabilization of microtubules and promotion of microtubule polymerization [15]. Tau isoforms containing either three or four microtubule-binding repeats (3R Tau or 4R Tau) exhibit approximately equal expression in the normal adult human brain; however, 4R Tau is the predominant isoform expressed in the adult murine brain. Tau was recently shown to be associated with the proliferation of newborn hippocampal granule neurons that require a high level of cytoskeletal plasticity to divide, proliferate and differentiate in response to external stimuli [16][17][18]. During early brain development, 3R Tau predominates in the neonatal brain and exhibits a lower a nity for microtubules than 4R Tau. Therefore, 3R Tau participates in the morphological differentiation and migration of immature neurons. In contrast, 4R Tau binds with higher a nity to microtubules, thereby maintaining the cytoskeletal stability and neuronal integrity [19,20]. To date, the role of Tau isoforms in the formation of drug-associated memory is unknown. In the present study, our ndings reveal a novel mechanism by which 4R Tau modulates cocaine memory formation by regulating AHN.

Animals
Male C57BL/6J wild-type (WT) mice (8-12 weeks old) were purchased from Vital River Laboratory Animal Technology Co. Ltd. (Beijing, China). Tau −/− knockout (Tau-KO) mice on the C57BL/6J background were obtained from Jackson Laboratories (#007251, USA). Male and female homozygous mice were bred to generate Tau −/− homozygous mice, and 8-12 week-old male homozygous Tau-KO mice and WT littermates (C57BL/6J) were used in this study. All mice were housed in the animal room on a standard 12-h light/12-h dark cycle with a constant temperature and food and water available ad libitum. All experimental procedures and use of the animals were conducted in accordance with the guidelines established by the Association for Assessment and Accreditation of Laboratory Animal Care and the Institutional Animal Care and Use Committee of Sichuan University. All efforts were made to minimize the suffering of the mice.

Behavioral paradigm
Mice were acclimated for one week before experiments and habituated to handling for 2 d before each behavioral test. Conditioned place preference (CPP) and self-administration were used to assess cocainecue memory formation. The home cage CPP, food CPP and locomotor activity were employed as control behavioral tests. All behavioral experiments were performed in a double-blind manner.
The CPP test was conducted using a standard three-chambered apparatus equipped with two large conditioning compartments (black and white) that differed in their ooring (bar and grid) and a small middle chamber (grey, smooth PVC oor) that connected the two large compartments. Prior to each session, animals were habituated to the chambers for at least 10 min for a continuous 2 d.

Pre-test session
Baseline preference was assessed by placing the mice in the middle chamber and allowing them to habituate the entire chambers freely for 15 min; an initial measurement of baseline preference was de ned as the time spent in the black chamber subtracted from the time spent in the white chamber (Time pre−test ). Animals were excluded from the test if they showed a strong unconditioned preference for either side chamber (chamber bias > 300 s).

CPP training
Animals were randomly assigned to two groups and trained for 6 d with alternating injections of cocaine (20 mg/kg or 2.5 mg/kg, i.p) or saline injections in both compartments. They were con ned to the conditioning or unconditioning chambers for 30 min after the injection and then returned to their home cages. On the test day, the animals were placed in the middle compartment and the time of spent in the two compartments was recorded for 15 min. We also de ned as the time spent in the black chamber minus the time spent in the white chamber as Time black . The CPP scores were calculated as follows: if Time black was positive (the mouse preferred the black chamber), namely, Time black minus Time pre−test , positive scores indicated a reinforced preference, and negative scores indicated a reversed preference. If Time black was negative (the mouse preferred the white chamber), Time black minus Time pre−test , negative scores indicated a reversed preference and positive scores indicated a reinforced preference. Generally, cocaine strongly reverses preference and results in a positive CPP score, and the CPP score was de ned as the extent of the shift in preference after the cocaine injection.
Home cage CPP Mice were exposed to cocaine as described for consolidation training, and then they were not con ned to the conditioning or unconditioning chambers, but were returned to their home cages without any context exposure to the test apparatus. On the test day, mice were placed in the middle connecting chamber, and the time spent in each chamber was calculated as described in previous experiments to evaluate the CPP score.

Food CPP
The food CPP test used a similar apparatus and methodology as described above. Mice were restricted from food for one week before the test, and their weight was maintained at 80% of their original body weight. During the food conditioning sessions, the food-induced CPP group was transferred to the foodpaired chamber (2-3 g of food were placed in the chamber) for 30 min. In the non-food conditioning sessions, mice were assigned to the non-food-paired chamber for 30 min. After conditioning training for 3 h, animals were only fed once daily (1 h). The alternating sessions of conditioning were repeated 3 times (a total of 6 d). On the test day, mice were placed in the central chamber and allowed to freely explore both compartments for 15 min; the time spent in each chamber was recorded to calculate the CPP score.
CPP and extinction Mice were exposed to the CPP protocol, and the test day course was repeated each day until the preference of the cocaine group for the conditioned compartment had returned to the habituation baseline.
Cocaine self-administration Mice were anesthetized with sodium pentobarbital (60 mg/kg) and implanted with a single sterilized silastic catheter (0.51 ID x 0.94 mm OD, BB518-20, Scienti c Commodities) into the right jugular vein. The distal end of the catheter was threaded through the skin on the back of the mice and exited the skin via a stainless steel guide cannula (RWD Life Science). After surgery, catheters were ushed with 0.1 ml of a saline solution containing penicillin (160000 U/ml) and heparin (30 U/ml) daily. One week after recovering from surgery, animals were trained to self-administer intravenous injections of cocaine (0.75 mg/kg/infusion) or saline during 2 h sessions daily over 10 d on a xed ratio 1 (FR1) schedule in an operant chamber. The response to the active poke produced a cocaine injection and was accompanied with a blue light stimulus for 20 s and an audible tone for a 5-s timeout period. At the same time, the inactive poke failed to inject the drug and produce conditioning stimuli.

