Abstract
Background:
iPSCs, with many mutations or epigenetic modifications, are in a unique aging state and have pluripotent properties. Previously, we reported that hsa-miR-520d-5p reverted fibroblasts to mesenchymal stem cells (MSCs) and converted approximately half of the mutations in cancer cells back to wild-type alleles.
Objective:
We focused on the genes involved in the hair life cycle comprising 4 phases and examined whether 520d-5p can convert iPSCs with mutation(s) to less-mutated status during the differentiation to MSCs.
Materials and methods:
Using next-generation sequencing analysis, we compared the effect of 520d-5p on the mutation level in 177 hair life cycle-related genes in iPSCs to that in iPSC-differentiated MSCs.
Results:
Transfection by hsa-miR-520d-5p induced genetic improvement in 79.7% (141/177) of the genes and reverted 17.7% (25/141) to wild type; the average conversion rate was 54.4%. The hsa-miR-520d-5p-induced genes, including the function of dermal papilla cells or bulge stem cells closer to intact status. Interestingly, the dramatic changes of nucleotide extended to non-target genes as well as target genes.
Conclusions:
Thus, hsa-miR-520d-5p is useful for studying hair or scalp conditioning and developing therapy against aging or hair diseases such as alopecia.
Introduction
Aging is a biological phenomenon that progresses not only through telomere shortening and associated dynamic genomic instability but also through the addition of many mutations and epigenetic modifications, ending with cell death.1
Hair growth and hair loss are closely connected to aging. Hair growth requires the normal functioning of dermal papilla (DP) cells, skin-derived precursor (SKP) cells,2 and bulge stem cells in old as well as young persons. Therefore, it is critical to maintaining the potential of the stem cells to differentiate into other cell lineages, such as blood, fibroblast, adipocyte, and neuron, which are crucial for fostering hair growth.3 DP cells are mesenchymal stem cells (MSCs) that interact with the SKP cells (3). Also, bulge stem cells, derived from the peripheral ring of the placode basal layer, are essential for hair growth.4 Moreover, induced pluripotent stem cells (iPSCs) are found in human hair follicles and may be used to reverse hair loss.5–7
iPSCs are unique in that they leave traces of aging even in young cells that have acquired pluripotency.8 Thus, a pluripotent cell with genomic mutations can be regarded as a special kind of aging cell, resulting in the irreversible mutations that cause tumor formation in ex vivo transplantation of iPSCs. Previously, we have reported that we can generate MSCs by introducing microRNAs (miRNAs) into adult fibroblasts, thereby extending their lifespan by about 4-fold and restoring their collagen-producing capacity.9 Also, we have found that human miRNA (miR-520d-5p) can be introduced into iPSCs and cancer cells to reduce the number of single-nucleotide mutations in cancer-related genes by half.10
Moreover, hsa-miR-520d-5p (520d-5p) is predicted to bind to more than 9,000 mRNAs, based on the online algorithm analysis (miRBase: http://www.mirbase.org/; RNAcentral: https://rnacentral.org/; TargetScanHuman: http://www.targetscan.org/vert_72/). In particular, 39 genes are implicated in the growth or functions of DP cells.11
Male and female androgenetic alopecia [(F)AGA] is caused by driving rapidly from anagen to catagen. Therefore, after we investigated the genes reported to be of pathological significance regarding (F)AGA or hair life cycle within the past decade, we focused on the 177 genes involved in the hair cycle comprising the catagen, telogen, exogen, and anagen stages (Fig. 1).12 In addition, we examined whether the introduction of 520d-5p during the invitro differentiation of iPSCs into MSCs could improve the sequence status of the genome. Our study likely contributes to a better understanding of aging and the development of antiaging strategies to extend life span or health span.
A scheme showing the relation of a phase in the life cycle of a hair follicle. Numbers ①–⑦ show the process between each phase of hair life cycle, and (◻) means the functional genes involved in each process. (*) means genes without nucleotide conversions. The genes without (*) mean with nucleotide conversion, as shown in Table 1.
