MicroRNA-195 rescues ApoE4-induced cognitive deficits and lysosomal defects in Alzheimer’s disease pathogenesis

Our recent findings link the apolipoprotein E4 (ApoE4)-specific changes in brain phosphoinositol biphosphate (PIP2) homeostasis to the susceptibility of developing Alzheimer’s Disease (AD). In the present study, we have identified miR-195 as a top micro-RNA candidate involved in the ApoE/PIP2 pathway using miRNA profiles in human ROSMAP datasets and mouse microarray studies. Further validation studies have demonstrated that levels of miR-195 are significantly lower in human brain tissue of ApoE4+/− patients with clinical diagnosis of mild cognitive impairment (MCI) or early AD when compared to ApoE4−/− subjects. In addition, brain miR-195 levels are reduced along with disease progression from normal aging to early AD, and cerebrospinal fluid (CSF) miR-195 levels of MCI subjects are positively correlated with cognitive performances as measured by mini-mental status examination (MMSE) and negatively correlated with CSF tau levels, suggesting the involvement of miR-195 in early development of AD with a potential impact on cognition. Similar differences in miR-195 levels are seen in ApoE4+/+ mouse hippocampal brain tissue and cultured neurons when compared to ApoE3+/+ counterparts. Over-expressing miR-195 reduces expression levels of its top predicted target synaptojanin 1 (synj1), a brain PIP2-degrading enzyme. Furthermore, elevating miR-195 ameliorates cognitive deficits, amyloid plaque burden, and tau hyper-phosphorylation in ApoE4+/+ mice. In addition, elevating miR-195 rescues AD-related lysosomal defects in inducible pluripotent stem cells (iPSCs)-derived brain cells of ApoE4+/+ AD subjects while inhibiting miR-195 exacerbates these phenotypes. Together, our data uncover a novel regulatory mechanism of miR-195 targeted at ApoE4-associated brain PIP2 dyshomeostasis, cognitive deficits, and AD pathology.

The functional roles of PIP 2 and synj1 have been implicated in AD pathogenesis by our laboratory and others [26][27][28][29][30][31][32]. For example, increased expression of synj1 is linked to early endosome enlargement [33], and ApoE4-associated cognitive deficits in AD [26]. The reduction of synj1 provides several beneficial effects in AD such as accelerating Aβ clearance via the lysosomal degradation pathway [30], ameliorating mild traumatic brain injury (TBI)-induced elevation in tau hyperphosphorylation [31], and rescuing ApoE4-associated cognitive impairments [26]. However, molecular signaling mechanisms that link ApoE4 with brain PIP 2 /synj1 pathways and impact on cognitive function remain elusive. We have previously shown that increased synj1 levels in ApoE4 brains are partially due to a failure in efficient synj1 mRNA degradation [26]. mRNA stability is often regulated by micro-RNA (miRNA) binding to 3′-UTR regions of mRNA [34]. We postulate that brain synj1 expression may be differentially regulated by ApoE isoforms through miRNA modulation.
Several miRNAs have been previously implicated in various AD processes [35]. Moreover, changes in various brain miRNA levels between AD subjects and normal-aged controls have been reported [36,37]. Therefore, in the present study, we leveraged the existing miRNA datasets from the Religious Orders Study and the Rush Memory and Aging Project (ROSMAP) [38,39], in combination with miRNA array studies we performed in ApoE4 + and ApoE4 − cortical neurons. We have identified a miRNA, miR-195, as a top candidate that is differentially expressed between ApoE4 + and ApoE4 − carriers and targeted at synj1 mRNA as predicted by multiple bioinformatics databases including mirDB [40,41]. The changes in miR-195 levels are further validated using postmortem human and mouse brain tissue as well as cultured neurons. Furthermore, we have characterized a regulatory role of miR-195 in ApoE4-associated brain PIP 2 dyshomeostasis, cognitive deficits, and AD pathology.

