Reducing miR485-3p ameliorates Alzheimer`s disease pathology by 1 regulation of amyloid beta and neuro-inflammation

Accumulation of amyloid-β (Aβ), neurofibrillary tangles and neuroinflammation play the important neuro-pathology in patients with AD. miRNA is multifunctional and involved in physiological and pathological processes. Recently, microRNAs have been linked to neurodegenerative diseases. However, it is little known whether miRNA dysregulation contributes to AD pathology progression such as A  processing, phagocytosis and neuroinflammation. Here, we identify miR485-3p as a novel modulator of AD pathology in 5XFAD mice. demonstrated that miR485-3p ASO significantly reduced Aβ plaque and amyloid biosynthetic enzyme. Importantly, the attenuation of Aβ  plaques through miR485-3p ASO was mediated through Aβ phagocytic activity of glial cells, by which it can directly target CD36. MiR485-3p ASO also decreased inflammatory responses. Collectively, these responses inhibited neuronal loss caused by Aβ lead to improvements of cognitive impairment. Our data provide evidence for the molecular mechanisms which underlie the miR485-3p ASO responses in an AD mouse model. These results suggest that attenuating miRNA 485-3p levels 70 might represent a novel therapeutic approach in AD. The miR485-3p ASO (AGAGAGGAGAGCCGUGUAUGAC) were synthesized by Integrated DNA Technologies (USA). Non-targeting ASO (negative control, Cat#AM17010) were purchased from ThermoFisher (USA). All animals were initially anesthetized with 3–5 % isoflurane in oxygen and fixed on a stereotaxic frame (JeongDo). For intracerebroventricular (ICV) injection, miR485-3p ASO or non‐targeting control oligonucleotides were formulated with in vivo jetPEI reagent (Polyplus). miR485-3p ASO (1.5 μg) or control oligonucleotide formulated with in vivo jetPEI reagent, was injected with a 10ul Hamilton syringe (26-gauge blunt needle) at 1.5ul/min. miR485-3p ASO were infused in a volume of 5 μl into 10 month old 5XFAD mice by intracerebroventricular (ICV). miR485-3p ASO or non‐targeting control oligonucleotides were given once a week for 2 weeks. Intracerebroventricular (ICV) position was identified using the coordinates from the bregma: AP=−0.2 mm, L= ± 1.0 mm, ventral (V)= 187 −2.5mm. Three The results were visualized using an enhanced system, and quantified by densitometric analysis (Image J software, All experiments were performed independently at least three times.

the AD brain could be neuroprotective because of its involvement in A processing and 114 inflammation. 115 In this paper, we describe a new factor that regulates SIRT1, and which plays an important 116 role in AD pathology. Notably, our data show that expression of SIRT1 is reduced in brains of 117 AD patients and 11-mo-old 5XFAD AD mice. We found miR485-3p among many miRNAs is  Brain precentral gyrus samples from patients with AD and from controls were purchased from 133 Netherlands brain bank. Information related to these patients and controls is shown in 134 (Additional file 1: Table S1).  Stranded mRNA Library Kit (Illumina) was used to build the library. Afterwards, data was 152 8 processed using 'Raw read' for mRNA sequencing. Raw reads were aligned to GRCm38.96 153 (NCBI) using STAR aligner v2.7.1 for calculation of 'RSEM' expression values [14]. We 154 performed the STAR aligner as the default option. Since the total number of reads for each 155 sample was different, normalization was performed by TMM method. Thirteen mouse samples 156 were processed in the same way. All data is available in the GEO (Gene Expression Omnibus, 157 https://www.ncbi.nlm.nih.gov/geo/) as GSE142633. Public database usage, reanalysis and network analysis 160 We used results from Weinberg et al. to confirm miRNAs that are highly related to cognitive 161 impairment [15]. The 100 genes shown in Figure S1 were extracted from Table 1   The "miRDB" was used to search for miRNA targeting specific genes [16]. The "Genecard" 164 database was used to search for genes related to disease or biological symptoms [17]. The 165 results in Figure S2a show search results from using keywords, "Inflammation", "Amyloid beta 166 degradation" and "Alzheimer" in August 2019. We used "VennDiagram" package of R for 167 analysis for Venn diagram. The "GeneMAINA" (version 3.5.1) package of Cytoscape (version 168 3.7.1) was used for protein to protein interaction analysis [18]. We used 265 common genes 169 that included hsa-miR-485-3p target genes and Alzheimer-related genes as inputs for protein 170 interaction analysis. Among them, 139 genes interacted without neighbor gene. In addition, 9 171 genes were highly associated with cerebral nervous system diseases (including AD) and at the      anti-actin (Santa Cruz, Cat#sc-47778). The results were visualized using an enhanced 253 chemiluminescence system, and quantified by densitometric analysis (Image J software, NIH).