Materials and Methods
Cells
We used three cell lines and lentiviral vectors for DNA extraction. Human induced pluripotent stem cells (hiPSCs) (HPS0002: 253G1) and human mesenchymal stem cell (hMSC) were provided by the RIKEN BioResource Center Cell Bank and the Cell Resource Center for Biomedical Research, Institute of Development, Takara Bio (Kusatsu, Shiga, Japan), respectively.13 hiPSC (253G1) and hMSC were cultured in ReproStem medium (ReproCell, Tokyo, Japan) with 10 ng/ml of bFGF-2 and MSC Growth Medium (Takara Bio, Kusatsu, Shiga, Japan), respectively. Additionally, the human mesangial cell line 293FT (Invitrogen Japan K.K., Tokyo, Japan) was used to produce 520d-5p expressing-lentiviral particles for transfecting them into hMSCs. The 293FT cells were cultured in Dulbecco's modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum (FBS), 0.1 mM minimum essential media (MEM) nonessential amino acid, 2 mM
520d-Expressing vector for the transfection
We examined the effect of miR-520d-5p overexpression using the viral vector produced from 293FT cells (5 × 106 cells/10-cm culture dish) with 20 μg of pMIRNA1-miR-520d-5p/GFP (System Biosciences, Mountain View, CA, USA) or the mock vector pCDH. We investigated hMSCs with miR-520d-5p expression by centrifuging the cells at 170,000 × g at 4°C for 120 min to harvest the viral particles produced in the 293FT cell culture media. The viral pellets were collected, and viral copy numbers were measured with a Lenti-X™ qRT-PCR Titration kit (Clontech, Mountain View, CA, USA). Meanwhile, one million lentiviral copies were used per 10-cm culture dish for 293FT cell infection. We transfected 50 nM synthesized oligonucleotides into 293FT cells with the FuGENE HD Transfection Reagent (Roche Diagnostics, Basel, Switzerland). pCDH/lenti/GFP-treated cells wereused as controls. Finally, 520d-5p-transfected hMSCs were generated by the lentiviral transfection of 520d-5p to hMSCs after the differentiation of iPSCs to hMSCs was induced using a hiPSC-derived MSC differentiation system and reagents (Veritas Corporation, Tokyo, Japan).
Next-generation sequencing analysis
iPSC (253G1), 520d-transfected and 253G1-derived progenitor hMSC (520d/hMSC), and hMSC were used innext-generation sequencing (NGS) analysis. DNA extraction was performed using a Qiagen DNeasy kit (QIAGEN, Tokyo, Japan) according to manufactures instructions. Exome sequencing was performed with a next-generation sequencer (Illumina, Inc., San Diego, USA). Three replicates per group were analyzed (n = 3). Genomic DNA was processed using SureSelectXT Human All Exon v5 + UTRs (Agilent Technologies, Inc.) and sequenced on the Illumina HiSeq 2500 platform with 101-bp paired-end reads (HSS, Sapporo, Japan). The paired-end reads in FastQ format were mapped to the reference genome HG19 using BWA-0.7.10.
The mapping files in the SAM (Sequence Alignment/Map) format were converted to the BAM (binary version of SAM) format and sorted by SAMtools-1.2. Local realignment around the known indels was performed by genome analysis toolkit (GA)-Lite-2.3.0 on the sorted BAM files. Picardtools-1.133 was used to remove polymerase chain reaction (PCR) duplicates.14,15 Finally, base quality score recalibration was performed using GATK. RNA sequencing was performed using the same apparatus as exome sequencing. Total RNA was processed using the TruSeq Stranded mRNA Library Prep Kit (Illumina) and sequenced on the Illumina HiSeq 2500 platform with 101 bp paired-end reads. Finally, the paired-end reads in FastQ format were mapped to reference genome HG19 using TopHat.
Samtools mpileup [options: -d 10000 -L 10000 -B -t DP, DV, SP, DP4, DPR] was piped with bcftools call[options: -A -v -m -f GQ] to produce variant call format (VCF) files. The VCF files were further filtered to select the variants with at least two supporting reads: a minimum allele frequency of 0.8 and a maximum read depth of 35 at the called sites. For high throughput, the deep sequencing of selected cDNA was subjected to Sanger sequencing to screen for nonsynonymous mutations.16–18
Results
pCDH/lenti/GFP-treated cells used as a control never induced genomic changes as reported in cancer cells or hiPSCs.10 We compared the genomic status or mutations induced by aging or the four genes used to generate hiPSC in hiPSCs, hMSCs, or hiPSC-derived 520d/hMSCs. A remarkable improvement in the mutation status of the 141 hair cycle-related genes was observed (p < 0.01 by t-test) (Table 1). Twenty-nine genes with mutations showed no genomic conversion. However, unlike our previous report,10 two-, three-, and four-base conversions were confirmed in 33 genes, eight genes, and one gene, respectively (Table 2). Seven genes had no mutations in iPSCs (Table 2). The relationship between the hair life cycle and the genes we examined was demonstrated using NGS analysis (Fig. 1).
