Ethanol Exposure Increases miR-140 in Extracellular Vesicles: Implications for Fetal Neural Stem Cell Proliferation and Maturation

Abstract

Background:

Neural stem cells (NSCs) generate most of the neurons of the adult brain in humans, during the mid-first through second trimester period. This critical neurogenic window is particularly vulnerable to prenatal alcohol exposure, which can result in diminished brain growth. Previous studies showed that ethanol exposure does not kill NSCs, but, rather, results in their depletion by influencing cell cycle kinetics and promoting aberrant maturation, in part, by altering NSC expression of key neurogenic miRNAs. NSCs reside in a complex microenvironment rich in extracellular vesicles, shown to traffic miRNA cargo between cells.

Methods:

We profiled the miRNA content of extracellular vesicles from control and ethanol-exposed ex vivo neurosphere cultures of fetal NSCs. We subsequently examined the effects of one ethanol-sensitive miRNA, miR-140–3p, on NSC growth, survival, and maturation.

Results:

Ethanol exposure significantly elevates levels of a subset of miRNAs in secreted extracellular vesicles. Overexpression of one of these elevated miRNAs, miR-140–3p and its passenger strand relative, miR-140–5p, significantly increased the proportion of S-phase cells while decreasing the proportion of G₀/G₁ cells compared to controls In contrast, while miR-140–3p knockdown had minimal effects on the proportion of cells in each phase of the cell cycle, knockdown of miR-140–5p significantly decreased the proportion of cells in G₂/M phase. Furthermore, miR-140–3p overexpression, during mitogen-withdrawal-induced NSC differentiation, favors astroglial maturation at the expense of neural and oligodendrocyte differentiation.

Conclusion:

Collectively, the dysregulated miRNA content of extracellular vesicles following ethanol exposure may result in aberrant neural progenitor cell growth and maturation, explaining brain growth deficits associated with prenatal alcohol exposure.

Introduction:

Prenatal alcohol exposure (PAE) can result in a cluster of craniofacial, neuro-cognitive, and growth deficits that are collectively termed Fetal Alcohol Spectrum Disorders (FASDs) (Streissguth and O’Malley, 2000). Despite published prevention guidelines (ACOG, 2011), PAE is common (Popova et al., 2017) and FASDs are thought to affect between 1.1 and 5% of school-aged children in the US (May et al., 2018), and may account for between 2 and 3% of all births, worldwide (Roozen et al., 2016). Self-reported PAE is particularly high during the 1^st and 2^nd trimesters of pregnancy (SAMHSA, 2013), due in part to the high prevalence of unplanned pregnancies in the US (Finer and Zolna, 2016). The period encompassing the mid-1^st trimester through the end of the 2^nd trimester, when inadvertent alcohol exposure is most likely to occur, is a particularly important period of fetal brain vulnerability, because, neural stem cells (NSCs) generate most of the neurons of the adult brain during this period (Bystron et al., 2008). Consequently, interfering with neurogenesis during this critical developmental window is likely to produce persistent deficits in brain function.

Previous research shows that PAE in preclinical models, during 1^st and 2^nd trimester-equivalent periods of brain development, decreases brain weight (Maier et al., 1999). However, while ethanol results in significant cell death in differentiating neurons (Cheema et al., 2000, McAlhany et al., 2000, Mooney and Miller, 2003), NSCs that give rise to these neurons are themselves resistant to cell death following ethanol exposure (Vangipuram and Lyman, 2010, Prock and Miranda, 2007, Santillano et al., 2005). Instead, we and others previously found that ethanol exposure results in increased cell proliferation and NSC growth, resulting in the depletion of NSCs in favor of more mature neural progenitor populations (Santillano et al., 2005, Vangipuram and Lyman, 2010, Miller, 1989, Miller and Nowakowski, 1991, Tingling et al., 2013, Camarillo and Miranda, 2008).

The mechanisms that mediate PAE’s effects on NSC growth and maturation are likely to be complex and remain to be elucidated. MicroRNAs (miRNAs), a class of small non-protein-coding RNAs that inhibit protein translation through their ability to guide the binding of RNA-induced silencing complexes (RISC) to target mRNAs (Miranda, 2014), have emerged as one promising candidate (as reviewed in (Mahnke et al., 2018)). Previous studies from our group and others have shown that certain miRNAs are responsive to ethanol exposure, and that these miRNAs control several ethanol-sensitive developmental processes, such as survival and proliferation in fetal NSCs, and may mediate the effects of PAE on early brain development (Sathyan et al., 2007, Tsai et al., 2014, Pappalardo-Carter et al., 2013, Tal et al., 2012).

Self-renewal and maturation within the fetal neuro-epithelial NSC niche may also be synchronized by paracrine signals that are vulnerable to ethanol exposure. Recent studies suggest that miRNAs are not only localized within the cell, but are also present in circulation and extracellular milieu (Pirola Carlos et al., 2012). Interestingly, cellular miRNAs may be packaged within small membrane-bound extracellular vesicles (EVs) for release (Sohel, 2016, Zhang et al., 2015). EVs are also known to contain protein, lipid, as well as RNAs, and may participate in both paracrine and endocrine intercellular communication, whereby EVs released from one cell fuse with, and release their content into, neighboring or distant cells without any direct cell-to-cell contact (Mathivanan et al., 2010, Osteikoetxea et al., 2015, Llorente et al., 2013, Subra et al., 2007, Taylor and Gercel-Taylor, 2013, Baglio et al., 2015, Lu and Risch, 2016). Currently, two of the heavily studied categories of extracellular vesicles are exosomes and microvesicles. Exosomes are EVs, in the size of 50–200 nanometers, generated from early endosomes that mature in the multi-vesicular bodies and are released to the extracellular space through exocytosis. In contrast, microvesicles are 50–2000 nanometer EVs, generated by direct budding of the cellular plasma membrane, and shed into the extracellular environment (Lawson et al., 2016, Théry et al., 2002, El Andaloussi et al., 2013, Thery et al., 2009, Iraci et al., 2016).