Locomotor activity
The locomotor activity was measured as the distance traveled. Animals were acclimated to the chambers (48 cm x 48 cm) equipped with a camera for 15 min once a day for a consecutive 2 d. Before the test, the baseline locomotor activity was not statistically different between groups. In the following week, animals were injected cocaine (20 mg/kg, i.p) or an equal volume of saline, immediately placed in the locomotor activity chamber, and allowed to explore for 15 min. The distance traveled was measured daily for one week, and automated tracking was performed with EthoVision 7.0 software (EthoVision 7.0; Noldus Information Technology, Leesburg, VA).

Drugs
Cocaine was purchased from the National Institute for the Control of Pharmaceutical and Biological Products (Beijing, China) and dissolved in saline. LY294002 (S1105, Selleck) were dissolved in saline containing 4% DMSO, 30% PEG300 and 5% Tween 80.

Stereotactic surgery and inhibitor administration
Animals were anesthetized with sodium pentobarbital (60 mg/kg) and mounted in a standard stereotaxic instrument (RWD Life Science). After shaving the hair and cleaning the incision site with medical-grade alcohol, the scalp was incised to expose the skull and permanent bilateral guide cannulas (RWD Life Science) were bilaterally implanted into the dorsal dentate gyrus (dDG) (AP, − 2.0 mm; ML, ± 1.4 mm; DV, − 2.2 mm) with stereotaxic instruments. Dental cement was used to anchor the guide cannula and the stainless steel stylet blocker was inserted into each cannula to prevent blockage and infection. All mice were used for subsequent training after one week of recovery from surgery. LY294002, a speci c inhibitor of the phosphoinositide 3-kinase (PI3K)-AKT signaling pathway, was also dissolved in saline containing 4% DMSO, 30% PEG300 and 5% Tween 80 at a nal concentration of 4.6 µg/µl. LY294002 were administered bilaterally (1 µl/side, 0.5 µl/min) with a micro-injector, 15 min before cocaine or saline administration during the cocaine-associated memory formation.
The coding regions of 2N4R Tau were ampli ed from cDNAs obtained from C57BL/6J mice by PCR and cloned into pAAV-CMV-MCS-3Flag and pAAV-CMV-MCS-2A-mNeongreen-3Flag to produce AAV-2N4R Tau or mutant 2N4R Tau, respectively. The vector was con rmed by sequencing. The recombinant plasmids were then packaged into AAV2/8 particles together with the AAV helper plasmid and puri ed using an iodixanol step-gradient ultracentrifugation method. Recombinant AAVs were produced by transient transfection in HEK293T cells and the expression of 4R Tau was analyzed using western blotting.
Mice (8-12 weeks old) were anesthetized with sodium pentobarbital (60 mg/kg) and placed on a stereotaxic apparatus (RWD Life Science) to inject the virus into the dDG (AP, − 2.0 mm; ML, ± 1.4 mm; DV, − 2.2 mm). After shaving the hair and cleaning the incision site with medical-grade alcohol, the scalp was incised to expose the skull and the connective tissue was gently removed from the skull surface with cotton swabs. Small craniotomy holes were drilled with a skull rotor (RWD Life Science) for virus injection. The micro syringe needles were used to bilaterally infuse the tissue with AAV2/8-CMV-2N4R-3Flag (0. The coding regions of 2N4R Tau and PI3K-P85α were ampli ed from murine cDNAs using PCR and cloned into pLenti-EF1a-EGFP-P2A-Puro-CMV-MCS-3Flag and pLenti-CMV-MCS-MYC-2A-mCherry-PGKblasticidin vectors to produce LV-2N4R Tau, LV-mutant 2N4R Tau and PI3K-P85α, respectively. The Tau shRNA was cloned into pLKD-CMV-EGFP-2A-Puro-U6-shRNA and con rmed by sequencing. The sequences of the scrambled control shRNA and Tau shRNA were 5'-TTCTCCGAACGTGTCACGT-3' and 5'-GGAGTTTGACACAATGGAA-3', respectively. Recombinant lentiviruses were also produced by transient transfection in the HEK293T cells and then the levels of 4R Tau and PI3K-p85 were analyzed using western blotting.

Tissue isolation
Mice were sacri ced by rapid decapitation at the end of the behavioral tests. The hippocampus or dentate gyrus (separated from the hippocampus) were removed from the brain, snap frozen in dry ice, and stored at -80 °C until the assay.

Western blot
Brain tissues and cells were lysed and proteins were extracted using a mammalian cell and tissue extraction kit (K269-500, Biovision) containing phosphatase inhibitors (4906845001, Roche) according to the manufacturer's protocols. The total protein concentration was analyzed with a Bradford assay kit (P0006, Beyotime). Twenty micrograms protein were loaded and separated on 10 or 12.5% sodium dodecyl sulfate-polyacrylamide gels. After separation, the gels were then transferred to a polyvinylidene di uoride (PVDF) membrane (IPVH00010, Millipore) in a mixed solution of Tris-glycine buffer and 20% (v/v) methanol. The membrane was blocked in TBST buffer containing 5% non-fat dry milk (9999, Cell Signaling Technology) for 1 h at room temperature, and then incubated and gently shaken overnight with the primary antibody at 4 °C. On the next day, after three washes with TBST for 15 min, the blots were incubated with the secondary antibody at room temperature for 2 h. Immunoreactivity was visualized using a chemiluminescence substrate (WBKLS0500, Millipore) with a chemiluminescence imagine system (CLINX, Shanghai, China). The optical density of each band was quanti ed using Chemi Analysis software (CLINX, Shanghai, China). The following antibodies were used for western blotting:

