Arctiin blocks hydrogen peroxide-induced senescence and cell death though microRNA expression changes in human dermal papilla cells

Background Accumulating evidence indicates that reactive oxygen species (ROS) are an important etiological factor for the induction of dermal papilla cell senescence and hair loss, which is also known alopecia. Arctiin is an active lignin isolated from Arctium lappa and has anti-inflammation, anti-microbial, and anti-carcinogenic effects. In the present study, we found that arctiin exerts anti-oxidative effects on human hair dermal papilla cells (HHDPCs). Results To better understand the mechanism, we analyzed the level of hydrogen peroxide (H2O2)-induced cytotoxicity, cell death, ROS production and senescence after arctiin pretreatment of HHDPCs. The results showed that arctiin pretreatment significantly inhibited the H2O2-induced reduction in cell viability. Moreover, H2O2-induced sub-G1 phase accumulation and G2 cell cycle arrest were also downregulated by arctiin pretreatment. Interestingly, the increase in intracellular ROS mediated by H2O2 was drastically decreased in HHDPCs cultured in the presence of arctiin. This effect was confirmed by senescence associated-beta galactosidase (SA-β-gal) assay results; we found that arctiin pretreatment impaired H2O2-induced senescence in HHDPCs. Using microRNA (miRNA) microarray and bioinformatic analysis, we showed that this anti-oxidative effect of arctiin in HHDPCs was related with mitogen-activated protein kinase (MAPK) and Wnt signaling pathways. Conclusions Taken together, our data suggest that arctiin has a protective effect on ROS-induced cell dysfunction in HHDPCs and may therefore be useful for alopecia prevention and treatment strategies.


Background
Reactive oxygen species (ROS) are reactive oxygencontaining endogenous byproducts that are produced during normal metabolism and play pivotal roles in maintaining homeostasis [1]. Under stressful environmental conditions (e.g., ionizing radiation, ultraviolet (UV) radiation, drugs, and smoke), the levels of ROS, including hydrogen peroxide (H 2 O 2 ), are significantly increased [2]. These ROS subsequently activate various signaling molecules, including p53 and mitogen-activated protein kinase (MAPK) to induce cell growth arrest and apoptosis [3,4]. In skin, exogenous ROS can cause cellular damage, impaired collagen synthesis, and keratinocyte apoptosis, which are mainly observed in aged skin [5]. One group reported that high ROS levels are involved in the etiologies underlying vitiligo skin disease and skin cancer [6]. Furthermore, accumulating evidence supports the hypothesis that oxidative stress caused by H 2 O 2 is a key factor in the onset and progression of hair loss, which is known as alopecia [7][8][9].
Alopecia is a hair loss skin disorder that typically causes baldness [10]. It was originally defined as an age-dependent or androgen-dependent mechanism in dermal papilla cells (DPCs) [11]. However, clinical reports have demonstrated that alopecia does not occur in an androgen-dependent manner; there are also androgen-independent mechanisms, including chemotherapy-dependent and stress-dependent effects [9,12,13]. A growing number of studies have shown that the levels of apoptosis and senescence are significantly increased in balding DPCs as compared with non-balding DPCs [7,14]. Interestingly, ROS has been known as an important inducer of the androgendependent and -independent alopecia [9]. The dermal papillae of balding scalps showed higher levels of ROS compared with non-balding scalp [15,16]. Moreover, increased ROS levels were associated with decreased DPC motility [7]. In addition, DPCs from balding scalp also exhibited higher levels of cell senescence [7]. More recent studies also demonstrated that cisplatin-induced alopecia is mediated by ROS production and ROS-mediated apoptosis in DPCs [17]. These finding have led to the hypothesis that ROS may be an important target when designing therapeutic strategies to prevent or treat alopecia.
Arctiin is a lignin chemical reagent isolated from Arctium lappa. This chemical reagent has been known to exert anti-inflammatory, anti-proliferative, and antimicrobial effects [18][19][20][21]. We recently reported that arctiin has a protective effect against UVB radiation in skin cells, including dermal fibroblasts and keratinocytes [22,23]. We subsequently observed that the arctiin-mediated antiphotoaging effect is functionally related with microRNA (miRNA)-mediated signaling pathways [22,23]. However, it is unknown whether arctiin exerts anti-oxidative effects and what biological effects arctiin has on human follicle cells. In present study, we studied the biological events of human hair dermal papilla cells (HHDPCs) in the presence of arctiin and/or H 2 O 2 and performed cell-based assays to determine whether H 2 O 2 -induced cell dysfunction could be inhibited by arctiin treatment. We also investigated the role of miRNA-mediated mechanisms in using microarrays and bioinformatic analysis.

