Introduction

CD4 T cells initiate an adaptive immunity upon antigen recognition by the cognate T cell receptors (TCR). The CARD11-BCL10-MALT1 (CBM) complex serves as a central signaling hub that channels TCR proximal events to the canonical NF-κB and JNK pathways1,2. In addition, CBM complex assembly triggers activation of the MALT1 protease, which modulates immune responses by cleaving substrates involved in cell signaling (e.g. A20, CYLD, HOIL-1), adhesion (Tensin-3), transcription (RelB) and mRNA regulation (e.g. Regnase-1, Roquins, N4BP1)3. CARD11 functions as a molecular seed and recruits preformed BCL10-MALT1 complexes, which induces BCL10 oligomerization and filament formation via its CARD (caspase recruitment domain)4,5. The MALT1 death domain (DD) binds to the CARD of BCL10 in a way that the MALT1 C-terminus, containing three immunoglobulin (Ig) domains, the paracaspase (PCASP) domain and two TRAF6 binding motifs (T6BM), protrudes away from the inner BCL10 core filament6. BCL10 oligomerization and BCL10-MALT1 binding are both necessary for the recruitment of BCL10 to CARD11, and thus for CBM complex formation after T cell stimulation6,7. In cells, assembly of BCL10 into cytoplasmic oligomers coincides with NF-κB activation, indicating the necessity of BCL10 filament formation4,5,6,8,9.

TCR stimulation induces binding of the E3 ligases TRAF6 and LUBAC (linear ubiquitin chain assembly complex) to the CBM complex, which has been suggested to activate NF-κB signaling10,11,12,13. Deletion of TRAF6 or destruction of T6BMs in MALT1 abrogates TCR-triggered NF-κB activation11,14,15. TRAF6 catalyzes conjugation of K63-linked ubiquitin chains to MALT1, thereby recruiting the TAB-TAK1 kinase complex, which phosphorylates and activates IKKβ to drive canonical NF-κB signaling16,17,18,19. Further, LUBAC, consisting of HOIP, HOIL-1, and SHARPIN, is recruited to the CBM complex in T cells, and decorates CARD11, BCL10, and MALT1 with M1-linked ubiquitin chains10,13,20. While HOIP knock-down (KD) or knock-out (KO) diminishes CBM-dependent NF-κB activation in Jurkat T cells, inconsistent and often contradictory results have been obtained regarding the role of other LUBAC subunits. Whereas one study suggested that HOIP acts independently of its catalytic activity and HOIL-1, but relies on SHARPIN, another report described that the function of HOIP relies on its ligase activity independent of SHARPIN10,13. For HOIL-1, a MALT1 substrate, either positive or negative regulatory functions for TCR-induced NF-κB activation have been described21,22,23. TCR-induced NF-κB signaling is mildly reduced in murine thymocytes and CD4+T cells with conditional LUBAC deficiencies. However, LUBAC ablation impaired thymocyte differentiation independent of NF-κB, and the severely reduced viability of peripheral T cells is not connected to CBM functions24,25.

BCL10 ubiquitination has been implicated in triggering NF-κB signaling as well as BCL10 degradation and concomitant CBM complex destruction9,13,26,27,28,29. Through mutagenesis, lysines 17, 31 and 63 in the BCL10 CARD were identified as potential acceptor sites for K63- or M1-linked polyubiquitin chains. Mutation of these residues resulted in impaired IKK recruitment and NF-κB activation13,29. M1-ubiquitinated BCL10 can serve as a platform for the association to the noncatalytic IKK subunit NEMO/IKKγ via its M1-ubiquitin binding UBAN (ubiquitin binding in ABIN and NEMO) domain30,31,32. Furthermore, conjugation of K48- or K63-linked ubiquitin chains on lysine 31 and 63 can promote removal of BCL10 filaments by selective autophagy9,29. Importantly, inducible BCL10 ubiquitination relies on the presence of MALT1 and CARD11, indicating that ubiquitin conjugation depends on CBM complex assembly13,29.

