Dual functions of Aire CARD multimerization in the transcriptional regulation of T cell tolerance

Aggregate-like biomolecular assemblies are emerging as new conformational states with functionality. Aire, a transcription factor essential for central T cell tolerance, forms large aggregate-like assemblies visualized as nuclear foci. Here we demonstrate that Aire utilizes its caspase activation recruitment domain (CARD) to form filamentous homo-multimers in vitro, and this assembly mediates foci formation and transcriptional activity. However, CARD-mediated multimerization also makes Aire susceptible to interaction with promyelocytic leukemia protein (PML) bodies, sites of many nuclear processes including protein quality control of nuclear aggregates. Several loss-of-function Aire mutants, including those causing autoimmune polyendocrine syndrome type-1, form foci with increased PML body association. Directing Aire to PML bodies impairs the transcriptional activity of Aire, while dispersing PML bodies with a viral antagonist restores this activity. Our study thus reveals a new regulatory role of PML bodies in Aire function, and highlights the interplay between nuclear aggregate-like assemblies and PML-mediated protein quality control.

The predicted secondary structure for the mAire CARD (based on the homology model in Fig. 1e) is shown on the top. The indicated residue numbers are based on mAire sequence. Green triangles indicate the locations of mAire mutations used in (b-d) and Fig. 1f-h. Blue diamonds indicate the APS-1 patient mutations analyzed in (e-h) and Fig. 1i, j. Conserved CARD-CARD interfaces identified for hNOD1 and hMAVS CARDs are shown at the bottom. Note that Ia, IIa and IIIa surfaces interact with Ib, IIb and IIIb, respectively. Dotted line for IIIa indicates shifted position in the predicted a-helix in Aire CARD relative to hNOD1 and hMAVS CARD. b. Purification of mAire CARD. Left, schematic of the protein purification protocol; mAire CARD was expressed as a NusA fusion in E. coli and was purified by Ni-NTA affinity purification. 3C protease was used to release CARD from the fusion construct. Right, SDS-PAGE gel of 3C protease-cleaved WT and mutant mAire CARD used in EM imaging of filaments in Fig. 1b, f. Note that the NusA tag alone does not form filaments. c. Transcriptional activity of WT mAire CARD and the mutants in 4D6 cells, as measured by the relative mRNA levels of Aire-dependent genes (represented by CD4, CELF2, IGFL1 and S100A9). The relative mRNA level of an Aire-independent gene, HPRT1, is shown as a negative control. Data are presented as mean ± s.d., n = 3. P-values (two-tailed t-test) were calculated in comparison to WT mAire. * p < 0.05; ** p < 0.01; p > 0.05 is not significant (ns). Exact p-values are provided in the Source Data File. Right, western blot (WB) showing the expression levels of FLAG-tagged mAire. d. Representative fluorescence microscopy images of WT and mAire mutants in 4D6 cells. e. Purification of hAire CARD. Left, schematic of the protein purification protocol. Unlike mAire CARD, hAire CARD fused to NusA-tag yielded very low levels of purified protein.
Successful purification of hAire CARD from E. coli required a minimal N-terminal His6 tag and a refolding step using 6 M guanidine hydrochloride (GdHCl). 3C protease was used to remove the His6 tag. Right, SDS-PAGE gel of 3C protease-cleaved WT and mutant hAire CARD used in (f). f. Representative EM images of WT and hAire CARD with APS-1 patient mutations indicated in (a). g. Transcriptional activity of WT and mutant hAire in 293T cells. Experiments were performed as in (c) and are presented as mean ± s.d., n = 3. P-values (two-tailed t-test) were calculated in comparison to WT hAire. * p < 0.05; ** p < 0.01; p > 0.05 is not significant (ns). Exact pvalues are provided in the Source Data File. h. Representative fluorescence microscopy images of FLAG-tagged hAire in 293T using anti-FLAG.