Arctiin inhibits H 2 O 2 -mediated cell proliferation loss in a dose-dependent manner in HHDPCs
Before we analyzed the inhibitory effect of arctiin against H 2 O 2 -induced cell dysfunction, we preferentially sought to evaluate the effect of arctiin on HHDPC growth. First, HHDPCs were exposed to different concentrations of arctiin for 24 h, and cell viability was analyzed with WST-1 assays. As shown in Figure 1A, 10, 20, and 30 μM arctiin significantly increased cell viability by 38.89 ± 3.99%, 42.48 ± 13.69%, and 48.96 ± 6.07% compared with dimethyl sulfoxide (DMSO)-treated control cells. However, exposure to larger doses of arctiin (≥40 μM) did not increase cell viability beyond that observed in cells treated with 10 μM arctiin, and 60 μM arctiin induced HHDPC cytotoxicity. Therefore, we concluded that an arctiin dose of 10 to 30 μM arctiin has maximum growthpromoting effect on HHDPCs, and those doses were used in further experiments. Next

Arctiin inhibits H 2 O 2 -mediated cell death and cell cycle arrest in HHDPCs
To confirm the result shown in Figure 1B, we performed cell cycle analyses using PI staining and flow cytometry. HHDPCs were treated with arctiin and H 2 O 2 under the same conditions used for Figure 1B, and then cells were stained with PI solution to analyze cell cycle patterns. As shown in Figure 2A, arctiin and H 2 O 2 treatment altered the cell cycle distribution of HHDPCs. Notably, H 2 O 2 only treatment led to accumulation of 7.45% in sub-G1, 12.84% in S phase and 7.45% in G2/M phase, with a corresponding decrease in the percentage of G0/G1 phase cells as compared with untreated control cells ( Figure 2A). We also found that those accumulations in sub-G1, S, and G2/M phases in H 2 O 2 -treated cells were significantly decreased by arctiin pretreatment in a dosedependent manner. Pretreatment with 10 and 20 μM arctiin led to reductions of 2.82% and 7.43% in sub-G1 phase cells as compared with H 2 O 2 -treated cells, respectively, indicating that arctiin inhibited H 2 O 2 -mediated sub-G1 accumulation, which is indicative of dead cells ( Figure 2B). Also, the proportion of G2/G1 cells was increased by H 2 O 2 treatment as compared with untreated control cells, indicating that H 2 O 2 induced G2 arrest in HHDPCs ( Figure 2C). However, the proportion of G2 arrest cells was significantly decreased by arctiin pretreatment ( Figure 2C). Overall, these results suggest that arctiin blocks H 2 O 2 -mediated cell death and G2 arrest in HHDPCs.

Arctiin inhibits H 2 O 2 -mediated ROS generation in HHDPCs
ROS generation mediated by H 2 O 2 is characterized by increases in cell death and cell cycle arrest in several cell lines [1]. To determine whether arctiin pretreatment inhibits H 2 O 2 -mediated ROS generation, we performed DCF-DA analyses to assess intracellular ROS production in HHDPCs. As shown in Figure 3A, arctiin did not alter intracellular ROS levels in untreated control cells, but it significantly abolished the H 2 O 2 -induced increase in intracellular ROS generation. Cells treated with 750 μM H 2 O 2 showed a 45.77% accumulation of M phase (DCFpositive) cells as compared with untreated control cells ( Figure 3B). However, pretreatment with 20 μM arctiin, led to reduction of 29.77% of cells in the M phase as compared with H 2 O 2 -treated cells ( Figure 3B). These results suggest that H 2 O 2 -mediated ROS production in HHDPCs is inhibited by arctiin.

