SUMOylation-disrupting WAS mutation converts WASp from a transcriptional activator to a repressor of NF-κB response genes in T cells

Wiskott-Aldrich syndrome (WAS) is an X-linked immunodeficiency-cum-autoimmunity disorder arising from mutations in WAS that encode WASp, whose deficiency in hematopoietic cells results in the human disease.¹  WASp polymerizes actin in the cytoplasm via the actin-related protein (ARP)2/3-complex and reprograms RNA polymerase II-dependent transcription in the nucleus via its effect on mixed lineage leukemia (MLL)- and switch/sucrose non-fermentable (SWI/SNF)-dependent chromatin remodeling of gene promoters.^2-4  Like human WASp, other nucleation-promoting factors (NPFs) of the WASp family, WAVE1, WASH, JMY, and N-WASp, also support transcription or other nuclear functions in different organisms.^5-11  Accordingly, the classical cytoplasmic NPFs are emerging as key players in the nucleus in roles that are mechanistically distinct from that in the cytosol. How then is the compartment-delimited role enabled for these dual-compartment, dual-function NPFs? For WASp, its cytosolic effect on ARP2/3-mediated actin polymerization is enabled by allosteric activation, wherein association with CDC42-GTP or Fyn-mediated tyrosine phosphorylation switches WASp conformation from auto-inhibitory to active.^12,13  This structural change facilitates dimerization/oligomerization of WASp, which optimizes ARP2/3 activation and cytosolic F-actin nucleation.¹⁴  In contrast, the cotranscriptional activity of human-WASp or Xenopus-WAVE1 does not seem to critically depend on ARP2/3 or the verprolin, cofilin, acidic (VCA) domain.^5,15  Accordingly, the regulatory inputs, including unique posttranslational modifications (PTMs) involved in instructing the transcriptional output of WASp family proteins, remain elusive, clarification of which we propose is central to understanding the differential effect of WASp on immunity vs autoimmunity/autoinflammation.

Given the substantial evidence for the involvement of small ubiquitin-related modifier (SUMO)ylation in transcriptional regulation,^16-19  we asked whether this PTM, not heretofore reported for any WASp family members, enables the transcriptional role of WASp. Testing a SUMOylation-linked hypothesis is tempting because (1) WASp associates with RanBP2,¹⁵  an E3-SUMO-ligase,²⁰  raising the possibility that RanBP2 might also SUMO modify WASp; (2) SUMOylation impacts subcellular localization and protein-protein interactions on chromatin,^19,21  events that could modulate WASp’s role as a transcriptional cofactor; and (3) SUMOylation targets entire groups of physically interacting proteins,¹⁹  this raises the possibility that nuclear WASp could be SUMOylated as part of collective SUMOylation of trans-regulatory factors (RBBP5, signal transducer and activator of transcription [STAT]1, nuclear factor [NF]-κB), nucleoporin (NUP358/RanBP2), or actin-myosin complex (actin, MYL12A, MYL6, ARP2/3), all proteins known to associate with WASp^3,4,15  and be SUMO modified.^21-26 

In SUMOylation, a SUMO1, 2, or 3 covalently attaches to the lysine(K) residue contained within a SUMO consensus, ψKxE/D, or phospho-SUMOyl consensus, ψKxExxS/T (ψ, hydrophobic-residue; x, any residue; D, aspartic acid; E, glutamic acid) in a multistep reaction involving E1-activating (SAE1/2), E2-conjugating (UBC9), and E3-ligating (RanBP2; PIAS) enzymes.^17,18,26  SUMOylation typically modifies only a small fraction of the substrate (∼10%), and is reversible through the effects of multiple SUMO-specific proteases. Substrates can be modified by 1 SUMO at 1 K-residue (monoSUMOylation) or 1 SUMO at multiple K-residues (multi-monoSUMOylation) or can form an oligomeric SUMO chain at 1 or multiple K-residues (polySUMOylation). Both SUMO1 and SUMO2/3 can catalyze oligomeric SUMO chains in vivo,^20,26,27  which fulfill multiple signaling functions including transcriptional regulation and genome organization.²⁸  Accordingly, SUMOylation of transcription factors has been linked to either gene activation or gene repression in both yeast and mammalian cells^29,30  via mechanisms that include altering chromatin dynamics.^16,31,32  For peroxisome proliferator-activated receptor (PPAR)-γ and Ikaros that function either as a transcriptional coactivator or corepressor, this switch in transcriptional output is effected by SUMOylation.^33,34  Beyond serving as a simple on-off switch, SUMOylation can also fine-tune gene expression as demonstrated for the nuclear receptor SF-1,³⁵  a paradigm that is relevant to testing whether WASp-SUMOylation differentially regulates genes that pattern immunity vs inflammation.