Immunohistochemistry
Animals that had undergone behavioral training were deeply anesthetized with sodium pentobarbital (60 mg/kg) and perfused transcardially with phosphate-buffered saline (PBS), followed by ice-cold 4% paraformaldehyde in 0.1 M PBS, pH 7.4. Brains were carefully extracted from the skull, post xed with 4% PFA overnight, and then dehydrated with 30% sucrose at 4 °C. Brains were sectioned into 45-µm coronal slices using a freezing microtome (Leica, Germany) and stored in 12-well plates lled with a cryoprotectant solution (glycerol, ethylene glycol and 0.1 M phosphate buffer, pH 7.4, 1:1:2 by volume) at -20 °C until processing for immunohistochemical staining.
The sections were washed with TBS 3 times (10 min each) and incubated with a blocking solution containing 0.3% TritonX-100 and 5% normal donkey serum (017-000-121, Jackson ImmunoResearch) in TBS for 2 h at room temperature. Sections were then incubated with primary antibodies in the blocking solution at 4 °C overnight. On the next day, the sections were also washed with TBS-0.3% TritonX-100 (TBST) 3 times, transferred to the blocking solution containing uorescent dye-conjugated secondary antibodies, and then incubated for 1.5 hours at room temperature. After another three washes with TBST, sections were cover slipped with anti-fade mounting medium containing DAPI (H-1200, Vector), and confocal images were acquired with a laser confocal microscope (Nikon, Japan). The corresponding DG areas and cell numbers were determined using Image J software (US National Institutes of Health, Bethesda, MD, USA) to assess the cell densities. The following antibodies were used for immunostaining: rabbit anti-Tau (1:100, Protech), mouse anti-3R Tau  For BrdU staining, DNA was denatured by incubating the section in 2 M hydrochloric acid for 30 min at 37 °C and then renatured with 0.05 M boric acid (28341, Thermo Scienti c) at room temperature for 30 min, followed by TBS rinses for 10 min. Immunohistochemistry was then performed.
The cannula placements were con rmed in 5-µm thick coronal sections using hematoxylin eosin (HE) staining, and images were acquired using a light microscope. Mice with misplaced cannulas were excluded from the statistical analysis. Immunocytochemistry Cells cultured on glass coverslips coated with poly-D-lysine (P6407, Sigma) were washed 3 times with PBS and then xed with 4% PFA for 10 min. After 3 washes with PBS, cells were blocked with blocking buffer (5% normal donkey serum and 0.3% TritonX-100 in PBS) for 1 h and then incubated with primary antibodies overnight. On the next day, coverslips were washed 3 times with PBS and incubated with the uorescent dye-conjugated secondary antibodies for 1 h at room temperature. Coverslips were then washed 3 times with PBS, stained with Hoechst (H1399, Life Technologies) for 5 min, and mounted with anti-fade mounting medium (H-1000, Vector). The following antibodies were used for immunostaining: rabbit anti-PI3K-p85α (1:400, Abcam), mouse anti-4R Tau (1:100, Biolegend), donkey anti-mouse (Alexa Fluor 488) (1:500, Invitrogen), and donkey anti-rabbit (Alexa Fluor 568) (1:500, Invitrogen).

Golgi staining
Golgi staining was performed using the FD Rapid Golgi Stain Kit (PK401, FD Neurotechnologies) to label neurons in the dDG. Brie y, animals were deeply anesthetized with sodium pentobarbital (60 mg/kg), and the brains were immediately removed and rinsed with MilliQ water. Afterwards, the retrieved brains were immersed in equal parts of Solutions A and B (containing potassium dichromate, mercuric chloride, and potassium chromate) and stored at room temperature for 2 weeks in the dark. Tissues were subsequently rinsed, placed in the cryoprotectant solution, and stored at room temperature for at least 72 hours in the dark before cutting. The brain slices were cut in the coronal plane at approximately a 100-mm thickness with the freezing microtome (Leica, Germany). Tissue sections were placed on poly-D-lysine-coated slides and dried in the dark. After drying, sections were transferred to distilled water, subsequently stained with a developing solution, and then successively dehydrated with 50, 75, 95, and 100% ethanol. Finally, the sections were defatted with a xylene substitute and cover slipped with DPX mounting medium (06522, Sigma).
Images were acquired from prepared slides using a confocal microscope (Olympus, Japan). Each neuron was scanned at high magni cation (100X, oil immersion) to ensure that all parts of the dendrites were intact. For each group, a minimum 5 neurons per slice were examined. At least 60 neurons were selected in one group. 3D neuronal reconstruction was also performed using a confocal microscope. The total length and dendrite spine density were measured using ImageJ software (US National Institutes of Health, Bethesda, MD, USA).

Co-immunoprecipitation analysis
Cells and tissues were harvested for the co-immunoprecipitation (co-IP) analysis using the simpli ed and reliable Pierce™ Crosslink Magnetic IP/co-IP Kit (88805, Thermo Scienti c). Brie y, the Tau or PI3K-p85 primary antibody was bound to 50 µl of Protein A/G magnetic beads (B23201, Bimake) for 15 min and washed three times. After the incubation, the protein supernatants extracted using the mammalian cell and tissue extraction kit were collected, and the protein concentration was determined using a Bradford assay kit. The supernatant from each sample was incubated with the antibody-crosslinked beads overnight at 4 °C. On the next day, beads were washed two times with IP lysis/wash buffer and one time with ultrapure water. Elution buffer was used to elute the bound antigen, and neutralization buffer was used to neutralize the low pH. The supernatants were collected for western blot analysis.

Cytoplasmic and nuclear fractionation
Harvested cells and tissues were extracted using the ProteoExtract® Subcellular Proteome Extraction Kit (539791, Millipore). The cytoplasmic and nuclear fractions were separated according to the manufacturer's instructions.
Preparation of the hippocampal single-cell suspension The brain tissue was dissected on ice and enzymatically digested using the Adult Brain Dissociation kit (130-107-677, Miltenyi). The single-cell suspension was prepared for the ow cytometry assay according to the manufacturer's instructions.