Here, we directly compare the contributions of LUBAC and TRAF6 to CBM complex assembly and function in primary human CD4 and Jurkat T cells. We reveal that TRAF6 acts upstream of LUBAC and is a prerequisite for linear ubiquitination of BCL10, but LUBAC is largely dispensable for TCR-induced NF-κB signaling. However, TRAF6 and LUBAC control recognition and cleavage of distinct substrates by MALT1. Thus, our functional and structure-guided analyses demonstrate how the coordinated activity of these E3 ligases controls CBM complex assembly and effector functions.

Results

TRAF6 but not LUBAC is the main driver of CBM downstream signaling in human CD4 T cells

Transcripts of TRAF6 and LUBAC subunits RNF31/HOIP and RBCK1/HOIL-1 are expressed in various subsets of human CD4 T cells (Supplementary Fig. 1a). To clarify the necessity of the E3 ligases TRAF6 and LUBAC, we generated knock-out (KO) cells by CRISPR/Cas9 gene editing of primary human CD4+ T cells purified from PBMCs from healthy donors. Using a sgRNA targeting TCRα for optimization of CRISPR/Cas9 KO in CD4+ T cells, we detected loss of TCRα/β surface expression indicative of homozygous TCRα KO in ~50% of CD4+ T cells (Supplementary Fig. 1b). For sgRNA targeting TRAF6, the same protocol achieved TRAF6 downregulation verified by flow cytometry, indicating that TRAF6 was ablated in a subset of CD4+ T cells (Supplementary Fig. 1c). Transfection of sgRNAs targeting TRAF6, HOIP or HOIL-1 resulted in a decline, but not a complete absence of the targeted proteins in pools of primary human CD4+ T cells from different donors (Supplementary Fig. 1c, d). On average, ~60% reduction of TRAF6 protein and ~85% reduction of HOIP protein was achieved with the respective sgRNAs (Fig. 1a).

Fig. 1: Comparison of TRAF6 and LUBAC-dependent CBM signaling in human CD4+ T cells.

a TRAF6 and HOIP expression normalized to β-Actin in CD4+ T cells from different donors after sgRNA transfection quantified based on Western blotting from 7 biological replicates (see Supplementary Figs. 1c, 2a). b Western blot analysis of sgControl, sgTRAF6, sgHOIP and sgHOIL-1 KO primary human CD4+ T cells after stimulation with P/I for the indicated time points. Expression of the target genes (TRAF6, HOIP and HOIL-1), NF-κB signaling and MALT1 substrate cleavage were evaluated by Western blot. c, d Representative flow cytometric analyses (left panel) and quantification of changes in mean fluorescence intensity (MFI; values normalized to untreated [UT], right panel) of IκBα levels in sgControl, sgTRAF6, sgHOIP and sgHOIL-1 primary CD4+ T cells after P/I (c) or TNF (d) stimulation (30 min) from 8 (sgControl) and 4 (targeting sgRNA) biological replicates. IκBα levels of UT samples were set to 1. Treated samples were normalized to the untreated samples. e Representative flow cytometric analyses (left panel) and quantification of changes in MFI (right panel) of p-p65 levels in sgControl, sgTRAF6, sgHOIP and sgHOIL-1 primary CD4 T cells after P/I stimulation (30 min) shown for donor A from 8 (sgControl) and 4 (targeting sgRNA) biological replicates. f Image stream analyses showing representative pictures. Scale bar: 7 µm. g Histograms showing similarity scores of p65 and NucBlue (nucleus) stain and quantification of p65 and NucBlue similarity scores and percentage nuclear p65 of sgControl, sgTRAF6 and sgHOIP transfected CD4+ T cells after P/I stimulation (30 min) from 3 biological replicates. All bars show the means ± SD and p-values were calculated by unpaired two-tailed t-test (a) one-way ANOVA (c, d, g) or two-way ANOVA (e) combined with Dunnett’s (c–e) or Tukey’s (g) multiple comparisons test. kDa: kilodalton.