254
All experiments were performed independently at least three times.   For thioflavin-S(ThS) staining, the sliced brains were stained with filtered 1% aqueous 283 Thioflavin-S solution for 8 min. The sections were then rinsed with 80%, 95% ethanol and 284 three washes with distilled water. Afterward, brain slices were mounted and slides allowed to 285 dry in the dark overnight. Images were taken on a Leica fluorescence microscope.    The test was done for 3 days. The first day adapts the mouse for 5 minutes in a bright zone.

345
The second day is the training phase. The study consists of two steps. The first step places each 346 mouse in the light zone which is then moved to the dark zone twice. One hour after the first 347 step, each mouse is placed in the light compartment. The door separating the two compartments 348 was opened 30 seconds later and after mice enter the dark compartment, the door was closed 349 and an electrical foot shock (0.3 mA/10 g) was delivered through the grid floor for 3 seconds.

350
If the mouse does not go into the dark zone for more than 5 minutes, it is considered to have 351 learned avoidance, and the training was done up to 5 times. Twenty-four hours after the training 352 trial, mice were placed in the light chamber for testing. Latency was defined as the time it took 353 for a mouse to enter the dark chamber after the door separating the two compartments opened.  All data are presented as the mean ± SD. NGS data were analyzed using R (version 3.5.2). Previous studies reported that SIRT1 levels were reduced in brains of human AD patients 382 and this reduction affected AD progression from early to late stages [28,29]. We hypothesized 383 that increasing SIRT1 might have prophylactic effects that could alleviate AD. For this, we first 384 examined levels of SIRT1 protein in AD patient postmortem brain (precentral gyrus) samples. 385 We found that SIRT1 protein levels were notably reduced in AD patient brains compared to 386 normal human brains (Fig. 1a, b and Additional file 1: Table S1). To confirm these results, we 387 obtained brain frontal cortex samples from 6 mo-old 5XFAD mice and 11 mo-old 5XFAD mice, 388 then analyzed the expression levels of SIRT1 using next generation sequencing (NGS).

389
Interestingly, we observed there was no difference in SIRT1 expression between 6 mo-old 390 5XFAD and WT mice, while it was markedly decreased in 11-mo-old 5XFAD mice compared 391 with WT mice (Fig 1c). SIRT1 expression was gradually reduced as the 5XFAD aged mice 392 (Fig. 1d). Taken together, these findings suggest that SIRT1 expression is down-regulated in 393 AD, and imply that SIRT1 may play a role in AD pathogenesis. We found that miR485-3p was increased in AD frontal cortex (Additional file 2: Figure S1a).

400
Consistent with this, miR485-3p was significantly higher in AD precentral gyrus tissue than in 401 normal tissue while miR485-5p was not significantly different between AD and normal 402 19 (Additional file 2: Figure S1b). Surprisingly, across the list of SIRT1 related miRNAs, the only 403 one upregulated in AD patients was miR485-3p (Additional file 2: Figure S1c). Based on three 404 public prediction algorithms, including targetscan, miRDB and miRbase, miR-485-3p has a 405 binding site in the 3`UTR of SIRT1. To evaluate the prediction, we first investigated whether 406 human miR485-3p mimic and ASO regulates miR485-3p expression in the mouse. Realtime 407 PCR showed that the expression of miR485-3p in mouse primary cortical neurons was 408 markedly reduced by transfection with human miR485-3p ASO (Additional file 2: Figure S1d). 409 We next transfected miR-control, human miR485-3p or miR485-3p ASO into mouse primary 410 cortical neurons, and measured the expression of SIRT1. SIRT1 protein was reduced in 411 miR485-3p transfected primary cortical neurons compared to miR-control transfected neurons 412 (Fig. 2a, b). In contrast, miR485-3p ASO transfection elevated SIRT1 expression (Fig 2a, b). 413 Prior studies reported that SIRT1 regulated PGC-1 transcription in C2C12 myoblasts [30] and containing either wild-type or mutated sequence of the potential miR485-3p site (Fig. 2c). We 421 transfected HEK293T cells with either wild-type or mutant sequence luciferase reporter, and 422 measured promoter activity. As expected, wild type promoter activity was significantly reduced 423 but the mutant form was not different in miR485-3p transfected cells (Fig. 2d). In addition, we 424 identified the physical binding of miR485-3p to the 3' UTR of SIRT1 by in vitro binding assay.