We previously demonstrated that nuclear WASp is required for transcriptional activation.^3,4  Here, we tested the hypothesis that WASp is a bi-functional transcriptional activator and repressor and investigated the mechanism by which WASp functions in transcriptional repression. Because SUMO1 coenriches with H3K4me3 at gene promoters to activate transcription³⁶  and because WASp is required to inscribe H3K4me3 at promoters,³  we explored the WASp:SUMO link as a chromatin-based mechanism for transcriptional reprogramming in T_helper (T_H) cells. We provide evidence for SUMO modification serving as a regulatory switch that controls activating vs repressive bi-functionality of WASp in transcription. Our study uncovers a mechanism by which SUMOylation-deficient WASp perturbs the transcriptional output of NF-κB response genes important in immunity and inflammation, identifies pathogenic mutation that co-opts this deranged pathway, and proposes potential therapeutic use of histone deacetylase (HDAC) inhibitors to restore inappropriate gene activation in WAS.

Primary T_H, B, natural killer (NK), and monocytoid dendritic cells generated from donor peripheral blood mononuclear cells, Jurkat T_H cells, HeLa cells, and WAS^null T_H cells from a WAS patient carrying a nonsense mutation were used. WASp-SUMO sites were predicted using the group-based prediction system (GPS)-SUMO algorithm³⁷  and single or combinatorial K>R mutations and disease-causing mutations introduced by QuickChange^IISite-Directed Mutagenesis (Stratagene) using primers shown in supplemental Table 1 available on the Blood Web site, and the constructs were stably expressed in the T_H cells by AmaxaNucleofector (Lonza) or Lipofectamine (Life Technologies) as previously described.^4,15  Transfected cells were cultured under T_H1-skewing (recombinant human interleukin [rhIL]-12, anti-IL-4 antibody, rhIL-2), T_H17-skewing (rhIL-6, rhTGFβ1, rhIL-2), or nonskewing T_H0 conditions for 7 days as previously described.³  Stable expression of mutants was verified by flow cytometry and western blotting at day 7 after transfection, which showed >90% T cells expressing the constructs. NK cells were propagated in rIL2 (100 IU/mL), dendritic cells were matured with lipopolysaccharide (LPS), and B cells were activated with immunoglobulin (Ig)M crosslinking. Expression of endogenous SUMO1 was suppressed in primary human CD4⁺ T_H cells by transfecting SUMO1 short hairpin RNA (shRNA) or scrambled shRNA (SantaCruz) (0.8 mL shRNA Transfection Medium, 200 μL shRNA solution A + solution B, incubated at 37°C for different time points). For HDAC inhibition assays, CD4⁺ T cells were incubated in the presence of 100 ng/mL trichostatin-A and 2 mM sodium butyrate for 2 hours.

Multiple mass spectrometry (MS) assays on immunoprecipitated WASp-enriched protein complexes were performed as previously described.^4,15  Briefly, lysates from micrococcal nuclease (MNase)-treated nuclei of human primary or Jurkat T_H1-skewed cells expressing WASp-Flag/Myc dual-tagged protein were incubated with anti-WASp, or anti-Flag and anti-Myc, or their isotype-Ig antibodies. The recovered polypeptides were analyzed by nano-liquid chromatography (LC)-MS on an LTQ-Orbitrap Velos mass spectrometer using the SEQUEST database.

Coimmunoprecipitations and immunoblottings were done as previously described^3,4,15  using the reagents and antibodies shown in supplemental Table 1. Electrophoretic mobility shift assay (EMSA) was carried out in 5 μg nuclear extract coincubated with double-stranded oligonucleotide probes for NF-κB, Octamer, and CCCTC-binding factor (CTCF) that were 5′end-labeled with [γ-³²P]ATP. EMSA assays were performed as previously described.⁴ 

Intracellular staining of cytokines/proteins was performed as previously described.^4,15  Briefly, cells were treated with protein transport inhibitors (GolgiPlug/GolgiStop), fixed, permeabilized, labeled with fluorochrome-conjugated antibodies (or control immunoglobulin antibody), and analyzed on a Becton-Dickinson LSRII using FACSDiva, and geometric mean (GM) fluorescence intensity was calculated. Deconvolution imaging of paraformaldehyde-fixed single T_H1 cells was performed with Zeiss digital microscopy integrated with SlideBook software, as previously described.³⁸ 

For in vitro SUMOylation, transfected FLAG immunoprecipitate (500 µg cell extract) from HeLa or T cells was coincubated with recombinant SUMO1 (or mutant SUMO1), SUMO2, SUMO3, or SUMO1 + 2 + 3 (1 µL) plus SUMOylation buffer (2 µL), and Mg-ATP (1 µL) in 20 µL total volume and incubated for 60 minutes at 37°C. Each assay was quenched by adding 20 µL of 2× sodium dodecyl sulfate-polyacrylamide gel electrophoresis gel-loading buffer followed by heating to 95°C for 5 minutes before subjecting to western blotting (Enzo). To capture in vivo SUMOylated WASp, T_H1-skewed cells were treated with 1 mM of N-ethylmaleimide (NEM; Sigma Aldrich) for 4 hours.