Flow cytometry
Single-cell suspensions from each group were centrifuged and the pellet was resuspended with a viability dye Fixable Viability Stain 780 (1:1000; BD) for cell live/dead discrimination, followed by an incubation with blocking buffer containing CD16/CD32 (1:100; BD) for 10 min. Cells were then incubated with Foxp3/Transcription Factor Staining Buffer Set (00-5523, Invitrogen) at 4 °C for 30 min in the dark. Finally, samples were incubated with the antibody mixture at 4 °C for 30 min in the dark, washed, centrifuged, resuspended in staining buffer containing 2% fetal bovine serum (FBS) and then sorted using a FACSAria SORP (BD, USA). Gating parameters and data analysis were performed using FlowJo 10 software (Tree Star, USA). The following antibodies were used for ow cytometry:

EdU or BrdU incorporation assays
For the analysis of newborn neuron proliferation in the hippocampus, the mice were injected with 50 mg/kg (i.p) bromodeoxyuridine (BrdU) (00-0103, Invitrogen) or 5-ethynyl-2'-deoxyuridine (Edu) (900584, Sigma) 3 times every 4 h on the last training day, and analyzed after the tests were complete [22]. BrdU and EdU immunostaining were performed to assess the proliferation of newborn hippocampal cells during cocaine memory formation.
For the ow cytometry assay of proliferation, single-cell suspensions were prepared from mice injected intraperitoneally with EdU as described above. Cultured N2a cells were serum-starved overnight, exposed to the external stimuli, and then 10 µM EdU was added to the culture medium for 2 h before cells were collected.
For the ow cytometry detection of EdU expression, collected cells were transferred to 5-ml FACS tubes, Quanti cation of the Tau-interacting proteome The dDG was removed from the hippocampus for co-IP, and the co-IP method was employed to detect the Tau-interacting proteome. The proteins were extracted and the co-IP procedures were conducted as described above. The collected protein supernatants were loaded and separated on 10% sodium dodecyl sulfate-polyacrylamide gels, and then subjected to gel digestion followed by the LC-MS/MS analysis.

In-gel digestion
For in-gel tryptic digestion, gel pieces were destained in 50 mM NH 4 HCO 3 in 50% acetonitrile (v/v) until they were clear. Gel pieces were dehydrated with 100 µl of 100% acetonitrile for 5 min, rehydrated in 10 mM dithiothreitol and incubated at 56 °C for 60 min. Gel pieces were again dehydrated in 100% acetonitrile and rehydrated with 55 mM iodoacetamide. Samples were incubated at room temperature for 45 min in the dark. Gel pieces were washed with 50 mM NH 4 HCO 3 and dehydrated with 100% acetonitrile, followed by rehydration with 10 ng/µl trypsin and resuspension in 50 mM NH 4 HCO 3 on ice for 1 h. The excess liquid was removed and gel pieces were digested with trypsin at 37 °C overnight. Peptides were extracted with 50% acetonitrile in 5% formic acid, followed by 100% acetonitrile. Finally, peptides were dried completely and resuspended in 2% acetonitrile in 0.1% formic acid.

LC-MS/MS analysis
The tryptic peptides were dissolved in 0.1% formic acid (solvent A), directly loaded onto a home-made reverse-phase analytical column (15-cm length, 75-µm i.d.). The gradient was comprised of 6-23% solvent B (0.1% formic acid in 98% acetonitrile) in 16 min, 23-35% in 8 min, increasing to 80% in 3 min, and then maintained at 80% for the last 3 min with a constant ow rate of 400 nl/min on an EASY-nLC The peptides were subjected to an NSI source followed by tandem mass spectrometry (MS/MS) in Q Exactive™ Plus coupled online to the UPLC. The electrospray voltage applied was 2.0 kV. The m/z scan range was 350 to 1800 for the full scan, and the intact peptides were detected in the Orbitrap at a 70000 resolution. Peptides were then selected for MS/MS using NCE at a setting of 28%, and the fragments were detected in the Orbitrap at a resolution of 17500. A data-dependent procedure that alternated between one MS scan followed by 20 MS/MS scans with 15.0-s dynamic exclusion was performed.
Automatic gain control (AGC) was set to 5E4.

Data processing
The MS/MS data were processed using Proteome Discoverer 1.3. Tandem mass spectra were searched in the SwissProt Mouse database. Trypsin/P was speci ed as the cleavage enzyme, allowing up to 2 missing cleavages. The mass error was set to 10 ppm for precursor ions and 0.02 Da for fragment ions. Carbamidomethylation on Cys was speci ed as the xed modi cation and oxidation on Met was speci ed as the variable modi cation. Peptide con dence was set at high, and the peptide ion score was set to > 20.

Protein interaction networks
All interesting gene name identi ers (mainly proteins involved in regulating cell proliferation) were searched against the STRING database version 10.5 for protein-protein interactions. Only interactions between the proteins belonging to the searched dataset were selected, thereby excluding external candidates. STRING de nes a metric called a con dence score to de ne the interaction con dence, and we retrieved all interactions with a con dence score ≥ 0.7 (high con dence). The interaction network identi ed using STRING was visualized with Cytoscape.

Statistical analysis
All data were analyzed with GraphPad Prism 7 software, presented as means ± SEM, and subjected to the Kolmogorov-Smirnov test to assess the normality of the distribution. For simple comparisons, an unpaired two-tailed Student's t test was used. For multiple comparisons, one-way or repeated-measures two-way ANOVA test was utilized for each experiment. In all cases, n refers to the number of animals. For all results, statistical signi cance was de ned as p < 0.05.

