Receptor tyrosine kinase ROR1 ameliorates Aβ1–42 induced cytoskeletal instability and is regulated by the miR146a-NEAT1 nexus in Alzheimer’s disease

Alzheimer’s disease (AD) involves severe cytoskeletal degradation and microtubule disruption. Here, we studied the altered dynamics of ROR1, a Receptor Tyrosine Kinase (RTK), and how it could counter these abnormalities. We found that in an Aβ1–42 treated cell model of AD, ROR1 was significantly decreased. Over expressed ROR1 led to the abrogation of cytoskeletal protein degradation, even in the presence of Aβ1–42, preserved the actin network, altered actin dynamics and promoted neuritogenesis. Bioinformatically predicted miRNAs hsa-miR-146a and 34a were strongly up regulated in the cell model and their over expression repressed ROR1. LncRNA NEAT1, an interactor of these miRNAs, was elevated in mice AD brain and cell model concordantly. RNA Immunoprecipitation confirmed a physical interaction between the miRNAs and NEAT1. Intuitively, a transient knock down of NEAT1 increased their levels. To our knowledge, this is the first instance which implicates ROR1 in AD and proposes its role in preserving the cytoskeleton. The signalling modalities are uniquely analyzed from the regulatory perspectives with miR-146a and miR-34a repressing ROR1 and in turn getting regulated by NEAT1.


Results
and key cytoskeletal proteins are deregulated in Aβ 1-42 treated cell model compromising the cytoskeletal architecture. To begin with, we looked at the deregulated levels of ROR1, both at the transcript and protein levels in SHSY-5Y cells treated with Aβ 1-42 and compared with DMSO control (considered as 1). In order to check if Aβ 1-42 elicited cell cytotoxicity and death, we performed cell viability assays. Both MTT and Trypan Blue experiments showed that the Aβ 1-42 concentration used here was sufficient to reduce the SHSY-5Y population by nearly 50% (Supplementary Fig. 1). Subsequently, both mRNA (fold change 0.43 Fig. 1a) and protein levels (fold change 0.37 Fig. 1b,c) showed that treatment with Aβ 1-42 elicited downregulation of ROR1. Owing to ROR1's association with cytoskeleton, we wanted to see if our cell model showed deregulation of cytoskeletal representative proteins, namely α-Tubulin (microtubule), Smooth Muscle Actin (SMA) (intermediate filament) and Vimentin (microfilament). On exposure to Aβ  , the levels of α-Tubulin (fold change 0.33), SMA (fold change 0.59) and Vimentin (fold change 0.62) decreased significantly (Fig. 1d,e). The same treatment was also sufficient to show visible phenotypic changes in the actin network of cells (assayed by phalloidin staining). In comparison to DMSO control, Aβ 1-42 exposure led to marked disruption of the mesh like actin assembly in cell clusters (Fig. 1f, panels i, iii). Higher magnification images showed in more detail that the fibril like actin mesh (DMSO) (Fig. 1f, ii) was absent in the Aβ 1-42 cells (Fig. 1f, iv), in which the actin were mostly present in punctate clusters. www.nature.com/scientificreports/ ROR1 over expression abrogates Aβ 1-42 induced degradation of cytoskeletal components. ROR1 having been down regulated in the study model, the next logical approach would be to see if ROR1 over expression produced significant phenotypic changes. Fluorescent confocal microscopy, 24 h post transfection with a ROR1-GFP Spark clone in SHSY-5Y cells, showed its sub-cellular distribution and marked alterations in the cellular structure (Fig. 2a., panels i, iii). A transient over expression of ROR1 led to the generation of multiple neurites in cells (marked with white arrows) limited to the cell terminals. Super-resolution microscopic images showed that in dividing cells, ROR1 was distinctly enriched in the cytokinetic bridge (Fig. 2a, panel ii and inset) and probably, the terminally located MTOCs (Fig. 2a, panel iv and inset, marked by white arrows). Following the observation that a transient increase in ROR1 promoted neurite generation, we could further show that ROR1 over expression prior to Aβ 1-42 treatment hindered the cleavage of MAP2, indirectly indicating that ROR1 helped preserve the microtubule network (Fig. 2b,c). Similar changes were also observed in the SMA and Vimentin levels (Fig. 2c,d), but Vinculin did not show any significant recovery.