We investigated the effects of TRAF6, HOIP or HOIL-1 KOs on NF-κB signaling. In line with previous results, sgRNAs against HOIP or HOIL-1 caused LUBAC instability33,34 (Fig. 1b). While IκBα degradation and p65 phosphorylation were mildly impaired in sgTRAF6 T cells upon P/I stimulation, NF-κB signaling was largely unaffected after targeting HOIP or HOIL-1. However, IκBα degradation following TNF stimulation was reduced in sgHOIP and sgHOIL-1 cells, indicating effective targeting of LUBAC in CD4+ T cells (Supplementary Fig. 1e). We observed diminished cleavage of HOIL-1 in sgHOIP, and to a lesser extent in sgTRAF6 CD4+ T cells, but otherwise cleavage of MALT1 substrates was not severely altered (Fig. 1b). There was a slight tendency for constitutive cleavage of CYLD and N4BP1 in sgTRAF6 human CD4+ T cells even in the absence of stimulation. The effect was not as pronounced as in mice with conditional TRAF6 deficiency in CD4+ T cells or TRAF6 KO Jurkat T cells15, likely due to only partial ablation of TRAF6 in the human CD4 T cells.

Bulk analyses by Western blotting may obscure the effects of TRAF6 or LUBAC KO due to residual expression of targeted genes in many cells, and we switched to single-cell analyses for NF-κB signaling. The strong decline of IκBα protein amounts after 30 min P/I stimulation in sgControl-transfected CD4+ T cells was partially abolished after sgTRAF6 transfection (Fig. 1c, Supplementary Fig. 1f). In contrast, sgHOIP or sgHOIL-1 transfection did not perturb IκBα degradation on single-cell level after P/I stimulation in multiple donors but impaired IκBα reduction after TNF stimulation, which was not observed in sgTRAF6-transfected CD4+ T cells (Fig. 1c, d, Supplementary Fig. 1f, g). We also determined p65 phosphorylation on single-cell level after P/I stimulation in human CD4+ T cells (Fig. 1e, Supplementary Fig. 1h). Again, TRAF6 depletion prompted a significant decrease in p-p65 after P/I stimulation in two independent donors, while HOIP and HOIL-1 KO only mildly diminished p65 phosphorylation.

To analyze the impact of TRAF6 and LUBAC on NF-κB translocation in human CD4+ T cells, we performed image stream analyses and determined p65 nuclear accumulation on single-cell level. P/I stimulation led to nuclear accumulation of p65 in CD4+ T cells transfected with sgControl or sgHOIP (Fig. 1f, g). There was a heterogeneous response after sgTRAF6 transfection, which prevented p65 translocation in ~50% of CD4+ T cells, which matches the KO efficiency in the CD4+ T cells (see Supplementary Fig. 1b). Overall, our comparative analyses provide evidence that TRAF6 is the main driver for initiating NF-κB signaling downstream of the CBM complex in human CD4+ T cells.

TRAF6 and LUBAC control optimal NF-κB gene induction in human CD4 T cells

To determine the effects of TRAF6 and LUBAC on gene induction in human T cells, we performed CRISPR/Cas9 KO of TRAF6 and HOIP in CD4 T cells of four donors for transcriptomic analyses using RNAseq (Supplementary Fig. 2a). The samples cluster according to stimulation condition (untreated (UT) versus CD3/CD28), but no significant clustering with regards to donors or sgRNA transfection was observed (Supplementary Fig. 2b). Almost 300 genes were upregulated after 90 min of TCR/CD28 stimulation (log2FoldChange ≥ 1.5 & padj≤ 0.05) (Supplementary Fig. 2c). The most significantly upregulated biological processes in response to TCR/CD28 stimulation were the Hallmark pathway ‘TNFA signaling via NFKB’ and Gene Ontology (GO) terms associated with T cell and lymphocyte activation (Fig. 2a). ‘TNFA signaling via NFKB’ was still induced in sgTRAF6 and sgHOIP transfected CD4 T cells, reflecting the partial KO only in a subset of cells (Fig. 2b). However, there was a significant reduction in the induction of ‘TNFA signaling via NFKB’ pathway in sgTRAF6 but not sgHOIP CD4 T cells. Certain NF-κB-controlled genes only rely on TRAF6 but not HOIP expression (e.g. DUSP1, BTG2, MYC), but induction of other classical NF-κB target genes like NFKBIA/IκBα, TNFAIP3/A20 and ICAM1 depend on both TRAF6 and HOIP (Fig. 2c, Supplementary Fig. 2d). Of note, TRAF6 but not HOIP KO CD4+ T cells showed mildly augmented ‘T cell activation’ GO, which may be related to the negative regulatory impact of TRAF6 in naïve CD4+ T cells, and this effect may confine transcriptomic analyses15,35.