Supplementary Figure 2: Chemically induced multimerization partially restores the transcriptional activity of AireDCARD.
a. Transcriptional activity of ∆CARD fused with tandem repeats of FKBP (FKBP1-4) in the presence and absence of chemical dimerizer (AP1903) in 293T cells. Experiments were performed as in Fig. 1g and are presented as mean ± s.d., n = 3. Note that KRT14 and S100A9 are Aire-dependent genes. P-values (two-tailed t-test) were calculated in comparison to empty vector (EV) + AP1903. * p < 0.05; ** p < 0.01. a. Examples of Aire foci that have no overlap, partial overlap, and complete overlap with endogenous PML bodies. We observed that some WT and Sp110-CARD swap foci were not completely overlapping with PML, but rather positioned adjacent to PML bodies. For systematic and quantitative analysis of the degree of co-localization of Aire foci and PML bodies, see (b). b. Fraction of Aire (WT Aire or Sp110-CARD swap) foci co-localized with PML bodies while defining co-localization as having greater than a certain threshold % overlapping area (xaxis). See Methods for details of automated image analysis workflow. Sp110-CARD swap foci are associated with PML bodies significantly more than WT mAire regardless of the choice of threshold. A total of 2022 and 383 Aire foci were examined for WT mAire and Sp110-CARD swap samples, respectively. For quantitation of Aire foci co-localized with PML, we arbitrarily chose the threshold definition of 50% overlap area. c. Representative fluorescence microscopy images of two independent monoclonal 4D6 cells stably expressing WT hAire-FLAG under a doxycycline-inducible promoter. 4D6 cells were induced at 1 µg/ml doxycycline for 24 hrs before immunostaining with anti-FLAG and anti-PML. Top right, quantification of Aire foci co-localized with PML bodies from the two clones of 4D6 cells stably expressing WT hAire-FLAG. Analysis was done as in (b) and shows minimal co-localization of WT hAire with PML bodies, independent of the choice of the threshold definition. Statistical significance comparison between WT hAire-FLAG foci overlapping with PML (50% threshold) from the two 4D6 clones was calculated using a twotailed Student's t-test (see Methods for details) and determined to be not significant (p = 0.9118). A total of 371 and 450 Aire foci were examined for WT clones A and B, respectively. Bottom right, WB analysis of WT hAire-FLAG expression in the presence of 0.2 µg/ml (+) or 1 µg/ml (++) of doxycycline.

Supplementary Figure 4: Effect of SIM-mAire, IE1 and MG132 on Aire foci localization.
a. Transcriptional activity of WT mAire (black circle) with and without co-expression of SIM-mAire (green circle) in 293T cells. Each circle represents 0.6 µg/ml DNA transfected. Experiments were performed as in Fig. 4d and presented as mean ± s.d., n = 3. Experiments were performed as in Fig. 1g and are presented as mean ± s.d., n = 3. P-values (two-tailed ttest) were calculated in comparison to WT mAire. ** p = 0.0015 and * p = 0.016. Note that KRT14 and S100A9 are Aire-dependent target genes.
b. Representative fluorescence microscopy images of SIMΔCARD-FLAG with or without coexpression of WT mAire-HA in 4D6 cells. c. Representative fluorescence microscopy images of IE1-HA in 4D6 cells. Cells that express IE1 show diffuse nuclear staining of endogenous PML, whereas cells with no IE1 expression have intact PML bodies. d. WB showing the expression levels of the proteins used in Fig. 4e, f. e. SUMO modification analysis of FLAG-tagged WT mAire and ΔCARD. mAire (0.6 µg/ml DNA) was co-expressed with HA-SUMO1 or -SUMO2 (0.6 µg/ml DNA) in 293T cells and the immunoprecipitation was performed as in Fig. 3c. f. Representative fluorescence microscopy images of FLAG-tagged WT mAire in the presence and absence of MG132 (10 µM). 4D6 cells were treated with MG132 for 4 hrs before fixation. Right, quantitative analysis of Aire foci co-localized with PML bodies. n = number of Aire foci examined per sample. Statistical significance comparison was calculated using a two-tailed Student's t-test for two population proportions where each population consists of all individual Aire foci examined.*** p = 2.159e-12. g. Transcriptional activity of WT mAire in the presence and absence of MG132 (10 µM). Cells were treated with MG132 for 16 hrs. The fold-change as a result of Aire expression was plotted in the right panel. Experiments were performed as in Fig. 1g and are presented as mean ± s.d., n = 3.  Supplementary Fig. 3c. b. Transcriptional activity of hAire WT, C302Y (top panel), and C311Y (bottom panel) stably expressed in 4D6 cells in the presence of 0.2 µg/ml (+) and 1 µg/ml (++) of doxycycline for 48 hrs. Cells were harvested and RT-qPCR was performed as in Fig. 1g. Transcriptional activity was normalized to samples without doxycycline. Data are presented as mean ± s.d., n = 3. Note that IGFL1 and S100A9 are Aire-dependent genes and the Aire-independent gene HPRT1 is shown as a negative control. c. Representative fluorescence microscopy images of FLAG-tagged hAire∆CARD C302Y variant (C302Y∆CARD) in 4D6 cells. Cells were immunostained with anti-FLAG and anti-PML. C302Y∆CARD does not form nuclear foci.