Over expressed ROR1 promotes neurite elongation in presence of Aβ  . We next investigated if the aberrant neurite generation due to ROR1 over expression could also occur on treatment of Aβ  , and if so, then how it would affect the cellular architecture. The following experiments were performed in the background of ROR1 over expression only, and additional controls like empty GFP were deemed counterintuitive. We found that a transient increase of ROR1 and Aβ 1-42 led to an increase of neurites (Fig. 3a, panel i, ii and inset). However, unlike only ROR1 over expression, here, the neurites were significantly elongated in length, but less in number (Fig. 3b). In order to a get a better estimate of the neurite dynamics, we measured the ratio of neurite length: neurite number for the ROR1, Aβ 1-42 and ROR1 + Aβ 1-42 group, which showed that the ratio was highest ROR1 + Aβ 1-42 , followed by ROR1 and then only Aβ 1-42 ( Supplementary Fig. 2). Another interesting observation was that the elongated neurites were directed towards juxtaposed cells where they made contacts (marked by white arrows) and ROR1 was specifically enriched in the neurite terminals ( Fig. 3a, panel ii inset). In order to better understand how ROR1 itself, or in conjunction with Aβ 1-42, were affecting the cytoskeletal dynamics, we performed a Filamentous: Globular (F: G) actin assay. On exposure to Jasplakinolide (actin stabiliser) the F: G ratio was > 1 (compared to DMSO control) (Fig. 3c, d(i)). Cytochalasin D (actin depolymeriser) had the reverse effect. Treatment with Aβ 1-42 markedly decreased the ratio, but on prior increase with ROR1 followed by Aβ 1-42 , there was a strong enrichment of filamentous actin (or inhibition of actin depolymerisation) which led to a subsequent increase of the F: G Actin ratio (Fig. 3d (ii). for the control experiments and quantitation in Supplementary Fig. 3).

Hsa-miR-146a-5p and 34a-5p are up regulated by Aβ 1-42 and target ROR1 and Vimentin.
To gain a mechanistic insight into the trigger of ROR1 deregulation in Aβ 1-42 treated cells, we pursued the regulatory RNA-protein network model and looked for miRNA interacting (and preferably repressing) components of ROR1 network using ENCORI (http:// starb ase. sysu. edu. cn/ index. php). The miRNA-mRNA tool was used with the following parameters-predicting program-5, miRNA-mRNA with Pan-Cancer analysis-2 and stringency of CLIP data-3. With these attributes, ROR1 was predicted to interact with hsa-miR-146a-5p and hsa-miR-34a-5p ( Supplementary Fig. 4). The same bioinformatics search predicted that that these two miRNAs also targeted a cytoskeletal protein of our interest-Vimentin. qRT-PCR, with primers designed against the human mature miRNA sequences showed that hsa-miR-146a was more abundant compared to hsa-miR-34a, (normalised against control U6snRNA) ( Supplementary Fig. 5). Both the miRNAs were strongly and significantly up-regulated in the Aβ 1-42 treated cell model (Fig. 4a). Subsequent assays using AD transgenic mice brain tissues revealed almost the same patterns of up regulation, although here, the fold change of increase of hsa-miR-146a was much greater than hsa-miR-34a (Fig. 4b). In order to validate the bioinformatics prediction, we transiently over expressed both the miRNAs individually (using miRNA clones in pMIR vector) and then looked at the transcript levels of ROR1, and indeed both of them targeted and strongly repressed ROR1 levels ( Fig. 4c), although the effect of hsa-miR-146a-5p was more pronounced. These two ROR1 targeting miRNAs also targeted and repressed Vimentin (Fig. 4d), validating the prediction data. Combining both the results, we found that hsa-miR-146a-5p was the stronger common repressor of both these proteins. Looking at the effect of these two miRNAs on cytoskeletal proteins, we posited that they would be involved in neurological processes which are governed by such components. Hence, we performed a Gene Enrichment analysis with the help of DIANA tools (miRPath module). Intuitively, GSEA revealed that both hsa-miR-146a and hsa-miR-34a were involved in core neurological pathways like Long Term Potentiation, Wnt signalling, Insulin signalling and Mapk signalling pathways (Fig. 4e). However, they were more enriched in the processes like-Regulation of actin cytoskeleton, Neurotrophin signalling and axon guidance, all of which were deregulated in AD. In this analysis too, hsa-miR-146a showed a stronger enrichment compared to hsa-miR-34a (Fig. 4e).
