A combination of three antioxidants decreases the impact of rural particulate pollution in Normal human keratinocytes

It is well established that exposure of human skin to airborne pollution, particularly in the form of particulate matter sized 2.5 μm (PM2.5), is associated with oxidative stress, DNA damage and inflammation, leading to premature signs of skin aging. Because much of the damage results from oxidative stress, we examined the effects of a topical composition containing three antioxidants in an in vitro model system to assess the potential for amelioration of premature aging. The use of multiple antioxidants was of interest based on the typical composition of therapeutic skincare products. It is important to determine the efficacy of multiple antioxidants together and develop a short‐term assay for larger scale efficacy testing.

amenable to high through-put screening.The results obtained reinforce the concept that a multiple antioxidant preparation, topically applied, may reduce proinflammatory signalling and cellular damage and thereby reduce premature skin aging due to exposure to rural-derived airborne pollution.

K E Y W O R D S
antioxidants, keratinocytes, oxidative stress, particulate pollution

INTRODUCTION
Air pollution is often associated with industrialized urban areas with dense populations; however, rural areas are also exposed to pollution, including particulate matter (PM), in part, due to air currents.A pioneering epidemiological study published in 2010 provided evidence that decades of exposure to air pollution was associated with signs of skin aging in women from the Ruhr and two rural counties north of the Ruhr district in Germany [1].Their primary exposure to airborne PM was from traffic and steel and coal industries.
Particulate matter, its sources and characteristics have been thoroughly reviewed.Open fires, industrial facilities, power plants and vehicle exhaust are primary sources [2,3], and PM can be divided into three types depending on size; ultrafine particles (UFP), fine particles (PM 2.5 ) and coarse particles (PM 10 ).There is sufficient evidence to establish that PM 2.5 penetrates skin cells and has a major role in the impact of air pollution on human health and is closely associated with PM-induced skin disease [4,5].Epidemiological studies have shown that exposure of humans to airborne particulate matter (PM) is associated with increased signs of extrinsic skin aging including wrinkle formation and pigment spot formation [1,[6][7][8][9].
The ability of PM to stimulate oxidative stress is a critical factor in the initiation of pathways leading to observed cytotoxic responses in the skin [10].Therefore, the primary therapeutic strategy advanced to counteract the adverse effects of PM on human skin is the design of safe and effective topical products containing functional antioxidant ingredients able to counteract pollution-induced oxidative stress.To this end, we conducted cell culture experiments to both confirm the specific signalling pathways involved in keratinocyte response to PM 2.5 and to determine the ability of three combined antioxidants (AOx) to counteract such pathways.Representative signalling molecules known to be associated with accelerated skin aging or response to polycyclic aromatic aryl hydrocarbons [11] were monitored following treatment of normal human keratinocytes (NHEK) with PM 2.5 ± AOx.As reviewed recently, pollution has been shown to induce IL-6, CXCL10, CYP1A1, MMP-1 and nuclear factor erythroid 2-related factor 2 (NRF2) in human skin in vivo, as well as normal human epidermal keratinocytes and dermal fibroblasts in vitro [12][13][14].Induction of IL-6 has been shown to occur following oxidative stress in vitro and in vivo by exposure to both PM 2.5 and UVR [15,16].CXCL10 is a keratinocytederived inflammatory mediator [17,18] that can stimulate keratinocyte migration and proliferation as well as wound re-epithelialization.Although not commonly monitored in work studying pollution effects on skin, we chose CXCL10 because of its association with skin inflammation, playing a key role in atopic dermatitis pathogenesis.CYP1A1 plays a critical role in the activation and detoxification of specific components of pollution [19].The NRF2 pathway is a key mediator in the response to oxidative stress [19,20].Interestingly, exposure to various pollutants has been linked to both induction and suppression, as well as nuclear translocation of NRF2 [13], however at least one report has shown that two pharmacological agents (SK-119 and SH-29) depend on activation of NRF2 to protect against damage to HaCat cells from particulates [21].
Therefore, AOx efficacy measures in this study included MMP-1, IL-6, CXCL10, NRF2 and CYP1A1.In addition, RNA sequencing (RNA-seq) was used to monitor transcription of genes that either participate in protective antioxidant functions or are known to be an integral part of aryl hydrocarbon receptor and signalling pathways.The treatment mixture (AOx) comprised three components (resveratrol, niacinamide and GHK), which individually have considerable published evidence for their antagonism of oxidative stress and inflammation in human skin.For example, resveratrol has been shown to inhibit the PM-induced inflammatory response in human keratinocytes [22], GHK (glycyl-L-histidyl-Llysine) has been shown to modulate skin repair processes [23] and niacinamide has been shown to reduce the proinflammatory response of keratinocytes to both UVR and PM 2.5 [24].