Tau downregulation is involved in cocaine-cue memory formation
Conditioned place preference (CPP) test, an associative memory model linking drug reward with environment cues, is widely used to assess the formation of drug-associated memory [23][24][25]. In the present study, we rst investigated the effect of cocaine on the expression of Tau in the hippocampus during the cocaine memory formation. Compared with the control group, the mice trained with CPP showed signi cantly reduced level of Tau in the hippocampus (Fig. 1a-c). We next performed two other reward tests, the home cage CPP and food CPP tests, to investigate whether the changes in Tau level was speci c to the formation of cocaine CPP memory. In the former test, mice were only injected with cocaine, but without cocaine-associated cue training (Fig. S1a, b). In the latter test, mice were only exposed to a palatable food reward to train the food-associated cue memory, but were not administered cocaine (Fig.   S1d, e). Importantly, changes in the level of Tau was not observed in the hippocampus of the mice subjected to these two tests (Fig. S1c, f). We further measured Tau level after cocaine CPP extinction.
Interestingly, the level of Tau clearly returned to baseline values (Fig. S1g-i).
We continued to investigate the role of Tau in cocaine self-administration, a paradigm that incorporates memories of drug experience and drug-associated environmental cues. Cocaine-associated memory was formed when an instrumental action (active pokes) resulted in cocaine delivery (unconditioned stimulus) paired with an audiovisual cue (conditioned stimulus). Mice were trained on a FR1 schedule, where a single active poke produced an infusion of cocaine or saline. Compared with saline-treated mice, cocaineself-administered mice showed reliable cocaine memory formation ( Fig. 1d-g). Consistent with these ndings, the level of Tau in the hippocampus was obviously reduced (Fig. 1h). We further detected Tau level in the hippocampus of cocaine-treated mice subjected to a locomotor activity test that lacks cocaine-associated contextual stimuli. Interestingly, Tau expression was not altered after cocaine injection for 7 days (Fig. 1i-k).
To further explore the role of Tau in cocaine memory formation, Tau-KO mice and non-transgenic wildtype (WT) mice were injected a low dose of cocaine (2.5 mg/kg) [26]. Notably, cocaine signi cantly induced a clear preference for the cocaine-paired chamber in Tau-KO mice that was not observed in WT mice (Fig.S2a-c). Collectively, Tau downregulation may play an important role in cocaine-associated memory formation.

Tau downregulation promotes AHN during cocaine memory formation
It has been known that AHN contributes to the formation of memory, including cocaine-associated memory [12,13]. To explore the role of Tau in AHN and cocaine memory formation, a ow cytometry assay was performed to investigate whether Tau ablation would affect the proliferation of newborn hippocampal neurons during cocaine memory formation. Notably, compared with WT mice, Tau-KO mice showed a qualitative increase in the proliferation of the neuronal subpopulations positive for SOX2 (a marker of neural stem cells and progenitor cells), SOX2/GFAP double-labelled cells (a marker of radial glia-like cells), and DCX (a marker of neuroblasts and immature neurons) (Fig. S2d-g). Thus, Tau de ciency may enhance AHN and facilitate cocaine memory formation.