Though NSCs are known to be rich producers of EVs, the role of EVs in NSC development and differentiation, and in the NSC response to pathological stress, is unknown (Morton et al., 2018, Janas et al., 2016). For this reason, we performed EV isolation from NSCs. We then compared the EV miRNA profiles of control and ethanol-exposed NSCs, identifying a subset of miRNAs whose expression levels were significantly induced by ethanol treatment. One of them, miR-140–3p, is a microRNA known to control mesenchymal stem cell fate commitment (Li et al., 2014, Salone et al., 2014, Karlsen et al., 2014), that we have previously demonstrated to be alcohol- and nicotine-sensitive (Balaraman et al., 2012). Using gain- and loss-of-function strategies, we show that over-expression of miR-140–3p increased NSC growth. We further showed that miR-140–3p influences expression of key differentiation-associated mRNA transcripts. Thus, the elevated expression of miR-140–3p, as observed in EVs derived from ethanol-treated NSCs, favors the accumulation of GFAP and reduction of GLAST glial progenitors, and is consistent with our previous findings that ethanol inhibits neurogenesis (Santillano et al., 2005, Camarillo and Miranda, 2008, Camarillo et al., 2007). Therefore, the dysregulated miRNA content of NSC-EVs, following ethanol exposure, may underlie some of the aberrant neural maturation associated with FASDs.

Materials and Methods

Ex Vivo Fetal Mouse Neurosphere Culture Model

Neural stem cells (NSCs) from the dorsal telencephalic vesicles of mixed male and female gestational day (GD) 12.5 C57BL/6N mouse fetuses (Envigo; Catalog #4440F) were propagated as non-adherent spheres in serum-free mitogenic media, as previously published (Santillano et al., 2005, Prock and Miranda, 2007, Sathyan et al., 2007, Camarillo et al., 2007, Tsai et al., 2014). All animal procedures were approved by the Texas A&M University Laboratory Animal Care Committee.

Ethanol Treatment

NSCs seeded at a density of 2 × 10⁶ cells per T25 flask, with each flask being defined as a single sample, were randomly assigned to a control (0 mg/dL, 0mM) or high (320mg/dl, 70 mM) ethanol exposure group. Before use, 190 proof ethanol (Sigma; Catalog # 493538) was diluted into fresh culture medium, and ethanol concentrations in culture-conditioned medium were measured by gas chromatography for each experiment. Both control and ethanol-treated flasks were tightly capped, with phenolic caps and parafilm sealed, to prevent ethanol loss in the culture medium throughout the three-day exposure period. Gas chromatographic analyses of our ethanol exposure group indicated that alcohol concentrations in the media (270–350mg/dL; 58mM-76mM) approximated the blood alcohol concentration attained in heavy drinkers (Adachi et al., 1991, Perper et al., 1986). Mitogen-withdrawal-driven differentiation was achieved by seeding neurospheres onto laminin-coated (ThermoFisher; Catalog # 23017015) culture dishes in the absence of EGF and LIF, but with FGF, as described previously (Camarillo et al., 2007, Camarillo and Miranda, 2008, Miranda et al., 2008). This paradigm results in the preferential formation of neuronal-lineage-committed migratory cells.

Extracellular Vesicle Isolation

Extracellular vesicle fractions were isolated from ethanol-treated and control neurospheres and cultures using the ExoQuick-TC™ Exosome Precipitation Solution according to manufacturer instructions (System Biosciences; Catalog # ExoTC10A-1). Samples were then passed through a 0.2um sterile filter with polyethersulfone membrane (VWR; Catalog # 28145–501) to exclude particles with diameters larger than 200 nanometers.

Nanoparticle Tracking Analysis

The size and concentration of extracellular vesicles were measured by nanoparticle tracking analysis (Nanosight LM10; Malvern). Briefly, the NanoSight instrument can be used to track, detect, and measure the size and concentration of EVs through detection of light scatter generated by nanoparticles in suspension (Dragovic et al., 2011, Gardiner et al., 2013). This light scatter is video-captured, with frame-by-frame analysis used to determine the particle Brownian motion and travel distance, allowing for calculation of particle diffusion and hydrodynamic diameter. Gibco® 1 x PBS buffer (Thermo Fisher; Catalog # 14190144) was used for dilution of isolated EV samples. 1ml of diluted EV sample was slowly syringe-injected loaded into the Nanosight LM10 chamber. For each sample, a minimum of three 60s videos, with a minimum of 1000 tracks per video, were recorded and analyzed. All settings were kept constant between samples for the sake of consistency and accuracy.

Western Blot Analysis

Protein from neurospheres and isolated EVs was extracted using 1xRIPA lysis buffer (EMD Millipore; Reference # 20–188), with addition of Halt protease & phosphatase inhibitor cocktail (Thermo Fisher Scientific; Product # 78442). Extracted protein concentration was determined using Pierce BCA protein assay kit (Thermo Fisher Scientific; Catalog # 23225). 25ug of protein was size-fractionated on a 4–12% Bis-Tris Gel, ran at 150V for 60 minutes, and blotted to a PVDF membrane using iBlot transfer system (Thermo Fisher Scientific; Catalog # IB301001). Subsequently, the membrane was blocked for one hour with 2.5% goat serum and 5% nonfat dry milk in Tris-buffered saline that contains 0.1% Tween-20 (TTBS), then incubated overnight with polyclonal rabbit anti-CD63 antibodies diluted 1:1000 (System Biosciences; Catalog # EXOAB-CD63A-1). CD63 is an established surface marker for EVs. The blot was washed and incubated with an HRP-conjugated goat anti-rabbit IgG (Invitrogen) at dilution 1:1000 for one hour, then developed using PerkinElmer Western Lightning Plus-ECL (PerkinElmer; Catalog # NEL103001EA) and visualized using a CCD camera (Fluorchem Q, Alpha Innotech).