Fig. 2: TRAF6 and HOIP-dependent gene expression in CD4 T cells.

a Gene set enrichment analysis (clusterProfiler/fgsea) after CD3/CD28 stimulation, using Hallmark and Gene Ontology Biological Process (GO:BP) gene sets; two-sided enrichment test, upregulated gene sets shown; permutation test p-values, Benjamini–Hochberg adjusted. b Gene set variation analysis (GSVA) scores of the Hallmark gene set “TNFA Signaling via NFKB” and GO:BP gene set “T cell Activation”. The interquartile range (IQR) around the median was represented, the bars were drawn to the medians, the whiskers extend to 1.5x IQR from the quartiles, and two-sided student’s t-test p-values are displayed for the GSVA scores. c TRAF6/HOIP dependency analysis (“Methods”) using DESeq2 interaction model, two-sided Wald test, Benjamini–Hochberg FDR–adjusted p-values. Higher values indicate weaker induction upon CD3/CD28 stimulation in sgTRAF6 or sgHOIP CD4+ T cells compared to control. All data of RNAseq experiments are from 4 biological replicates (donors).

LUBAC is not a major component of CBM-dependent NF-κB signaling in Jurkat T cells

We have previously shown defective NF-κB activation in TRAF6 KO Jurkat T cells15. We decided to study the impact of LUBAC on CBM signaling more closely by generating KOs for each individual LUBAC component in duplicate clones in Jurkat T cells. Again, remaining LUBAC subunits were destabilized in the absence of HOIP, HOIL-1, and to a milder degree SHARPIN (Fig. 3a-c). We transduced an NF-κB-EGFP reporter gene into LUBAC KO clones and measured NF-κB reporter gene activation after P/I, CD3/CD28 or TNF stimulation (Fig. 3d-f, Supplementary Fig. 3a-c). Ablation of HOIP or HOIL-1 nearly abolished, while SHARPIN depletion impaired NF-κB activation by TNF. Accordingly, TNF-triggered IκBα phosphorylation and degradation were decreased in HOIP-, HOIL-1- or SHARPIN-deficient cells, demonstrating the critical necessity of LUBAC for TNFR signaling (Fig. 3g-i). In contrast, NF-κB reporter induction and IκBα degradation after P/I or CD3/CD28 stimulation was normal or only very mildly reduced in LUBAC KO cells (Fig. 3d-i, Supplementary Fig. 3a-f). Further, neither ERK nor JNK phosphorylation was significantly changed upon P/I or CD3/CD28 treatment after HOIP KO (Supplementary Fig. 3g, h). To confirm these results, we virally reconstituted HOIP and HOIL-1 KO cells (Fig. 3j, l). While reconstitution of FSS-HOIP and FSS-HOIL-1 enhanced TNF-induced NF-κB activation, P/I or CD3/CD28-dependent activation was unaffected or even decreased in the case of CD3/CD28 stimulation after HOIP rescue (Fig. 3k, m). We analyzed NFKBIA/IκBα and TNFAIP3/A20 expression levels in HOIP and TRAF6 KO Jurkat T cells after viral reconstitution and confirmed that HOIP but not TRAF6 is required for induction of both NF-κB target genes after TNF stimulation (Fig. 3n). In contrast, TCR/CD28- and P/I-induced expression of both genes strongly relies on TRAF6 (Fig. 3n, Supplementary Fig. 3i). HOIP did not control immediate induction of NF-κB target genes, but mildly enhanced TNFAIP3 expression after prolonged TCR/CD28 stimulation.