Human miRNA expression profile and data preprocessing
The miRNA expression profile was downloaded from the ROSMAP study (Synapse https://doi.org/10.7303/ syn3388564). miRNAs that had a call rate less than 95% and an absolute value of lower than 15 in <50% of the samples were removed. miRNA expression values were normalized using a variance stabilization normalization method. Cartridges were specified as batches and were corrected with the Combat function in the R package sva (V3.20.0). The data pre-processing resulted in 309 miRNAs in 511 samples. Pre-processed RNA-seq FPKM gene expression abundance data were also downloaded from the ROSMAP study (Synapse https://doi.org/10.7303/ syn3388564). Genes with at least 1 FPKM in at least 10% of the samples were selected, and data were then corrected for confounding factors including batch, PMI, and RIN scores. The pre-processed gene expression profile contains 16,235 genes and 619 samples.

Differential expression and miRNA-gene correlation analysis
Differential expression analysis was performed on all miR-NAs between 111 APOE4 −/− (ε3/3) and 24 APOE4 +/− (ε3/4) carriers using the R package limma (V3.34.0) [42]. Multiple tests were adjusted using the Benjamini-Hochberg's (BH) FDR method. Correlation analysis was performed between miRNAs and genes using spearman's correlation test. The miRNA-gene correlation was also examined in each of the subgroups of AD diagnosis, sex, and APOE genotype. miRNA array studies of mouse primary neuron samples Embryonic 17 days old ApoE −/− cortical neurons were cultured for 7 days in vitro in the presence of conditioned media derived from ApoE3 and ApoE4 primary astrocyte cultures as described [26,43] (N = 3/group). miRNAs were extracted using miRCURY extraction kits (Exiqon Inc.) and then labeled using miRCURY LNA microRNA Hi-Power Labeling kit, Hy3/Hy5, and hybridized on the miRCURY LNA microRNA Array. The quality assessment using control spike-in oligo nucleotides produced signals in expected range indicated successful labeling. Following normalization of quantified signals after background correction using global Lowess regression algorithm, unsupervised and supervised data analysis were performed. The microRNA profiling identified a subset of microRNAs that are differentially expressed in the ApoE3 versus ApoE4treated neurons.

miRNA target prediction
Targets of the miRNAs were predicted with the R package multiMiR (V2.2.0), which is a miRNA-target interaction database complying nearly 50 million human and mouse data from 14 different databases [44].

Human brain and CSF sample preparation
Equal amounts of postmortem human parietal cortex brain tissues (50 μg by net weight) from the NIH brain and tissue repository (NBTR), as well as cerebrospinal fluid (1 ml by net weight) from mild cognitive impairment (MCI) subjects enrolled in the James J Peters VAMC study ("Markers of Transition to AD in the Veterans with MCI") and Icahn School of Medicine at Mount Sinai (ISMMS) AD Research Center participants were used for studies. The experimental procedures involving human sample handling were approved by the appropriate committees at James J Peters VA Medical Center (JJP VAMC) and ISMMS. Patient demographic information with N/group provided in Table S1.

Animal models
Human ApoE4 +/+ or ApoE3 +/+ knock-in (KI) mouse models without [45][46][47] or with 5xFAD background [48,49] were genotyped as described [50]. All animal experiments were performed in accordance with NIH guidelines and were approved by the JJPVAMC and ISMMS Institutional Animal Care and Use Committees (IACUC). Sex as a biological variable was taken into considerations with inclusion of both male and female mice in all experiments.

Phospholipid analysis
Samples were subjected for lipid extraction, followed by the quantification by anion-exchange HPLC as described previously [30,51].