Transmission Electron Microscopy (TEM) and Immuno-gold Labeling of EVs

For the validation of our EV isolation, TEM and immunogold labeling were used. TEM sample preparation and imaging were performed at the Texas A&M Microscopy and Imaging Center, following an established protocol (Théry et al., 2006). For immuno-gold labeling, EV surface marker CD63 (polyclonal rabbit anti-CD63 antibodies diluted 1:100 System Biosciences; Catalog # EXOAB-CD63A-1) was used.

Briefly, 5ul of resuspended EV-containing pellet was deposited on Formvar-carbon coated electron-microscopy grids for 20 minutes and fixed in 1% glutaraldehyde for 5 minutes. Grids were washed by floating on 100ul of PBS/50 mM glycine. The grids were then transferred to a drop of blocking buffer (PBS/5% (w/v) BSA) for 10 minutes and floated on a 5-μl drop with the first antibody diluted at 1:1000 in the PBS/1% (w/v) BSA and incubated for 30 minutes. Grids were washed by floating on washing buffer (PBS/0.1% (w/v) BSA). Grids were washed with PBS and incubate on 5-μl drops of 12 nm goat anti-Rabbit colloidal gold (Jackson ImmunoResearch Lab; Catalog # 111–205-144) diluted 1:5000 in the PBS/1% (w/v) BSA for 20 minutes. EV samples were contrasted with a mixture of 2% methyl cellulose and 4% uranyl-acetate by floating on a 50-μl drop for 10 minutes on ice. Residual staining solution was wicked off by blotting with Whatman no. 1 filter paper. Grids were air-dried before being examined in a JEOL 1200 EX transmission electron microscope (JEOL USA Inc.) at an acceleration voltage of 100 kV. Images of EVs were recorded at calibrated magnifications using 3k slow-scan CCD camera (model 15C, SIA).

RNA Isolation, miRNA qPCR Array, and mRNA qPCR

RNA from neurospheres and EVs was isolated using the miRNeasy mini kit (Qiagen; Catalog # 217004). cDNA synthesis was performed using the qScript™ cDNA SuperMix (Qiagen; Catalog # 95048) or the Universal cDNA Synthesis Kit II (Exiqon; Catalog # 203301). EV miRNA was reverse transcribed using a 10ul mix of cDNA and ExiLENT SYBR® Green master mix (Exiqon: Catalog # 203421), which was loaded into each well of a microRNA Ready-to-Use PCR Mouse & Rat panel I+II, V4.M qPCR array (Exiqon; Catalog # 203615) to assess and measure 752 unique miRNAs as described previously (Balaraman et al., 2013, Balaraman et al., 2014, Balaraman et al., 2016). qPCR analysis was done on an Applied Biosystem ViiA 7 Real-time PCR system (ABI/Life Technologies, Grand Island, NY) (Balaraman et al., 2014, Selvamani et al., 2014). miRNA expression values were normalized to the 5 most invariant miRNAs across samples for qPCR arrays (mmu-miR-99b-5p, mmu-miR-384–5p, mmu-miR-151–3p, mmu-miR-25–3p, mmu-miR-676–3p), or U6 for the ex vivo differentiation or overexpression and antagomir studies.

For mRNA transcript quantification, the presented data correspond to the mean 2^-ΔΔCt after being normalized to β-actin. Primers were designed to span exon-exon junctions. For each primer pair, thermal stability curves were assessed for evidence of a single amplicon. The length of each amplicon was verified using agarose gel electrophoresis, and amplicon identity was verified by Sanger sequencing. A list of primers and their sequences is presented in Table 1.

Table 1:

List of Primers Used

Targets (Mus musculus)	Forward Sequence	Reverse Sequence	Product Length	NCBI Accession Number
ACTB	CTCTGGCTCCTAGCACCATGAAGA	GTAAAACGCAGCTCAGTAACAGTCCG	200bp	NM_007393
Nestin	CTCAGATCCTGGAAGGTGGG	GCAGAGTCCTGTATGTAGCCA	81bp	NM_016967
Olig2	GAACCCCGAAAGGTGTGGAT	TTCCGAATGTGAATTAGATTTGAGG	105bp	NM_016967
NeuN	AACCAGCAACTCCACCCTTC	CGAATTGCCCGAACATTTGC	118bp	NM_001285437
DCX	ACCTGACCCGATCCTTGTCT	ACATAGCTTTCCCCTTCTTCCA	121bp	NM_001110222
PDGFRα	CGTGCTTGGTCGGATTTTGG	CAGGTTGGGACCGGCTTAAT	83bp	NM_001083316
MAP2	CTGGAGGTGGTAATGTGAAGATTG	TCTCAGCCCCGTGATCTACC	84bp	NM_001039934
GFAP	CTAACGACTATCGCCGCCAA	CAGGAATGGTGATGCGGTTT	297bp	NM_001131020
GLAST	CAACGAAACACTTCTGGGCG	CCAGAGGCGCATACCACATT	260bp	NM_148938

miRNA Transfection

Mimics for the differentially expressed EV miRNAs, at a concentration of 25nM (Dharmacon miRNA mimics), were transfected into NSCs, using Lipofectamine RNAiMAX (ThermoFisher; Catalog # 13778) according to the manufacturer’s instructions. Similarly, hairpin inhibitors for mmu-miR-140 were transfected into NSCs at a concentration of 25nM (Dharmacon). miRIDIAN microRNA Mimic Negative Control #1 (Dharmacon; Catalog # CN-001000–01) and miRIDIAN microRNA Hairpin Inhibitor Negative Control #1 (Catalog # IN-001005–01-05) were used as controls.

MTT Cell Viability Assay

The MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) colorimetric assay (ThermoFisher; Catalog # M6494) for cell metabolic activity, measured by formation of a colored formazan reaction product, was used to assess the number of viable cells. 72 hours following transfection with miRNAs, NSCs were incubated for 3 hours with 6mM MTT. The reaction product was subsequently solubilized with 10% SDS in 0.01N HCl for 3 hours. Absorbance intensities were measured at 570nM, using Tecan Infinite 200 microplate reader (Tecan; SKU# 8344–50-0005).