Arctiin inhibits H 2 O 2 -mediated senescence in HHDPCs
Cell cycle arrest in the G2 phase and ROS generation are functionally related with cellular senescence [24,25]. Therefore, we next determined whether G2 cell cycle arrest and ROS generation contribute to senescence and whether arctiin pretreatment can block H 2 O 2 -mediated senescence in HHDPCs. Using SA-β-galactosidase (SA-βgal) assays, we evaluated cellular senescence by counting the SA-β-gal-positive blue-stained senescent cells after arctiin and/or H 2 O 2 treatment under the same experimental conditions shown in Figure 3A. H 2 O 2 treatment increased the percentage of senescent cells by 24.89% compared with untreated control cells, however, 20 μM arctiin pretreatment the percentage of senescent cells by 10.89% compared with H 2 O 2 -treated cells ( Figure 4). These results indicate that arctiin negatively regulates H 2 O 2 -mediated senescence in HHDPCs.

Arctiin alters H 2 O 2 -mediated changes in miRNA expression
miRNAs are important small non-coding RNA molecules, and exerts their biological functions by posttranscriptionally regulating those of their target genes [26,27]. Numerous studies have demonstrated that miRNAs regulate 4 major biological functions, such as development, proliferation, differentiation and apoptosis [28,29]. Furthermore, accumulating results have shown that altered miRNA expression profiles are involved in UVB-or H 2 O 2 -protective effects and even androgenetic alopecia in human skin cells [23,[30][31][32]. Therefore, we next sought to understand the putative molecular mechanism underlying the protective effect of arctiin against H 2 O 2 in HHDPCs by analyzing miRNA expression changes. We performed miRNA microarrays and identified 30 miRNAs that were differentially expressed following arctiin pretreatment and H 2 O 2 exposure as compared with H 2 O 2 -treated cells. Eighteen and 12 miRNAs were upregulated and downregulated more than 2.0-fold, respectively ( Table 1). The most dysregulated miRNAs were miR-602 (5.74-fold increase) and miR-1290 (5.80-fold decrease). These findings indicate that arctiin regulates the expression levels of specific miR-NAs in HHDPCs.
To investigate the cellular effect of the altered miR-NAs on HHDPCs, we selected the altered specific miR-NAs and gathered the list of putative target mRNAs of the miRNAs using a target prediction tool (MicroCosm Targets ver. 5), and then determined the biological functions associated with the target genes by Gene Ontology (GO) analysis using AmiGO bioinformatic tools. Our data show that arctiin has a protective effect against H 2 O 2induced cellular senescence and apoptosis in dermal papilla cells; therefore, we analyze GO of the target genes and categorize them into cellular processes including aging, skin development, apoptosis and cell proliferation. As shown in Tables 2 and 3, the altered miRNAs were functionally related in the four biological categories. Mir-602 was the most upregulated miRNA and has many targets including EDN1 and SOD2 (aging); APC (skin development); ERBB4, PPARG, and TP53BP2 (apoptosis); and STAT3, CDK9, and ID4 (cell proliferation). Mir-1290 was the most downregulated miRNA and targets SLC1A2 (aging), APC and COL8A1 (skin development), NOTCH1 and BMI1 (apoptosis), and ROBO1, CDC27 (cell proliferation).
Previous studies have indicated that arctiin-mediated alterations in miRNAs may be involved in regulated the four cellular mechanisms mentioned above. Also, as a matter of fact, the GO term contains bi-directional processes for each term. The term of 'apoptosis' includes both anti-apoptotic and pro-apoptotic processes. Therefore, GO analysis is not sufficient for understanding the biological functions of altered miRNAs in our study. To improve the accuracy of the biological meaning of the  target genes, the targets were analyzed using the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathwaybased enrichment analysis program in the bioinformatic database DAVID. As shown in Tables 4 and 5, the altered miRNAs may be involved in regulating pathways involved in cancer, cell cycle, and Wnt and MAPK signaling, among others. For example, miR-602 is putatively involved in regulating MAPK and insulin signaling pathways; however, miR-1290 is involved in cancer, focal adhesion, and insulin signaling pathways. Overall, the results indicate that the protective effect of arctiin against H 2 O 2 -induced alterations in HHDPCs may be regulated by arctiin-specific miRNAs and pathways that are possibly affected by miRNAs.