435
Since we found SIRT1 is up-regulated in neurons exposed to miR485-3p ASO, we explored 436 whether human miR485-3p ASO exerts beneficial effects on the pathology of AD as judged by 437 amyloid plaque formation and insoluble A levels in 10 mo-old 5XFAD. 5XFAD transgenic 438 mice show amyloid plaque deposition starting at 2 months [32]. Aβ in brain, aggregates into injection, and the injections were repeated at day 7 (Fig. 3a). The efficacy of miR485-3p ASO 442 in brain was confirmed using qRT-PCR and its functional outcomes assessed by miR485-3p 443 expression (Additional file 4: Figure S3). We next investigated the effect of a miR485-3p ASO 444 in 10-mo-old 5XFAD using 6E10 staining and thioflavin S. The number of amyloid plaques 445 was markedly decreased in 5XFAD that were given 1.5g intracerebroventricular injections, 446 compared to control miR-injected mice, suggesting that miR485-3p ASO may have 447 ameliorated the amyloid burden (Fig. 3b, c, Additional file 5: Figure S4). To further investigate 448 the effect of the miR485-3p ASO on A production, we examined the level of insoluble A 1- -injected 5XFAD. However, miR485-3p ASO did not significantly affect total APP levels ( Fig   454   3f, g). Furthermore, the brain levels of -CTFs and sAPP, the main products of BACE were 455 markedly reduced in 5XFAD mice that received miR485-3p ASO (Fig 3f, g). In complete 456 agreement, miR485-3p ASO injection in the brain of 5XFAD mice also significantly decreased 457 BACE1 protein levels compared with control mice (Fig 3f, g). Interestingly, miR485-3p ASO

458
-injected 5XFAD mice markedly increased Adam10 protein and sAPP protein expression (Fig   459   3f, g). In addition, administration of miR485-3p ASO significantly reduced the levels of SIRT1 460 and PGC-1 protein in the same situation. Together, these results indicate that the miR485-3p 461 ASO triggers a strong reduction in both A production and plaque formation in vivo.  (Fig. 4b). To further assess the phagocytic effect of miR485-3p ASO in glia, we  higher in miR485-3p ASO injected 5XFAD (Fig. 4e). In addition, CD36 expression was 491 meaningfully higher in Iba-1-positive microglial cells using immunohistochemistry 492 (Additional file 6: Figure S5e). Based on the above observations, we next examined whether 493 CD36 expression in mouse primary glial cells could be altered by miR485-3p or miR485-3p 494 ASO transfection. As shown in Fig. 4f, CD36 expression was markedly decreased in miR485-495 3p transfected primary glial cell compared to miR-controls. In contrast, miR485-3p ASO 496 transfection elevated CD36 expression. 497 We examined miR-485-3p targeting within the 3`-UTR of CD36 by constructing reporter 498 plasmids containing either wild-type or mutant forms of the potential miR485-3p target site. 499 We transfected HEK293T cells with luciferase reporter plasmids, and measured promoter 500 activity. Wild type promoter activity was significantly reduced but the mutant form was not 501 different in miR485-3p transfected cells (Fig. 4g). In addition, we identified a direct interaction 502 between miR485-3p and the 3' UTR of CD36 by in vitro binding assay. Relative binding 503 efficiency was significantly reduced in 3' UTR-containing mutant seed sequences (Fig. 4h) transfected cells compared to control transfected cells, but this effect was significantly 510 abrogated in cells treated with CD36 blocking antibody (Fig. 4i). These results clearly support 511 the conclusion that CD36 expression is regulated in a miR485-3p dependent manner and affects 512 A phagocytosis.