Chromatin immunoprecipitation (ChIP) assays were performed with micrococcal nuclease (MNase)-digested chromatin derived from NEM-treated and -nontreated T_H cells as described^3,4,15  using antibodies/reagents shown in supplemental Table 1. Nonspecific signals obtained with control IgG-ChIP were subtracted. For reverse transcriptase-polymerase chain reaction (RT-PCR) assays, cDNA was synthesized from total RNA prepared using Quick-RNA MiniPrep (Zymo), and NF-κB transcriptome analysis was performed with the RT²ProfilerPCR Array (Qiagen). The calculated fold-change [2^(−ΔΔC_t)] is the normalized gene expression [2^(−ΔC_t)] in the test sample (T_H1 skewed/T_H17 skewed) divided by the normalized gene expression [2^(−ΔC_t)] in the control sample (T_H0 nonskewed). Normalization was performed using C_t values derived for GAPDH, ACTB, or RPLP0.

To investigate SUMO modification of WASp, we queried our LC-MS/MS dataset of WASp-associated proteins generated from human T_H1-skewed cells, primary and Jurkat, with the latter transfected with Flag/Myc-tagged WASp.¹⁵  Polypeptides of SUMO enzymes involved in E1 activation (SAE1/2), E2 conjugation (UBC9), and E3 ligation (RanBP2) were recovered from cytosolic and nuclear WASp complexes, whereas those of other SUMO enzymes (PIAS1-4, NSE2, MAPL, PC2, HDAC4, TOPORs) were not (Figure 1A; supplemental Figure 1). The WASp:SUMO enzyme association yielded by MS was validated by reciprocal, sequential coimmunoprecipitation, which showed that the WASp:RanBP2 complex associates with SUMO1 but not SUMO2/3 in T_H1 cells (Figure 1B). The collective presence of these SUMO enzymes raises the possibility that WASp is SUMO modified.

To investigate WASp-SUMOylation in vitro, FLAG/Myc-tagged WASp was expressed in HeLa cells, and the FLAG immunoprecipitate was coincubated with recombinant SUMO1 (or mutant SUMO1), SUMO2, SUMO3, or SUMO1 + 2 + 3. The presence of an ∼85 kDa band (∼20 kDa heavier than ∼65 kDa unmodified WASp) and multiple chain-like, slower-migrating bands (>85 kDa) suggests SUMOylated-WASp (multi-monoSUMOylated or polySUMOylated), observed with SUMO1 but not SUMO2, SUMO3, or mutant SUMO1 (Figures 2A-B).

Endogenous WASp is SUMOylated in vivo in the absence of exogenous SUMO enzymes in primary T_H1-skewed and T_H17-skewed but not in T_H0-nonskewed cells treated with NEM, an inhibitor of SUMO isopeptidases (Figure 2C). SUMO*WASp conjugates are first captured at 4 hours in cytosol and 8 hours in the nucleus on T_H1 skewing (supplemental Figure 2A). Besides unmodified WASp (∼65 kDa), multiple slower-migrating forms (>85 kDa) were captured in the nuclear fraction of T_H1 and T_H17 cells (Figure 2C). For reactions that did not include NEM, additional, slower-migrating WASp bands were undetected, upholding that these additional bands are SUMO*WASp conjugates. The cytosolic fraction, which lacked the chain-like patterning seen in the nucleus, displayed 2 prominent bands at ∼130 and >300 kDa in NEM-treated cells (Figure 2C), which are sizes that correspond to the WASp dimer/oligomer.¹⁴ 

Because NEM also inhibits deubiquitinases, it is possible that some of these slower-migrating bands could represent hybrid SUMO ubiquitin chains (∼30 kDa heavier; SUMO: ∼20 kDa, Ub: ∼10 kDa), especially given the cooperative roles of ubiquitin/SUMO dual modification in nuclear functions.³⁹  Significantly, RNAi-mediated SUMO1 knockdown, which does not impair polyubiquitination, precludes WASp-SUMOylation without affecting β-actin SUMOylation (Figures 2D; supplemental Figure 2B); the latter is catalyzed by SUMO2/3.²⁵  These results establish WASp as a physiologic SUMO1 substrate in T_H1 and T_H17 cells, including in activated B cells, NK cells, and mature dendritic cells (supplemental Figure 2C).