LncRNA NEAT1 exerts a protective effect by repressing miR146a and miR-34a. Continuing with the ncRNA regulatory networks governing ROR1, we introduced another layer of complexity. We used the ENCORI database to look for the potential lncRNA interactors of hsa-miR-146a-5p and hsa-miR-34a-5p. We employed the miRNA-lncRNA tool which had data from Ago-CLIP seq experiments and predictive data from miRanda, with the search parameters-CLIP data, high stringency (≥ 3) and Degradome data, high stringency (≥ 3). Following these search criteria, hsa-miR-146a-5p and hsa-miR-34a-5p were predicted to interact with a lncRNA -NEAT1 ( Supplementary Fig. 6). In order to implicate NEAT1 in AD, we first looked for its deregulation in our cell model and mice model, with human and mouse primers, respectively. NEAT1 transcript levels were up regulated in both disease models (Fig. 5a), but the mice model showed a stronger increase, probably hinting at the fact that the transgenic mice AD model better mimicked late stage AD. Having ascertained that    Graph depicting three (n = 3) independent biological replicates quantifying levels of hsa-mir-146a and hsa-mir-34a by qRT-PCR in transgenic AD mice or age matched wild type mice brain tissues. (c) Graph depicting three (n = 3) independent biological replicates quantifying levels of ROR1 by qRT-PCR in SHSY-5Y cells treated with hsa-mir-146a-5p and hsa-mir-34a-5p pMIR clones or corresponding empty vector controls. (d) Graph depicting three (n = 3) independent biological replicates quantifying levels of Vimentin by qRT-PCR in SHSY-5Y cells treated with hsa-mir-146a-5p and hsa-mir-34a-5p pMIR clones or corresponding empty vector controls. Levels of U6snRNA were taken as endogenous control for the miRNAs and levels of GAPDH were taken as endogenous control for the mRNA levels. The levels of individual miRNAs or mRNA were normalized by the corresponding U6snRNA or GAPDH levels. Fold changes were computed by considering the relative levels of lncRNA in corresponding controls to be 1. Error bars indicate ± SD. Significance level between different experimental pairs is shown (NS, not significant; *p < 0.05; **p < 0.01; ***p < 0.001). (e) KEGG analysis of the 2 de regulated miRNAs; bar graphs indicate Log (p value) and ranges from (-)10 to 0. Pathways deregulated in AD are marked with an asterisk.  Error bars indicate ± SD. Significance level between different experimental pairs is shown (NS, not significant; *p < 0.05; **p < 0.01; ***p < 0.001). www.nature.com/scientificreports/ NEAT1 indeed was deregulated in AD, to validate the prediction data, we transiently knocked down NEAT1 levels (using siRNA) and checked for the down regulation ( Supplementary Fig. 7). We probed for the levels of hsa-miR-146a and hsa-miR-34a after NEAT1 knock down. Compared to a negative control siRNA, treatment with NEAT1 siRNA led to a concomitant increase of both the miRNAs, with hsa-miR-146a showing a higher increase (Fig. 5b). Conversely, we also tested if this putative interaction and suppression was bi-directional. A transient over-expression of the mature miRNA clones in cells (Fig. 5c) failed to elicit a response in the NEAT1 levels. We employed combined Immunocytochemistry (ICC) plus RNA -Fluorescence In Situ Hybridisation (RNA FISH), and RNA Immuno Precipitation (RIP) to study lncRNA-miRNA interaction. NEAT1 lncRNA was observed in nuclear locations different from that of the DNA marker, in cell populations (Fig. 5d, panel i). A higher magnification image (Fig. 5d, panel ii) showed its distinct distribution in defined spots called nuclear paraspeckles. Further, NEAT1 was predicted to interact with a RNA Binding protein (RBP) FUS using the lncRNA-RBP tool from ENCORI ( Supplementary Fig. 8). Combined ICC of FUS with RNA FISH (Fig. 5d, panel iii), using NEAT1 specific probes, showed a strong overlap between the two in the cell nucleus. From the theoretical prediction and co-localisation analysis, we next designed a RIP experiment using FUS as the bait. Compared to control IgG, FUS pull down from cell lysates and subsequent assay by qRT-PCR showed a strong enrichment of NEAT1 (Fig. 5e). A reanalysis using mature miRNA specific probes from the same FUS pull down RNA also subsequently showed a clear enrichment of the NEAT1 interacting miRNAs-hsa-miR-146a and hsa-miR-34a.