4R Tau overexpression reduces AHN and attenuates cocaine memory formation
To investigate whether Tau overexpression affects AHN and weakens cocaine memory formation, we rst compared the expression of Tau isoforms in the dentate gyrus (DG), where AHN generally occurs [27].
Because the neuroanatomy of the hippocampus supports the segregation of neuronal outputs along a dorsal-ventral axis [28], the dorsal dentate gyrus (dDG) and ventral dentate gyrus (vDG) were separated from the hippocampus for immunoblotting analysis after CPP training, respectively. Both Tau and 4R Tau levels were signi cantly decreased in the dDG after cocaine CPP conditioning, whereas 3R Tau levels were not altered (Fig. S3a). In addition, the levels of the memory formation-related proteins phosphorylated CREB and phosphorylated co lin were signi cantly increased; however, the levels of phosphorylated ERK remained unchanged (Fig. S3b). No signi cant differences in the levels of Tau, Tau isoforms, phosphorylated CREB or phosphorylated co lin were observed in the vDG after cocaine CPP conditioning (Fig. S3c, d). Immunostaining further veri ed the decreased expression of both Tau and 4R Tau in the dDG of cocaine-conditioned mice (Fig. S3e-g). Collectively, 4R Tau, but not 3R Tau, may speci cally modulate cocaine-induced memory formation.
We next explored the function of genetically overexpressed 4R Tau in AHN during cocaine memory formation. To this end, we constructed AAVs speci cally expressing 2N4R Tau (AAV2/8-CMV-2N4R-2A-mNeonGreen-3Flag), the longest brain 4R Tau isoform containing an intact Tau amino acid sequence [29]. AAV-2N4R Tau was stereotactically microinfused into the dDG before cocaine CPP training, and 2N4R Tau expression was visualized directly based on mNeonGreen expression (Fig. 2a). As expected, overexpression of AAV-2N4R Tau (AAV2/8-CMV-2N4R-3Flag) in the dDG signi cantly decreased the cocaine CPP score (Fig. 2b, c). Furthermore, the proliferation rates of the neuronal subpopulations positive for SOX2, both SOX2 and GFAP, and DCX were reduced (Fig. 2d-g). We further determined the effect of 4R Tau overexpression on the proliferation of newborn hippocampal neurons in the dDG during cocaine memory formation. Mice were administered three consecutive injections of BrdU (50 mg/kg, i.p.), a marker of the exogenous proliferation of newborn cells, in one day before the cocaine CPP test (Fig. 3a). Importantly, AAV-mediated 4R Tau overexpression markedly reduced the density of BrdUlabelled cells in the dDG of the hippocampus (Fig. 3b, c). In addition, the density of Ki67/BrdU doublelabelled cells was clearly reduced following the overexpression of 4R Tau (Fig. 3b, d). By quantifying the densities of SOX2/BrdU double-labelled cells (Fig. 3b, e), SOX2/GFAP/BrdU triple-labelled cells (Fig. 3b, f) and DCX/BrdU double-labelled cells (Fig. 3b, g), we found that 4R Tau exerted a deleterious effect on the proliferation of the aforementioned neuronal subpopulations. Furthermore, immunostaining and immunoblotting analyses revealed increased expression of the immediate early protein c-fos in the dDG during cocaine memory formation; nevertheless, this increase was obviously attenuated by 4R Tau overexpression (Fig. 3h-j). These results strongly support the hypothesis that 4R Tau overexpression inhibits the proliferation and activity of adult hippocampal neurons in the dDG during cocaine memory formation.
We continued to investigate the effect of 4R Tau overexpression on the levels of memory formationrelated proteins and the synaptic structure of cells in the granule layer of the dDG. The phosphorylation levels of CREB and co lin were obviously increased in WT mice after cocaine CPP paradigm; however, these increases were clearly attenuated by AAV-mediated 4R Tau overexpression. In contrast, the levels of total CREB and co lin did not change (Fig. 4a, b). Golgi staining revealed a signi cant increase in the dendritic spine density of granule neurons after the cocaine CPP paradigm, whereas this increase was markedly inhibited by 4R Tau overexpression (Fig. 4c). In the cocaine self-administration model, the active pokes and numbers of cocaine infusions were signi cantly increased during cocaine memory formation; however, this increase was obviously attenuated by 4R Tau overexpression (Fig. 4d-g).
Similarly, cocaine-induced increases in the levels of phosphorylated CREB and co lin were decreased by 4R Tau overexpression (Fig. 4h, i). However, 4R Tau overexpression did not attenuate cocaine-induced hyperlocomotion in the absence of cocaine contextual stimuli (Fig. 4j-l). Thus, 4R Tau upregulation disrupts the formation of cocaine-associated memory and dendritic structural plasticity. 4R Tau modulates AHN by altering PI3K-AKT signaling during the cocaine memory formation Tau is able to bind to a variety of other proteins, including a number of proteins functioning in cell signaling transduction [21,30]. Tau-interacting proteome was analyzed to investigate the underlying mechanisms by which Tau regulates AHN. The dDG tissue was harvested, subjected to coimmunoprecipitation (co-IP), and analyzed using liquid chromatography coupled with tandem mass spectrometry (LC-MS/MS). Interestingly, 20 speci c Tau-interacting proteins, which are intimately involved in regulating cell proliferation, were identi ed (Table S1). Eight of these proteins, such as PI3K-p85, a critical PI3K regulatory subunit, were enriched in the PI3K-AKT and GSK-3β-β-catenin pathways (Fig. 5a). Using an immunoblotting analysis, we con rmed that PI3K-AKT and GSK-3β-β-catenin signaling in the dDG were altered by cocaine CPP conditioning; however, AAV-mediated 4R Tau overexpression inhibited PI3K-AKT signaling, but not GSK-3β-catenin signaling (Fig. 5b, c). We employed a co-IP analysis to further determine whether the interaction of Tau with PI3K-p85 would be altered by the cocaine CPP paradigm and found that both 3R Tau and 4R Tau bound endogenous PI3K-p85 in the dDG. Notably, the interaction of 4R Tau with PI3K-p85 was markedly reduced by cocaine CPP conditioning and reversed by the genetic overexpression of 4R Tau. However, the interaction of 3R Tau with PI3K-p85 was unchanged by cocaine CPP conditioning (Fig. 5d). Based on these results, the interaction of 4R Tau with PI3K-p85 may speci cally modulate cocaine memory.
