Volume 26, Issue 3, Pages 860-873 (March 2018) IFN-γ and TNF-α Pre-licensing Protects Mesenchymal Stromal Cells from the Pro- inflammatory Effects of Palmitate  Lauren Boland, Anthony J. Burand, Alex J. Brown, Devlin Boyt, Vitor A. Lira, James A. Ankrum  Molecular Therapy  Volume 26, Issue 3, Pages 860-873 (March 2018) DOI: 10.1016/j.ymthe.2017.12.013 Copyright © 2017 The American Society of Gene and Cell Therapy Terms and Conditions

Figure 1 MSCs’ Metabolic Activity Decreases and Cell Death Increases after Palmitate, but Not Glucose Exposure (A) The metabolic activity of MSCs was determined using XTT after 48 or 96 hr of BSA or 0.4 mM Palm-BSA exposure in high (25 mM) or low (5.5 mM) glucose media (mean ± SEM, two-way ANOVA at 48 hr with Tukey correction for multiple comparisons: *p < 0.05; palmitate effect, not significant (n.s.), p = 0.8094; glucose effect, n.s., p = 0.4255; at 96 hr: palmitate effect, p = 0.0019; glucose effect, n.s., p = 0.612; n = 3 independent experiments). (B) Viability of MSCs after 96 hr of BSA or 0.4 mM Palm-BSA exposure in high or low glucose was determined using PI staining followed by flow cytometry analysis (mean ± SEM, two-way ANOVA with Tukey correction for multiple comparisons: palmitate effect, p < 0.0001; glucose effect, n.s., p = 0.9160; n = 3 independent experiments). PI+ population gating was determined using a staurosporine-treated positive control. Representative flow plot histograms of PI− (left of red line) versus PI+ (right of red line) MSC populations. Experiments were performed in MSC donor 00081. Molecular Therapy 2018 26, 860-873DOI: (10.1016/j.ymthe.2017.12.013) Copyright © 2017 The American Society of Gene and Cell Therapy Terms and Conditions

Figure 2 Palmitate Inhibits MSC Proliferation and Dramatically Alters MSC Morphology (A) Proliferation was assessed by direct cell counting of Hoechst 33342-stained MSCs via fluorescence microscopy after 48 or 96 hr of BSA and/or Palm-BSA exposure in high-glucose (25 mM) media. 10 images were taken at 5× magnification per treatment condition for each independent experiment (mean ± SEM, two-way ANOVA with Tukey correction for multiple comparisons, *p < 0.5, n = 3 independent experiments). (B) Nuclear area was determined on a per-cell basis using ImageJ software after staining MSCs with Hoechst 33342 and ActinGreen 488 ReadyProbe. Quantification was performed on 10 images per treatment condition for three independent experiments. Data are represented via violin plot with notched boxplots showing the 25th, 50th, and 75th percentiles. (C) Cumulative frequency distribution of cell’s nuclear area curve. Horizontal black dotted lines represent 25% and 75% relative frequency, respectively. Vertical gray dotted lines represent the 25th and 75th percentile of nuclear area in BSA controls. (D) Representative images of Hoechst 33342- and ActinGreen 488 ReadyProbe-stained MSCs taken at 20× magnification. Images were chosen to display the same approximate number of cells in each image (nuclei counts range from 21 to 25 cells per image). All experiments were performed with MSC donor 00081 at P3–P5. Blue represents BSA, green represents 0.1 mM Palm-BSA, orange represents 0.2 mM Palm-BSA, purple represents 0.4 mM Palm-BSA (color designation is maintained throughout all figures). Molecular Therapy 2018 26, 860-873DOI: (10.1016/j.ymthe.2017.12.013) Copyright © 2017 The American Society of Gene and Cell Therapy Terms and Conditions

Figure 3 IDO Protein Levels Are Unaffected by Palmitate Exposure, but Enzymatic Activity Decreases Significantly (A) Graphical representation of relationship between IFN-γ and IDO. (B) Representative western blot of IDO protein levels after 48 and 96 hr of simultaneous exposure to BSA, Palm-BSA, and/or IFN-γ. β-Actin provided as loading control (n = 3 independent experiments). (C) Densitometric quantification of IDO to β-actin protein levels (±SD, two-way ANOVA with Sidak correction for multiple comparisons to BSA control, n = 3 independent experiments). (D) IDO activity was measured via quantification of kynurenine secretion into the supernatant of MSCs exposed to 48 and 96 hr of BSA and/or Palm-BSA in high-glucose media (mean ± SEM, two-way ANOVA with Dunnett correction for multiple comparisons relative to BSA control, *p < 0.5; n = 3 independent experiments). Experiments were performed with MSC donor 00081. Molecular Therapy 2018 26, 860-873DOI: (10.1016/j.ymthe.2017.12.013) Copyright © 2017 The American Society of Gene and Cell Therapy Terms and Conditions