The GPS-SUMO algorithm³⁷  predicted SUMO-acceptor K-residues in all WASp domains, although most occurred in WH1 and Basic domains (Figure 3A). Some of these K-residues occurred in the nonclassical SUMO consensus configurations previously identified in the global mapping of SUMOylation sites.⁴⁰  These are S*motif-1, 75-ψKDxxxxSY-83, a variant of phospho:SUMOyl consensus (ψKxDxxSx) and S*motif-2, 142-D/ExKψ-145, an inverted SUMO consensus (Figure 3B). In both S*motifs, K-residues (K76, K144) and ψ-residues (V75, I145) are conserved through evolution and are in-frame with the SUMO consensus (ψKxD/E). In phospho:SUMOyl-consensus, the K76-residue is preceded by four evolutionarily conserved, hydrophobic (ψ) amino acids (75-VFCV-72), known as hydrophobic cluster SUMOylation motif (HCSM) and followed by Ser82 (+6 position) and Tyr83 (+7 position), phosphorylation of which functions to augment SUMOylation, as does the HCSM.^41,42  Additionally, WASp contains the potential SUMO interaction motif (SIM) 50-VVQL-53, which follows the classical SIM consensus [V/I][V/I] × [V/I/L] that is also conserved through evolution (Figure 3B). Accordingly, WASp contains target sequences for both SUMOylation and SUMO interaction, the functional significance of which is implied in the clustering of disease-causing mutations (∼61 patients reported) within these sites (Figure 3C; supplemental Figure 3D).

To test whether the predicted K-residues are SUMO targets, we generated K>R deletion/substitution mutants with FLAG/MYC dual-tag, stably expressed in HeLa, and immunoprecipitated WASp mutants with anti-FLAG antibody for in vitro SUMOylation (supplemental Figure 3). FL-WASp (wild-type) coincubated with recombinant SUMO1, E1/E2-enzymes, and ATP revealed multiple slower-migrating WASp forms (∼85 to ∼220 kDa), which is consistent with SUMO modification (multi-monoSUMOylation or polySUMOylation cannot be distinguished). In contrast, K>R single mutants demonstrated that K76, K144, and K147 residues are more critical than K66 or K235 for in vitro SUMOylation based on the intensity of SUMO-modified forms (Figure 3D). Notably, K235R results were mixed, displaying both preserved and partially compromised SUMO*WASp conjugates in separate assays (Figure 3D).

To verify the above findings in vivo, normal T_H cells were stably transfected with FLAG-MYC-tagged FL-WASp or K>R single mutants with comparable expression (Figure 3E). After T_H1 skewing/T-cell receptor (TCR) activation and NEM treatment, nuclear/cytosolic fractions were examined by western blot. Transfected FL-WASp (FL-wt) gave multiple slower-migrating bands in the nucleus that corresponded with those of SUMO1 (∼85-220 kDa) (Figure 3F). This implies that transfected FL-WASp is SUMOylated by endogenous SUMO enzymes, and the heavier bands are consequent to posttranslational modification rather than alternative splicing. Although neither K>R single mutant showed complete abrogation of SUMOylation, some gradation of SUMOylation defect was observed, wherein 1 or more slower-migrating form(s) was lost in nuclear and/or cytosolic fractions. This indicates alternative usage of various K-residues for SUMOylation. We therefore generated a penta-SUMO site mutant (PSM: lacking all 5 K-residues) (supplemental Figure 3C), which we show abrogates WASp-SUMOylation in vitro and in vivo (Figures 3D,F), implying that these lysines are required for WASp-SUMOylation.

In sequential coimmunoprecipitation assays, the PSM mutant does not bind SAE2 and UBC9, establishing that failed WASp-SUMOylation is consequent to impaired interaction of the PSM mutant with SUMO enzymes (Figure 3G). Although the PSM mutant does not associate with SUMO1, its association with ubiquitin is unaffected (Figure 3G), implying that the deleted K-residues are SUMO specific. This finding complies with the paradigm that only about one-quarter of SUMOylation sites are also ubiquitinated in the human proteome.⁴¹  It is possible, however, that some of these K-residues, upon SUMOylation, may become targets of ubiquitination, acetylation, or methylation. Notably, the PSM mutant binds importin-β1 and CRM1 and translocates to the T_H1-cell nucleus (Figure 3H), assuming the same 4′,6-diamidino-2-phenylindole (DAPI)-dim distribution as normal-WASp (supplemental Figure 4),³  implying that SUMOylation and nuclear import are not linked events for WASp, as they are for actin.²⁵ 

We next examined V75M, 1 of 3 disease-causing mutations occurring in HCSM (72-ψψψψKDxxxxSY-83) (Figure 3C) and found that, although the monoSUMOylated-WASp*S1 form (∼85 kDa) is present in the nucleus, multiple high-molecular-weight WASp*SUMO conjugates (>85 kDa) are deficient, suggesting a selective defect in SUMO chain (or SUMO:ubiquitin chain) formation, a finding also captured in K76R and K235R mutants (Figure 3F). Unexpectedly, the unmodified form (∼65 kDa) of V75M and K235R WASp mutants is also deficient in the nucleus. Why this might occur is unclear, although a similar finding was reported for another polySUMOylation-deficient mutant of yeast SMT3 (hSUMO1 homolog).²⁸ 

