The kinase DYRK1A reciprocally regulates the differentiation of Th17 and regulatory T cells

Abstract

The balance between Th17 and T regulatory (Treg) cells critically modulates immune homeostasis, with an inadequate Treg response contributing to inflammatory disease. Using an unbiased chemical biology approach, we identified a novel role for the dual specificity tyrosine-phosphorylation-regulated kinase DYRK1A in regulating this balance. Inhibition of DYRK1A enhances Treg differentiation and impairs Th17 differentiation without affecting known pathways of Treg/Th17 differentiation. Thus, DYRK1A represents a novel mechanistic node at the branch point between commitment to either Treg or Th17 lineages. Importantly, both Treg cells generated using the DYRK1A inhibitor harmine and direct administration of harmine itself potently attenuate inflammation in multiple experimental models of systemic autoimmunity and mucosal inflammation. Our results identify DYRK1A as a physiologically relevant regulator of Treg cell differentiation and suggest a broader role for other DYRK family members in immune homeostasis. These results are discussed in the context of human diseases associated with dysregulated DYRK activity.

Keywords: T cell differentiation; computational biology; dual-specificity tyrosine-regulated kinase signaling; human; immunology; inflammation; mouse; systems biology.

Figure 1.. Chemical biology approach to identify novel T_reg enhancers.

All data representative of at least 2 independent experiments. (A) Overview of our approach, including key methods applied. (B) Dose-response curves showing fractional enhancement (Fr enhance) of compounds (LD₅₀/EC₅₀ > 2) for T_reg (blue), Th1 (orange) and Th17 (red) lineages. (C) Plot of LD₅₀/EC₅₀ vs maximal fractional T_reg cell enhancement showing all 21 T_reg-specific enhancers (9RA, 9-cis retinoic acid; 13RA, 13-cis retinoic acid; ADQ, amodiaquine; AMIO, amiodarone; AMR, amrinone; ATRA, all-trans retinoic acid; ART, artemisinin; CIS, cisapride; CLO, clotrimazole; HAR, harmine; LOV, lovastatin; MBCQ, 4-((3,4-methylenedioxybenzyl)amino)-6-chloroquinazoline;4-quinazolinamine); PEN, pentamidine; PG, proguanil; PYR, pyrvinium pamoate; RAPA, rapamycin; RIB, ribavirin; ROT, rotenone; SER, sertaconazole; SIM, simvastatin; WM, wortmannin; Supplementary file 2). Retinoic acids are in green; compounds with LD₅₀/EC₅₀ < 2 are in gray and LD₅₀/EC₅₀ > 2 are in blue. The orange cluster is described in the text. (D) Ability of selected compounds (LD₅₀/EC₅₀ > 2 and controls) to inhibit mTOR activity, as measured by S6 phosphorylation (***p < 0.001, 1-way ANOVA with Dunnett correction). (E) Fractional (Fr) inhibitory activity of all 21 T_reg-specific enhancers on Th1 and Th17 cell differentiation. (F) SEA-predicted relationships between all 21 T_reg enhancers. Black lines predict binding of compounds (red circles) to proteins (blue diamonds) with likelihood proportional to line width. Green lines denote connection via curated KEGG pathways. See also Figure 1—figure supplements 1–8. DOI: http://dx.doi.org/10.7554/eLife.05920.003

Figure 1—figure supplement 1.. Titrating Th differentiation conditions.

Titrating cytokines for T_reg, Th1, and Th17 conditions identifies sub-maximal (blue arrowheads) and near-maximal (red arrowheads) lineage-promoting conditions. DOI: http://dx.doi.org/10.7554/eLife.05920.004

Figure 1—figure supplement 2.. Culture cellularity affects T_reg differentiation.

T_reg^low (blue) stimulation with varying initial cell numbers recapitulates the inverse relationship between final culture cellularity and percentage of FOXP3-expressing cells. T_reg^hi (red) conditions generate uniformly high levels of FOXP3-expressing cells independent of culture cellularity. DOI: http://dx.doi.org/10.7554/eLife.05920.005

Figure 1—figure supplement 3.. Schematic of analytic and hit-calling pipeline.

DOI: http://dx.doi.org/10.7554/eLife.05920.006

Figure 1—figure supplement 4.. Effect of compounds (LD₅₀/EC₅₀ < 2) on Th differentiation.

Dose-response curves showing compound effect on fractional enhancement of T_reg (blue), Th1 (orange), and Th17 (red) lineages. DOI: http://dx.doi.org/10.7554/eLife.05920.007

Figure 1—figure supplement 5.. Modeling analyses to calculate EC₅₀ values, indicated in parentheses, for all 21 T_reg-specific enhancers.

DOI: http://dx.doi.org/10.7554/eLife.05920.008

Figure 1—figure supplement 6.. Modeling analyses to calculate LD₅₀ values, indicated in parentheses, for all 21 T_reg-specific enhancers.

DOI: http://dx.doi.org/10.7554/eLife.05920.009

Figure 1—figure supplement 7.. Similarity clustering analysis of combined T_reg, Th1 and Th17 phenotypic data.