Figure 4 Exposure to Palmitate Severely Inhibits PBMC Proliferation and Causes MSCs to Convert from an Immunosuppressive to a Pro-inflammatory Phenotype (A–C) CFSE-stained PBMCs for donor 1 (A), donor 2 (B), and donor 3 (C) were stimulated with CD3/CD28 Dynabeads and 100 U/mL rh-IL-2. Stim Ctrl is CFSE-stained PBMCs stimulated with CD3/CD28 Dynabeads and 100 U/mL rh-IL-2 without BSA or Palm-BSA. Gate for % proliferated was determined using an unstimulated CFSE-stained PBMC control (mean ± SEM, one-way ANOVA with Dunnett correction for multiple comparisons relative to BSA control, *p < 0.5; n = 3 independent experiments for each donor). (D–F) CFSE-stained PBMCs were co-cultured with MSCs at a 1:4 MSC/PBMC ratio (allogeneic co-culture pairings: D, PBMC donor 1 and MSC donor 7083; E, PBMC donor 2 and MSC donor 00055; F, PBMC donor 3 and MSC donor 00081). Proliferation of CFSE-stained PBMCs in co-cultures was normalized to proliferation in matched PBMC-only BSA and/or Palm-BSA-treated conditions (dashed line is proliferation in respective PBMC-only BSA or Palm-BSA condition) to demonstrate suppressive potential (mean ± SEM, one-way ANOVA with Dunnett correction for multiple comparisons relative to BSA control, *p < 0.5; n = 3 independent experiments for each allogeneic donor pairing). Molecular Therapy 2018 26, 860-873DOI: (10.1016/j.ymthe.2017.12.013) Copyright © 2017 The American Society of Gene and Cell Therapy Terms and Conditions

Figure 5 MSC Blockade of Pro-inflammatory Cytokine Production Is Completely Abrogated by Palmitate Exposure (A–G) A multiplex Th1/Th2/Th17 cytokine bead array (CBA) ELISA was performed on supernatants isolated from PBMC-only (open square with dashed line) and MSC PBMC co-cultures (closed circle with solid line) for (A) IFN-γ, (B) TNF-α, (C) IL-6, (D) IL-17a, (E) IL-4, (F) IL-10, and (G) IL-2. Mean ± SEM, two-way ANOVA with Tukey correction for multiple comparisons was performed; significance tables of all comparisons are provided in Tables S1–S3; n = 3 independent experiments. Ctrl (for PBMC only), CFSE-stained PBMCs stimulated with CD3/CD28 Dynabeads and 100 U/mL rh-IL-2 without BSA or Palm-BSA; Ctrl (for MSC PBMC co-culture), 1:4 ratio of MSC to CFSE-stained PBMCs stimulated with CD3/CD28 Dynabeads and 100 U/mL rh-IL-2 without BSA or Palm-BSA. (H–K) Radial plot demonstrating differences in overall cytokine milieu between PBMC-only and MSC PBMC co-cultures in BSA (H), 0.1 (I), 0.2 (J), and 0.4 (K) Palm-BSA condition. Vertices represent magnitude of individual cytokines that were normalized to the maximum value, with highest value (1.0) at the ring farthest from the center and 0 for each cytokine at the center. All data points are from the allogeneic pairing of PBMC donor 3 and MSC donor 00081. Molecular Therapy 2018 26, 860-873DOI: (10.1016/j.ymthe.2017.12.013) Copyright © 2017 The American Society of Gene and Cell Therapy Terms and Conditions