Cell culture
NHEK (normal human neonatal epidermal keratinocytes, pooled) were purchased from Thermo-Fisher; experiments were completed within 6 passages after primary culture was established.Cells were cultured in EpiLife Medium without phenol red with Growth Supplement HKGS (Thermo-Fisher).Experiments were initiated 48 h after establishing test NHEK cultures.

Exposure to rural PM 2.5
Rural PM 2.5 (1 mg/mL in water) was provided by Dr. Terry Gordon and was collected during the summer months of 2009 from Goshen, NY [25].The metal composition of the Goshen-derived particulate matter was the following (ng/mL): arsenic 3.026, cadmium 0.362, chromium 1.117, cobalt 0.023, lead 5.914, nickel 1.173, iron 8.039, magnesium 112.697, calcium 277.780 and zinc 27.354.Prior to adding to cells, rural PM 2.5 was sonicated at a low power setting (160 Watts) for 5 min.To determine the optimal dosage of PM, cells were treated with increasing concentrations, viability monitored and a minimally cytotoxic concentration (cytotoxicity ≤10%) used for the remainder of the work.

Antioxidants
The AOx blend was composed of 3 mM of GHK peptide (Lucas Meyer Cosmetics), 1.6 mM of niacinamide (Sigma-Aldrich) and 0.1 mM of resveratrol (Sigma-Aldrich).AOx blend concentrations of 1%, 0.5%, 0.1% and 0.01% of were tested initially to determine a noncytotoxic range.The concentration of 0.01% was selected for future experiments based on >90% viability after 20 h of exposure.Cells were treated with AOx for 30 min before exposure to PM 2.5 for 5 h.To initiate exposure to pollution, the initial growth media were replaced with media containing vehicle only, AOx only, PM 2.5 only or AOx plus PM 2.5 .

Cell viability
Mitochondrial metabolic activity served as an indicator of cell viability and cytotoxicity.Cell viability/cytotoxicity was measured by MTT (Sigma-Aldrich) according to manufacturer's instruction.

Assessment of oxidative stress, reactive oxygen species
Cells were pre-treated with AOx or vehicle control for 30 min, and then media were replaced with media containing vehicle only, AOx only, PM 2.5 only, or AOx plus PM 2.5 .Cell lysates were collected 5 h following exposure.Reactive oxygen species were quantified using 7′-dichlorofluorescin diacetate (DCFDA) (Abcam) as according to manufacturer instructions.Fluorescence was normalized to protein concentration measured by BCA protein assay (Millipore).

IL-6 and MMP-1 measured by ELISA assays
Cell media collected 5 h following the same treatments as above were analysed by ELISA for IL-6 (R&D Systems) and MMP-1 (Thermo-Fisher) according to the manufacturer's instructions.Values were normalized to protein concentration measured by BCA protein assay (Millipore).

RNA sequencing analysis of differentially expressed genes
Total RNA samples were extracted from NHEK using TRIzol reagent (Life Technologies).Following TRIzol isolation, RNA was purified using the RNeasy Plus Micro Kit (Qiagen).RNA-seq library preparation and sequencing were performed at the NYU Langone's Genome Technology Center.Briefly, RNA-Seq libraries were prepared using Illumina TruSeq Stranded mRNA sample preparation kit according to the manufacture's instruction.Sequencing was performed in the paired-end 50 bp run using Illumina NovaSeq 6000 system.
RNA-Seq data analysis was performed using CLC Genomics Workbench Version 21.0.5 (Qiagen).Raw sequence data (FastQ) was trimmed to remove any remaining adaptors and ambiguous nucleotides.The trimmed sequence files were aligned to human genome (Hg38) allowing two mismatches.Reads mapped to the exons of a gene were summed at the gene level.Gene expression levels were quantified as total read per million (TPM).Differential gene expression was analysed using the Advance RNA-seq plug-in tool (Qiagen) by comparing each treated group versus corresponding control group.TMM (trimmed mean of M values) normalization was performed to adjust library sizes before differential expression analysis.The genes with a false discovery rate (FDR) <0.05 between control and treated group and an average mean of total counts no less than 10 were defined as differentially expressed genes (DEGs).