Mouse Neuronal and human iPSC culture
Primary cortical neurons were cultured as described [26,43], before being fixed and stained for confocal microscopy analysis (Zeiss LSM) [52]. In some conditions, bovine serum albumin (BSA) or receptor-associated protein (RAP) was included in the cultured media (N = 3/group). Alternatively, cultured neurons were transfected with adeno-associated virus 2 (AAV2)-containing miR-195 or miR-374 or scramble controls for 5 days before subjected to analysis. The ApoE4 +/+ and ApoE3 +/+ iPSCs were differentiated into neural progenitor cells (NPCs) by dual SMAD inhibition followed by neural rosette selection and forebrain-specific patterning by 20 ng/ml FGF2 exposure as described [53,54]. These NPCs were purified by MACS for CD271 − /CD133 + [53] and differentiated to cortical neurons [55,56] and a homogeneous population of astrocytes [54,57] before subjected to viral transfection and confocal microscopy analysis. Alternatively, mouse cortical neurons derived from ApoE4 +/+ and ApoE3 +/+ KI mice with synj1 +/+ or synj1 −/− genotypes were co-cultured with ApoE4 +/+ and ApoE3 +/+ iPSC-derived pure astrocytes. In some experiments, cultured neurons were transfected with AAV2-containing miR-195, scramble controls, or miR195 inhibitors that specifically prevent miR-195 binding to its target mRNA before co-culturing with iPSC-derived astrocytes. Cultured iPSC-derived neurons and astrocytes were then incubated with lysotracker red for various time periods before fixation and double-stained for pTau and a nuclear marker DAPI (blue) for confocal microscopy analysis (Zeiss). N = 4-5/condition.

Stereotaxic injection and behavior studies
Eight to nine weeks old male and female ApoE3 +/+ and ApoE4 +/+ KI mice without 5xFAD background (N = 19-23/group), or with 5xFAD background (N = 15-17/ group) were placed in the stereotaxic apparatus with AAV2 or scramble virus administered into the dorsal CA1 regions of bilateral hippocampal brain regions using pressure injection as described [58,59]. Injection volumes (0.5-2.0 µl) were delivered over 10 min to avoid tissue damage. Six to nine months after viral delivery, mice were tested with the NOR task as described [26,60,61]. Mice were randomized for genotype and sex, and blinded throughout the behavior data collection and analysis, surgical manipulations, and sample collection followed NIH practice guidelines. Animals were excluded from behavior analysis if the total exploration time was less than 4 s or if they had an illness that prevented them from reliably completing the behavior tests.

Brain and neuronal sample preparation and biochemical analysis
Snap-frozen mouse hemi-brains or cultured neurons were harvested in lysis buffer [62] and processed via step-wise solubilization [62,63], followed by SDS-PAGE to determine levels of synj1, dyn1, holoAPP, and CTFs. Levels of Aβ 42 , Aβ 40 , pTau, Tau and ApoE were determined using high-sensitive ELISA kits. Some tissue was used for miRNA and RNA extraction followed by qPCR and RNA-seq analysis. Some animals underwent perfusion followed by brain tissue section for immunohistochemical staining of amyloid plaque, synj1 and pTau.

Differential gene expression analysis for miR-195 treated mice
The RNA-seq samples collected from mouse brains were profiled on the Illumina HiSeq platform. Quality control of generated reads was performed using FASTQC (0.11.8).
The raw sequencing reads were aligned to the GRCm38 mouse genome (release 95) using star aligner (V2.5.0b). Following the read alignment, gene expression was quantified at the gene level based on Ensembl gene model GRCm38.95 using FeatureCounts [64]. Genes with at least one count per million in all samples were considered as expressed and hence retained for further analysis. The trimmed mean of M-values normalization method [65] was used to adjust for sequencing library size differences. Differential expression analysis was then performed on the quality controlled and normalized gene expression data using the R package limma (V3.34.0). The comparisons were carried out between miR195-treated and scramble control samples stratified by sex and APOE genotype. Multiple tests were adjusted using the BH FDR method.