Discussion
The identification of novel anti-oxidant chemo-reagents is required to effectively treat or prevent alopecia due to androgen dysregulation or oxidative stress. Here, we used cell-based approaches, to demonstrate the potent antioxidant activity of the lignin derivative arctiin. Interestingly, pretreatment with arctiin drastically inhibited H 2 O 2induced decreases in viability in HHDPCs. Flow cytometry demonstrated that arctiin pretreatment also increased HHDPC resistance against cell death and G2 cell cycle arrest mediated by H 2 O 2 . Furthermore, those effects of arctiin were related to its anti-oxidative effect. Upregulation of intracellular ROS levels, which was mediated by H 2 O 2 , was dramatically inhibited following arctiin pretreatment. Recent reports have demonstrated that the loss of proliferative capacity in balding dermal papilla cells is associated with expression changes SA-β-gal and markers of oxidative stress [7]. We also found that arctiin inhibited H 2 O 2 -mediated upregulation of SA-β-gal activity in HHDPCs. These findings suggest that arctiin is a novel potent therapeutic agent for oxidative stress-induced cell dysfunction in HHDPCs.
Using miRNA microarrays, we identified 30 miRNAs that may have important roles in the arctiin-mediated protective effect against H 2 O 2 in HHDPCs. Although further experiments are needed to validate miRNA expression levels to confirm the microarray data, we focused on the biological meaning of the altered miRNAs in our study because the cellular functions of miRNA are dependent on their target mRNAs. Therefore, we used several bioinformatic tools to understand the biological meaning of the altered miRNAs. Our bioinformatical analysis showed that the miRNAs altered in response to arctiin pretreatment before H 2 O 2 stimulation are commonly involved in MAPK and Wnt signaling pathways. TAK1/MAP3K7 is a member of the MAP3K family, and it has been reported that impaired hair follicle morphogenesis and hair loss are mediated by TAK1/MAP3K7 deletion in mice [33]. Also, TAK1/MAP3K7 deficiency upregulates ROS levels, resulting in skin keratinocyte cell death [34]. Extracellular signal-regulated kinase (ERK), a member of the MAPK family, plays an important role in HHDPC proliferation. ERK signaling is activated by minoxidil, which is a widely used drug for treating androgenetic alopecia, and ERK inhibition blocks the anti-alopecia effect of the minoxidil [35]. In addition, our bioinformatic results showed that MAPK signaling was the most commonly targeted pathway for the downregulated miRNAs mediated by arctiin in HHDPCs (Table 5). This result indicates that MAPK pathway activation is important for HHDPC proliferation. However, our bioinformatic analysis revealed that the MAPK signaling pathway is also targeted by the upregulated miRNAs ( Table 4), indicating that inhibition of MAPK signaling pathway might be involved in protective effects against ROS in HHDPCs. It has been reported that ROS activates ERK/MAPK, and ROS-mediated ERK activation induces apoptosis and senescence in several cell lines [36]. Although the molecular relationship between alopecia and MAPK signaling pathway remains to be Our bioinformatic analysis also revealed that WNT signaling pathway is putatively targeted by the miRNAs altered following arctiin treatment. Wnt signaling has also been implicated in alopecia. A recent study demonstrated that Wnt10A, which is a member of the Wnt family, is involved in the etiology of androgenetic alopecia [37]. A Wnt10A deficiency causes deregulation of the hair cycle by shortening the anagen phase, which is observed in androgenetic alopecia hair follicles [37]. Funato et al. demonstrated that H 2 O 2 -induced ROS can regulate Wnt/β-catenin signaling pathways [38]. Also, it was recently reported that the minoxidil-mediated anagen prolongation effect is due to β-catenin pathway activation [39]. Although further investigations are necessary to clarify the molecular interplay between ROS and Wnt signaling pathway in hair follicles and in patients with alopecia, our results suggest that arctiin-mediated antioxidative effects in HHDPCs may be involved in regulating Wnt signaling.