We subsequently performed immuno uorescence staining to track the cellular localization of these two proteins in the cultured N2a cells in vitro and to further clarify the interaction of 4R Tau with PI3K-p85. 4R Tau was primarily located in the cytosol of N2a cells and was clearly co-localized with PI3K-p85 (Fig.S4a). Co-IP assays further revealed the interaction of endogenous Tau with PI3K-p85 (Fig.S4b). An association of human Tau with PI3K-p85 has been shown in COS-7 cells, and the replacement of prolines 216 and 219 and arginine 221 of human Tau with alanine drastically decreased its binding to PI3K-p85 [21]. By comparing the amino acid sequences of human and mouse Tau, prolines 216 and 219 and arginine 221 in human Tau correspond to prolines 205 and 208 and arginine 210 in mouse Tau, respectively. We then replaced these three amino acid residues of murine Tau with alanine and performed a co-IP assay. Notably, the interaction of mutant murine Tau with PI3K-p85 was markedly reduced (Fig.S4c), con rming the direct interaction of 4R Tau with PI3K-p85.
We further explored the function of the 4R Tau interaction with PI3K-p85 in the cultured N2a cells. N2a cells were serum-starved overnight, and then cultured with EdU, an exogenous proliferation marker, for 2 h before the cells were collected for ow cytometry detection. LV-mediated 4R Tau overexpression signi cantly inhibited the proliferation of N2a cells, whereas mutant 4R Tau exerted no effect (Fig.S4d).
Moreover, the level of AKT phosphorylated at serine 473, a marker of PI3K-AKT activation, was markedly decreased by the overexpression of 4R Tau, but not mutant 4R Tau (Fig. S4e, f).
We further utilized a Tau shRNA to silence Tau expression in N2a cells and determined whether the downregulation of Tau expression promotes cell proliferation by reducing its association with PI3K-p85. The siRNA oligo pairs targeting Tau were transfected into N2a cells and e ciently knocked down Tau expression (Fig. 6a). The co-IP analysis revealed a decrease in the association of 4R Tau with PI3K-p85 (Fig. 6a). Not surprisingly, the Tau protein de ciency signi cantly increased N2a cell proliferation and the percentage of phosphorylated AKT-positive cells (Fig. 6b-d).
4R Tau restricts PI3K-p85 to the cytoplasm Because the nuclear accumulation of PI3K-p85 is required for PI3K-AKT signaling activity [31], we examined whether 4R Tau could trap endogenous PI3K-p85 to reduce its nuclear translocation in cells. Immuno uorescence staining revealed the even distribution of endogenous PI3K-p85 (red) in the cytosol and nucleus of N2a cells. Strikingly, endogenous PI3K-p85 was predominantly expressed in the nucleus when the Tau protein was deleted by a Tau shRNA (Fig. 6e). An exogenous PI3K-p85-expressing LV (red) was transfected into the N2a cells to further explore the effect of 4R Tau on the cellular distribution of PI3K-p85. Similarly, exogenously expressed PI3K-p85 was also mainly expressed in the cytoplasm. Importantly, the Tau-de cient cells exhibited a clear increase in the nuclear translocation of exogenous PI3K-p85 (Fig. 6f). We further separated cytosolic and nuclear fractions for western blot analysis. Tau deletion signi cantly increased the level of nuclear PI3K-p85 and the cytosolic level of phosphorylated AKT (Fig. 6g), suggesting that Tau, a cytoskeletal protein, may restrain PI3K-p85 in the cytoplasm, eventually blocking PI3K-AKT signaling. Besides, Tau-silenced N2a cells were pretreated with LY294002, a PI3K-AKT inhibitor, at a nal concentration of 10 µM to determine the role of PI3K-AKT signaling in N2a cell proliferation. The levels of PI3K-p110α and phosphorylated AKT were decreased by LY294002 in Tausilenced N2a cells (Fig. S5a). The ow cytometry analysis showed a clear decrease in the subpopulations of Tau-silenced N2a cells expressing both EdU and phosphorylated AKT (Fig. S5b-g). Based on these results, 4R Tau regulates PI3K-AKT signaling by binding to the PI3K-p85 subunit, subsequently modulating neuronal proliferation.
The interaction of 4R Tau with PI3K-p85 modulates cocaine memory formation The 2N4R Tau-expressing AAV (AAV2/8-CMV-2N4R-2A-mNeonGreen-3Flag) was infused into the dDG of Tau-KO mice prior to the cocaine CPP paradigm to investigate the role of the 4R Tau and PI3K-p85 interaction in cocaine memory formation. Not surprisingly, 2N4R Tau overexpression signi cantly reduced the cocaine CPP score of Tau-KO mice; moreover, the co-IP assay showed an enhanced interaction of 4R Tau with PI3K-p85 (Fig. 7a-c). In Tau-KO mice, ow cytometry assays revealed a signi cant increase in the hippocampal cell populations labelled with EdU or antibodies against phosphorylated AKT; however, this effect was obviously reversed by AAV-mediated 2N4R Tau re-expression in the dDG (Fig. 7d-f).
Similarly, the levels of nuclear PI3K-p85 and cytosolic phosphorylated AKT were also reduced by 2N4R Tau re-expression (Fig. 7g). However, re-expression of 2N4R Tau in the dDG did not reverse cocaineinduced hyperlocomotion in Tau-KO mice (Fig. 7h-j).
The mutant 2N4R Tau-expressing AAV (AAV2/8-CMV-mutant 2N4R-2A-mNeonGreen-3Flag) was stereotactically infused into the dDG of Tau-KO mice before cocaine CPP training. As expected, mutant 2N4R Tau exerted little effect on cocaine memory formation, and only a small amount of 2N4R Tau bound to PI3K-p85 (Fig. S6 a-c). Mutant 4R Tau also failed to reduce the number of the cells labelled with EdU or antibodies against phosphorylated AKT, or the nuclear PI3K-p85 levels and cytosolic phosphorylated AKT levels ( Fig. S6d-g).
Finally, we applied LY294002 to pharmacologically suppress PI3K-AKT signaling in the dDG of Tau-KO mice. LY294002 was micro-infused into the dDG of mice through an implanted cannula before CPP training (Fig. S7a). As expected, the mice pretreated with LY294002 displayed a signi cant decrease in the cocaine CPP score (Fig.S7b, c). In addition, cocaine-induced changes in the levels of PI3K-p110α and phosphorylated AKT were attenuated by LY294002 (Fig.S7d). The proliferation rates of the neuronal subpopulations labelled with Ki67/SOX2 and Ki67/DCX were also reduced by the LY294002 treatment (Fig. S7e, f).
In summary, downregulation of 4R Tau reduces its association with PI3K-p85 and thus promotes PI3K-p85 nuclear translocation to activate PI3K-AKT signaling, which eventually promotes AHN and evokes cocaine-cue memory formation.