Functional enrichment analysis
The functional enrichment analysis was carried out for genes significantly correlated with miRNAs in human ROSMAP dataset, predicted target genes of each miRNA of interest, and differentially expressed genes identified from miR195-treated mouse RNA-seq dataset. These genes were queried against the molecular signatures database (MSigDB v6.1) using Fisher's Exact Test and gene set enrichment analysis (GSEA) [66].

Statistical analysis
The sample sizes of each experiment were chosen based on power calculations derived from previous similar studies, which allowed us to determine group sizes needed to achieve statistically significant results. All experiments including controls were performed in randomly assigned groups. Sample collection and data analysis followed NIH practice guidelines. Experimenters were blinded to the experimental condition of the animal while conducting experiments. The conditions were revealed after quantification was completed. Levels of miR-195 and mi-374 were normalized to U6 and RNU6B (internal controls) while synj1 mRNA normalized to GAPDH and 18s, and then expressed as Log 2 fold of changes when compared to controls. Levels of synj1, dyn1, pTau, Tau, and holoAPP were normalized to β-actin levels and expressed as a percentage of the control. Absolute Aβ 42 , Aβ 40 , pTau, Tau, and ApoE concentrations were quantitatively determined by ELISA and expressed as a percentage of the control. Independentsamples t-tests were used to determine significant mean differences (the threshold for significance sets at p < 0.05). ANOVA with post-hoc tests were used to determine group differences for multiple comparisons. Pearson correlation coefficients were calculated to determine the linear relationship between the two variables. Equality of variance was checked for all statistical comparisons. When independent-samples t-tests were used and equality of variances of compared groups were not the same, the Welch's corrections were applied. All statistical analysis was performed using Prism 8.0.
Furthermore, we examined the correlation between synj1 mRNA and all miRNAs in ROSMAP dataset using Spearman's correlation test. MiR-195 is negatively correlated with synj1 mRNA levels in all human subjects ( Fig. 1c: r = −0.115, p = 0.0093). Similarly, negative correlations between miR-195 and synj1 mRNA levels are seen in all female subjects (r = −0.167, p = 0.0026) and in ApoE3 +/+ carriers (r = −0.127, p = 0.027). A negative correlation trend can be seen in ApoE4 +/− carriers as well but with no statistical significance (r = −0.178, p = 0.057), suggesting a possible weakened or perturbed network regulation between miR-195 and synj1 in the presence of ApoE4 allele. Separately, miR-374b also showed negative correlation with synj1 mRNA (r = −0.09, p = 0.04). In addition, mmu-miR-374b-5p is differentially expressed between ApoE3 and ApoE4-treated conditions but the differences in hsa-miR-374-5p levels between human ApoE4 +/− and ApoE4 −/− carriers are not statistically significant (Fig. S1B, To better understand the molecular pathways miR-195 regulates, we investigated the functions of predicted miR-195 targeted genes and those significantly correlated with miR-195 in the ROSMAP dataset. The top enriched functions for target genes and genes negatively correlated with miR-195 include regulation of neuronal and synaptic function, neurogenesis, and differentiation, while functions of genes positively correlated with miR-195 are enriched in the circulatory system and vasculature development (Table S2).
Together, these results support a role of miR-195 as a top candidate miRNA in regulating the ApoE-synj1-PIP 2 pathways.