Scientific Reports
In the RIP assay, hsa-miR-146a showed near double enrichment compared to hsa-miR-34a, indicating that the former had a stronger interaction with NEAT1. In order to validate that the effect of NEAT1 knock-down was not just restricted to the miRNA levels, but their target ROR1 as well, we looked at the transcript levels of ROR1 after NEAT1 silencing (Fig. 5f) and indeed, ROR1 levels went down significantly on transient NEAT1 suppression, thereby confirming the hypothesis that NEAT1, its interacting miRNAs and their target ROR1, essentially constitute a single entity (Fig. 6).

Discussion
In this study, we focus on ROR1 with the motivation that cytoskeleton disruption in AD due to Aβ 1-42 is a well recognised hallmark [5][6][7][8][9] and we could establish the same through biochemical assays and confocal imaging. Further, in recent times, microtubule associated ROR1 has been implicated in reinforcement of neuronal network [15][16][17][18] , which we find to be true on ROR1 over expression and subsequent neuritogenesis with the caveat that AD involves significant disruption of the same. ROR1 was specifically localised in the cytokinetic bridge www.nature.com/scientificreports/ and MTOCs. However, use of specific microtubule markers would confirm this hypothesis. But, as we wanted to look at the effects of ROR1 in AD (which affects post-mitotic neurons), this line of investigation was not pursued. Intuitively, we find ROR1 levels to decrease in our AD model. The same transient over expression of ROR1 in presence of Aβ 1-42 is found to be necessary and sufficient to hamper cytoskeletal degradation of key proteins, promote neuritogenesis and drastically alter the F: G actin dynamics. In search for the small molecule regulators of ROR1, we could identify two miRNAs-miR-146a and miR-34a, which were theoretically predicted to target ROR1. Subsequent validation in our cell model and transgenic mice AD model revealed significant up regulation of both. Using mature miRNA clones, we substantiated the hypothesis that both miR-146a and miR-34a targeted and repressed ROR1 levels in cells, miR-146a being the stronger repressor. Fortuitously, both of these also targeted Vimentin, a cytoskeletal protein of importance in AD. It was surprising that the repression of Vimentin was in fact stronger than ROR1, which leads us to believe that up regulation of these two miRNAs cumulatively affects the cytoskeleton disruption in AD by dual repression of ROR1 and Vimentin. It is however important to note that the decrease of ROR1 levels might also act due to a transcriptional repression pathway and not just by repression by miRNAs, which would be part of a future study. It is not surprising, therefore, to find that miR-146a and miR-34a are parts of core neurobiological pathways implicated in AD, like LTP, axon guidance and regulation of actin cytoskeleton. These novel results are also backed up by literature reports that show miR-146a and miR-34a govern the regulators of actin pathways, namely RhoA and ROCK1 [22][23][24] . To further understand how these miRNAs were themselves regulated, we deciphered their interaction with the lncRNA NEAT1, which recently has been shown to be deregulated in a plethora of neurodegenerative scenarios [46][47][48][49] . Using the same AD cell and mice model, we validated the up regulation of NEAT1. The direct interaction of miR-146a and miR-34a with NEAT1 was characterised with subsequent transient knock down, RIP and combined ICC with RNA-FISH experiments. We could also show a direct repercussion of perturbation of the NEAT1 level on ROR1 transcript levels, completing the proposed RTK-miRNA-lncRNA regulatory loop. Such repressive effects of lncRNAs on miRNAs have been shown recently in colorectal cancers 52 , pancreatic cancer 53 and cardiomycetes apoptosis 54 . Moreover, it was also found that the protein component of the miRNA machinery, Ago2, co-localised strongly with the NEAT1 interactor FUS in distinct nuclear clusters ( Supplementary Fig. 9, Panels i and ii) pointing to the fact that the miRNA-lncRNA machinery possibly interact in a closed loop. Such instances of the nuclear shuttling of Ago2 have also been reported 55,56 .