Fig. 3: LUBAC is required for TNFR- but not TCR-induced NF-κB activation in Jurkat T cells.

a–c Expression analyses of LUBAC components HOIP, HOIL-1 and SHARPIN in parental and LUBAC KO Jurkat T cell clones by Western blot. d–f NF-κB-EGFP reporter activation in parental and LUBAC KO Jurkat T cell clones following stimulation with TNF, P/I and CD3/CD28 stimulation for 5 h. Quantification of flow cytometric EGFP analyses was done by calculating the median fluorescence intensity (MFI) from 4 (d, f, e for P/I) or 3 (e for TNF and CD3/CD28) biological replicates. g–i NF-κB signaling was analyzed in parental and KO Jurkat T cells after stimulation with TNF (left panel) and P/I (right panel) for the indicated time points. Western blots showing IκBα phosphorylation and degradation are shown. j Reconstitution of HOIP after transduction in HOIP KO Jurkat T cells analyzed by Western blot. k NF-κB-EGFP reporter activation in HOIP WT and mock transduced HOIP KO Jurkat T cells following TNF, P/I and CD3/CD28 stimulation for 5 h was analyzed by flow cytometry. Quantification of flow cytometric EGFP analyses was done by calculating the median fluorescence intensity (MFI) from 4 biological replicates. l Reconstitution of HOIL-1 after transduction in HOIL-1 KO Jurkat T cells analyzed by Western blot. m NF-κB-EGFP reporter activation and quantification was done as in k from 3 biological replicates. n RT-PCR analyses of NFKBIA/IκBa and TNFAIP3/A20 mRNA in HOIP WT and mock transduced HOIP KO (upper panel) and TRAF6 WT and mock transduced TRAF6 KO (lower panel) Jurkat cells following TNF and CD3/CD28 stimulation for 1 and 4 hours from 3 biological replicates. o NF-κB signaling was evaluated in parental and TRAF6 KO, HOIP KO and TRAF6/HOIP DKO Jurkat T cells without and with P/I stimulation using Western blot analysis. All bars show the means ± SEM (d–f, k, m) or ±SD (n) and p-values were calculated by unpaired two-tailed t-test (k, m), one-way ANOVA combined with Dunnett’s multiple comparisons test (d–f), or two-way ANOVA combined with Sidak’s mutiple comparisons test (n). Only significant p-values (<0.05) are shown. kDa kilodalton.

Since TRAF6 is required for NF-κB signaling in primary CD4+ and Jurkat T cells (Fig. 1)15, we determined a potential contribution of LUBAC in the absence of TRAF6 and generated HOIP/TRAF6 double (D)KO Jurkat T cells (Fig. 3o). While individual ablation of TRAF6 prevented degradation of IκBα and phosphorylation of p65 after P/I stimulation, the TRAF6/HOIP DKO mirrored the effect of individual TRAF6 deficiency. Taken together, the data suggest that TRAF6 is the main driver for connecting initial CBM activation to the canonical NF-κB signaling pathway after T cell co-stimulation, and LUBAC may contribute to more sustained NF-κB transcriptional responses in Jurkat and human primary CD4+ T cells.

HOIP and TRAF6 regulate MALT1 substrate selectivity in T cells

Besides activation of downstream signaling pathways, CBM complex formation leads to induction of MALT1 protease activity, which is constitutively enhanced in the absence of TRAF615. To compare the role of TRAF6 and LUBAC for MALT1 protease activity, we performed biotin pull-downs from Jurkat T cell lysates after incubation of a biotinylated MALT1 activity-based probe (bio-MALT1 ABP) that selectively binds to the active form of MALT136. We quantified the amounts of active MALT1 in unstimulated and P/I treated cells comparing parental, TRAF6, HOIP, TRAF6/HOIP and HOIL-1 KO Jurkat T cells (two clones each) (Fig. 4a). As reported, TRAF6 deletion induced constitutive MALT1 protease activity in Jurkat T cells15. HOIP and HOIL-1 ablation did not change constitutive MALT1 protease activity, which was maintained in TRAF6/HOIP DKO cells (Fig. 4a). MALT1 protease was induced to a similar extent in all KO cells after P/I stimulation, despite a mild decrease in the absence of TRAF6.