DOI: http://dx.doi.org/10.7554/eLife.05920.010

Figure 1—figure supplement 8.. Euclidean distance clustering analysis of gene expression data from cell lines treated with T_reg enhancers first analyzed by principal component analysis.

DOI: http://dx.doi.org/10.7554/eLife.05920.011

Figure 2.. Harmine-enhanced T_reg cells and harmine attenuate inflammation.

(A–E) Experiments comparing the suppressive activity of T_reg cells generated under either T_reg^HAR (blue) or T_reg^hi (red) conditions; conditions without T_reg cells are shown in black. All data representative of at least 2 independent experiments; in vivo experiments used at least 3 mice per cohort. (A) In vitro suppression assay, with unstimulated tT_reg cells in orange. (B) T_reg^hi-T_reg cells and T_reg^HAR-T_reg cells similarly delay onset of diabetes in the NOD-BDC2.5 model of type 1 diabetes. (C) Comparable inhibition of inflammation by T_reg^hi-T_reg cells and T_reg^HAR-T_reg cells in the CD45RB^hi transfer model of colitis. Representative images are shown in (D). Bars represent 100 μm. (E) Similar inhibition of airway inflammation by T_reg^hi-T_reg cells and T_reg^HAR-T_reg cells, as measured by number of eosinophils (Eos) in bronchoalveolar lavage (BAL) fluid, in a model of asthma. (F) Comparison of T_reg cells (as a percentage of total CD4⁺ T cells) in the thoracic lymph nodes of mice treated with intranasal harmine, vs mock treatment. (G) Intranasal administration of harmine prior to immunization inhibits recall airway inflammation in the asthma model. Right, representative images of inflammation around airway vessels. **p < 0.01, *p < 0.05, ns, not significant, Mantel-Cox test (B, C) and Student's t-test (E, F). See also Figure 2—figure supplements 1–2. DOI: http://dx.doi.org/10.7554/eLife.05920.012

Figure 2—figure supplement 1.. Effects of harmine treatment in vivo on T cell populations.

Mice were administered either intranasal harmine HCl (blue) or water (gray). T cell populations in thoracic lymph nodes were quantitated as shown, demonstrating relative percentages (above) and absolute numbers (below). All statistically significant differences are indicated, **p < 0.01, ***p < 0.001, Student's t-test with Holm-Sidak correction. DOI: http://dx.doi.org/10.7554/eLife.05920.013

Figure 2—figure supplement 2.. Effects of harmine treatment in vivo on antigen-presenting cell populations.

Mice were administered either intranasal harmine HCl (blue) or water (gray). Migratory and classical dendritic cells (mDC and cDC, respectively) in thoracic lymph nodes were quantitated (top right), and their respective expression of costimulatory molecules (CD40, CD80, and CD86) compared. ns, not significant, Student's t-test. DOI: http://dx.doi.org/10.7554/eLife.05920.014

Figure 3.. Harmine's effects on canonical T_reg/Th17 pathways.

All data representative of at least 2 independent experiments. (A) Effect of harmine on murine T_reg, Th1 and Th17 differentiation. (B) Effect of harmine on absolute numbers of total live and T_reg cells. (C) Effect of harmine on human T_reg differentiation. (D) Effect of harmine on Th differentiation under Th0 conditions as shown by percentage of cells with indicated markers. (E) Harmine's pro-T_reg effect when added or removed at different times after T_reg^low stimulation as shown by percentage of maximal T_reg enhancement (% max T_reg enh). (F and G) Volcano plots comparing p value vs fold change in gene expression in naïve CD4⁺ T cells, T_reg^hi-T_reg cells and T_reg^HAR-T_reg cells as indicated. Previously reported signature genes for T_reg cells (F) and activated T cells (G) are highlighted in red (upregulated) and green (downregulated). Numbers on the right and left reflect genes that are up- and down-regulated in the indicated comparison respectively with χ² test p values in the middle. (H) Median fluorescence intensity (MFI) of FOXP3 in T_reg cells generated under indicated conditions. (I) Time-course analysis of FOXP3 expression in cells cultured under indicated conditions. (J and K) Western blot analyses showing effect of T_reg enhancers on S6 kinase, SMAD2 and SMAD3 phosphorylation when added under T_reg^low conditions. Numbers denote fractional phosphorylation relative to T_reg^low conditions. (L) Effect of harmine (blue) on RORγT expression and STAT3 phosphorylation in Th17^hi conditions (gray). (M) qPCR analyses showing effects of harmine on key Th17 genes at 4 days (left) and 2 hours (right) after stimulation. Gray and blue bars represent Th17^hi and Th17^hi + harmine conditions, respectively. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, ns, not significant, Student's t-test with Holm-Sidak correction (B, C, D, M). See also Figure 3—figure supplements 1–5. DOI: http://dx.doi.org/10.7554/eLife.05920.015

Figure 3—figure supplement 1.. Effect of harmine on cellular proliferation.