Although there is evidence of the disruption of the cytoskeletal machinery in Alzheimer's Disease due to the Aβ 1-42 (and its effects on the proteins α-Tubulin, Vimentin and SMA) 57,58 , to our cognizance, this is the first consolidated network study to undisputedly connect ROR1 to Aβ 1-42 treatment in AD, by showing its deregulation both at the transcript and protein levels. In-vivo over expression of ROR1 exerts protective effects on the gross cytoskeletal assembly and neurite formation. A functional link between ROR1 and its targeting miRNAs is established. Eventually, we could also show a regulatory paradigm of ROR1-miRNA 146a/34a-NEAT1 in AD.

Conclusion
To summarise, effect of Aβ 1-42 on cells were diverse (Fig. 6). First, Aβ 1-42 deregulated the expression of RTK ROR1 and relevant cytoskeleton associated components. Second, cytosolic Aβ 1-42 affected the mature miRNAs-miR-146a and miR-34a (through a hitherto unknown mechanism), which in turn repressed ROR1 and finally, nuclear Aβ 1-42 differentially regulated NEAT1, which in turn regulated the miRNAs. Comprehending this RTK-miRNA-lncRNA network promises to initiate further studies involving other RTKs as potential therapeutic targets in abrogating AD pathophysiology.

Materials and methods
Ethics statement. All animal experiments were conducted following the institutional guidelines for the use and care of animals and approved by the Institutional Animal and Ethics Committee of the National Brain Research Centre (NBRC/IAEC/2012/71) and carried out in compliance with the ARRIVE guidelines.
Cell culture and transfection. Human neuroblastoma cell lines SHSY-5Y were procured from NCCS, Pune, India and were cultured routinely in DMEM-F12 (Gibco) supplemented with 10% (v/v) heat-inactivated FBS (Gibco), antibiotics penicillin/streptomycin PS 1% (v/v) and 400 μg/ml G418 (Invitrogen, USA) at 33 °C in humidified condition and 5% CO 2 . All transfections were carried on 70-80% confluent cells using Lipofectamine 2000 (Invitrogen) as per manufacturer's protocol. Unless otherwise mentioned, for single transfection experiment 1 µg (30 mm plate), 2.5 μg (60 mm plate) or 5 μg (100 mm plate) of plasmid DNA constructs as well as 5 μl, 10 μl or 15 µl of Lipofectamine 2000 respectively were used. Normalisation of transfection was performed using pEGFP-C1 (Clontech) and using the same protocol above followed by quantification of GFP positive cells using a microscope.  www.nature.com/scientificreports/ of 1 µM and the cells were exposed to this Aβ 1-42 for 24 h. Only DMSO in the same concentration and amount was used as a control.

Constructs, reagents and siRNAs. Constructs
APP/PS1 mice. Trangenic AD mice (APP/PS1 or B6C3-Tg (/APPswe, PSEN1dE9/) 85Dbo/J) were procured from the Jackson Lab. AD transgenic mice have human APPswe mutations (at positions K670N and M671L) and human presenilin gene with exon 9 deletion 1(PSEN1dE9) under a mouse prion gene promoter. Mice were supplied with water and food as often as necessary. AD mice, along with controls at their age of 12 months, were anaesthetized with xylazine (10 mg/kg body weight) and ketamine (100 mg/kg body weight) and perfused transcardially with PBS followed by 4% paraformaldehyde (w/v) in PBS. Brains were collected and further placed in 4% paraformaldehyde for 24 h and then treated with 10, 20 and 30% sucrose (in PBS) followed by sectioning in a cryo-microtome (20 μm thickness).

Isolation of RNA from cells and FFPE tissue.