Cell Cycle Analysis

At 48 hours post-transfection, DNA synthesis was assessed by pulse-labeling cells with 10μM EdU (5-ethynyl-2’-deoxyuridine) for 1 hour. Cells were immediately harvested, and cell cycle analysis was performed with the Click-iT® EdU Alexa Fluor® 488 Flow Cytometry Assay Kit (Thermo Fisher, Cat # C10420), in conjunction with 7-Amino-Actinomycin D (Thermo Fisher, Cat # 00–6993-50), according to manufacturer instructions, using the Beckman Coulter® Gallios 2/5/3 Flow Cytometer. Data was analyzed using Kaluza software (Beckman Coulter).

Cell Death Analysis

At 48 hours post-transfection, the Promega Caspase-Glo® 3/7 Assay Systems (Promega, Cat # G8091) was used to quantify apoptotic cell death.

Ingenuity Pathway Analysis

Ingenuity pathway analysis was conducted as previously described (Balaraman et al., 2016). Briefly, IPA miRNA Target Filter® was used to identify experimentally validated miRNA-target gene interactions or potential miRNA gene targets with a high predicted confidence of interaction (context score < −0.4 [11, 12]). Subsequently, IPA’s Core Analysis workflow was used to conduct functional network analysis, to identify gene regulatory networks overrepresented amongst predicted miRNA targets.

Single Cell Gene Expression

Confluent neurospheres were trypsinized using 0.25% EDTA-free trypsin (Thermofisher #15090046) for 5 minutes at room temperature with occasional pipetting. Trypsin was subsequently inactivated with 10% FBS and cells were filtered through a 40-micron filter to ensure single-cell suspensions. The cells were then spun down followed by an ice-cold PBS wash, and 10,000 cells were reserved for ice-cold methanol fixation. Cells were rehydrated according to manufacturer instructions immediately preceding the 10X Genomics Single cell Protocol.

We used the chromium single cell 3’ reagent kit (10X Genomics; CG000183) to prepare a single cell master mix using an input of 7000 rehydrated cells. The single cell master mixes were subsequently loaded into Chromium Chip B (10X Genomics; PN-1000074) and gel beads in emulsion were prepared according to the manufacturer protocol followed by RT, cDNA amplification and library preparation step. The libraries were sequenced on NextSeq 500 (Illumina) (400 million single-end reads sequencing depth).

The Cell Ranger analysis pipeline was used to demultiplex raw base call (BCL) files generated by the sequencer into FASTQ files, perform genome alignment, filtering, and unique molecular index (UMI) quantification utilizing the Chromium cellular barcodes to generate gene-barcode UMI matrices. The gene-barcode UMI matrices were used as input for analysis using R’s “Seurat” package, during which UMIs were log normalized.

For the correlation analyses, we calculated the spearman’s correlation between Wwp2 (the miR140 host gene, miR140HG) and a set of markers for stem cell identity and differentiation in Wwp2/miR140HG-positive cells (a total of 972 Wwp2/miR140HG-positive cells out of 13,770 assessed cells, with a range of Wwp2/miR140HG expression from 0.11 to 2.99 log₁₀ normalized UMI counts, representing a 753-fold range in expression). The correlation plot was constructed using the “corrplot” package in ‘R’.

Statistical Analysis

Statistical analyses, including linear regression, student’s t-test, or one-way ANOVA with either Sidak’s or Dunnet’s post-hoc analysis, were conducted using the GraphPad Prism software, version 6.00 for Windows.

Results

Mouse Neural Stem Cells are Abundant Producers of Extracellular Vesicles

We isolated EVs from the culture supernatant of murine NSCs in neurosphere cultures. An immunoblot showed that, compared to parent cells, the presumptive EV fraction from our NSCs supernatant was enriched for the EV marker CD63 (Figure 1A). Electron micrographs indicated that the NSC-derived EVs were mainly round-shaped vesicles, with immuno-gold staining demonstrating positive staining for CD63 on the surface of our isolated EVs (Figure 1B). NSCs secreted an abundant number of extracellular vesicles, ranging from ~ 10¹⁰ to 10¹¹ vesicles per initial cell count of ~2 × 10⁶ cells, after 72 hours of culture (Figure 1C and Supplementary Figure 2B). There was no significant difference in vesicle size between EVs that were positively stained for CD63 and those that were not (Supplementary Figures 1A and 1B). Nanoparticle tracking analysis (NTA) indicated the NSC EV diameters ranged in size from 50–150nm, with the diameter median and mode being ~130nm, consistent with the known size range (50–200nm) of exosomes (Figure 1C) (Mathivanan et al., 2010, Théry et al., 2002, Janas et al., 2016).

Figure 1. NSCs are Abundant Producers of EVs.

1A) Immunoblot of CD63 expression in EVs and NSCs. Differences in CD63 molecular weight between EV and NSC fraction are due to polyethylene glycol in the Exoquick-TC™ preparation.

1B) TEM image of EVs immunolabelled with CD63. Yellow circles indicate puncta with CD63 labelling.

1C) Nanosight image of EVs derived from NSCs Mkr=molecular weight marker

Ethanol Exposure Alters the microRNA Content of NSC-Derived EVs

Ethanol exposure did not result in a significant change in the size and concentration of EVs derived from neurosphere cultures, or the expression of CD63 on EVs (Supplementary Figures 2A-2C). However, ethanol exposure of NSCs at 320mg/dL altered the profile of miRNAs packaged within EVs. Out of the 652 expressed miRNAs, 53 miRNAs were differentially expressed (Unpaired t-test, p<0.05) in the EV fraction of ethanol-treated NSCs compared to controls, not accounting for FDR correction (Figure2A, Table 2). We observed that the majority of differentially expressed miRNAs in EVs, 47 out of 53 miRNAs, were significantly upregulated by ethanol treatment relative to controls, whereas only 6 miRNAs were significantly downregulated. Interestingly, of the top 30% most abundantly expressed EV miRNAs exhibiting a greater than 1.5-fold change between treatment conditions, only 4 were significantly increased in the EV fraction of ethanol-treated NSCs compared to controls when controlling for the false discovery rate (Unpaired t-test, q<0.05; Table 3). Collectively, our data suggests that a small cluster of miRNAs in NSC-derived EVs are significantly altered with ethanol exposure.