Data analysis
Data analysis employed one-way ANOVA for the effects of PM 2.5 and AOx on primary keratinocytes with an n between 3 and 6 for each experimental group.At least three separate experiments were performed for each endpoint measured.Differences were judged to be significant "*" with a p-value of <0.05.

Determination of optimal noncytotoxic exposure conditions for PM 2.5 and AOx
Experiments were performed to determine the optimal concentration for subsequent studies.Dose response curves for cytotoxicity/viability are shown in Figure 1.Using a concentration with minimal impact on cellular viability, treatment with 5 μg/mL of particulate pollution significantly induced ROS (Figure 2a) when compared to control unexposed cells.Pre-treatment with AOx significantly attenuated this response and AOx alone did not induce ROS (Figure 2a).Based on both a low cytotoxicity and induction of reactive oxygen species (ROS) by PM 2.5 , and prevention of ROS by 0.01% AOx, it was determined that 0.01% AOx and 5 μg/mL PM 2.5 would be used for the remainder of the studies.In addition, we selected a short treatment time frame so that future screening assays are more practical.

IL-6 protein release by NHEK
Accumulation of ROS is associated with skin inflammation via the production of pro-inflammatory cytokines such as IL-6 [26].Therefore, we examined changes in the inflammatory response by measuring IL-6 production in NHEK cells following exposure to rural PM 2.5 for 5 h.The exposure to PM 2.5 significantly stimulated IL-6 extracellular protein release (Figure 2b).Pre-treatment of NHEK cells with AOx attenuated this response, in addition to significantly reducing basal release of IL-6.

Induction of CXCL10 mRNA by PM 2.5 and inhibition by AOx in NHEK
Support for the use of AOx for prevention of pollution damage is well illustrated by the results for CXCL10 mRNA (Figure 2b).Treatment of NHEK with AOx alone tended to suppress CXCL10 mRNA, but not statistically significantly.Treatment of keratinocytes with particulates significantly increased the level of CXCL10 mRNA, as expected, and pre-treatment of cells with AOx significantly inhibited this induction.
F I G U R E 1 Evaluation of viability and ROS in NHEK following treatment with antioxidants or PM 2.5 .Primary human keratinocytes were incubated for 5 h with varying concentrations of (a) AOx or (b) PM 2.5 and then analysed for viability by MTT.(c) Cells were pre-treated with AOx or vehicle control for 30 min, then exposed to vehicle, Aox, 5 μg/mL PM 2.5 or PM 2.5 plus Aox for 5 h and then monitored for ROS.Data are from at least three separate experiments and presented as means ± SD.A statistical analysis was performed with a one-way analysis of variance (ANOVA).Significant differences (*p < 0.05) between experimental groups are indicated by an asterisk with n ≥ 3.

NHEK expression of CYP1A1 mRNA
Components of PM 2.5 have been shown to upregulate aryl hydrocarbon (AhR) pathways, including CYP1A1, an enzyme involved in host response to inflammation and cellular stress (reviewed in Ref. 13) and an important enzyme in the metabolic transformation of molecules such as benzo(a)pyrene into carcinogens.Here, PM 2.5 significantly increased CYP1A1 mRNA expression, as expected (Figure 2c).Pre-treatment with AOx suppressed not only basal expression of CYP1A1 but also significantly reduced its induction by PM 2.5 .

MMP-1 protein release by NHEK
Figure 2d illustrates the changes in MMP-1 in the experimental groups compared with vehicle control.Treatment with the antioxidant mix significantly reduced MMP-1 release into the media while treatment with PM 2.5 elevated MMP-1.Levels of MMP-1 induced by PM 2.5 were significantly reduced by pre-treatment of cells with AOx.