Coimmunoprecipitation demonstrates that, although the binding of SAE2 to the V75M mutant is unaffected, that of UBC9 is lost, which is consistent with the paradigm that S*motifs are bound by UBC9.⁴³  The finding that mutation of either lysine- or hydrophobic-residue perturbs WASp-SUMOylation lends further support to the functionality of S*motif-1. Significantly, in contrast to V75- and K76-residue mutations, other disease-causing mutations involving S24-, T45-, R86-, and K476-residues, although stably expressed (Figure 3E), do not impair WASp-SUMOylation (Figure 3F). This implies that not all pathogenic WAS mutations disrupt WASp-SUMOylation.

Because SUMO chain modification of both endogenous and transfected WASp is prominent in the nucleus and because SUMO chains are linked to chromatin signaling functions of its substrates,^27,28  we sought to clarify its contribution to WASp’s role as a cotranscriptional factor. We focused on NF-κB signaling because of its relevance to T_H1 differentiation⁴⁴  and also because we previously showed that WASp deficiency impairs NF-κB activity in T_H cells.⁴  EMSA assays demonstrate that, in PSM-expressing T_H1 cells, DNA-binding activity of NF-κB and its downstream effector Octamer (OCT) proteins⁴⁵  is diminished compared with that in T_H cells expressing FL-WASp (Figure 4A). DNA-binding activity of CTCF is unaffected, which denotes the specificity of the WASp*SUMO effect on chromatin signaling.

To explicate the mechanism underlying impaired NF-κB:DNA-binding activity, we asked whether COMMD1, an inhibitor of NF-κB signaling at the DNA level,^46,47  contributes to this defect. In sequential coimmunoprecipitation assays, we show that the WASp:SUMO1:NF-κB(p65) complex formed with endogenous WASp or transfected FL-WASp in T_H1 cells does not contain COMMD1 (Figures 4B-C). Similarly, LC-MS/MS failed to recover COMMD1 polypeptides associated with normal WASp in primary T_H1 cells.¹⁵  In contrast, SUMOylation-deficient WASp mutants (PSM, V75M) associate with COMMD1 and NF-κB(p65) (Figure 4C). In vivo, sequential ChIP assays showed increased coenrichment of COMMD1 with SUMOylation-deficient mutants compared with FL-WASp at the IFNG promoter in T_H1 cells (Figure 4D). This finding was also captured at promoters of other WASp target genes, STAT1 and TLR1, but not TNFAIP2 or CSF2, which are both WASp nontarget genes (Figure 5A, no-Tx panel). Accordingly, single ChIP showed decreased promoter enrichment of NF-κB(p65) in T_H1 cells expressing SUMOylation-deficient WASp, which is consistent with COMMD1’s function of removing NF-κB from DNA.^46,47  Increased COMMD1 and decreased p65 enrichment at the IFNG promoter was also captured in normal T_H1 cells depleted of endogenous SUMO1 by RNA interference (Figure 5B), as well as in normal T_H0 cells (Figure 5C), where endogenous-WASp is non-SUMOylated and nonnuclear (Figure 2C). These findings propose a chromatin signaling function for WASp*SUMO1 in p65-dependent IFNG activation during T_H0 to T_H1 differentiation (Figure 5D).

Because histone acetyltransferase p300 regulates NF-κB-dependent gene activation^48,49  including via impacting COMMD1⁴⁷  and HDAC6,⁵⁰  we wondered whether a concomitant defect of p300 contributes to the NF-κB defect linked to non-SUMOylatable-WASp. Using sequential coimmunoprecipitation, we show that the endogenous WASp:SUMO1:p300 complex nucleated in primary T_H1 cells does not contain HDAC6 (Figure 4B), suggesting that SUMOylatable-WASp associates with coactivator histone acetyltransferase activity. In contrast, SUMOylation-deficient WASp binds HDAC6 (Figure 4C), suggesting its ectopic association with corepressor HDAC activity.⁵⁰  Furthermore, sequential ChIP assays showed HDAC6 coenriched with p300 at the IFNG promoter in PSM or V75M but not in FL-WASp-expressing T_H1 cells, where this coassociation was nominal (Figure 4D). Similar findings were also captured at STAT1 and TLR1 promoters (Figure 5A, nontreated HAT/HDAC panel). Notably, promoter enrichment of p300 is unaffected (Figure 4D), suggesting that WASp-SUMOylation status does not impact p300 recruitment/retention but rather its functional output: activation vs repression.