Reduction of brain miR-195 levels is associated with ApoE4 genotype, disease progression and cognitive decline To validate differential expression patterns of miR-195 between ApoE4 +/− and ApoE4 −/− carriers, we next Fig. 1 MiR-195 is identified as a top miRNA candidate involved in APOE-regulated synj1 expression. a Venn diagram shows miR-195 as the only miRNA in common shared among 4 groups: miRNAs differentially expressed between ApoE4 + and ApoE4 − carriers in the human ROSMAP dataset, miRNAs differentially expressed between ApoE4 + and ApoE4 − in the mouse miRNA array studies, miRNAs negatively correlated with synj1 mRNA in ROSMAP, and miRNAs predicted to target at synj1 mRNA by multiMiR database. Numbers of miRNAs overlapping among subgroups are indicated (red numbers). b Log Fold of changes (LogFC) and p values of differences in miR-195 levels between ApoE4 + and ApoE4 − carriers, between female ApoE4 + and ApoE4 − carriers in ROSMAP dataset, as well as differences in miR-195 levels between mouse ApoE4 + and ApoE4 − treated neurons. c Analysis of correlation between miR-195 and synj1 mRNA in human subjects of the ROSMAP database. examined miR-195 levels in human brain tissue and CSF samples (subject demographic information provided in Table S1). We found that miR-195 levels were reduced in parietal cortex tissues derived from ApoE4 +/− mild cognitive impairment (MCI) and early AD subjects with clinical dementia rating (CDR) scores between 0.5 and 1 compared to levels in ApoE4 −/− donors (Fig. 2a). Interestingly, we observed a pattern of reduction in miR-195 levels along with disease progression from normal aging to MCI and early AD (Fig. 2b), similar to what we previously seen with PIP 2 and phosphoinositol (PI) changes in early AD development [26]. We also found a significant reduction in miR-195 in female subjects compared to male subjects (Fig. S2A), with differences also noted between male ApoE4 −/− versus female ApoE4 +/− subjects. Consistently, we observed a reciprocal elevation of synj1 mRNA levels in ApoE4 +/− subjects when compared to levels in ApoE4 −/− subjects (Fig. S2B). A positive correlation was noted between brain miR-195 and PIP 2 levels in ApoE4 −/− carriers with CDR 0.5-1, and a positive correlation trend was seen in CDR 0-1 subjects regardless of ApoE genotypes (Fig. S2C). No correlation was seen between brain miR-195 and other phospholipid species, e.g., PI and phosphoinositol phosphate (PIP). No correlation was seen between brain miR-195 and other variables such as post-mortem interval (PMI) and age. A negative correlation was observed between brain miR195 and another known target of miR-195, beta-secretase 1 (BACE-1) [67] expression in the CDR 0.5-1 cohort (Fig. S2D). However, we did not observe any correlation between miR-195 and Aβ levels (data not shown).
There were statistically significant differences in human miR-374 levels between ApoE4 +/− and ApoE4 −/− subjects (hsa-miR374-5p; Fig. S3A). However, there were no significant changes in miR-374 levels along the disease progression except a transient elevation at MCI stage (Fig. S3A). Consistently with ROSMAP data (Fig. S1B), we also observed statistically significant differences in human miR-155 levels with higher levels in ApoE4 +/− subjects (hsa-miR155-5p; Fig. S3B). No significant differences were seen in miR-155 levels along disease progression (data not shown). Moreover, no significant differences were seen in miR-195 or miR-374 levels between ApoE4 +/− and ApoE4 −/− subjects of normal aging or advanced AD either (CDR 0 or 3 and above; data not shown), suggesting the functional involvement of miR-195 in early disease development and acceleration by ApoE4 genotype.
Using cerebrospinal fluid (CSF) samples from a cohort of MCI subjects with CDR 0.5 (MCI defined by clinical examination and neuropsychological assessments), we found that CSF miR-195 levels were positively correlated with cognitive performance measured by mini-mental status examination (MMSE; Fig. 2c), and negatively correlated with pTau levels (Fig. 2d). While CSF PIP 2 levels were below the detectable range, a positive correlation was seen between CSF cardiolipin and miR-195 (Fig. S3C), suggesting a potential involvement of miR-195 in mitochondrial function [68,69].
Together, our results suggest that reduction of brain and CSF miR-195 levels is associated with ApoE4 genotype, cognitive decline, and tau pathology during early AD development.