Summary of hair life cycle-related genes with conversion effects of nucleotide by 520d-5p
Summary of hair life cycle-related genes without the conversion effects, or with 2-base, 3-base, or 4-base conversion by 520d-5p
In the genes involved in signaling pathways regarding hair growth, the conversion rate (CR; converted number/total number of mutations) of mutations by 520d-5p in eight representative genes, DKK2 (100%; 12/12), FGF5 (100%; 21/21), TGFB1 (100%; 15/15), FZD1 (100%; 50/50), BMP2 (83.9%; 25/31), NOG (100%; 53/53), NOTCH2 (46.2%; 34/73) (Fig. 2), and AR (71.4%; 5/7) (Fig. 3), was analyzed. Also, two representative genes, WNT9B (94.7%; 17/18) and VEGFA (88.0%; 44/50), had two-base sequential conversions, two representative genes, SOX9 (94.7%; 54/57) and AQP3 (77.8%; 14/18), had three-base sequential conversions, and the gene, LRIG1 (77.8%; 28/36), had four-base sequential conversions (Fig. 3).
Representative six genes (DKK2, FGF5, TGFB1, FZD1, BMP2, and NOG) showed a remarkable reduction of mutation level. Top, middle, and bottom in each panel mean hiPSCs, hMSCs, or 520d/hMSCs derived from hiPSCs, respectively. The square (NOTCH2) indicates the site where the base sequence is markedly closer to the wild type by 520d and is difficult to distinguish.
Representative six genes (AR, WNT9B, VEGFA, SOX9, AQP3, and LRIG1) also showed a remarkable reduction of mutation level. Top, middle, and bottom in each panel mean hiPSCs, hMSCs, or 520d/hMSCs derived from hiPSCs, respectively. The square (AR) indicates the site where the base sequence is markedly closer to the wild type by 520d and is difficult to distinguish. Also, representative genes with base sequential improvement show two, three, or four sequential base set of changes for WNT9B and VEGFA, for SOX9 and AQP3, or LRIG1, respectively. Several nucleotides in LAMA5 damaged at the nucleotide level by extrinsic stress were converted by 520d-5p.
However, although 520d-5p transfection reverted five out of six mutations except for the repeat sequences in AR gene to wild type (Supplementary Fig. 1, top), the repeat sequences, 22 CAG repeats and 17 GGC repeats, of AR in iPSCs were identical to those in 520d/hMSCs (Supplementary Fig. 1, bottom).
Thirty-three out of 39 genes involved in human DP cells had remarkable CR, and 15 genes were algorithmically targeted by 520d-5p (Supplementary Fig. 2) (TargetScanHuman 8.0; https://www.targetscan.org). CR in 10 out of 23 genes involved in the placode of bulge stem cells was more than 80%, and three genes were algorithmically targeted by 520d-5p (Supplementary Fig. 3). CR of LAMA5 playing a central role in the signaling involved in ultraviolet (UV) or external stress was less than 10% (8.5%; 7/82) (Fig. 3). With the transfection of 520d-5p, we did not find any genes that had obvious worsening changes at the nucleotide level. Finally, the average conversion rate of the 177 genes was 54.4% (Tables 1 and2).
Discussion
Cells in hair life cycle
Hair growth requires the normal functioning of DP cells, SKP cells,2 and bulge stem cells in old as well as young persons. It is critical to maintaining the potentials of the stem cells which differentiate into other cell lineages, such as blood, fibroblast, adipocyte, and neuron, that are crucial for fostering hair growth.3 DP cells are MSCs that interact with the SKP cells.3 Also, bulge stem cells are functionally essential bearers for hair growth and are derived from the peripheral ring of the placode basal layer.4 In addition, Lrig1 is a crucial regulator of epidermal stem cell quiescence; the downregulation of its gene triggers the proliferation of keratinocytes.19
Expressive molecules and 520d-5p involved in hair life cycle
Recently, a handful of studies have demonstrated the impact of miRNAs on common hair loss-related disorders,20 but the exhaustive molecular mechanisms are still unclear. 520d-5p is predicted to bind to more than 9,000 mRNAs using the online algorithm analysis as mentioned above. In particular, 39 genes are implicated in the growth or functions of DP cells.11 520d-5p potentially targets 15 of them, based on their target scores of more than 50% corresponding to a 7-base match to the target mRNA sequences. 520d-5p improved the sequence status of 35 out of the 39 DP cell-related genes. On the other hand, 520d-5p induced improvement in the sequence toward wild type in 13 of the 15 genes.