Fig. 4: LUBAC and TRAF6 modulate MALT1 substrate recognition.

a Detection of catalytically active MALT1 in parental, TRAF6 KO, HOIP KO, TRAF6/HOIP DKO or HOIL-1 KO Jurkat T cells under untreated or P/I stimulated conditions (30 min) by pulldown of biotin-MALT1-ABP. Substrate cleavage was quantified by determining the ratio of active MALT1 (MALT1 ABP-PD) to total MALT1 (lysate) from 3 biological replicates. b–d Cleavage of MALT1 substrates was detected in parental, as well as two independent clones of TRAF6 KO, HOIP KO, TRAF6/HOIP DKO or HOIL-1 KO Jurkat T cells, either untreated or P/I stimulated (30 min) by Western blot. For quantification the ratios of cleavage products to full-length proteins were determined for HOIL-1 (b), CYLD (c), and N4BP1 (d) from 3 biological replicates. All bars represent the means ± SEM and p-values were calculated by one-way ANOVA combined with Dunnett’s multiple comparisons test. Only significant p-values (<0.05) are shown. kDa kilodalton.

Next, we tested for MALT1 substrate cleavage after stimulation and noticed that cleavage of HOIL-1 was reduced, while CYLD cleavage was enhanced in HOIP KO cells but unaffected in HOIL-1 KO cells (Supplementary Fig. 4a, b). Therefore, we decided to quantify constitutive and inducible cleavage of MALT1 substrates HOIL-1, CYLD, Regnase-1 and N4BP1 in the panel of Jurkat KO cells (Fig. 4b-d, Supplementary Fig. 4c). Ablation of HOIL-1 did not significantly affect cleavage of any other substrate. Constitutive HOIL-1 cleavage in the absence of TRAF6 as well as P/I-inducible HOIL-1 cleavage, strictly relied upon HOIP expression (Fig. 4b). In contrast, inducible CYLD cleavage was enhanced in the absence of HOIP, which in turn relied on TRAF6 (Fig. 4c). HOIP deficiency did not influence cleavage of Regnase-1 or N4BP1 (Fig. 4d, Supplementary Fig. 4c, d). Further, even though loss of TRAF6 induced constitutive cleavage of all four substrates, inducible HOIL-1 or N4BP1 cleavage was significantly diminished in the absence of TRAF6 (Fig. 4b, d).

To take a closer look by which mechanism HOIP and TRAF6 influence MALT1 substrate selection, we analyzed the cleavage of distinct substrates in KO Jurkat T cells reconstituted with the WT or mutant HOIP or TRAF6 constructs (Supplementary Fig. 4e). While inducible HOIL-1 cleavage was regained by expression of HOIP WT or catalytically inactive C885S, the HOIP-binding mutant ΔUBA failed to rescue cleavage, demonstrating that LUBAC assembly and not E3 ligase activity is required for optimal HOIL-1 cleavage (Fig. 5a). In contrast, augmented CYLD cleavage in the absence of HOIP relied on LUBAC assembly (ΔUBA) and E3 ligase activity (C885S) (Fig. 5b). SPATA2 bridges CYLD to LUBAC37, but SPATA2 deficiency did not significantly affect IκBα degradation or CYLD cleavage in Jurkat T cells, suggesting that SPATA2 is not involved in the recognition of CYLD by the CBM complex and MALT1 (Supplementary Fig. 4f). In line with the KO data, Regnase-1 cleavage was not influenced by the expression of HOIP WT or mutants (Fig. 5c). In case of TRAF6, expression of TRAF6 WT but neither the dimerization mutant R88A/F118A nor the UBC13-binding mutant C70A rescued the diminished inducible cleavage of HOIL-1, CYLD and N4BP1 (Fig. 5d-f). Since both mutants abolish K63-ubiquitin ligation by TRAF638, the catalytic activity of TRAF6 is critical for optimal recognition and cleavage of these substrates.