CFSE-labelled CD4⁺ T cells were stimulated under T_reg^low conditions and proliferation assessed daily. The effect of adding harmine, rapamycin, or high TGFβ (T_reg^hi) was compared as shown. DOI: http://dx.doi.org/10.7554/eLife.05920.016

Figure 3—figure supplement 2.. Effect of harmine on absolute numbers of Th17 cells.

***p < 0.001, Student's t-test. DOI: http://dx.doi.org/10.7554/eLife.05920.017

Figure 3—figure supplement 3.. Effect of addition or removal of harmine at different times on Th17 differentiation.

DOI: http://dx.doi.org/10.7554/eLife.05920.018

Figure 3—figure supplement 4.. Correlation between gene expression in T_reg^hi-T_reg cells and T_reg^HAR-T_reg cells, relative to naïve CD4⁺ T cells.

DOI: http://dx.doi.org/10.7554/eLife.05920.019

Figure 3—figure supplement 5.. Qualitative analyses of genomewide expression in T_reg^HAR-T_reg cells.

Previously described methods were used to identify similarities between T_reg^HAR-T_reg cells and other specialized T_reg subsets (Joller et al., 2014). Volcano plots compare relative expression of genes in T_reg^hi-T_reg cells and T_reg^HAR-T_reg cells. Overlaid are genes of previously identified signatures; red and green reflect genes up- and down-regulated in these signatures respectively. Numbers on the right and left reflect genes that are up- and down-regulated in the indicated comparison respectively with χ² test p values in the middle. DOI: http://dx.doi.org/10.7554/eLife.05920.020

Figure 4.. Mechanistic dissection of harmine.

(A) Comparison of expression profiles suggesting a harmine-relevant T_reg signature. The black bar separates 2 independently row-normalized experiments. (B) Harmine signature genes are enriched for genes in IBD loci. (C) Validation of top signature genes by qPCR (all p < 0.05, Student's t-test), including genes with BACH2 binding sites (B). (D) Effect on harmine on T_reg/Th17 differentiation in BACH2-deficient (orange) vs -sufficient (black) cells. Conditions in the presence and absence of neutralizing antibodies are indicated by solid and dotted lines, respectively. (E) Effect of compounds that inhibit different harmine targets (indicated in parentheses) on T_reg cell differentiation. (F) DYRK inhibitors suppress Th17 cell differentiation. (G) Increased levels of DYRK1A in Th17 vs T_reg cells. Upper left, Western blot analyses of sorted T_reg/Th17 cells with relative expression enumerated below. Histograms show FACS analyses of DYRK1a in T_reg cells (blue) compared to either Th17 (red) or non-Th17 (black) cells. (H) Knock-down of Dyrk1a (D1a) enhances T_reg (left) and inhibits Th17 (right) cell differentiation. Cells treated with non-targeting shRNA (Ctrl) and no shRNA (None) are shown for comparison. (I) Amnis analyses showing nuclear-overlapping NFAT1 signal before (black) and after stimulation in T_reg^low conditions with (blue) or without (red) harmine. Representative images (right) illustrate cytoplasmic and nuclear NFAT1 localization in cells outside and within the gate, respectively. (J) Western blot analyses quantitating nuclear fraction of NFAT1 in cells stimulated in T_reg^low ± harmine conditions. All data in D–J representative of at least 2 independent experiments. See also Figure 4—figure supplements 1–3. DOI: http://dx.doi.org/10.7554/eLife.05920.021

Figure 4—figure supplement 1.. Effect of harmine on NFAT1 nuclear localization with time.

Time-course Amnis analyses showing effect of harmine (blue) relative to control T_reg^low conditions (black). DOI: http://dx.doi.org/10.7554/eLife.05920.022

Figure 4—figure supplement 2.. Comparing effects of DYRK1A deficiency to harmine treatment.

RNAseq experiments in primary CD4⁺ T cells comparing the effects of transduction with Dyrk1a shRNA (shDyrk1a) to harmine treatment in control shRNA-transduced cells (left). Further statistical comparison uses volcano plots comparing p value vs fold change in gene expression in harmine-treated and -untreated (control) cells (middle and right panels). Genes differentially regulated by Dyrk1a knockdown are highlighted in red (upregulated) and green (downregulated). Numbers on the right and left reflect genes that are up- and down-regulated, respectively, in harmine vs control conditions, with χ² test p values in the middle. DOI: http://dx.doi.org/10.7554/eLife.05920.023

Figure 4—figure supplement 3.. Secondary analyses of effects of DYRK1A deficiency compared to harmine treatment.

Volcano plots comparing p value vs fold change in gene expression in harmine-treated (left panels) and Dyrk1a shRNA-treated (right panels) cells. Previously reported signature genes for T_reg cells (top panels) and T_reg^HAR-T_reg cells (bottom panels) are highlighted in red (upregulated) and green (downregulated). Numbers on the right and left reflect genes that are up- and down-regulated, respectively, in harmine vs control conditions (left) or in shDyrk1a vs control conditions (right), with χ² test p values in the middle. DOI: http://dx.doi.org/10.7554/eLife.05920.024