We next asked whether the aberrant chromatin enrichment of HDAC6 impairs p300 HAT function of acetylating H3-lysine14 required for transcriptional activation.⁵¹  Sequential ChIP shows that increased HDAC6 enrichment occurs contemporaneously with decreased enrichment of acetylated H3K14 at IFNG, STAT1, and TLR1 promoters in PSM- and V75M-expressing T_H1 cells (Figures 4D and 5A). Increased HDAC6 and decreased H3K14ac enrichment at the IFNG promoter was also captured in primary T_H1 cells depleted of SUMO1 by RNA interference (Figure 5B) and in normal T_H0 cells (Figure 5C). Because promoter occupancy/activity of NF-κB is attenuated by histone-H3 hypo-/deacetylation,⁵²  together our data suggest that, for certain NF-κB response genes, HDAC6-mediated attenuation of the p300 HAT activity⁵⁰  may contribute to NF-κB destabilization at chromatin in T_H1 cells expressing SUMOylation-deficient WASp (Figure 5D).

RT-PCR-based transcriptome profiling was used to broadly assess the effects of WASp-SUMOylation on the functional modulation of NF-κB response genes (n = 84; Figure 6A; supplemental Figure 5). Compared with FL-WASp, SUMOylation-deficient mutants showed decreased expression of most NF-κB response genes (n = 80) in T_H1 cells. These genes fall into 4 gene ontology (GO) categories: (1) cytokines/chemokines, (2) cell survival/proliferation, (3) transcription, and (4) signal transduction. Whereas mRNA and protein expression of IFNG, a T_H1 cytokine, is decreased, that of the proinflammatory cytokines granulocyte macrophage–colony-stimulating factor⁵³  and tumor necrosis factor, alpha-induced protein 2 (TNFAIP2)⁵⁴  is paradoxically increased (Figure 6, non-Tx panel). Similarly, expression of proinflammatory cytokine IL-1β (proIL-1β or mature IL-1β cannot be distinguished) is increased (Figure 6B, non-Tx panel) but without a concomitant mRNA increase (Figure 6A), a finding that could be linked to impaired NF-κB activity of preventing maturation of IL-1β.⁵⁵  Furthermore, mRNA expression of BCL3 that imparts stability to the T_H1 phenotype by preventing spontaneous conversion to the T_H17 phenotype⁵⁶  is low in T_H1-skewed cells expressing SUMOylation-deficient WASp (Figure 6A). Accordingly, expression (mRNA and/or protein) of T_H1 genes (IFNG, TBX21 STAT1, RELB⁵⁷ ) is deficient, whereas that of T_H17 genes (IL17A, IL21, IL22, IL23R, RORC, CSF2) is paradoxically increased in T_H1-skewed cells expressing SUMOylation-deficient WASp (Figures 6B and 7). In contrast, upregulation of T_H17 genes is unaffected in T_H17-skewed cells expressing SUMOylation-deficient WASp (Figure 7). These finding imply that WASp-SUMOylation is necessary not only for activation of T_H1 immunity but also for checking ectopic activation of proinflammatory pathways that destabilizes T_H1 specification.

Because NF-κB activity is downmodulated by HDAC1, 2, 3, and 6^58-60  and because the pan-histone deacetylase inhibitor (pHDACi) was shown to displace HDAC6^61,62  and relieve p300 transcriptional repression,⁵⁰  we tested the hypothesis that pHDACi may be sufficient to neutralize COMMD1- and HDAC6-mediated transcriptional repression imposed by SUMOylation-deficient WASp.

T_H1-differentiating, WASp^null patient T_H cells untransfected (UT) or transfected with FL-WASp, PSM-WASp, or V75M-WASp were cocultured with or without pHDACi (trichostatin A and Na butyrate), and their mRNA levels were quantified in the RT-PCR-based NF-κB array. T_H1 cells expressing normal WASp showed that about two-thirds of the NF-κB response genes were neither up- nor downregulated (>2.0 fold) by pHDACi (supplemental Figure 5C, blue-shaded column), which agrees with prior reports demonstrating that <20% of genes become misregulated by pHDACi treatment.^63,64  Of the genes downregulated by pHDACi in normal T_H1 cells, most belonged to cell survival/proliferation and chemokines/cytokines GO categories, with CSF3 and CCL2 demonstrating marked suppression by pHDACi. In T_H1 cells expressing SUMOylation-deficient WASp, however, pHDACi treatment displayed a more dramatic switch in mRNA expression patterns compared with their nontreated controls. Specifically, pHDACi restored to normal (or increased) mRNA expression of most NF-κB response genes, and correspondingly also some of the tested proteins, which were deficient in nontreated controls (Figure 6; supplemental Figure 5). Interestingly, pHDACi treatment normalized (or decreased) the expression of multiple proinflammatory factors (granulocyte macrophage–colony-stimulating factor, TNFAIP2, IL-1β, IL17A) that were aberrantly increased in nontreated controls (Figure 6; supplemental Figure 5C), a finding that aligns with the known inhibitory effect of pHDACi on proinflammatory cytokine signaling.⁶³ 