MiR-195 expression is reduced in ApoE4 mouse brains and cultured neurons
Next, we investigated if differences in miR-195 levels can be recapitulated in mouse models and primary neurons. We found that the levels of miR-195 were lower in 12month old ApoE4 +/+ mouse brains compared to ApoE3 +/+ mice (Fig. 3a). A nominal reduction in miR-195 levels was seen in ApoE −/− brains. Similarly, miR-374 was decreased in ApoE4 +/+ mouse brains when compared to ApoE3 conditions (Fig. S4A) with a nominal reduction in ApoE −/− mice. In cultured ApoE −/− hippocampal neurons, levels of miR-195 were consistently lower with ApoE4 CM from astrocytes compared to those with ApoE3 CM (Fig. 3b). A nominal difference was noted in miR-374 levels in neurons treated with ApoE4 CM when compared to those treated with ApoE3 CM (Fig. S4A) but failed to achieve statistical significance due to large variations among samples.
Interestingly, treatment of ApoE-receptor-associated protein (RAP), an inhibitor of ApoE receptors [70,71] abolished differential expression patterns of miR-195 relative to control (BSA: bovine serum albumin). The RAP treatment in the presence of ApoE3-CM led to a reduction in miR-195 levels (Fig. 3c), whereas in ApoE4-CM treated conditions, miR-195 levels were much lower at baseline with a trend of improvement following RAP treatment. These results suggest that astrocyte-derived ApoE likely binds to ApoE receptors on neurons leading to changes in neuronal miR-195 and ApoE4 exhibits loss-of-function effects on neuronal miR-195 expression.
We also examined if miR-195 over-expression leads to changes in gene expression patterns and downstream pathways in EFAD mouse brains. Again, most differentially expressed genes (DEGs) are enriched in regulation of neuron, synapse, and immune functions (Table S3). Further GSEA studies [66] suggested top pathways perturbed by miR-195 over-expression are mitochondrial related pathways, consistent with studies in human brain dataset with top pathways enriched for genes negatively correlated with miR-195 involved in mitochondrial function (Table S2).
Together, our results suggest that elevating miR-195 levels in ApoE4 mouse models without and with AD background can rescue ApoE4and AD-related cognitive deficits and pathological changes.
Together, these results suggest that elevating miR-195 levels in human iPSC-derived ApoE4 +/+ AD brain cells can rescue lysosomal defects, whereas inhibiting miR-195 can exacerbate these phenotypes.

