A high-throughput, 28-day, microfluidic model of gingival tissue inflammation and recovery

Abstract

Nearly half of American adults suffer from gum disease, including mild inflammation of gingival tissue, known as gingivitis. Currently, advances in therapeutic treatments are hampered by a lack of mechanistic understanding of disease progression in physiologically relevant vascularized tissues. To address this, we present a high-throughput microfluidic organ-on-chip model of human gingival tissue containing keratinocytes, fibroblast and endothelial cells. We show the triculture model exhibits physiological tissue structure, mucosal barrier formation, and protein biomarker expression and secretion over several weeks. Through inflammatory cytokine administration, we demonstrate the induction of inflammation measured by changes in barrier function and cytokine secretion. These states of inflammation are induced at various time points within a stable culture window, providing a robust platform for evaluation of therapeutic agents. These data reveal that the administration of specific small molecule inhibitors mitigates the inflammatory response and enables tissue recovery, providing an opportunity for identification of new therapeutic targets for gum disease with the potential to facilitate relevant preclinical drug efficacy and toxicity testing.

Fig. 1. A triculture model of human gingival tissue, comprised of human gingival fibroblasts (hGFs), human oral keratinocytes (hOKs), and human microvascular cells (hMVECs) have been cultured in PREDICT96 for 1 month under recirculating fluid flow.

a Draper’s high-throughput microfluidic culture platform contained 96 microfluidic devices with a top and bottom channel (inset) separated by a porous membrane (1 µm pore size, 1e6 porosity). b The microfluidic-based pump exists in the lid of the plate and delivers independent fluid flow to the top and bottom channels of each microfluidic device. c A maximum intensity projection of the top channel of a single PREDICT96 device showed a confluent layer of hOKs after 32 days of culture. d Mean TEER measurements from three separate experiments and up to 288 replicates per day indicated a general trend of barrier formation over the experimental duration. The tissue barrier increased over the first 10 days of culture as layers proliferated and junctions formed, reaching a plateau sustained between 250–360 Ohms cm² for the following ~21 days. Error bars represent standard deviation. e A confocal orthogonal view of the MOUTH tissue showed >100 µm thick hOK tissue on the top side of the membrane. f Calcein live stain of hMVEC layer at day 32 demonstrates a robust endothelial layer can be maintained through extensive culture with g von Willebrand factor (red), indicating vascular-specific staining on the bottom side of the membrane. h–j Changes in tissue morphology and structure could be seen at different z-planes of the tissue, including h a layer with visible granules in the upper z-plane, i a distinct actin structure in the middle z-plane, and j a basal proliferative layer of hOKs.

Fig. 2. MOUTH tissue responded to a single dose of inflammatory stimuli.

a MOUTH samples were stimulated with an inflammatory trigger of 300 ng/mL each of TNF-α and IL-1β (Stim) around day 10 of culture and evaluated compared to vehicle controls. b Secreted levels of PGE2 increased in bottom channels of MOUTH Stim devices compared to vehicle control devices within 24 h (p < 0.0332) and remained higher than vehicle controls for at least 48 h (p < 0.0021). Data from b are averaged across three experiments (N = 3–12 per experiment). c–e Devices stimulated with inflammatory cytokines IL-1β + TNF-α produced an increase in MIP-3α, IL-10, and IFN-γ production at 24 and 48 h following stimulation. Data from c–e are from one experiment with N = 3. In all cases, error bars represent standard deviation. Significance was determined by Tukey’s test. Significance: *p < 0.0332, **p < 0.0021, ***p < 0.0002, ****p < 0.0001.

Fig. 3. MOUTH tissue responded to inflammatory stimuli at different time points over the 20-day testing window.

a An early and late stimulation were applied to MOUTH devices at 10 and 21 days, respectively, and evaluated compared to vehicle controls. b Similar trends of secreted levels of PGE2 were observed between Early and Late Stim conditions compared to corresponding vehicles, but only Early Stim was significant by 48 h compared to controls. Data represent at least three replicates from one experiment. c–e MOUTH tissue has a similar inflammatory response to stimulation during the testing window. Devices stimulated with inflammatory cytokines TNF-α + IL-1β presented significant increases in MIP-3α, IL-10, and IFN-γ at 24 h under Early Stim conditions, and trended similarly in MIP-3α and IFN-γ secretion under Late Stim conditions. In all cases, error bars represent standard deviation. Significance was determined by Tukey’s test for b, and Dunnett’s T3 multiple comparisons test for c–e. Significance: *p < 0.0332, **p < 0.0021, ***p < 0.0002, ****p < 0.0001.

Fig. 4. MOUTH inflammatory response was modified with a small-molecule inhibitor.

a Devices were treated with a MAPK inhibitor (SB) 2 h prior to stimulating with inflammatory triggers to evaluate the modulation of tissue response compared to controls. b Twenty-four and 48 h post-stimulation, secreted PGE2 levels in both SB and SB + stim devices were significantly lower than Stim devices. The data presented is averaged across three experiments (N = 3–6 per experiment). c–e Preventative treatment of devices with SB prior to stimulation significantly prevented IL-10 secretion but did not significantly prevent the secretion of MIP-3α or IFN-γ. Devices treated with just SB only decreased IFN-γ secretion relative to vehicle controls. In all cases, error bars represent standard deviation. Significance was determined by Dunnett T3 multiple comparisons test for b, c and Tukey’s test for d, e. Significance: *p < 0.0332, **p < 0.0021, ***p < 0.0002, ****p < 0.0001.

Fig. 5. MOUTH tissue responded to multiple doses of inflammatory stimuli and/or inhibitors.

a A dosing scheme illustrates consecutive doses of inflammatory stimulation and inhibitory treatment with a MAPK inhibitor (SB) versus controls. b Barrier function of Stim devices and Pre-SB + Stim devices trended similarly through the first dose of inflammatory stimulants, in which TEER decreased for several days post-stimulation before increasing and plateauing at sub-baseline levels by Day 21. The first dose of inflammation stimulant administered on Day 11 of culture (Stim 1x) decreased TEER values by 31% within 3 days and recovered to 82% of baseline TEER within 10 days. TEER values of devices pre-treated with SB prior to the first dose of stimulant (Pre-SB + Stim), decreased by 23% within 3 days and recovered to 71% of baseline TEER within 10 days. The trend of the two conditions diverged after the second dose, which was administered 10 days after the first dose. The second dose (Stim 2x), did not significantly reduce TEER in the first 48 h, but ultimately TEER was reduced by 60% by 7 days post-second dose relative to Day 11 values. By the end of culture on Day 28, compared to the average TEER measured on Day 11 before Stim 1x, TEER of Stim devices had decreased by 61%, and TEER of Pre-SB + Stim devices had decreased by 31%. TEER of vehicle control devices had decreased by 16%. c Secreted PGE2 levels increased 24 hours after devices were stimulated the first time (Stim 1x). These levels recovered to pre-stimulation levels by the second round of dosing, on day 21, when the same devices were dosed again with IL-1β and TNF-α (Stim 2x). With the second dose, secreted PGE2 levels increased within 24 h, following the same trend as the first dose. SB 2x devices had a decrease in PGE2 release, also similar to the first dose of SB on day 11. Pre-incubation with SB inhibited inflammatory response for 24 h after the first and second inflammatory doses, as indicated by the preservation of PGE2 levels compared to Stim conditions. Data represent at least three replicates from one experiment. In all cases, error bars represent standard deviation. Significance was determined by unpaired Welch’s t-test. Significance: *p < 0.0332, **p < 0.0021, ***p < 0.0002, ****p < 0.0001.