Identification of a group of 9-amino-acridines that selectively downregulate regulatory T cell functions through FoxP3

Abstract

FoxP3⁺ regulatory T cells (Tregs) are responsible for immune homeostasis by suppressing excessive anti-self-immunity. Tregs facilitate tumor growth by inhibiting anti-tumor immunity. Here, we explored the targeting of FoxP3 as a basis for new immunotherapies. In a high-throughput phenotypic screening of a drug repurposing library using human primary T cells, we identified quinacrine as a FoxP3 downregulator. In silico searches based on the structure of quinacrine, testing of sub-libraries of analogs in vitro, and validation identified a subset of 9-amino-acridines that selectively abrogated Treg suppressive functions. Mechanistically, these acridines interfered with the DNA-binding activity of FoxP3 and inhibited FoxP3-regulated downstream gene regulation. Release from Treg suppression by 9-amino-acridines increased anti-tumor immune responses both in cancer patient samples and in mice in a syngeneic tumor model. Our study highlights the feasibility of screening for small molecular inhibitors of FoxP3 as an approach to pursuing Treg-based immunotherapy.

Keywords: Cancer; Immune response; Immunology.

Graphical abstract

Figure 1

Validation of quinacrine and quinacrine-like acridines that downregulate FoxP3 expression in human T cells (A–C) CD3⁺ T cells were treated with drugs for 24 h, then stained with CD4 and FoxP3 antibodies for flow cytometry analysis. (A) FACS plots shows the representative gating strategy. Concentration-response curves for QDD (B) and MP4 (C) for downregulation of FoxP3 are shown, as well as live cells (%) at tested concentrations. Normalized response: FoxP3% in CD4⁺ T cells normalized to mock treatment (n = 3 donors). Viability data are presented as mean ± SEM. (D) Overview of structure-activity relation (SAR) for 9-amino-acridine structures. The acridine backbone is set as the starting point in the illustration (left). The purple lines represent compounds with potent ability to downregulate FoxP3, while the red lines represent compounds with no or poor ability to regulate FoxP3. The blue arrow represents the group of compounds with the replacement of H in the 9-N-R1 position from the backbone. n = number of compounds in each group.

Figure 2

QDD and MP4 impair the suppressive functions of Tregs toward Teff cells Isolated CD4⁺CD25⁺CD127^dim/- Tregs from healthy donor buffy coats were stimulated with CD2/CD3/CD28 beads while treating with QDD/MP4 at specified concentrations for 48 h and subjected to the indicated analyses. (A–C) Expression of FoxP3, CTLA-4, LAG3, CD25, PD-1, and CD73 in Tregs were measured by flow cytometry. (A) Representative histograms show expression level of each protein with or without compound treatment. (B) The expression level of each marker was represented as mean fluorescence intensity (MFI) on the y axis. Boxplots show all data points with median and min to max (n = 3 donors). (C) Correlation of expression of Treg markers is presented as a heatmap (left), and a plot with FoxP3 MFI (x axis) vs. MFI of other markers (right). The Pearson correlation coefficient, r, is calculated and the plots are generated in GraphPad Prism. (D) After 48 h of pre-activation and compound treatment, Tregs were co-cultured with CellTrace Far Red-stained effector CD4 or CD8 Teff at 1:2 ratio for 96 h. Proliferation of effector CD4 or CD8 Teffs was analyzed by flow cytometry and compared to the proliferation of Teffs alone, which is set to 100%. The FACS histograms on the right show the proliferating Treff assessed by CellTrace Far Red signal. (E) Production of TNFα in CD4/CD8 Teff cells was determined by intracellular flow cytometry staining after co-culture with Tregs and is presented as the percentage of TNFα producing cells normalized to Teff only (set to 100%). FACS plots on the right shown representative gating of TNFα in CD4 Teff. Data are represented as mean ± SEM (n = 3–5 donors). p value was determined by ordinary one-way ANOVA (B) or two-way ANOVA (D and E). ∗p ˂ 0.05, ∗∗p ˂ 0.01, ∗∗∗p ˂ 0.001, ∗∗∗∗p ˂ 0.00001. ns, not significant.

Figure 3

Selective inhibitory effects of QDD and MP4 on Tregs over Teffs (A) CD3⁺ T cells from healthy donors were treated with compounds for two days under TCR stimulation. Different Treg populations were determined by gating for CD45RA versus FoxP3 in CD4 T cells (left FACS plot) by flow cytometry. FoxP3 positive cells (%) in each Treg population (middle) and cell viability (right) are shown. (B) Purified CD4 Teff, CD8 Teff, and Tregs from healthy donor CD3⁺ T cells were stained with CellTrace Far Red and treated with QDD or MP4 at indicated concentrations for 4 days, under stimulation by CD2/CD3/CD28 beads. Cell proliferation was measured by flow cytometry for CellTrace Far Red-positive cells and normalized to that of mock-treated cells. IC₅₀ was calculated in each population. (C–E) Representative phosphorylation signals were measured by phospho-flow cytometry in healthy donor CD3⁺ T cells that were treated with QDD or MP4 for 30 min at specified concentrations, by gating on CD8 Teff (CD8⁺), CD4 Teff (CD4⁺CD25⁻FoxP3^-) and Treg (CD4⁺CD25⁺FoxP3⁺) populations. Phospho-flow staining was performed right after 30 min treatment of compounds (C and D) or following IL-2 stimulation for 45 min (E). MFI of each protein phosphorylation level is represented as the arcsin ratio normalized to the mock-treated group. n = 3 healthy donors. Mean ± SEM is shown, and statistics represent calculation of compound groups against mock in each population. ∗p ˂0 .05, ∗∗p ˂ 0.01, ∗∗∗p ˂ 0.001, ∗∗∗∗p ˂ 0.00001; two-way ANOVA.

