^{* T Cell Receptor Signals to NF-κB Are Transmitted by a Cytosolic p62-Bcl10-Malt1-IKK Signalosome}

Cytosolic Bcl10-p62 clusters colocalize with phosphorylated IKK and TRAF6 in response to TCR stimulation

In experiments with primary effector T cells and the murine T helper 2 (T_H2) clone, D10, we previously demonstrated that stimulation with cognate antigen or anti-CD3 antibody stimulates the formation of cytosolic aggregates containing Bcl10 and Malt1. These structures, which we have named POLKADOTS, coalesce around pre-existing “speckles” of the adaptor protein p62 in a manner dependent on the TCR-dependent, K63-linked polyubiquitination of Bcl10 (7, 8). To ascertain the subcellular location of key mediators in the TCR–NF-κB pathway with respect to the cytosolic Bcl10 clusters, we stimulated D10 cells with antigen-loaded CH12 cells, which are an I-A^k-expressing murine B cell line. We then used confocal microscopy to image cyan fluorescent protein (CFP)-tagged PKCθ (PKCθ-CFP), yellow fluorescent protein (YFP)-tagged Bcl10 (Bcl10-YFP), and endogenous Carma1 and p62. At 10 and 20 min after stimulation of the D10 cells, PKCθ and Carma1 translocated to the immunological synapse (Fig. 1A), consistent with previous findings (3). Whereas some conjugates exhibited prominent colocalization between Bcl10 and PKCθ in D10 cells at the immunological synapse, as we previously reported (13), all conjugates (defined as D10-CH12 couples with PKCθ translocation at the immunological synapse) exhibited localization of Bcl10 within cytosolic aggregates that colocalized with p62 (7) in the antigen-activated cells (Fig. 1, A and B).

Our previous fluorescence resonance energy transfer (FRET) analyses demonstrated that Bcl10 is closely associated with TRAF6 in the POLKADOTS structures (8). Because TRAF6 is involved in recruiting and ubiquitinating IKK (1, 14), and because phosphorylated IKK (pIKK) is detected solely in the cytosol (5), we hypothesized that IKK is activated upon association with the cytosolic Bcl10-Malt1-p62–containing POLKADOTS structures. Staining with an antibody that specifically recognizes pIKKα/β (IKKα and IKKβ phosphorylated at residues Ser¹⁷⁶ and Ser¹⁸⁰) revealed no apparent specific signal in unstimulated cells (Fig. 1C), although non-specific staining of the plasma membrane was observed in all cells. At 20 min and 2 hours after stimulation, we observed that pIKKα/β co-localized with the cytosolic Bcl10 clusters, but not with PKCθ at the immunological synapse (Fig. 1, C and D). Total IKKβ was present in cytosolic aggregates both before and after stimulation with antigen (Fig. 1D). Close inspection of the imaging data revealed that pIKKα/β staining was consistently adjacent to, but rarely overlapping with, total IKKβ staining (Fig. 1D and fig. S1A), suggesting that the ρIKKα/β at POLKADOTS was distinct from the pool of IKKβ aggregates found in T cells before stimulation. The lack of overlap between pIKKα/β and total IKKβ staining was further confirmed by staining with a distinct IKKβ monoclonal antibody (fig. S1B). Staining with an anti-TRAF6 antibody also showed the TCR-induced redistribution of TRAF6 from a diffuse and primarily cytoplasmic localization in unstimulated cells to a punctate distribution that overlapped with Bcl10 and Malt1 in POLKADOTS 20 min after TCR stimulation with anti-CD3 antibody (Fig. 1E), with less prominent colocalization at the 2-hour time point (fig. S1C).

To confirm these data in primary T cells, we stimulated in vitro differentiated CD4⁺ T_H2 effector cells with anti-CD3 and anti-CD28 antibodies (anti-CD3/28), followed by staining for endogenous Bcl10, p62, and pIKK proteins. At 20 min after stimulation, endogenous Bcl10 clusters were co-localized with p62 speckles, and pIKK was specifically localized to these p62-Bcl10 aggregates (Fig. 1F). Thus, the phenomenon of TCR-dependent colocalization of pIKK and cytosolic Bcl10 clusters was not only observed in D10 cells that over-expressed Bcl10 protein, but also occurred at clusters of endogenous Bcl10 in primary effector T cells. Together, these data suggest that IKK is recruited to and phosphorylated at the POLKADOTS structures shortly after TCR stimulation.

Activation of NF-κB occurs at the POLKADOTS signalosome

Once activated, the IKK holoenzyme phosphorylates IκBα, targeting IκBα for proteasomal degradation, thereby enabling the nuclear translocation of NF-κB. Because of the rapid kinetics of proteasomal degradation, ρIκBα is abundant only for a short period of time after TCR stimulation. Accordingly, at 10 min after stimulation of D10 cells with anti-CD3 antibody, we detected pIκBα that co-localized with Bcl10, Malt1, and pIKK (Fig. 2A). The pIκBα signal was undetectable by 20 min after stimulation, presumably because of its rapid proteasomal degradation, consistent with a previous report of IκBα degradation kinetics in D10 cells and primary T cells (7). These data suggest that IκBα is transiently recruited to the POLKADOTS structures, where it is phosphorylated by IKK, which stimulates its proteasomal degradation.