Because pHDACi functions in the nucleus to either deacetylase histone tails or evict HDAC-containing corepressors from gene promoters,^62,65,66  we asked whether the salutary effect of pHDACi on misregulated transcription is consequent to these epigenetic modes of actions. First, we show that the dose and duration of pHDACi treatment is sufficient to augment promoter enrichment of acetylated H3K14 in FL-WASp-expressing T_H1 cells by ChIP (Figure 5A, histoneH3 modification column). Significantly, in T_H1 cells expressing SUMOylation-deficient WASp, low H3K14ac and high HDAC6 enrichment observed in nontreated cells is restored to near-normal levels on pHDACi treatment at the IFNG, STAT1, and TLR1 promoters (Figure 5A, HAT/HDAC column). Contemporaneously, promoter occupancy of NF-κB is increased but that of its inhibitor COMMD1 is decreased by pHDACi (Figure 5A, NF-κB column), implying that acetylated H3K14 stabilizes NF-κB on gene promoters in T_H1 cells, a paradigm that was previously established in the HeLa and SIRT6 murine model.⁵²  Conversely, neither HDAC6 nor COMMD1 is enriched at the CSF2 and TNFAIP2 promoters, which could explain the chromatin “licensing” for its ectopic activation in T_H1 cells expressing SUMOylation-deficient WASp. We conclude that HDAC-associated activity imposed by SUMOylation-defective WASp is required for trans-repression of a subset of NF-κB response genes in T_H1 cells, the perturbation of which may underlie imbalanced activation of pathways supporting inflammation over immunity in WAS T_H cells.

A coactivator role for WAS family proteins in transcription is established for nucleus-located WASp, Wave1, N-WASp, WASH, and JMY in different organisms.^{3,5,8,11,67,68}  The present study identifies a corepressive role for WASp and provides the initial insights into the molecular pathway that converts nuclear WASp from a transcriptional coactivator to a promoter-specific corepressor of NF-κB response genes in T_H1 cells.

WASp bound to promoters instructs a dichotomous transcriptional outcome depending on whether it is SUMO modified or not. A model emerges wherein SUMOylatable-WASp transactivates and non-SUMOylatable-WASp transrepresses gene activation during T_H1 differentiation (Figure 5D). Implicit in this model is the idea that WASp participates in the process that controls the exchange between coactivator and corepressor occupancy at target loci and that WASp-SUMOylation informs this outcome. Because coactivators and corepressors are continuously exchanged at transcriptionally active genes in T_H cells,⁶⁹  whether loss of WASp-SUMOylation actively recruits corepressors (repression model) or it functions to evict promoter-bound corepressors (derepression model)⁷⁰  cannot be distinguished. The constitutive presence of corepressive complexes on the IFNG locus in normal T_H0 cells, where endogenous WASp is physiologically non-SUMOylated, favors the derepression model. Furthermore, because the corepressive activity of WASp is manifest in a disease context where WASp is mutated, this begs the question of whether wild-type WASp is intrinsically a coactivator or corepressor. Our identification of SUMOylation as a physiologic PTM of WASp in multiple immune cells favors WASp to be inherently a coactivator. This idea, however, could be challenged if physiologic processes that disrupt WASp-SUMOylation converting normal (nonmutated) WASp from a coactivator to corepressor are uncovered.

Pending clarification of these working models, accumulating evidence proposes WASp family proteins (WASp, WASH/FAM21) as nuclear factors that modulate transcriptional output of NF-κB response genes,^4,67  with the present study elucidating the molecular underpinnings of the WASp effect on NF-κB chromatin signaling. Although a detailed molecular analysis of all NF-κB response genes affected by disrupted WASp-SUMOylation is beyond the scope of this study, a picture emerges wherein T_H1-skewed cells expressing SUMOylation-deficient WASp exhibit a proinflammatory signature in the face of deficient T_H1 function. First, multiple T_H17 genes are ectopically upregulated under T_H1 conditions. Second, the expression of two proinflammatory genes CSF2 and TNFAIP2 is inappropriately increased. Third, the expression of the IL-1β protein is increased but without a concomitant increase in mRNA expression, a finding that is consistent with the enhanced processing of mature IL-1β from proIL-1β, a caspase-1-dependent process that is negatively regulated by NF-κB.⁵⁵  Because these proinflammatory cytokines have been linked to autoimmunity/autoinflammation,^53,54,71  our data propose that part of the cellular defect underling development of autoimmunity/autoinflammation in a subset of WAS patients expressing non-SUMOylatable-WASp is contributed by the ectopic activation of inappropriate immune function genes.