Figure 2. Ethanol Alters the miRNA Profile in NSC-derived EVs.

2A) Volcano plot of miRNA expression in EVs from control and ethanol treated NSCs. X-axis depicts Log₂ transformed fold change of Ethanol/Control miRNA expression. Orange data points above the dotted line represent miRNAs significantly affected by ethanol exposure. The red dot signifies miR-140–3p and the blue dot signifies miR-140–5p. n=5 samples per group

2B) MTT absorbance values following control, miR-140–3p, and miR-310a-3p overexpression in NSCs. n=10 samples per group.

2C) Relative abundance of miR-140–3p (red-dotted line) and miR-140–5p (blue-dotted line) compared to other miRNAs within NSCs (top) and NSC-derived EVs (bottom).

*p<0.05, ***p<0.001 by Student’s t-test

Table 2:

Table of Significantly Altered miRNAs in EVs from Ethanol-Treated NSCs

miRNA	Log2(Etoh/Ctrl)	p-value
mmu-miR-140–3p	1.038047	0.002772
rno-miR-338–3p	0.5185305	0.003492
mmu-miR-340–5p	0.6101994	0.005452
mmu-miR-709	0.4990811	0.005491
mmu-miR-15b-3p	0.7437596	0.006361
mmu-miR-378a-5p	0.5595576	0.006606
mmu-miR-181a-5p	0.4760198	0.006757
mmu-miR-187–3p	0.5162219	0.006777
rno-miR-667–3p	−1.796813	0.006934
mmu-miR-378a-3p	0.4153968	0.008311
mmu-miR-486–3p	0.9300453	0.009593
mmu-miR-674–5p	0.9715856	0.010089
mmu-miR-185–5p	0.3336599	0.0108
mmu-miR-301a-3p	0.3874377	0.011227
mmu-miR-467d-3p	1.302656	0.011264
mmu-miR-669l-5p	−1.072414	0.012656
rno-miR-344a-3p	0.3923947	0.012793
mmu-miR-29a-5p	0.9157254	0.014249
mmu-miR-668–3p	−4.322045	0.015463
mmu-miR-127–5p	3.628419	0.01561
mmu-miR-15a-5p	0.2828906	0.016326
mmu-miR-137–3p	1.097612	0.017092
mmu-miR-331–3p	0.4010352	0.017681
mmu-miR-30c-2–3p	1.131432	0.018304
mmu-miR-339–5p	0.4689127	0.01912
mmu-miR-322–3p	0.6379156	0.020104
mmu-miR-503–3p	0.2956948	0.020702
rno-miR-501–5p	0.4597853	0.021145
mmu-miR-30a-5p	0.5354863	0.02288
mmu-miR-16–5p	0.3169613	0.023537
mmu-miR-467c-5p	2.060594	0.026732
rno-miR-125b-2–3p	2.074389	0.026822
mmu-miR-872–5p	0.4429964	0.028883
mmu-miR-23a-3p	0.4438826	0.029259
rno-miR-17–1-3p	5.706707	0.030089
mmu-miR-338–5p	0.6141916	0.032902
mmu-miR-30a-3p	0.5727178	0.033067
mmu-miR-195a-5p	0.4338778	0.036908
mmu-miR-139–5p	0.5740866	0.037107
rno-miR-340–3p	0.4686742	0.037293
rno-miR-465–5p	0.8976653	0.037783
mmu-miR-132–3p	0.3586375	0.03788
mmu-miR-30e-3p	0.5565696	0.038382
rno-miR-345–5p	0.4033093	0.040369
mmu-miR-146b-5p	0.8567061	0.045176
mmu-miR-135a-5p	0.2700475	0.045312
mmu-miR-684	−3.171892	0.045519
rno-miR-324–3p	0.3913724	0.045842
mmu-miR-671–5p	0.8974715	0.046466
mmu-miR-1968–5p	1.504539	0.047055
mmu-miR-150–5p	−1.468246	0.047752
mmu-miR-1936	−4.304381	0.049028
mmu-miR-540–3p	2.762398	0.049127

Table 3:

Table of Most Abundant miRNAs in EVs from Ethanol-Treated NSCs

miRNA	Log2(Etoh/Ctrl)	p-value	q-value
mmu-miR-140–3p	1.08176	0.002772	0.03785
mmu-miR-15b-3p	0.754359	0.006361	0.03785
mmu-miR-340–5p	0.60876	0.005452	0.03785
mmu-miR-674–5p	0.942559	0.010089	0.045024
mmu-miR-29a-5p	0.88916	0.014249	0.05087
mmu-miR-322–3p	0.62676	0.020104	0.059809
mmu-miR-30a-3p	0.60816	0.033066	0.084319
mmu-miR-30e-3p	0.59616	0.038382	0.08564
mmu-miR-29a-3p	0.78976	0.051292	0.096661
mmu-miR-1983	0.76416	0.054152	0.096661
mmu-miR-129–5p	0.90616	0.063557	0.103135
mmu-miR-326–3p	0.77416	0.108994	0.140517
mmu-miR-17–3p	0.676559	0.10497	0.140517
mmu-miR-34b-5p	0.636559	0.117668	0.140517
mmu-miR-129–1-3p	0.614159	0.118081	0.140517
rno-miR-99a-3p	0.63276	0.181754	0.190842
mmu-miR-29c-3p	0.60136	0.180988	0.190842
mmu-miR-500–3p	0.61616	0.215239	0.213446
mmu-miR-488–3p	3.29636	0.296455	0.278512
mmu-miR-98–5p	3.04196	0.350318	0.312659