Induction of NRF2 protein
We next examined changes in the antioxidant master regulator, NRF2.Cells were pre-treated with AOx or vehicle for 30 min and then treated with PM 2.5 for 5 h.Western blot analysis revealed that AOx significantly up-regulated NRF2 protein expression compared to control (Figure 3).There was not a significant change in NRF2 due to exposure to PM.

RNA sequencing analysis identified DEGS
To explore the gene expression changes induced by PM 2.5 , antioxidant treatment, or the combination, RNA sequencing analysis (RNA-seq) was performed in NHEKs treated with vehicle, AOx, PM 2.5 or a combination of AOx and PM 2.5 (AOx + PM).Four biological replicates were collected from each treatment group.A total of 288, 54 and 281 DEGs (FDR < 0.05) were identified in cells treated with AOx, PM and AOx plus PM 2.5 , respectively (Figure 4a).Interestingly, the majority of genes altered by PM 2.5 are upregulated genes (80%) while more than 70% DEGs induced by AOx treatment are down-regulated (Figure 4a).Further analysis revealed that 30/45 of PM 2.5 up-regulated DEGs and 4/9 of PM 2.5 down-regulated DEGs (PM) were altered by the combination treatment (AOx + PM 2.5 , Figure 4b), suggesting that co-treatment of AOx could attenuate or reverse more than half of DEGs induced by PM 2.5 .These changes were more pronounced in pro-inflammatory related genes (Table 1).For example, PM 2.5 alone significantly upregulated the genes related to AhR signalling and cytokine production.However, a noticeable reduction (e.g.CYP1A1, CYP1A2, CYP1B1, IL8 and CCL20) or a complete reversal (e.g.IL23A, SOD2 and CSF2) of gene expression was observed in cells pre-treated with AOx.Additionally, A statistical analysis was performed with a one-way analysis of variance (ANOVA).Significant differences (*p < 0.05) between experimental groups are indicated by an asterisk with n ≥ 3. consistent with up-regulated NRF2 protein, several NRF2inducible genes were upregulated in cells treated with AOx alone, including HMOX1, TXNRD1, GSR and EPHX1.These genes were inversely regulated by PM 2.5 compared to PM 2.5 plus AOx (Table 1).Moreover, many MMPs exhibited opposite expression changes between PM 2.5 alone and PM 2.5 plus AOx.Thus, pre-treatment of NHEKs with AOx suppressed many pro-inflammatory related genes induced by PM 2.5 .

DISCUSSION
Particulate matter (PM), an important component of air pollution, has been associated with inflammatory skin disorders, including atopic dermatitis, eczema, psoriasis and extrinsic skin aging [1,[27][28][29][30].In this study, we first determined doses of PM 2.5 in normal human keratinocytes in culture that would result in reproducible oxidative stress with minimal cytotoxicity.We then treated cells F I G U R E 3 Nrf-2 protein expression.NHEK were treated with AOx or vehicle control for 30 min and then exposed to vehicle, 5 μg/mL PM 2.5 , AOx or both PM 2.5 and AOx.Cell lysates were then collected after 5 h and total protein determined before analysis.NRF2 protein was detected by Western blot after application of equal protein in each sample.A statistical analysis was performed with a one-way analysis of variance (ANOVA).Significant differences (p < 0.05) between experimental groups are indicated by an asterisk with n ≥ 3. T A B L E 1 Expression of NRF2-related, AhR-related, pro-inflammatory cytokine-related and MMP-related genes in NHEK cells treated with AOx alone, PM 2.5 alone or the combination of AOx and PM 2.5 .The colour from green to red indicates the fold change of DEGs relative to untreated control from low to high.