Figure 4

Quinacrine and MP4 interfere with the DNA binding activity of FoxP3 (A and B) EMSAs were performed to assess FoxP3 binding to DNA in vitro. QDD (A) or MP4 (B) at different concentrations were pre-incubated with His-FoxP3-ΔN for 30 min before introducing a double-stranded DNA sequence to allow binding. Bands marked by red brackets “{” represent the FoxP3-DNA complex, which is retarded in the gel compared to the mobility of free DNA denoted by “∗”. (C and D) AlphaScreen assays determined the ability of QDD (C) or MP4 (D) to compete DNA binding to FoxP3 in vitro. The signal from His-FoxP3 interacting with biotinylated-DNA was normalized to that from the counter screen. (E) qPCR was performed with total RNA isolated from CD3⁺ T cells treated with QDD or MP4 at indicated concentrations for 24 h. Relative mRNA expression levels of FoxP3, FoxP1, and FoxO1 were normalized against RPS9 internal control. (F) The binding of FoxP3 to FoxP3, STAT3 and CD25 gene promoter regions was measured by ChIP-qPCR in total chromatin DNA isolated from Tregs that were treated with MP4 for 4 h. Relative binding levels were normalized to input control (1% of total chromatin DNA). (G) RNA from Tregs treated with MP4 for 4 h were analyzed by qPCR to measure the gene expression levels of STAT3, CD25, and CTLA4, normalized to RPS9 internal control. Mean ± SEM (n = 3) is shown in the graphics. One-way ANOVA (E), two-way ANOVA (F), and unpaired t test (G) were used to determine p values. ∗p ˂ 0.05, ∗∗p ˂ 0.01, ∗∗∗p ˂ 0.001, ∗∗∗∗p ˂ 0.00001. ns, no significant.

Figure 5

Effects of QDD and MP4 on boosting T cell activation in cancer patient samples (A and B) Cells isolated from breast cancer patient lymph nodes (LNs) were treated with QDD or MP4 at specified concentrations for 48 h. BFA was added to the culture 6 h before harvest. (A) The representative gating strategy of FoxP3 in CD4⁺ T cells. (B) Expression of activation marker CD25 and intracellular IL-2 in CD4⁺ or CD8⁺ T cells were determined by flow cytometry analysis (n = 6 patients). (C) PBMCs isolated from CLL patients’ blood were treated with QDD or MP4 for 48 h and FoxP3 and CD69 expression measured in CD4⁺ or CD8⁺ T cells by flow cytometry (n = 8 patients). Normalized responses: FoxP3% in CD4⁺ T cells was normalized to DMSO control and FoxP3 MFI normalized to DMSO control. Boxplots show all data points with median and Min to Max. Significant differences between mock and compound-treated groups were determined by two-way ANOVA. ∗p ˂ 0.05, ∗∗p ˂ 0.01, ∗∗∗, p ˂ 0.001, ∗∗∗∗p ˂ 0.00001. ns, not significant.

Figure 6

In vivo effects of QDD-like acridines in a mouse tumor model (A) Titration of QDD and MP4 in C57BL/6 mice healthy mice by intraperitoneal injections is illustrated in the upper figure. Blood samples collected at each specified time point were measured for FoxP3% in CD4⁺ T cells by flow cytometry. (B and C) Panc02 tumor-bearing C57BL/6 mice were administered Kolliphor HS15-formulated-MP4 or vehicle for 14 days, from 7 days after tumor cell injection. (B) The treatment scheme is illustrated in the upper figure, while the tumor growth curve below shows the relative tumor size of each time point normalized to that of day 7 in each group. (C) FoxP3% in CD4 in blood, spleen, and tumor samples were measured by flow cytometry (left), illustrating FoxP3 gating in the pseudo color plots (right). (D) Mice bearing Panc02 tumors were treated with vehicle, MP4 or PD1 single treatment, and MP4+PD1 treatment following the illustrated scheme. Proportions of tumor infiltrating CD8⁺ T cells and activated CD8⁺T cells (CD8⁺CD69⁺) in CD45⁺ cells were determined by flow cytometry. n = 5–6 mice per group. Boxplots show all data points with median and min to max. Two-way ANOVA (A–D, left) and one-way ANOVA (D, right) were used to determine p values. ∗p ˂ 0.05, ∗∗p ˂ 0.01,∗∗∗, p ˂ 0.001, ∗∗∗∗p ˂ 0.00001. ns, no significant.