Figure 6 MSCs Can Regain Immunosuppressive Potency after Palmitate Exposure but Show a Dose-Dependent Delay in Recovery that Can Be Partially Modulated by TLR4 Inhibition (A) Timeline of recovery experiment involving exposure of MSCs to Palm-BSA prior to MSC PBMC Co-culture. (B) Division index (average number of cell divisions from a cell in the original CFSE-stained PBMC population, includes parent peak in calculation) was calculated using FlowJo software in 1:4 MSC:PBMC co-cultures after MSC exposure to BSA and/or Palm-BSA for 96 hr (mean ± SEM, one-way ANOVA with Dunnett correction for multiple comparisons to BSA control, *p < 0.5; n = 4 independent experiments). PBMC ctrl, PBMCs stimulated with CD3/CD28 Dynabeads and 100 U/mL rh-IL-2 without BSA and/or Palm-BSA; MSC+PBMC Ctrl, 1:4 ratio of MSC:PBMCs stimulated with CD3/CD28 Dynabeads and 100 U/mL rh-IL-2 without BSA and/or Palm-BSA. (C) Representative peak-fit analysis of flow cytometry histograms showing the number of PBMC generations proliferated in MSC PBMC co-culture after MSC exposure to 96-hr BSA and/or Palm-BSA. All representative histograms were from the same experiment to demonstrate suppression relative to stimulated PBMC Ctrl. (D) Percent proliferation of PBMCs from MSC:PBMC coculture (1:4) with MSCs pre-exposed to BSA and/or Palm-BSA in the presence or absence of the TLR4 inhibitor, C34 (10 μg/mL). Data show mean ± SEM, two-way ANOVA with Sidak correction for multiple comparisons to non-TLR4 inhibitor-treated condition; n = 4 independent experiments. (E) Representative flow cytometry histograms showing MFI of CFSE-stained PBMCs co-cultured with MSCs pre-exposed to 0.4 mM Palm-BSA ± TLR4 inhibition. (F) Quantification of MFI from CFSE-stained PBMCs in MSC:PBMC co-cultures (1:4) with MSCs pre-exposed to 0.4 mM Palm-BSA ± TLR4 inhibition (mean ± SEM, two-way ANOVA with Sidak correction for multiple comparisons to non-TLR4 inhibitor-treated condition, *p < 0.5, n = 4 independent experiments). Molecular Therapy 2018 26, 860-873DOI: (10.1016/j.ymthe.2017.12.013) Copyright © 2017 The American Society of Gene and Cell Therapy Terms and Conditions

Figure 7 IFN-γ and TNF-α Pre-licensing Completely Blocks the Pro-inflammatory Skewing of MSCs in the Presence of Palmitate (A and B) Representative flow cytometry histograms showing proliferated PBMCs (proliferated population left of the red line) in MSC PBMC co-cultures with naive (A) and pre-licensed (B) MSCs in BSA and/or Palm-BSA conditions. (C) Cytokine production in PL-MSC PBMC co-cultures was assessed using a seven-cytokine bead array (IL-6 levels were higher than the top ELISA standard and were therefore excluded from this analysis). Cytokine levels were normalized to the naive MSC PBMC co-culture cytokine levels (dashed line) in order to demonstrate suppression of pro-inflammatory cytokine release in the presence of PL-MSCs. All experiments were conducted with MSC donor 00081 with PBMC donor 3 (mean ± SEM, n = 3 independent experiments). Molecular Therapy 2018 26, 860-873DOI: (10.1016/j.ymthe.2017.12.013) Copyright © 2017 The American Society of Gene and Cell Therapy Terms and Conditions

Figure 8 IFN-γ and TNF-α Pre-licensing Requires Intact IFNGR Signaling and Fixes MSCs into an Anti-inflammatory Phenotype, Even after Extended Exposure to Palmitate (A) MSC:PBMC co-cultures were performed with naive PL-MSCs, ruxolitinib-treated (500 nM, JAK1/2 inhibitor) PL-MSCs, and tofacitinib-treated (500 nM, JAK3 inhibitor) PL-MSCs. JAK inhibitors were added to pre-licensing media 2 hr before adding in pre-licensing cytokines (10 ng/mL IFN-γ and 1 ng/mL TNF-α), and MSCs were subsequently cultured for 24 hr (mean ± SEM, one-way ANOVA with Dunnett correction for multiple comparisons to naive co-culture, *p < 0.5; n = 5 independent experiments of MSC donor 00081 with PBMC donors 1–5.) (B) Gene expression analysis of naive and PL-MSCs after 96 hr of BSA and/or Palm-BSA exposure followed by replacement with 24 hr of IFN-γ (10 ng/mL) media (data from one representative experiment, mean ± SD, one-way ANOVA with Dunnett correction for multiple comparisons with naive MSCs treated with BSA followed by 24 hr of IFN-γ stimulation, *p < 0.5; n = 3 independent experiments with MSC donor 00081). Molecular Therapy 2018 26, 860-873DOI: (10.1016/j.ymthe.2017.12.013) Copyright © 2017 The American Society of Gene and Cell Therapy Terms and Conditions