Regardless of how many pathway genes become misregulated by non-SUMOylatable-WASp, the linking of the WASp-SUMOylation defect to the imbalance between T_H1 immunity and inflammation has immediate clinical relevance for WAS, because disease-causing mutations, many resulting in XLT-to-WAS clinical progression including V75M, reside in SUMO and SIM motifs of WASp (Figure 3C).⁷²  Whether these mutations similarly disrupt WASp-SUMOylation to imbue proinflammatory qualities to T_H1-deficient cells remains to be determined. We do, however, show that not all disease-causing mutations that result in XLT-to-WAS progression (eg, missense mutations of S24, T45, and R86 residues that do not reside in SUMO or SIM motifs) impair WASp-SUMOylation and yet impede gene activation, the latter via a SWI/SNF-linked mechanism.⁴  In contrast, V75M impairs WASp-SUMOylation but does not disrupt SWI/SNF promoter activity (Figure 4D). The SWI/SNF-disrupting T45M mutation that phenocopies V75M clinically (XLT-to-WAS progression) does not display ectopic enrichment of HDAC6 or COMMD1 at the IFNG promoter observed with V75M but yet impairs NF-κB enrichment (supplemental Figure 6). These observations imply that pathogenic mutations have differing proclivity to disrupt different steps of transcription along the DNA translational axis. Because we show that loss of WASp phenocopies non-SUMOylatable-WASp in causing T_H1 immunity/inflammation imbalance, our findings are also relevant for some patient T cells lacking WASp expression.

Interestingly, of the disease-causing mutations that do impair WASp-SUMOylation (missense mutations of V75-, K76-, and K476-residues), not all display the identical SUMOylation defect. Mutations of V75- and K76-residues both impair polySUMO chains without affecting monoSUMOylation, a defect sufficient to recruit corepressors (COMMD1, HDAC6) at promoters of NF-κB response genes that are also H3K14 hypoacetylated. Certainly, besides HDAC6 and COMMD1, other HDAC-containing corepressors⁷⁰  may also contribute to the repressive chromatin signaling module associated with deficient polySUMO chain modification of WASp. We also cannot exclude the possibility that V75M, besides disrupting SUMOylation, may introduce conformational defects, which may contribute to altered function. At the minimum, our findings linking a selective SUMO chain defect to aberrant gene repression suggest that polySUMO modification support a specific chromatin signaling function distinct from that of the monoSUMO modification of WASp. Although the causality of this association remains speculative, emerging evidence linking polySUMO chain defects to impaired chromatin-to-nucleolar targeting and DNA instability,⁷³  altered chromatin organization and transcriptional repression,²⁸  and Alzheimer’s disease⁷⁴  together propose a physiologically relevant role for WASp polySUMO chains in immunity.

Future challenges in the WASp-SUMO field will be to determine (1) what fraction of these polymeric chains comprises a SUMO ubiquitin hybrid vs SUMO or ubiquitin alone and how their individual loss impairs function; (2) the effect of WASp-SUMOylation on immunological synapse formation; and (3) how WASp-SUMOylation influences regulatory networks controlling immunity/inflammation balance in health and disease. The latter is topical because many causal variants in autoimmune diseases map to NF-κB consensus motifs in the T_H-cell genome,⁷⁵  and SUMOylation-deficient WASp, we show, disturbs this NF-κB:DNA interaction. Notwithstanding, our identification of HDACi-mediated correction of gene misregulation in SUMOylation-impaired WASp T_H cells offers chromatin alteration using panHDACis (or the selective HDAC6 inhibitor) as a new target for individualized therapy to restore immunodeficiency and attenuate autoimmunity in WAS patients, as previously entertained.⁷⁶ 

The authors thank Jessica Hook of the University of Iowa for assistance with microscopy and imaging, M. Balasubramani of the University of Pittsburgh Proteomic Core Facility for analyzing the MS data, and the University of Iowa Dance Marathon for providing laboratory space.

This research was supported by National Institutes of Health, National Institute of Allergy and Infectious Diseases grants R01 AI073561 and R01 AI084957.

Contribution: K.S. performed the majority of the experiments; S.S. generated all SUMO mutants; S.-S.H. performed the EMSA assays; and Y.M.V. conceived the study, designed the experiments, analyzed the data, and wrote the paper.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Current affiliation for S.S. is Department of Botany, Raiganj University, Raiganj 733134, West Bengal, India.

Correspondence: Yatin M. Vyas, Division of Pediatric Hematology-Oncology, Stead Family Department of Pediatrics, University of Iowa Children’s Hospital, Iowa City, IA 52242; e-mail: yatin-vyas@uiowa.edu.