Fig. 5: E3 ligase-dependent and -independent modulation of MALT1 substrate recognition by HOIP and TRAF6.

a–c Cleavage of MALT1 substrates HOIL-1 (a), CYLD (b) and Regnase-1 (c) in parental HOIP KO Jurkat T cells transduced with either mock vector, HOIP WT, HOIP C885S or HOIP ΔUBA by Western blot under unstimulated and P/I stimulated (30 min) conditions. Cleavage was quantified as a ratio of cleavage products to full-length protein from 3 biological replicates. d–f Cleavage of MALT1 substrates HOIL-1 (d), CYLD (e) and N4BP1 (f) in parental TRAF6 KO Jurkat T cells transduced with either mock vector, TRAF6 WT, TRAF6 R88A/F118A or TRAF6 C70A by Western blot under unstimulated and P/I stimulated (30 min) conditions. Cleavage was quantified as a ratio of cleavage products to full-length protein from 3 biological replicates. g Analysis of A20 cleavage in parental, TRAF6 KO, HOIP KO and TRAF6/HOIP DKO Jurkat T cells by Western blot, either untreated or after stimulation with P/I (30 min and 2 h). The non-specific (ns) band below A20 full length is indicated. A20 full-length amounts were normalized to expression level of β-Actin, and the ratio of cleavage product to full-length was determined from 4 biological replicates. h Western blot analyses of A20, HOIL-1 and CYLD cleavage in A20 KO Jurkat T cells reconstituted with A20 WT or ZnF4, ZnF7, ZnF4/7 mutants untreated or after P/I stimulation (1 h). The non-specific (ns) band below A20 full-length is indicated. The ratios of cleavage products to full-length were determined for 4 biological replicates. All bars represent the means ± SEM and p-values were calculated by one-way ANOVA (a–f, h) or two-way ANOVA (g) combined with Dunnett’s multiple comparisons test. Only significant p-values (<0.05) are shown. kDa kilodalton.

We also determined the effects of TRAF6 and HOIP deficiencies on A20 cleavage (Fig. 5g). Basal A20 cleavage was enhanced in TRAF6 KO or TRAF6/HOIP DKO cells. Confirming previous results, HOIP facilitates inducible A20 cleavage by MALT112. However, A20 levels are prone to complex regulations, including proteasomal degradation, MALT1 cleavage and TRAF6-NF-κB-dependent resynthesis39 (Supplementary Fig. 3i). Proteasomal removal of A20 is also impeded in HOIP and TRAF6 KO cells (Fig. 5g), which may indirectly influence the ratio of A20 full length to cleavage fragment. A20 zinc finger (ZnF) 4 and 7 bind to K63- and M1-linked ubiquitin chains, respectively40. Thus, we asked if A20 recruitment to ubiquitin chains via the ZnFs controls cleavage of A20 or even HOIL-1 and CYLD by MALT1. We reconstituted A20 KO Jurkat T cells with A20 WT and ZnF mutants (Supplementary Fig. 4g). Cleavage of A20 was severely impaired in ZnF4, ZnF7 or ZnF4/7 mutant expressing Jurkat T cells, suggesting that TRAF6 and HOIP conjugation of K63- and M1-chains, respectively, controls A20 recognition by MALT1 (Fig. 5h). HOIP- and TRAF6-dependent cleavage of HOIL-1 or CYLD cleavage by MALT1 was not affected by A20 ZnF mutations.

Thus, TRAF6 E3 ligase but not LUBAC restricts basal MALT1 protease activity in resting T cells. In addition, TRAF6 catalytic activity enhances cleavage of distinct substrates in activated T cells. LUBAC assembly via HOIP controls cleavage of HOIL-1, but LUBAC assembly and activity also mildly counteract cleavage of CYLD by MALT1.