Category
Gene with 5 μg/mL PM 2.5 sourced from a rural location [25].It should be noted that the PM 2.5 used in this study was rich in heavy metals.It is known that PM exposure causes a set of proinflammatory signalling in human epidermal skin cells.Therefore, because damage from PM 2.5 is associated with ROS generation and expression or release of inflammatory and remodelling mediators such as IL-6, NRF2, MMP1 and CYP1A1 [13,15,[19][20][21], we used those endpoints to monitor the efficacy of the AOx to block these signs of damage.CXCL10 was also used as an endpoint because of its association with inflammatory skin conditions [14,18,31].Several novel observations were generated from this study.First, in these experiments, PM 2.5 was shown to increase oxidative stress in normal human keratinocytes over a relatively short period of time following one exposure.Many previous studies examining the effects of PM2.5 in normal human keratinocytes required 24 h of exposure or even longer term with multiple applications [13].In addition, particulate matter showed evidence of disrupted epidermal cellular integrity, which was also counteracted by pre-treatment with a blend containing three antioxidants.Most skin care products have multiple active ingredients, and so we employed a model that replicated this characteristic.The model used here also had a relatively unique aspect in that a rural source of pollution was used.It has been established that PM varies in chemical composition based on a number of factors including location [32] and many consumers associate pollution primarily with urban areas.However, it has been established that rural areas can have high levels of toxic hydrocarbons and toxic metals in the ambient air, and for that reason we chose a rural source for the PM used here, and one that has not been used before for this type of study.
A brief pre-treatment of human keratinocytes with the specific AOx formulation used in this study significantly reduced the generation of ROS by PM extracellular release of the pro-inflammatory marker, IL-6, induced by PM 2.5 , indicating that this combination of AOx is an effective agent for reduction of environmental stress in skin cells.Treatment of cells with AOx also lowered baseline (no PM 2.5 ) levels of CXCL10 and CYP1A1 as well as counteracting the increases induced by to PM 2.5 .Taken together, this is a strong argument that relatively modest levels of therapeutic antioxidant mixtures can ameliorate oxidative stress due to air pollution in skin cells.The significance of NRF2 modulation is particularly interesting in that NRF2 is a master transcription factor that binds to antioxidant response elements (ARE) and regulates the expression of antioxidant enzymes that protect epidermal cells from other categories of environmental stress, such as ultraviolet light [33].A recent review of sun protection noted that extract of Glycyrrhiza inflata root acts to inhibit UV-induced ROS generation by activating NRF2 in primary human fibroblasts [34].In addition, a recent report suggested that NRF2 activation can reverse the deleterious action of air pollution [21].Those findings reported for NRF2 involvement in the response to environmental stress are reinforced by the observations in this study.Here, we show that specific antioxidants, in combination, can induce NRF2 both in unexposed human epidermal keratinocytes and additionally in PM 2.5 -exposed cells.
In agreement with previous studies, PM 2.5 induced CYP1A1.This induction was suppressed by pre-treatment with AOx, therefore, at least in theory, reducing activation of specific toxic and carcinogenic metabolites of PM 2.5 such as benzo(α)pyrene transformation to benzo(α)pyrene diol epoxide.As shown by the RNA-seq findings, AOx inhibited the expression of major enzymes involved in phase I xenobiotic metabolism (e.g.CYP1A1 and CYP1B1) and protected cells from the potential harmful active intermediates generated from the metabolism of PM 2.5 .Together, these data suggest antioxidant intervention may protect keratinocytes from PM 2.5 -induced AhR-mediated inflammation and carcinogenesis.These results make a strong argument for including multiple antioxidant ingredients, such as those in AOx, in topical formulations.The data presented also demonstrate a group of short-term in vitro assays that could help screen potential therapeutic formulations for counteracting skin inflammation and damage due to air pollution.

F I G U R E 2
Molecular markers generated by NHEK treated with AOx or vehicle control, followed by exposure to 5 μg/mL PM 2.5 .Cells were pre-treated with AOx or vehicle control for 30 min and then exposed to vehicle, AOx, 5 μg/ mL PM 2.5 or PM 2.5 plus AOx for 5 h.Cells were then monitored for (a) IL-6 release, (b) CXCL10 mRNA, (c) CYP1a1 mRNA and (d) MMP-1 protein.IL-6 AND MMP-1 were normalized by total protein.Data are from at least three separate experiments and presented as means ± SD.

F
I G U R E 4 RNA-seq analysis.RNA-seq analysis of NHEKs treated with AOx alone, Goshen PM 2.5 alone or the combination of AOx and Goshen PM 2.5 .(a) Overview of differentially expressed genes (DEGs) in three treatment groups compared to untreated controls.(b) Venn Diagram showing the number of up-and down-regulated DEGs in three treatment groups.The colour from green to red indicates the fold change of DEGs relative to untreated control from low to high.

symbol Gene name Fold change/control
2.5 .Pre-treatment of keratinocytes with AOx blocked T A B L E 1 (Continued)