Total RNA was extracted following manufacturer's protocol using TriZol Reagent (Invitrogen, USA). We isolated RNA from paraffin-embedded tissue samples of AD mice; along with controls. In brief, isolation method for RNA from paraffin-embedded tissues consists of the following steps: De-paraffinization: For RNA extraction from tissue sections obtained from AD mice or controls, two sections (biological replicates, 20 µm thick) were put in 1.5 ml tubes, deparaffinized by double xylene washes (5 min each) followed by two centrifugations at room temperature (10,000 g, 10 min). Rehydration: The supernatant was discarded and the pellets were washed with absolute ethanol (1 ml) and 95% ethanol (1 ml) in DEPC water (successively). Following each step, the tissue was pelleted (10,000 g, 10 min). For digesting tissue proteins, following the final wash, alcohol was decanted; the pellets were dried in a drybath at 37 °C and put in 500 µl of digestion buffer (10 mMNaCl, 500 mMTris, pH 7.6, 20 mM EDTA and 1% SDS). Tissue proteins were removed using proteinase K (500 µg/ml) followed by incubation at 45 °C (16-18 h). Before RNA isolation, proteinase K was deactivated at 100 °C (7 min Quantitative real-time PCR. 2 μg total RNA was treated by DNase (Sigma) followed by cDNA preparation by primers (oligo dT or random hexamers), dNTPs and Reverse transcriptase (Fermentas). qRT-PCR was performed using Sybr green 2X Universal PCR Master Mix (ABI) in StepOne Real-time PCR system (ABI). For each gene, NTC was used at the same condition to determine the baseline and threshold value. Corresponding C t values were used for the relative quantification (fold change) of a target gene in a sample compared to the parental cell is expressed in terms of 2 −ΔΔCt values after normalizing w.r.t. house-keeping gene (internal control).
Gene-specific primers. PCR primers designed for and employed in this study are listed in Supplementary RNA immunoprecipitation (RIP) assay. RNA Immunoprecipitation was performed on fixed cells (4% formaldehyde) following the Abcam RIP protocol (https:// www. abcam. com/ proto cols/ RIP) following the manufacturer's instruction with modifications. SHSY5Y cells were harvested by trypsinization and resuspended in PBS, freshly prepared nuclear isolation buffer (1.28 M sucrose, 40 mM Tris-HCl, pH 7.5, 20 mM MgCl 2 , 4% Triton X-100) and water, and kept on ice for 20 min with frequent mixing. Next, a centrifugation step (2,500 G, 15 min) was performed to pellet the nuclei. Then the nuclear pellet was resuspended in freshly prepared RIP buffer (150 mM KCl, 25 mM Tris pH 7.4, 5 mM EDTA, 0.5 mM DTT, 0.5% NP40, 100 U/ml RNAase inhibitor, Protease inhibitors). The nuclei fraction was sonicated with the following parameters-30% amplitude-10 s-1 min gap, and the process was repeated 4 times. Following this, the solution was nutated at 20 rpm for 90 min at 4 °C. After nutation, the lysate was centrifuged at 12,000 RCF for 20 min at 4 °C. The pellet was discarded and the supernatant was used for protein estimation using Bradford reagent. 5 mg of total protein in RIP buffer was used for each sample to which 3 μg of antibody (FUS or IgG) was added and incubated at 4 °C with gentle rotation overnight. After that, to each tube 40 μl protein A/G beads were added and incubated for 2 h at 4 °C with gentle rotation. Then, beads were pelleted by centrifugation at 2,500 rpm for 30 s, the supernatant was removed, and the beads resuspended in 500 μl RIP buffer. Beads were washed for a total of three RIP washes, followed by one wash in PBS. The co precipitated RNAs were isolated by resuspending beads in TRIzol RNA extraction reagent (1 ml) according to manufacturer's instructions. RNA was collected in 15 μl DEPC water. The total RNA was used to make cDNA with Random Hexamer or miRNA specific stem loop primers. The following steps were same as the Quantitative Real-time PCR (qRT-PCR) protocol stated above. www.nature.com/scientificreports/ night. Following this, cells were inverted again in culture plate and washed with 1 ml of 1X PBS for 10 min, and repeated 2 more times. Next, 1 ml secondary antibody (1:300) in 1X PBS was added and kept at room temperature (2 h, in the