Discussion
AD is a complex, multifactorial neurodegenerative process, and accumulating evidence indicates the importance of miRNAs in AD pathogenesis. Our studies characterize the functional involvement of a miRNA, miR-195 in ApoE4associated pathology. Utilizing unbiased systems biology approaches as well as multiple computational prediction tools, we leverage existing human ROS-MAP miRNA datasets in combination with mouse micro-array studies and identify miR-195 as the top miRNA candidate in regulating the ApoE-synj1-PIP 2 pathways and that this regulation is differentially modulated by ApoE genotypes. More importantly, our data reveal a novel regulatory role of miR-195 in the ApoE4 genotype-associated cognitive and lysosomal defects that contribute to AD development.
Dysregulation in brain miRNAs has been described in human subjects and mouse models of AD with proposed involvement of Aβ-dependent and Aβ-independent pathways [36,37,73]. Our studies here demonstrate miR-195 reduction during AD development, which correlates with early disease progression but not with advanced stages of AD. ApoE4 genotype accelerates miR-195 reduction, which coincides with cognitive decline and increased tau pathology (Fig. 2). Previously, we reported patterns of changes in phosphoinositol (PI) metabolites correlated with disease conversion from normal aging to early AD [26]. Our results further indicate miR-195 changes are consistent with alternations in brain PIP 2 levels (Fig. S2C). These findings highlight the potential of utilizing miR-195 levels as surrogate biomarkers to monitor brain PIP 2 homeostasis and cognitive performance and detect early AD development and progression. Our studies of differentially expressed miRNAs between ApoE4 +/− and ApoE4 −/− carriers in ROSMAP and mouse microarray datasets also identified another miRNA in common, miR-155. A prior report suggested a protective role of miR-155 in AD with increased brain miR-155 levels in aged AD transgenic mice (APP/PS1 or Tau Tg) and in AD human subjects compared to levels in controls [73]. On the other hand, our data show that miR-155 is up-regulated in human ApoE4 +/− carriers but downregulated in mouse ApoE4 condition (Fig. 1c). Nevertheless, the involvement of miR-155 in ApoE4 pathogenic function remains unclear with opposing changes in human and mouse datasets.
Interestingly, functional enrichment studies of human dataset (Table S2) and mouse transcriptomic dataset (Table S3) implicate the roles of miR-195 in regulating neuronal and synaptic function, neurogenesis, and differentiation. Consistently, our data indicate that restoring miR-195 levels in vivo rescues ApoE4-associated cognitive deficits (Fig. 4a, c), ameliorates amyloid plaque burden (Fig. 4e, f) and pTau levels (Fig. 4b, d), and improves lysosomal defects in cultured human iPSC-derived brain cells (Fig. 5). While we did not see any reduction in endogenous mouse Aβ in ApoE KI mouse models (Fig. S5B) or soluble Aβ levels in EFAD mice (Fig. S5F), we observed dramatic decreases in oligomer Aβ levels and plaque burden with over-expression of miR-195 (Fig. 4e, f). In addition, we found a negative correlation between miR-195 and BACE1 expression in human brain tissue (r = −0.516, p = 0.004; Fig. S2D), consistent with a previous report [67]. However, no changes in BACE-1 levels were seen with miR-195 over-expression (data not shown). Together, these data suggest that miR-195 most likely regulates Aβ clearance instead of Aβ generation, and restoration of lysosomal function may facilitate these processes, as indicated in our prior reports that synj1 reduction accelerates lysosomal clearance of Aβ [43].
Moreover, a functional role of miR-195 in regulating pTau levels is implicated ( Fig. 4 and Fig. S6A and S6B), consistent with our recent findings that down-regulation of synj1 prevents mild TBI-induced tau hyper-phosphorylation [31]. It would be interesting to further investigate the molecular mechanisms by which miR-195 regulates tau hyper-phosphorylation. It was previously reported that the exosomal secretion of tau may play an important role in tau spread, which could be regulated by miRNAs [74]. We speculate that miR-195 might serve as the key in modulating tau pathology secondary to impaired clearance through the lysosomal pathway and/or accelerated spread through the exosomal secretory pathway.
The results suggest that differential regulation of miR-195 expression by ApoE isoforms is mediated through binding to ApoE receptors on neurons, and the ApoE4 genotype loses the ability to regulate miR-195 levels (Fig. 3c). In addition, our data showing ApoE −/− with lower miR-195 expression like ApoE4 (Fig. 3a), further supports loss-of-function effects of ApoE4 on miR-195 expression leading to increased synj1 mRNA and protein expression. Over-expression of miR-195 in synj1 −/− neurons failing to exhibit any additive effects on lysosomal enlargement (Fig. S6F) further strengthens the concepts that miR-195 rescues AD-related phenotypes through its target gene, synj1. It should be noted that ApoE4 +/− carriers exhibit higher sensitivity to miR-195 manipulations than ApoE4 −/− subjects (Figs. 4 and S5; Figs. 5 and S6), possibly due to much lower baseline miR-195 levels. Furthermore, differences in miR-195 levels between ApoE4 +/− and ApoE4 −/− conditions are not only seen in neurons but also in other brain cells such as astrocytes (Fig. S6C). The cell-type specific changes in miR-195 may contribute to different aspects of disease pathogenesis. We speculate that reduction in neuronal miR-195 levels may contribute to cognitive and synaptic dysfunction, while reduction in astrocytic miR-195 levels may play a role in defects in the secretory pathways leading to accelerated tau accumulation and spread.
We also examined sex impact in miR-195 expression. In ROSMAP dataset, the differences in miR-195 levels between ApoE4 +/− and ApoE4 −/− carriers persist in female subjects (Fig. 1b), similarly to our previous report of sex-specific effects on brain PIP 2 homeostasis [26]. Sex dimorphism is also noted in miR-195 expression with much lower levels in female subjects, particularly in ApoE4 −/− carriers (Fig. S2A). EFAD mice also exhibited sex dimorphic responses to miR-195 manipulations with improved cognitive function and reduced oligomer Aβ levels in male but not female EFAD mice (Fig. 4c, e). Age-related changes in miR-195 expression and synj1 mRNA levels have also been noted in ApoE KI and EFAD mouse brains, with differential expression between ApoE4 +/+ and ApoE4 −/− mice more prominent at 12 months of age compared to a younger age, whereas differences in miR-195 in EFAD mice are already evident at 4 months of age (data not shown), suggesting that ApoE-genotype associated miR-195 changes might be exacerbated by aging and/ or manifestations of AD pathologies.
While AD manifests as a multi-faceted disease process, targeting a specific miRNA to restore dysregulated networks and pathways at multiple levels could provide a promising avenue for future drug development. Therapeutic strategies directed at ApoE4 have been and are actively explored in several preclinical and clinical studies such as immunotherapies, antisense oligonucleotide treatments, gene editing, modulators of ApoE expression, as well as small molecules to enhance ApoE lipidation, to correct its structures, to compete receptor binding, and to inhibit ApoE-Aβ interaction [75,76]. Our findings here suggest a novel therapeutic direction that modulates ApoE4 pathogenic function by a miRNA miR-195 through brain PIP 2 lipid signaling pathways with multiple beneficial effects besides impact on Aβ and tau pathology.
In summary, our studies provide a mechanistic link between ApoE4 genotype-specific changes in brain miR-195 expression with AD-related phenotypes including brain phospholipid dysregulation, cognitive deficits, lysosomal defects, and tau pathologies. These studies may uncover novel therapeutic strategies targeted at a specific miRNA miR-195.