Discussion
Tau accumulation plays an important role in memory impairment [32,33], and a decrease in Tau expression prevents neuronal loss, amyloid beta-induced axonal transport de cits as well as memory impairment in tauopathy models [34][35][36]. In the present study we found a decreased expression of hippocampal Tau during cocaine memory formation; however, Tau expression remained unchanged in response to non-cocaine-conditioned stimuli. Furthermore, 4R Tau overexpression in the dDG disrupted cocaine memory formation, hippocampal neuronal proliferation and activity, and cocaine-seeking behaviors. In addition, cocaine-induced downregulation of 4R Tau clearly increased the expression of phosphorylated CREB, which is preferentially recruited or allocated to cocaine engram cells to encode memory consolidation [37]. These ndings suggest a novel mechanism that 4R Tau modulates cocaineassociated memory formation.

Distinct functions of Tau isoforms in the DG during cocaine memory formation
The segregation of neuronal outputs along the dorsal-ventral axis and their connectivity affect neurogenesis in the DG and hippocampus-associated memory [38]. During the cocaine contextual memory formation, 4R Tau expression was reduced in the dDG, but not in the vDG. The difference observed in these two brain regions may be attributed to the functional dissociation that exists along the dorsal-ventral gradient in the hippocampus [28]. Indeed, the dorsal hippocampus is more important for spatial learning and contextual discrimination than the ventral region [38,39]. In contrast, the ventral hippocampus is strongly associated with negative affective symptoms, such as emotional behavior and stress responses [40].
The functional neuronal expression of Tau isoforms has been a subject of controversy and debate. During brain development, the expression of 3R Tau and 4R Tau isoforms shifts during AHN [41], and 3R Tau provides a dynamic microtubule network in DCX-positive cells, allowing proper axonal growth during AHN [42]. Interestingly, cocaine decreased hippocampal 4R Tau expression, but had little effect on 3R Tau expression, suggesting different functions of Tau isoforms in speci c neuron populations. Indeed, 3R Tau binds with a lower a nity to microtubules, potentially contributing to the cytoskeletal plasticity. In contrast, 4R Tau is mainly expressed in the adult brain, exhibits higher a nity for microtubules, and functions in the establishment and maintenance of the synaptic structure of newly integrated neurons [20]. Therefore, based on our ndings, 4R Tau may contribute to the structural remodeling and integration of newborn neurons into the hippocampal circuit in response to drug conditioning. 4R Tau regulates cocaine-associated memory by modulating AHN AHN is also a process in which adult-born hippocampal neurons are functionally integrated into the addiction network by interacting with addiction-related brain regions to contribute to cocaine memory formation [11]. In our study, AHN in the dDG was tightly regulated by 4R Tau during cocaine memory formation, and a reduction in cell proliferation inhibited memory formation in both the cocaine-paired CPP and self-administration paradigms. Intriguingly, a few studies reported that cocaine is a potent suppressor of AHN and might frequently reduce cell proliferation in rodents [43,44]. However, these results were obtained from experiments in which rodents were forcibly administered cocaine in their home cage; moreover, the measurement of cell proliferation was usually performed immediately or soon after cocaine administration in an open-eld environment [45]. More importantly, these experiments lacked the cocaine-associated contextual stimuli that is critical for cocaine-cue memory formation. AHN induced by a genetic, pharmacological or environmental approach is able to exacerbate cocaineseeking behavior, showing a role of AHN in increasing the vulnerability to the action of drugs [12,13,46].
For example, through reducing AHN, mice exhibited higher motivation to cocaine self-administration and drug-seeking behaviors in the phase of reinstatement [46]. On the contrary, exercise before CPP training, which increases AHN, promotes cocaine memory formation [14]. Therefore, we infer that that if cocaine CPP conditioning is conducted in the animals with increased numbers of adult-born hippocampal neurons, more young neurons might be recruited for CPP training. It would explain why the cocaineassociated memory is more easily formed after increasing AHN [14]. Accordingly, reducing AHN in the acquisition phase of cocaine memory formation may suppress the survival, maturation, and/or function integration of the new neurons involved in the memory circuits and eventually impair cocaine memory formation. However, once cocaine memory has been formed, such as in cocaine withdrawal phase, reducing AHN may fail to weaken the previous cocaine memory, which trigger animal drug seeking and relapse during the period of abstinence [46]. Our ndings highlight the existence of a dynamic population of proliferating neurons in the DG, and AHN clearly emerges as a robust phenomenon during the cocaine memory formation.
The interaction of 4R Tau with PI3K-p85 modulates PI3K-AKT signaling By analyzing the Tau-interacting proteome, we revealed that PI3K-p85 is the key regulator of Taumediated AHN. Although accumulating studies have shown that PI3K-AKT signaling regulates hippocampal neuron survival and proliferation in the DG upon exposure to external chronic stress, the role of PI3K-AKT signaling in cocaine memory formation has not been completely elucidated [17,47]. Only one study published to date reported an association of human Tau with PI3K-p85 in COS-7 cells, and the replacement of a few amino acid residues of human Tau drastically decreased its binding with PI3K-p85 [21]. Here, we provide strong evidence that 4R Tau directly binds to PI3K-p85 and may restrict its nuclear translocation, thus modulating PI3K-AKT signaling. In turn, the cocaine-induced downregulation of 4R Tau may diminish the cytoplasmic localization of PI3K-p85 and subsequently increase its nuclear transportation, which is essential for cell proliferation and AHN.
In fact, as a critical regulatory subunit, PI3K-p85 interacts with several proteins to regulate the activation of the PI3K-AKT signaling pathway. For instance, the interaction of PI3K-p85 with bromodomaincontaining protein 7 (BRD7) [31], C-type lectin-like receptor 2 or leucine zipper tumor suppressor 2 suppresses PI3K-AKT signaling [48,49]. In contrast, the interaction of PI3K-p85 with CD133 activates PI3K-AKT signaling [50]. Therefore, through its interacting proteins, PI3K-p85 appears to exert distinct functions in the regulation of PI3K-AKT signaling. In the present study, the cellular distribution of PI3K-p85 was tightly regulated by 4R Tau, and the 4R Tau de ciency induced PI3K-p85 nuclear translocation, thus promoting neuronal proliferation. Further studies are needed to investigate the nuclear transport dynamics of PI3K-p85 and its precise function in neuronal proliferation during cocaine memory formation.

Conclusions
Collectively, 4R Tau regulates cocaine-associated memory through an AHN-dependent mechanism. This nding suggests the presence of unexplored mechanisms of cocaine-induced circuit plasticity in the hippocampus, and may have implications for the development of novel therapeutic strategies aimed at restoring a normal level of AHN or the disruption of the Tau-mediated synaptic plasticity that occurs in response to cocaine.

Consent for publication
All authors have approved of the contents of this manuscript and provided consent for publication.

Availability of data and materials
All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials. Additional data related to this paper may be requested from the authors.

Competing interests
The author declare that they have no competing interests.      Analysis of Tau-interacting proteins involved in regulating cell proliferation, and AAV-mediated 4R Tau overexpression weakens PI3K-AKT singling in cocaine CPP test. a The genes involved in regulating cell proliferation are mainly distributed the PI3K-AKT (light blue) and GSK-3β-β-catenin (light red) signaling pathway. Prediction gene refer to the typical genes involved in PI3K-AKT and GSK-3β-β-catenin signaling, which were identi ed by bioinformatics analysis. b, c Western blot analysis for the expression of PI3K-