The human skin scaffold model used in this study was commercially obtained from a certified supplier (MatTek Corporation, EpiDerm). All tissues were derived from de-identified human donors, provided under compliance with ethical standards and the Declaration of Helsinki. No living human participants were recruited, and no identifiable information was used. As such, institutional review board (IRB) approval and informed consent were not required. Swine PBMC isolation was done in accordance with and approved by the North Carolina State University Institutional Animal Care and Use Committee (IACUC ID# 18-084-B).

The following disperse (D) dyes were used in this study: D. Blue 1 (Sigma 215643-5G), D. Blue 124 (Sigma 21620-5G), D. Blue 79.1 (Sigma 75497-74-4), and D. Blue 183 (ACM2309946). Solubility was tested in different solvents and at different concentrations to determine the optimal solubility of all dyes and with the highest cell viability and low toxicity from the solvent. Acetone and DMSO (Corning 25-950-CQC) were selected as solvents. The dissolved dyes were centrifuged to remove insoluble particles or possible contamination for the cell culture.

Swine naive PBMC viability was measured via flow cytometry, and they were also stained with antibodies for cytokines TNF-α (Biolegend 502936) and IFN-γ (BD Biosciences 559812). Precisely, PBMCs from adult pigs, isolated by density centrifugation using Ficoll-Paque-Premium (GE Healthcare, Uppsala, Sweden) in SepMate tubes (StemCell Technologies, Cambridge, MA, USA) were cultured in 96-well plates at 2 × 10⁵ cells/well in RPMI-1640 (Corning, Corning, NY, USA) supplemented with 10% FBS (VWR, Radnor, PA, USA) and 1× antibiotic-antimycotic (Corning, USA) in the absence or presence of 483.8 ug/well (6) of each textile dye (concentration calculated according to Cortes et al. 2025) and 2.5 μg/mL Con A as a stimulant. After 24 h of culture in the incubator at 37^๐ C and 5% CO₂, octuplicate wells from each dye were pooled and stained for flow cytometry analysis using eBioscience Fix/Perm kit (ThermoFisher Scientific, USA) according to the manufacturer's protocol. Antibodies for CD3, clone PPT3 (Southern Biothec), CD4 clone 74-12-4 (BEI Resources) were titrated and used 10 uL of prediluted antibody/well at concentrations ranging from 0.1 µL/well (=20 ng/well) to 0.2 uL/well. PBMC viability was measured by flow cytometry using LIVE/DEAD™ Fixable Near-IR Dead Cell Stain Kit (Invitrogen, USA), according to the manufacturer's protocol. Once PBMCs were gated for single live cells, they were gated on CD3+CD4+ cells to determine expression of TNF-a and IFN- γ. Data was acquired on a Beckman Coulter CytoFlex using the CytExpert software (Beckman Coulter, Brea, CA, USA). Data analysis was performed using FlowJo v10.5.3 (FLOWJO LLC, Ashland, OR, USA).

To answer the question if the different textile dyes lead to an overall downregulation or upregulation of proteins involved in epithelial inflammation or compromised barrier integrity, IPEC cells exposed to dyes at a concentration of 2,424 µg/mL in 15 mL of culture medium, for a total of 36.36 mg per T75 flask (6) and exposed to the different dyes in DMEM-F12 supplemented with 10% FBS (VWR, Radnor, PA, USA) for 24 h (this time was sufficient to observe meaningful transcriptional changes). Dye uptake was confirmed by microscopy. After detaching the cells and a round of washes, RNA from IPEC cells was isolated using (Zymo RNA purification kit, USA) according to the manufacturer's protocol. RNA preps were diluted to a concentration of 20 ng/μL. Later, IPEC cells underwent molecular profiling using the nCounter Analysis system. Using a uniquely designed codeset (nCounter XT CodeSet Gene Expression Assay NanoString Technologies Seattle, WA), hybridization buffer was added to the reporter CodeSet to create a master mix. An aliquot of the master mix was mixed with 5 μL of each RNA sample, followed by the capture probe. The samples were placed at 65 °C in a pre-heated thermal cycler for at least 18 h. One sample per lane was loaded and then processed in the nCounter (NanoString Technologies) according to instructions and using 555FOVs counts. CLDN-4, VEGF, CCL-20, CXC9, IL-1RA, IL-16 were analyzed with the custom nanostring panel for each dye.