Data availability
The human miRNA and RNA-seq datasets from ROSMAP studies are available via the AD Knowledge Portal (https://www.synapse.org/#!Synapse:syn5856115 and https://www.synapse.org/#!Synapse:syn3505720). The AD Knowledge Portal is a platform for accessing data, analyses, and tools generated by the Accelerating Medicines Partnership (AMP-AD) Target Discovery Program and other National Institute on Aging (NIA)-supported programs to enable open-science practices and accelerate translational learning. Data are available for general research use according to the following requirements for data access and data attribution (https://adknowledgeportal.synapse.org/ DataAccess/Instructions). The mouse microarray studies and RNA-seq datasets after miR-195 perturbation experiments generated in this paper have been deposited to the AD Knowledge Portal as well (https://www.synapse.org/#!Syna pse:syn22123695).
Acknowledgements We thank Dr. Guojun Bu for generously providing us the RAP for our studies. This work was supported in whole or in part, by funding from NIH R01 (1R01AG048923) to DC, RF1 (RF1AG054014) and R56 (R56AG058655) to DC and BZ, Department of Veteran Affairs BLRD (I01BX003380) and RR&D (I01RX002290) to DC, RRD (IO1RX002660) to GE, CSRD (I01CX000619) to MS, NIH K01 (AD062683) to J. TCW, NIH U01 (UO1AG046170) to BZ and VH, RF1 (RF1AG057440) to BZ and VH, UO1 (UO1AG052411) to BZ and AG. Confocal microscopy studies and animal behavioral analysis were supported by James J. Peters VA Medical Center Research Core Facilities. Postmortem human brain tissue for validation studies were provided by NIH NeuroBioBank and CSF samples provided by JJP VAMC and ISMMS MCI research cohort.

Compliance with ethical standards
Conflict of interest The authors declare that they have no conflict of interest.
Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons. org/licenses/by/4.0/.