AR gene and miRNAs
Interestingly, 520d-5p strongly targeted AR mRNA (Supplementary Fig. 2), suppressing AR function due to the overexpression of AR and ARA55 (TGFB1 induced transcript 1) in the hair parts or hair papilla of patients with (F)AGA. Accumulated mutations in the second exon of AR sequences are converted to wild type. In DP cells, AR remains a key factor regarding (F)AGA and is regulated via various signaling pathways. AR is promoted by EGF, TGF-β1 (TGFB1), NCOA4, and RUNX1 signaling and inhibited by the Notch pathway.21,22 We found that 520d-5p improved the mutations in all the genes above. Because 520d-5p as well as 133b in scalp samples from AGA patients are upregulated, we have to explore the reason the upregulation is not sufficient to improve AGA, by 520d-5p transfection into cells derived from AGA patients.23
520d-5p in diverse signaling for hair growth
In 21 out of the 23 bulge stem cell-related genes, 520d-5p induced nucleotide conversion in them; also, 520d-5p targeted three genes, SOX9, PTHLH, and SMOC2, at the target score of more than 50% (Supplementary Fig. 3). SOX9, NFATC1, LHX2, and KRT15 are known adult stem cell markers of bulge cells.24–28 The mutated nucleotides in these genes were all improved by 520d-5p.
WNT, frizzled-related protein, beta-catenin (CTNNB1), LEF1, cadherin (CDH), NOG, BMP, LRP, and SHH are involved in hair bud formation or growth.29 Hair buds maintain their density by inhibiting the formation of other hair buds and via the negative feedback system of NOG, BMP, SHH, and LEF1 or the NOTCH-Delta system. We found that mutated nucleotides in these genes were all improved by 520d-5p.
On the other hand, Laminin-511 protects epidermal stem cells with the melanoma-associated chondroitin sulfate proteoglycan marker from UV stress and suppresses the reduction of stem cells. LAMA5 functions most upstream in the extrinsic damage system and bindsthe integrin beta1 receptor, repressing the inhibition of hair growth by BMP via NOG signaling.30 520d-5p has a therapeutic effect on ultraviolet-B wave (UVB)-irradiated cells via a recovery mechanism fromDNA damage or fragmentation.31 However, it could not reduce or change many intrinsic mutations in LAMA5.
NOTCH signaling-related genes are of interest due to their direct and exclusive suppression of AR. Human postnatal development requires intact NOTCH signaling in hair bulbs and outer root sheets.32 Mutations of these genes may reduce the suppression of AR, resulting in AR overexpression and subsequently alopecia.
520d-5p as a regulator for the genetic stabilization
In this study, using NGS analysis, we compared the effect of 520d-5p on 177 genes involved in the hair life cycle or (F)AGA in hiPSCs to that in hiPSC-derivatives (520d/hMSCs), in terms of the conversion rate of nucleotide sequences to wild type. Surprisingly, the transfection of 520d-5p into the hiPSCs-derivatives induced the collateral effects on genes that 520d-5p was not predicted to target by the algorithms. This effect of 520d-5p highlights the impact of unbalanced genomic dynamics and abnormalities because the effective range reached 82.9% (141/170) of the genes implicated in the hair life cycle, except seven genes, ASCL4, CD34, ENG, IL6, NANOGNB, SHISA2, and STAT2, without any mutations in hiPSCs. The genetic or epigenetic modifications, including DNA and RNA demethylation induction, by 520d-5p, likely cause recovery from lethal damage.33 These data are compatible with the results from our previous report that metabolome analysis in cancer cells demonstrates demethylation induction of DNA and RNA, and that the shutdown of all the metabolites by 520d-5p, followed by resetting the metabolomic status, leads to a benignant or normal status.34
520d-5p as a therapeutic agent
These data suggest that the genes upregulated by 520d-5p contribute to the improvement of hair and skin conditions; thus, 520d-5p can be used as therapeutic agents for skin diseases such as (F)AGA or female pattern hair loss.35 While reversion to wild-type nucleotide sequences does not always result in the functional recovery or structural restoration of the encoded protein, improvement of the accumulated or sequential abnormality of a nucleotide sequence may contribute more to the functional or structural normalization of protein than in cases without slight recoveries. Considering the results obtained from the iPSCs with traces of aging in their nucleotide sequences, the application of 520d-5p in (F)AGA will be expected in the near future.
Conclusion
miR-520d-5p or the compounds upregulating it are potentially useful for treating scalp aging and hair loss or maintaining hair growth via antiaging mechanisms, including both genetic and epigenetic modification processes. We should clinically confirm the effectiveness of miR-520d-5p using nanoparticles as drug delivery system (DDS).36
Footnotes
Acknowledgments
We all deeply thank Noriko Itaba, PhD, of Tottori University who supported for hiPSC differentiation procedure, and Mai Kimoto, PhD, of Hokkaido System Science Co., Ltd. who supported NGS analysis.