miR-140–3p Promotes NSC Proliferation

Given that ethanol exposure has profound effects on NSC and neuroepithelial growth (Santillano et al., 2005), we next investigated whether the ethanol-sensitive EV miRNAs themselves had effects on NSC growth. To this end, we overexpressed 20 of the most significantly altered EV miRNAs in NSCs and found that only two of these miRNAs, mir-140–3p and miR-301a-3p, significantly increased NSC growth (Unpaired t-test, p=0.012 and p=0.0006, n=8 samples per group; Figure 2B and Supplementary Figure 3). Interestingly, miR-140–3p is also the most significantly increased miRNA in the EVs of ethanol treated NSCs (Unpaired t-test, p=0.0028), with both miR-140–3p and its passenger strand, miR-140–5p, also expressed at high levels within NSCs and NSC-derived EVs (Figure 2C). Consistent with our array data, follow-up qRT-PCR analysis in neurosphere culture samples, confirmed significant elevation of miR-140–3p within NSCs and the NSC-derived EV fraction following 3-days of ethanol exposure (ANOVA F_(1,16)=26.04, p=0.0001, Supplementary Figure 4). However, at 5 days of ethanol exposure, we did not observe any change in miR-140–3p (2 independent biological replicates, all p’s >0.5) in contrast to our previous observations (Balaraman et al., 2012), that prolonged ethanol suppressed intracellular miR-140–3p expression, a difference that may perhaps be attributable to differences in sourcing of animals for each study.

We further investigated the effects of miR-140 overexpression or inhibition on NSC growth through its effects on the cell cycle (for data on efficacy of overexpression and knockdown, see Supplementary Figures 5A-5D). After pulse-labeling cells with the nucleotide analog EdU for 1-hour, we found that both miR-140–3p and miR-140–5p overexpression significantly increased EdU incorporation, suggesting an increased rate of DNA-synthesis (ANOVA F_(2,12)=19.58, p=0.0002; Figure 3A). Likewise, transfection with miR-140–3p and miR-140–5p antagomirs significantly decreased EdU incorporation (ANOVA F_(2,12)=6.309, p=0.013), with post-hoc analysis indicating that miR-140–5p antagomir significantly reduced EdU incorporation (Dunnett’s post-hoc, p=0.009; Figure 3B). We also observed that both miRNA 140–3p and miR-140–5p overexpression significantly increased the proportion of actively dividing cells in S-phase (ANOVA F_(2,12)=13.42, p=0.0009), while decreasing the proportion of cells in G₀/G₁-phase (ANOVA F_(2,12)=6.421, p=0.013; Figure 4A). Contrastingly, the antagomir to miR-140–3p had minimal effects on cell cycle, whereas only the antagomir to miR-140–5p had a significant effect of decreasing the proportion of cells in G₂/G_M-phase (ANOVA F_(2,12)=5.616, p=0.019; Dunnett’s post-hoc, p=0.047; Figure 4B). Furthermore, neither overexpression nor inhibition of miR-140–3p or miR-140–5p resulted in increased NSC apoptosis (Figures 5A and 5B) as measure by a Caspase 3/7 activation assay. Collectively, our data suggests that miR-140–3p, mirroring the effects of ethanol exposure (Santillano et al., 2005), promotes NSC proliferation through its actions on the cell cycle, without inducing cell death.

Figure 3. miR-140–3p and miR-140–5p Regulate the Rate of DNA Synthesis.

3A) EdU Intensity of NSCs following miR-140–3p or miR-140–5p overexpression

3B) EdU Intensity of NSCs following miR-140–3p or miR-140–5p inhibition

n=5 samples per group, ^#p<0.05, ^###p<0.001 by ANOVA. *p<0.05, ***p<0.001 by Dunnett’s post-hoc.

Figure 4. miR-140 Influences Cell Cycle Dynamics.

Proportion of cells in S, G₀/G_1, or G₂/M phase of the cell cycle following miR-140–3p and miR-140–5p overexpression (4A) or inhibition (4B). Representative flow cytometry images are displayed on the right panel.

n=5 samples per group, ^#p<0.05, ^##p<0.01 by ANOVA. *p<0.05, ***p<0.001 by Dunnett’s post-hoc.

Figure 5. miR-140 Does Not Influence Apoptotic Cell Death.

Quantification of caspase 3/7 activity following miR-140–3p and miR-140–5p overexpression (4A) or inhibition (4B).

n=5 samples per group, n.s. indicates no significance.

miR-140–3p Overexpression During Neural Differentiation Promotes an Aberrant Astrogliogenic Fate

To determine whether miR-140 might also influence NSC maturation, we first examined miR-140 levels during a 3-day mitogen (EGF/LIF)-withdrawal-induced NSC differentiation paradigm. Following the withdrawal of FGF, and in the presence of the extracellular matrix protein laminin, neurospheres become adherent. Subsequently, progenitors with a bi-polar morphology, resembling migratory neurons, proceed to migrate away from the neurosphere core and express neuronal markers like neurofilament (Figure 6A). In previous studies, we showed that these migratory cells express neuronal lineage markers and represent early neuronal lineage commitment (Camarillo et al., 2007, Camarillo and Miranda, 2008). We found that, in control, non-ethanol-exposed cultures, intracellular miR-140–3p expression significantly decreased during differentiation (ANOVA F_(2,42)=6.76, p=0.003), whereas miR-140–5p expression was not affected by differentiation (Figures 6B and 6C).

Figure 6. miR-140–3p Expression Decreases During NSC differentiation.

6A) Photomicrographs of a neurosphere over a 3-day mitogen-withdrawal-induced differentiation paradigm. Inset images depict the bipolar morphological appearance and immunolabeling with anti-neurofilament antibody of maturing, migratory neurons. Arrow heads demarcate axonal processes of bipolar neurons and arrows indicate a neurofilament labeled axon.