515
Studies have shown that inflammation is up-regulated in the brains of AD patients and 516 transgenic mice with AD-like pathology and inflammatory mediators are secreted by fA-517 stimulated-glia, where they contribute to neuronal loss and cognitive decline [42,43]. in fA treated primary glial cells, but this reduction was significantly recovered in miR485-3p 524 ASO transfected cells (Fig. 5a, b). SIRT1 is well known to reduce inflammation by regulation 525 of NF-B expression [45][46][47][48][49]. Accordingly, we investigated NF-B expression in primary glial 526 cells. The cells were transfected with miR485-3p ASO, and incubated under 1M fA for 6 h.

527
As shown in Fig. 5a and 5b, NF-B was considerably increased in control transfected cells  Figure S6b). NF-B expression was also markedly reduced in miR485-3p ASO-injected 538 5XFAD mice. Importantly, the expression of the pro-inflammatory cytokines, TNF- and IL-539 1 were lower in the miR485-3p ASO injected mice (Fig. 5c, d). We next investigated the effect 540 of the miR485-3p ASO on anti-inflammation responses using immunohistochemistry.

541
Immunohistochemistry showed that TNF- and IL-1 in glial cells was markedly attenuated 542 in miR485-3p ASO injected 5XFAD (Fig. 5e, f, Additional file 7: Figure S6c, d). These findings 543 strongly suggest that knocking down miR485-3p affects the activation of glial cells and reduces 544 proinflammatory cytokine production via regulation of SIRT1/NF-B signaling.

545
MiR485-3p ASO prevents neuronal loss and behavior deficits in 5XFAD mice 546 5XFAD transgenic mice show neuronal loss in cortical layer V at 9 months [32]. Previous 547 studies report that synaptic and neuronal loss in 5XFAD is correlated with A accumulation 548 and neuroinflammation [32,50,51]. Given our observation that SIRT1 and CD36 expression 549 through miR485-3p ASO controls A processing, phagocytosis and inflammation in 5XFAD 550 mice, we investigated the effect of the antisense inhibitor on neuronal cell death by assessing 551 NeuN (a neuronal cell marker) and cleaved caspase-3. Western blot revealed that the expression 552 of NeuN was increased in the miR485-3p ASO treated 5XFAD cortical region, whereas protein 553 expression of cleaved caspase-3 was reduced. However this was not seen in hippocampus under 554 the same conditions (Fig 6a, b). As well, immunohistochemistry showed the expression of 555 cleaved caspase-3 in neuronal cells was markedly elevated in control treated 5XFAD, whereas 556 this elevation was dramatically attenuated in miR485-3p ASO treated 5XFAD (Fig 6c, d). We indicating that levels of general motor and exploratory activity in the Y-maze were not changed 567 (Fig. 6g). In addition, we examined associative memory in the passive avoidance task, based 568 on the association formed between an electrical foot shock and a spontaneously preferred 569 26 specific environmental context (darkness vs light).
Step-through latency was similar between 570 control and miR485-3p ASO treated 5XFAD. However, miR485-3p ASO mice showed a 571 significant reduction in the latency to spend time the dark compartment 24hr after receiving an 572 electrical shock (Fig. 6h). In sum, these findings suggest that the miR485-3p/SIRT1 axis may silico finding, we confirmed that the expression of miR485-3p was increased in AD patient 624 postmortem tissue using realtime PCR. Interestingly, only miRNA485-3p emerged as a 625 candidate after we had combined miRNA profiles that specifically regulate SIRT1 and overall 626 increased miRNAs in AD patients.

627
The above results raised the possibility that miR485-3p might play important roles in AD 628 pathology through the regulation of SIRT1. Our data identified that miR485-3p binds to the 629 3`UTR region of SIRT1 and regulated its protein expression in vitro. Moreover, frontal cortex 630 obtained from miR485-3p-inhibitor injected 10-month old 5XFAD mice showed a reduced 631 amyloid burden as well as lower APP processing enzymes. In contrast, SIRT1 expression 632 increased in miR485-3p-inhibitor injected 5XFAD. These findings suggest that SIRT1 acts as 633 a downstream mediator of miR485-3p that may influence the production of A.

634
Brain resident glial cells, which are major immune cells in the brain, gather around Aβ 635 accumulations and remove Athrough phagocytosis [36]. in improvements in memory deficits in 5XFAD mice (Fig. 7). Accordingly, miR485-3p ASO 666 may contribute to the future development of efficient therapeutic approaches against AD. The data generated during this study is available upon request.