PI-based cell cycle analysis
To analyze cells in different phases of the cell cycle, treated HHDPCs (4 × 10 3 ) were gathered by trypsinization and fixed by adding cold 70% ethanol at −20°C for 1 h. After fixation, cells were stained by incubating with PI staining solution (50 μg/ml PI, 0.5% Triton X-100, and 100 μg/ml RNase) at 37°C for 1 h. The distribution of each cell cycle phase was determined by evaluating the intensity of fluorescence PI staining using the FL2-H channel of a FACSCalibur (BD Biosciences, Franklin Lakes, NJ, USA).

DCF-DA-based ROS analysis
To analyze intracellular ROS levels in HHDPCs, treated cells were washed, trypsinized, and collected. Cells were diluted in 20 μM DCF-DA/phosphate-buffered saline (PBS) and incubated at room temperature for 1 h in the dark. After incubation, cells were washed once with PBS and subjected to flow cytometer-based fluorescence analysis using a BD FACSCalibur flow cytometer (BD Biosciences).

β-galactosidase (β-Gal)-based cellular senescence analysis
To analyze the level of cellular senescence in HHDPCs after arctiin and H 2 O 2 treatment, treated cells were gathered and fixed by the addition of 2% formaldehyde/0.2% glutaraldehyde. After fixation, senescence-associated β-galactosidase (SA-β-Gal) staining solution (Biovision, Milpitas, CA, USA) was added to the fixed cells and incubated at 37°C overnight. Senescent cells (positive blue color) were observed and counted using a bright-field microscope at × 200 magnification, and the percentages were determined.

Microarray-based miRNA expression analysis
To investigate which miRNAs are altered in our study, treated HHDPCs were gathered and lysed using TRIzol reagent (Life Technologies) for total RNA purification. Total RNA was extracted from the lysed cells according to the manufacturer's protocol and estimated its integrity and purity was estimated using an Agilent 2100 Bioanalyzer® (Agilent Technologies, Santa Clara, CA, USA) and a MaestroNano® microvolume spectrophotometer (Maestrogen, Las Vegas, NV, USA). We confirmed that the RNA samples had values integrity values higher than 8.0 and A260/280 and A260/230 values greater than 1.8. The qualified RNA samples were subjected to miRNA microarray analysis as described previously [40]. Briefly, RNAs were dephosphorylated and labeled with cyanine 3-pCp (Agilent Technologies). The labeled samples were dried and treated with GE Blocking Agent (Agilent Technologies) to reduce background or nonspecific binding to the probe onto the microarray. Then, the samples were hybridized to the SurePrint G3 Human v16 miRNA 8x60K microarray (Agilent Technologies) in the Agilent Microarray Hybridization Chamber (Agilent Technologies) for 20 h. After hybridization, the array was scanned using an Agilent SureScan Microarray Scanner (Agilent Technologies) and quantitated using Agilent Feature Extraction Software (version 10.7, Agilent Technologies). Derived data were analyzed using GeneSpring GX software, version 11.5 (Agilent Technologies). The data were filtered using flagpresent and t-tests to identify miRNAs for further analysis. miRNA expression was evaluated by assessing the fluorescence ratio between two samples. Those displaying >2.0fold increases or decreases were selected for further bioinformatic analysis.

Bioinformatical tool-based biological analysis of miRNAs
To investigate the biological meaning of miRNAs with significantly altered expression, we used three kinds of bioinformatical tools: MicroCosm Targets Version 5 (http:// www.ebi.ac.uk/enright-srv/microcosm/htdocs/targets/v5/), AmiGo 2 GO analysis tool and DAVID (Database for Annotation, Visualization and Interrogate Discovery, http:// david.abcc.ncifcrf.gov/home.jsp) bioinformatics resources v6.7 [41]. Using the first tool, we predicted the putative target genes of the altered miRNAs, and the second tool was used to group target genes into four categories: aging, skin development, apoptosis, and cell proliferation. Finally, the targets were analyzed using the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway-based enrichment analysis program in the bioinformatic database DAVID. The Ensembl transcript ID lists of target genes were gathered and subjected to pathway analysis using the KEGG program in DAVID bioinformatic resources. The Ease score, which is a modified Fisher's extract P-value, was fixed at 0.5 and meaningful KEGG pathways showing a value of >0.9% (percentage of involved target genes/total target genes) were selected.