6B) miR-140–3p and 6C) mir-140–5p expression in NSCs over a three-day mitogen-withdrawal-induced differentiation paradigm

n=15 samples per group, ^##p<0.01 by ANOVA, **p<0.01 by Dunnett’s post-hoc. D0, D1, and D3 indicate days 0, 1, and 3 of differentiation, respectively.

Interestingly, when miR-140–3p was overexpressed during NSC differentiation (Supplementary Figures 6A and 6B), we noted that on day 3 of differentiation, the mRNA expression of GFAP, an astrocytic marker, was increased (day x treatment interaction, ANOVA F_(2,23)=6.468, p=0.0059; Sidak’s post-hoc, p=0.0006), while the expression of GLAST mRNA was decreased (day x treatment interaction, ANOVA F_(2,23)=16.79, p<0.0001; Sidak’s post-hoc, p<0.0001, Figures 7A and 7B). Mir-140–3p overexpression also significantly decreased mRNA expression for the early neuronal differentiation marker, DCX (day x treatment interaction, ANOVA F_(2,23)=7.88, p=0.003; Sidak’s post-hoc, p=0.0005), as well as for the later neuronal lineage commitment marker, NeuN (day x treatment interaction, ANOVA F_(2,23)=9.49, p=0.001; Sidak’s post-hoc, p=0.002, Figures 7C and 7D). There was also a significant decrease in mRNA expression for the oligodendrocytic marker, PDGFRα (day x treatment interaction, ANOVA F_(2,23)=5.98, Sidak’s post-hoc, p=0.0081; Figure 7E). While there was only a marginally significant interaction effect (ANOVA F_(2,23)=2.85, p=0.078) between differentiation and miR-140–3p overexpression on expression of a second oligodendrocyte marker, Olig2, a planned comparison indicated that miR-140–3p overexpression also significantly decreased Olig2 mRNA expression on day 3 (Sidak’s post-hoc, p=0.027; Figure 7F).

Figure 7. miR-140–3p Influences NSC Differentiation.

Expression of 7A) GFAP, 7B) Glast, 7C) DCX, 7D) NeUN, 7E) PDGFRα, 7F) Olig2, and 7G) Nestin in NSCs following miR-140–3p overexpression in a three-day mitogen-withdrawal-induced differentiation paradigm.

n=5 samples per group, ⁺⁺p<0.01, ⁺⁺⁺p<0.001, ⁺⁺⁺⁺p<0.0001 interaction effect between differentiation and miR-140–3p overexpression, ANOVA. **p<0.01, ***p<0.001, ****p<0.0001 by Sidak’s post-hoc. D0, D1, and D3 indicate days 0, 1, and 3 of differentiation, respectively.

Compared to miR-140–3p, miR-140–5p overexpression had a more restricted effect on differentiation. MiR-140–5p decreased GLAST expression on day 3 of differentiation (day x treatment interaction, ANOVA F_(2,22)=12.87, p=0.0002, Sidak’s post-hoc, p<0.0001), but did not affect GFAP expression (Figures 8A and 8B). While there were minimal effects on neural differentiation (Figures 8C and 8D), Olig2 (day x treatment interaction, ANOVA F(_2,22)=4.80, p=0.019; Sidak’s post-hoc, p=0.005) and PDGFRα (day x treatment interaction, ANOVA F(_2,22)=4.12, p=0.03,; Sidak’s post-hoc, p=0.04) were also significantly decreased on day 3 of differentiation compared to controls (Figures 8E and 8F). Interestingly, we observed a main effect of miR-140–5p overexpression, resulting in decreased Nestin mRNA expression throughout the differentiation period (ANOVA F_(1,22)=5.807, p=0.025; Figure 8G). Collectively, our data show that miRNA-140–3p overexpression results in aberrant astroglial-directed differentiation, while suppressing neuronal and oligodendrocytic differentiation.

Figure 8. miR-140–5p Influences NSC Differentiation.

Expression of 8A) GFAP, 8B) Glast, 8C) DCX, 8D) NeUN, 8E) PDGFRα, 8F) Olig2, and 8G) Nestin in NSCs following miR-140–5p overexpression in a three-day mitogen-withdrawal-induced differentiation paradigm.

n=5 samples per group, ⁺⁺p<0.01, ⁺⁺⁺p<0.001, ⁺⁺⁺⁺p<0.0001 interaction effect between differentiation and miR-140–5p overexpression, ANOVA. ^#p<0.05 main effect of miR-140–5p overexpression, ANOVA. **p<0.01, ***p<0.001, ****p<0.0001 by Sidak’s post-hoc. D0, D1, and D3 indicate days 0, 1, and 3 of differentiation, respectively.

miR-140–3/5p are encoded as intronic miRNAs within the host gene, Wwp2 (miR-140HG). We therefore used Wwp2/miR-140HG expression to identify presumptive miR140-transcribing NSCs and assess, in this subpopulation, using a single-cell RNAseq approach, the relationship between miR-140HG expression and markers of maturation. We found that Wwp2/miR-140HG was expressed in 972 out of a total of 13,770 assessed cells, suggesting that ~7% of all neural progenitor cells contribute to the high levels of miR-140–3p in EVs. Moreover, we found that even in Wwp2/miR-140HG-positive cells, there was a large, 753-fold range (0.11–2.99 log₁₀ RPKM) in Wwp2/miR140HG expression, suggesting the possibility of a similarly large dynamic range in miR140–3p expression as well. Consistent with our overexpression studies, the expression of miR-140 host transcript was significantly negatively correlated with Nestin expression and positively correlated with GFAP expression indicating that cells that highly express miR-140HG also highly express GFAP (Figure 9A).

Figure 9. miR-140 host gene expression is correlated with glial markers.

Correlation plot of WWP2, the mir-140 host gene (miR-140 HG), with different markers of neural and glial maturation.