dark). Again they were washed with 1 ml of 1X PBS for 10 min, and repeated 2 more times. Then, 1 ml of fixation buffer was added and incubated at room temperature for 10 min followed by two washes 1 ml of 1X PBS. The 1X PBS was aspirated off the cover glass containing adherent cells within the 35  The cell lysate was separated on SDS gel according to molecular weight then it was transferred to PVDF membrane (Millipore Corporation) which was blocked by 5% skimmed milk in TBST (50 mMTris-HCl, 150 mM NaCl, pH 7.5 containing 0.05% Tween 20). After that membrane was probed with primary antibody, followed by the incubation with HRP conjugated secondary antibody. The membranes were then developed with ECL kit (Pierce or Abcam). Band intensities were measured by Quantity One (Bio-Rad). Experiments were repeated thrice. Significance testing (p values) was performed by unpaired t-test. Immunocytochemistry. Immunocytochemistry was performed on fixed cells following the abcam ICC protocol (https:// www. abcam. com/ proto cols/ immun ocyto chemi stry-immun ofluo resce nce-proto col) following the manufacturer's instruction with slight modifications. Briefly, cells were fixed using 4% paraformaldehyde in PBS pH 7.4 for 12 min at room temperature. This was followed by cell washes in chilled PBS (thrice). Cells were then permeabilized with 0.1-0.25% Triton X-100 for 10 min at room temperature followed by PBS washes for three times for 5 min. Cells were then blocked with 1% BSA, 22.52 mg/ml glycine in PBST (PBS + 0.1% Tween 20) for 30 min to block unspecific binding of the antibodies. Then, the cells were incubated with diluted primary antibody in 1% BSA in PBST in a humidified chamber for overnight at 4 °C. The solution was decanted and the cells washed three times in PBS, 5 min each wash. Cells were then incubated with the secondary antibody in 1% BSA for 1 h at room temperature in the dark, following which the secondary solution was drained and cells washed in PBS (thrice, 5 min each in the dark). Finally, the cells were incubated with 0.1-1 μg/ml DAPI (DNA stain) for 5 min. The DAPI solution was discarded and cells were rinsed twice with PBS. Coverslips were mounted on fresh, cleaned and dried slides with a drop of mounting medium and sealed with nail polish to prevent drying and movement under microscope. F/G actin assay. After appropriate treatments, cells were scrapped from the petri dishes and washed twice in PBS. Cells were then centrifuged at 800 RCF for 3 min at 4 °C. The cell pellets were then resuspended in 200 μl PBS with 0.1% Triton-X-100 (with protease inhibitors). After incubation for 15 min with slight agitation, cells were again centrifuged at 15,000 RCF at 4 °C for 5 min. The soluble supernatant (which contained G-actin) was separated and the Triton-X-100 insoluble pellet (predominantly F-actin) was resuspended in 200 μl RIPA buffer. The soluble and insoluble fractions were mixed with 5X Loading dye, heated at 98 °C for 10 min, equal volumes of the two fractions were loaded and separated on a 12% SDS gel using standard electrophoresis protocol. Actin levels were assayed using the pan actin antibody (Clone C4).

Antibodies.
Trypan blue exclusion assay. For the Trypan Blue Exclusion Assay, SHSY-5Ycells were seeded in 6 well clear bottom plates (Thermo). After the appropriate treatments, media was discarded from cell cultures and washed twice with PBS. Cells were trypsinised and resuspended in 1 ml fresh media (without FBS). Following www.nature.com/scientificreports/ this, the experiment was performed as per manufacturer's instruction. Cell viability was plotted as % viability compared to controls.
MTT assay. For the MTT cell viability assay, standard protocols were followed as per the manufacturer's instruction. Cell viability was plotted as % viability compared to controls.
Gene set enrichment analysis and interaction predictions. GSEA, miRNA target analysis and miRNA-lncRNA predictions were performed as outlined in 57 .
Statistical analysis. Statistical analysis and significance testing were performed as given in 57 .
Consent for publication. The authors declare that they give consent for publication with the contents of this article.