N-terminal HOIL-1 cleavage fragment partially rescues LUBAC functions in T cells

Since HOIL-1 is a substrate of MALT1, CBM signaling may affect LUBAC composition and function. To test the effects in a near to endogenous setting, we virally rescued HOIL-1 KO Jurkat T cells with HOIL-1 1-165 (N-term) and 166-511 (C-term) fragments generated by MALT1 cleavage as well as cleavage resistant (R165A), ligase inactive (C460A) and HOIP-binding defective (ΔUBL) mutants (Fig. 6a). Co-expression of the surface marker ΔCD2 verified successful transduction of all constructs (Supplementary Fig. 5a)30. While full length HOIL-1 proteins were slightly above endogenous HOIL-1 compared to parental Jurkat, expression of HOIL-1 N-term or C-term fragments was lower or higher, respectively. Blockade of protein synthesis by cycloheximide (CHX) did not reveal significant differences in post-translational stability of the fragments, suggesting that other mechanisms counteract strong HOIL-1 N-term expression (Supplementary Fig. 5b). As expected, (see Fig. 3), expression of HOIL-1 WT or mutants did not affect P/I or CD3/CD28-induced NF-κB activation in Jurkat T cells (Fig. 6c). However, HOIL-1 WT, R165A and C460A stabilized HOIP and rescued TNF-triggered NF-κB activation (Fig. 6b, c). HOIL-1 C-term and ΔUBL, which were unable to interact with and failed to stabilize HOIP, did not rescue TNF signaling to NF-κB. In contrast, HOIL-1 N-term, despite low expression, yielded a mild HOIP stabilization and slightly augmented TNF-induced NF-κB response. Thus, the N-terminal HOIL-1 fragment is still able to partially mediate TNFR responses to NF-κB.

Fig. 6: MALT1 cleavage generates a partially active HOIL-1 N-terminal fragment.

a Scheme of HOIL-1 WT and mutant proteins. LTM: LUBAC-tethering motif, UBL: Ubiquitin-like, NZF: Npl4-type zinc-finger, RING: Really interesting new gene, IBR: In-between-RING. b LUBAC subunit expression after transduction of HOIL-1 WT and mutants into HOIL-1 KO Jurkat T cells analyzed by Western blot. c NF-κB-EGFP reporter induction in HOIL-1 KO Jurkat T cells with HOIL-1 WT and mutant constructs after P/I, CD3/CD28 and TNF stimulation for 5 h. EGFP expression was analyzed by flow cytometry and quantification of median fluorescence intensity (MFI). d Cell viability determined by CellTiter-Glo assay after 14 h TNF treatment in parental, HOIP KO and HOIL-1 KO Jurkat T cells. e CASP3/7 activation was assessed after 3.5 h TNF treatment in parental, HOIP KO and HOIL-1 KO Jurkat T cells. f Parental and HOIL-1 KO Jurkat T cells were treated as indicated. CASP and PARP cleavage was assessed by Western blot. g Cell viability determined by CellTiter-Glo assay in HOIL-1 KO Jurkat T cells expressing HOIL-1 WT and mutants following 14 h TNF stimulation. h Induction of CASP-8 cleavage in HOIL-1 KO Jurkat T cells after reconstitution with mock or HOIL-1 WT and mutants after 6 h TNF or TNF/CHX stimulation was analyzed by Western blot. i Expression of HOIL-1 WT as well as HOIL-1 N-term and HOIL-1 C-term alone or in combination after transduction into parental (left panel) and HOIL−1 KO (right panel) Jurkat T cells analyzed by Western blot. j Cell viability determined by CellTiter-Glo assay in parental (upper panel) or HOIL-1 KO (lower panel) Jurkat T cells transduced with HOIL-1 WT and mutants after TNF or TNF/CHX stimulation for 14 h. All quantifications were done from 3 (c, d, g, j) or 4 (e) biological replicates after normalization of median fluorescence intensity (MFI; c) or relative luminescence units (RLU; d, e, g, j) to untreated (UT) samples. All bar graphs represent means ± SEM and p-values were calculated by one-way ANOVA combined with Dunnett’s multiple comparisons test. Only significant p-values (<0.05) are shown. kDa kilodalton.

In non-lymphoid cells, LUBAC is well established to mediate anti-apoptotic NF-κB activation and to counteract the formation of apoptosis and necroptosis-inducing TNFR1 complexes41. Ablation of HOIP or HOIL-1 strongly sensitized Jurkat T cells to TNF-induced cell death, which involved activation of Caspases (CASP) 3, 7 and 8 and poly (ADP-Ribose)-Polymerase (PARP) (Fig. 6d-f). Apoptosis inhibitor zVAD-FMK impedes cell death and induces phosphorylation of RIP1, indicating a shift to necroptosis when apoptosis is blocked in LUBAC-deficient Jurkat T cells (Fig. 6f, Supplementary Fig. 5c, d). Next,