Recombinant Human Keratinocyte Growth Factor Ameliorates Cancer Treatment-Induced Oral Mucositis on a Chip

Abstract

Oral mucositis (OM) is a severe complication of cancer therapies caused by off-target cytotoxicity. Palifermin, which is recombinant human keratinocyte growth factor (KGF), is currently the only mitigating treatment available to a subset of OM patients. This study used a previously established model of oral mucositis on a chip (OM-OC) comprised of a confluent human gingival keratinocytes (GIE) layer attached to a basement membrane-lined subepithelial layer consisting of human gingival fibroblasts (HGF) and human dermal microvascular endothelial cells (HMEC) on a stable collagen I gel. Cisplatin, radiation, and combined treatments are followed by a recovery period in the OM-OC to determine possible cellular and molecular mechanisms of OM under effects of KGF. Cancer treatments affected the keratinocyte layer, causing death and epithelial barrier loss. Both keratinocytes and subepithelial cells died rapidly, as evidenced by propidium iodide staining. In response to radiation exposure, cell death occurred in the apical epithelial layer, predominantly, within 24h. Cisplatin exposure predominantly promoted death of basal epithelial cells within 32-36h. Presence of KGF in OM-OC protected tissues from damage caused by cancer treatments in a dose-dependent manner, being more effective at 10 ng/mL. As verified by F-actin staining and the Alamar Blue assay, KGF contributed to epithelial survival and induced proliferation of GIE and HGF as well as HMEC within 120h. When the expression of eighty inflammatory cytokines is evaluated at OM induction (Day 12) and resolution (Day 18) stages in OM-OC, some cytokines are identified as potential novel therapeutic targets. In comparison with chemoradiation exposure, KGF treatment showed a trend to decrease IL-8 and TNF-a expression at Day 12 and 18, and TGF-β1 at Day 18 in OM-OC. Taken together, these findings support the utility of OM-OC as a platform to model epithelial damage and evaluate molecular mechanisms following OM treatment.

Keywords: chemotherapy; inflammatory cytokines; oral mucositis on a chip; photocrosslinking; radiation therapy; recombinant human keratinocyte growth factor.

Figure 1.. Oral mucositis on a chip (OM-OC) assembly and analysis.

(A) Overview of the three-channel microfluidic device for triculture tissue assembly and immunofluorescence micrograph showing GIE (green), HGF (red), and HMEC (blue). (B) Timeline of triculture tissue assembly, showing cell number and collagen concentration added in the chip channels, followed by cancer treatment exposure and inflammatory cytokine analysis. Information about cell culture media, polymerization and photo-crosslinking process conditions as well as when cells were exposed to cancer treatments and the inflammatory profile performed is indicated in the boxes. GIE: human gingival keratinocyte; HGF: human gingival fibroblast; HMEC: human dermal microvascular endothelial cell; Ru/SPS: Ruthenium/Sodium persulfate.

Figure 2.. Cell and tissue effects of cancer treatment-induced OM-OC.

Epifluorescence images of the triculture tissue chip (A) untreated and treated with (B) chemo, (C) radiation, and (D) chemoradiation at D12 (24h). The epithelial layer was divided into apical and basal epithelium regions by a dashed blue line as indicated in (A-D). Dashed white regions outline the voids within epithelium caused by cancer treatments, while solid rectangles outline the PDMS pillars. (E) F-actin area fraction of apical versus basal epithelial regions within four groups. (F) GIE coverage and (G) F-dextran permeability of the tissue chip exposed to cancer treatments versus untreated chips. The GIE coverage and F-dextran permeability were measured as described in Figure S1. Data = mean±SEM. Dunnett’s test was used to compare treatment groups to control in (E) or later timepoints to a pre-treatment timepoint within each group in (F, G). Results of pairwise comparisons are indicated with ∗ (p<0.05), ∗∗ (p<0.01), and † (p<0.001).

Figure 3.. Tissue layer-specific cell death response of the OM-OC.

(A) Timelapse images of the rad-treated tissue chip cultured in media supplemented with PI dye to detect dead cells (stained in red). The cell death profile is expressed as a death area fraction and measured as described. The dashed green region outlined the epithelial region-of-interest (ROI), and the dashed yellow region outlined the sub-epithelial ROI for death area fraction measurements. (B) Cell death profiles of (i) epithelial and (ii) sub-epithelial cells in response to different cancer treatments. Data = mean±SEM. Dunnett’s test was used to later time points to a pre-treatment time point within each group. Results of pairwise comparisons are indicated with ∗ (p<0.05), ∗∗ (p<0.01), and † (p≤0.001).

Figure 4.. Morphological alterations of OM-OCs in response to chemorad/KGF treatment.

Representative time-lapse images showing the responses of tissue chips (A) before, (B) during peak OM damage, and (C) recovery due to chemoradiation in the presence of KGF of varied concentrations: (i) 0, (ii) 1, (iii) 4, and (iv) 10 ng/mL. Dashed yellow regions outlined the voids within the epithelium during OM induction. Yellow arrows indicated damaged HGFs/HMECs with altered morphology, while red arrows indicated dead HGFs/HMECs. Notably, GIEs within (C)-(i-ii) were in stretched/irregular shapes, indicating more stretching and spreading, while GIEs within (C)-(iii-iv) were in more regular shapes, with the epithelial layer almost the same as pre-exposure.

Figure 5.. OM-OC responses to chemorad in the presence of KGF.

(A) Cell death profiles of (i) epithelium and (ii) sub-epithelium of tissue chips in response to chemoradiation (chemorad) with KGF of varied concentrations (chemorad/KGF; 0, 1, 4, and 10 ng/mL). (B) GIE coverage and (C) F-dextran permeability of chemorad-treated and chemorad/KGF-treated groups during OM induction and recovery. Data = mean±SEM. Dunnett’s test was used to later time points to a pre-treatment time point within each group. Results of pairwise comparisons are indicated with ∗ (p<0.05), ∗∗ (p<0.01), and † (p≤0.001).

Figure 6.. Inflammatory responses of OM-OC to cancer and KGF treatment.

Cytokine expressions were assessed through culture media collected from OM-Ocs using cytokine antibody arrays, averaged over n=3–6 independent trials (sampling size of each condition is indicated in B). The cytokine expressions of each condition were normalized to the untreated chip on Day 12. (A) Colormaps of 80 inflammatory cytokines where high expression is depicted in green, low expression is depicted in magenta, and (x) is undetected compared to untreated chips on Day 12. (B) The table summarizes the expression of cytokines associated with OM (e.g., IL-1α, IL-1β, IL-2, IL-6, IL-8, IL-13, TNF-α, and TGF-β) of each condition compared to that of the untreated group. The expression comparison was defined as the treated group’s log2 expression value of specific cytokines normalized to the untreated control (i.e., D12_UNTX log2 expression values were all detectable and set to 1, indicated by -). The up green arrow indicates higher expression (log2>1), down magenta arrow indicates lower expression (log2<1), and (x) indicates undetected (log2=N/A) levels of specific cytokines in the treatment group compared to the untreated counterpart. D12/D18_C: chemo-treated group on Day 12/18; D12/D18_R: rad-treated group on Day 12/18; D12/D18_CR: chemorad-treated group on Day 12/18; D12/D18_K: chemorad/KGF-treated group on Day 12/18; D12_UNTX: untreated group on Day 12.