Discussion

Extracellular vesicles represent an important mode of paracrine communication due to their ability to interact with various cell types and tissues. Specifically, tumor invasion and metastases can be promoted by metalloproteases released from tumor cell-derived EVs (Muralidharan-Chari et al., 2010). Cancer cells have also been observed to propagate oncogenes and their cancer-associated transforming phenotype through EVs (Al-Nedawi et al., 2008). Recent studies have also shown that EVs within the central nervous system serve as a mode of intercellular communication for both neurons and glia by trafficking growth factors and other molecules (Morel et al., 2013, Ratajczak et al., 2006, Morton et al., 2018, Muralidharan-Chari et al., 2010).

Stem cells also produce an abundance of EVs (Drago et al., 2013, Katsuda et al., 2013, Tetta et al., 2013). Studies have shown that EVs released from stem cells, including NSCs, traffic cargo (including proteins, mRNAs, and miRNAs) to target cells, influencing their behavior. A recent study demonstrated that CD63-GFP-expressing NSC-derived EVs were readily up-taken by embryonic NSCs and astrocytes in vitro (Yoshimura et al., 2018). Functionally, NSC-derived EVs have been observed to transfer IFN-γ to activate Stat1 signaling in target cells (Cossetti et al., 2014). EVs may also play an important role in maintaining the stem cell phenotype. Indeed, EVs released by stem cells have been credited for their pro-regenerative ability by enhancing cell proliferation, inhibiting apoptosis, and promoting immune tolerance of recipient cells (Grange et al., 2017, Gai et al., 2016, Zhan et al., 2015, Bruno et al., 2016). Our previous studies have documented that ethanol exposure significantly reduced the numbers of cells expressing stem cell markers CD117, CD133, Sca-1, and ABCG2, and suggested that ethanol depletes neuroepithelial cells by promoting premature maturation (Santillano et al., 2005). Interestingly, EVs are known to be released from cells as a response to physiologic stress and environmental stimuli (H Rashed et al., 2017). Thus, it is feasible that ethanol’s effects on the developing neuroepithelium may be mediated through its actions on NSC-derived EVs.

In this study, we characterized the effects of ethanol exposure on our NSC-derived EV cargo and the implication of these effects on NSC growth and maturation. We found that while ethanol exposure did not alter numbers or sizes of NSC-derived EVs, it significantly altered their miRNA content, with miR-140–3p being the most significantly increased miRNA. We, and others, previously reported that intracellular expression of miR-140–3p is both ethanol and nicotine sensitive (Balaraman et al., 2012, Huang and Li, 2009). In this study, we found that a 72-hour period of ethanol exposure increased miR140–3p levels in both NSCs, and in NEC-derived EVs.

Furthermore, we observed that miR-140–3p overexpression significantly increased NSC proliferation through its effects on the cell cycle, mirroring observed effects of ethanol exposure (Santillano et al., 2005) on NSC growth. In the context of a stereotypic mitogen-withdrawal-stimulated NSC maturation paradigm, miR-140–3p promoted aberrant GFAP-mRNA^hi/GLAST-mRNA^lo astroglial differentiation, while suppressing neuronal and oligodendroglial lineage markers. Single-cell RNAseq analysis demonstrated a similar positive association between expression of mRNA transcripts from the Wwp2/miR-140HG locus which encodes miR-140–3p, and the expression of GFAP, but not GLAST, suggesting that alcohol amplifies a normally existing relationship between miR-140–3p and gliogenesis. The presence of neurons is required for GLAST expression in astrocytes (Swanson et al., 1997, Perego et al., 2000). Consequently, the loss of neuronal linage following miR-140–3p overexpression may indirectly result in aberrant astrocytic maturation and contribute to aberrant astrocyte function that has been associated with prenatal alcohol exposure (Wilhelm et al., 2018). The loss of oligodendroglial markers following miR-140–3p overexpression is also consistent with the appearance of white matter abnormalities in FASD (Norman et al., 2009) and loss of oligodendrocytes (Newville et al., 2017) in models of PAE. Thus, dysregulated miRNA content of neural progenitor EVs following ethanol exposure may underlie some of the aberrant brain maturation associated with FASD. While we did not investigate the direct targets of miR-140–3p and miR-140–5p that mediated their effects on NSC differentiation, it is likely that these miRNAs influence multiple pathways involved in neural maturation. In fact, Ingenuity Pathway Analysis™ indicates that the predicted targets of miR-140–3p are overrepresented in several developmental pathways including for example, the PTEN and 14–3-3 protein pathways (Supplementary Figure 7), which is extensively involved in NSC maturation, migration, and cell-division (Duan et al., 2015, Wen et al., 2013, Lee et al., 2019, Gregorian et al., 2009, Toyo-oka et al., 2014). Furthermore, this study examined the effects of miRNAs on neural development in isolation. Emerging research has shown that miRNAs are likely to work in concert to mediate disease pathology, including in FASD growth parameters (Tseng et al., 2019).

MicroRNAs in EVs have recently been implicated in ethanol activation of inflammatory neurotoxicity (Coleman et al., 2017) and hepatotoxicity (Massey et al., 2018) in the adult, and extracellular miRNAs have been identified as biomarkers for the severity of prenatal alcohol effects in infants (Balaraman et al., 2016). This initial study to characterize the effects of ethanol exposure on the miRNA content of EVs derived from fetal NSCs likely feeds into an emerging hypothesis that packaged extracellular RNAs are significant contributors to ethanol effects in both development and the adult. Given that EVs are an important mode of paracrine communication for NSCs and may play an important role in NSC renewal and differentiation, alterations in the miRNA content of EVs may underlie aberrant neural maturation in response to environmental factors including PAE. Future studies should focus on determining how ethanol influences packaging of NSC-derived EVs as well as their trafficking to target cells. Ultimately, exploiting the paracrine role of EVs may be a means to mitigate the effects of PAE, as has been shown with other pathologies (Wang et al., 2017, Bian et al., 2014, Kim et al., 2016).