Human full thickness epidermal skin scaffolds (Mattek, USA) derived from differentiated 3D tissue model from normal human epidermis (EFT-412-BW) under air-liquid interface (ALI) were first upon received, equilibrated with hydrocortisone free media with 2.5 mL in 6 well plates overnight (EFT-400-MM-HCF). Considering that a T-shirt is dyed with ∼2 g of dye for a surface of∼4,948 cm²∼and that the surface of the Epiderm tissues is 1 cm², we used 404 ug of the mentioned dyes (6) and incubated them with the cultures for 24 h of each textile dye (concentration and time calculated according to Cortes et al. 2025). Dye uptake was confirmed by microscopy. Tissues were then harvested, and histological sections were stained for H&E and were scored for epidermal damage (scale 0–3) by a blinded licensed pathologist. The scoring was done as follows: Epidermis scores 0 = Normal thickness with differentiated layers, 1 = Normal thickness with differentiated layers and scattered epithelial apoptosis, 2 = Loss of one or more layers of epidermis with scattered epithelial degeneration or apoptosis, 3 = Only stratum basal and stratum corneum with scattered epithelial degeneration or apoptosis.

Data are expressed as mean ± standard deviation (SD) from at least three independent experiments. For comparisons involving more than two groups, a one-way analysis of variance (ANOVA) was performed. Where ANOVA indicated significant differences, Dunnett's multiple comparisons test was used to compare each textile dye treatment group to the media control. A p-value of <0.01 was considered significant. All analyses were done with GraphPad Prism version 10.2.3 (San Diego, CA, USA).

PBMC assays revealed distinct cytokine and viability profiles: Disperse Blue 124 (MW: 377.42 g/mol, high hydrophobicity, complex aromatic functional groups) strongly activated IFNγ and TNF-α production, maintaining moderate viability. Blue 1 (MW: 268.27 g/mol, anthraquinone structure, moderate hydrophobicity) displayed lower cytokine production but high cytotoxicity. Blue 79.1 (MW: 625.4 g/mol, intermediate hydrophobicity and structure complexity) and Blue 183 (MW: 473.32 g/mol, stable azo dye, lower hydrophobicity) elicited none to low immune responses with minimal cytotoxicity (Tables 1, 2; Figures 1a–c).

Nanostring analysis for IPEC cells highlighted significant CCL20, IL-1RA, CLDN-4, and VEGF upregulation for dyes Blue 1 and Blue 124, correlating with their chemical features favoring cellular stress and inflammation. CLDN-4 and VEGF were upregulated, and this could suggest alterations in epithelial barrier integrity and angiogenesis-related processes (Table 2; Figure 1d). CXCL9 and IL-16 were notably downregulated, across treatments with dyes Blue 1 and Blue 124, which could have implications for immune signaling and inflammation, especially in barrier tissues like the gut or skin indicating potential suppression or regulation of these cytokines suggesting immunomodulatory effects linked to structural characteristics. CXCL9 is a chemokine induced by IFN- γ that attracts activated T cells. Its downregulation could imply reduced T cell recruitment to the epithelial barrier and potential dampening of the Th1 mediated inflammation. IL-16 is a chemoattractant for CD4+ T cells involved in modulating epithelial cell signaling. Its downregulation could imply reduced CD4+ T cell chemotaxis and local immune activation, as well as possible reduced immune surveillance at the epithelial cell surface. It could also indicate epithelial stress or damage.

In all slides, the dermis lacked well-formed collagen bundles; however, scattered spindle shaped cells (likely fibroblasts) were present. Several sections had acutely affected epithelium at the margins of sections. This was suspected to be from tissue handling. Histopathological analysis of human skin scaffolds revealed consistent trends (Table 3). Blue 124 induced the highest epidermal damage (Score = 3), with extensive keratinocyte necrosis and structural disorganization. Blue 1 produced moderate damage (Score = 2), while Blue 79.1 and Blue 183 resulted in only mild focal necrosis (Score = 1). These results strongly correlated with cytokine induction and gene expression data. Histology confirms that both immune activation and cytotoxicity in PBMC/IPEC models translate into visible tissue damage in skin. The correlation between dye structure and epidermal damage may strengthen the hypothesis that chemical properties (size, hydrophobicity, reactivity) are central to skin toxicity. The damage pattern (e.g., vacuolization, nuclear fragmentation) matches what we expect from chemical irritants or sensitizers.

A one-way ANOVA revealed a significant effect of textile dye treatment on IFN-γ and TNF-α production (p < 0.0001). Post-hoc Dunnett's test showed that Blue 124 significantly increased IFN-γ and TNF-α+ T cells compared to media control (p < 0.0001) and a significant reduction of cell viability for D. Blue 1 (p < 0.0001).