The NF-κB family consists of five proteins that exist as homo- or heterodimers (15), with RelA(p65)-p50 being the most common activating heterodimer. In unstimulated T cells, NF-κB resides in the cytosol, bound to IκBα. The degradation of IκBα unmasks a nuclear localization signal, which enables NF-κB to enter the nucleus, where it promotes the transcription of target genes. Because we detected pIκBα co-localized with Bcl10-Malt1 clusters, we hypothesized that RelA should be present contemporaneously. Imaging analysis revealed a homogenous cytosolic distribution of RelA before T cell stimulation (Fig. 2B). After 10 min of stimulation, either with antigen-loaded APCs or with anti-CD3 antibody, we observed that RelA co-localized with cytosolic Bcl10-Malt1 aggregates, but was not yet translocated to the nucleus (Fig. 2, B and C). By 20 min after stimulation, RelA was no longer colocalized with the POLKADOTS structures, but was enriched in the nucleus, consistent with our previous observations that IκBα degradation and NF-κB activation occur between 10 and 20 min after stimulation of the TCR on D10 cells (7, 10). Thus, RelA is transiently present at the POLKADOTS structures before it undergoes nuclear translocation.

The inhibition of either IKK or TAK1 blocks IKK phosphorylation at cytosolic Bcl10 signalosomes, causing reduced RelA activation

To further establish a mechanistic link between the IKK phosphorylation observed at the p62-Bcl10-Malt1 clusters (POLKADOTS) and the nuclear translocation of RelA, we treated D10 cells with BAY 11-7082 (BAY 11) or 5Z-7-oxozeaenol (5 oxo). BAY 11 is an inhibitor of IKK, which we used because a previous study suggests that trans-autophosphorylation of IKKβ participates in its activation (16). 5 oxo is a specific inhibitor of TAK1, a kinase identified as an essential activator of IKK phosphorylation in the TCR-NF-κB pathway (17). After stimulation with anti-CD3 antibodies, D10 cells treated with either BAY 11 or 5 oxo showed reduced phosphorylation of IKKα/β without having altered TCR-dependent phosphorylation of extracellular signal–regulated kinases 1 and 2 (ERK1/2) (Fig. 3A), consistent with a previous study (17).

Confocal microscopy analysis of antigen-stimulated D10 cells revealed that although inhibition of either IKK or TAK1 did not affect the translocation of PKCθ to the immunological synapse or the formation of Bcl10 clusters (Fig. 3B), BAY 11– and 5 oxo–treated D10 cells exhibited decreased pIKKα/β staining at Bcl10 clusters compared to those in vehicle-treated cells (Fig. 3B). These data also demonstrated that the observed anti-pIKK plasma membrane staining is non-specific, as it was not affected by BAY 11- or 5 oxo (Fig. 3B and fig. S2A). Additionally, both BAY 11 and 5 oxo blocked the nuclear translocation of RelA in D10 cells stimulated with antigen-loaded APCs or anti-CD3 antibody (Fig. 3, C and D), without apparently altering the translocation of PKCθ to the immunological synapse (Fig. 3C) or the formation of Bcl10 clusters (Fig. 3, C and D). Unstimulated, 5 oxo–treated D10 cells exhibited no change in the distribution of PKCθ-CFP, Bcl10-YFP, or endogenous RelA as compared to cells treated with vehicle (DMSO) alone (fig. S2, B and C). Quantification of microscopy data demonstrated a statistically significant reduction in the ratio of nuclear to cytosolic RelA in BAY 11– or 5 oxo–treated cells after stimulation with either antigen-loaded APCs or anti-CD3 antibody (Fig. 3E). Together, these observations suggest that inhibition of IKK or TAK1 activity prevents the efficient TCR-dependent phosphorylation of IKK. These data furthermore demonstrate that phosphorylation of IKK and activation of RelA are downstream of the translocation of PKCθ to the immunological synapse and the clustering of Bcl10 in POLKADOTS.

Loss of p62 blocks Bcl10 clustering, IKK phosphorylation, and RelA nuclear translocation in effector T cells

Studies suggest that p62 is required for the activation of NF-κB in differentiated effector T cells, but not in naïve T cells (12), which may reflect the fact that naïve cells have little or no p62 (18). Indeed, upon stimulation of wild-type primary CD4⁺ T cells, we observed that p62 was low in abundance in primary CD4⁺ T cells, and that there was a substantial increase in its abundance at 24 and 48 hours after stimulation with anti-CD3 and anti-CD28 antibodies (Fig. 4A), consistent with previous studies (12, 18). Moreover, we observed similar clustering of Bcl10 and ρIKKα/β beneath the capped TCR of naïve wild-type and p62^−/− T cells, further suggesting that p62 does not play an essential role in the TCR-dependent activation of NF-κB in naïve T cells (Fig. 4B).

Two observations suggest that the TCR-dependent activation of NF-κB changes from being p62-independent in naïve T cells to being p62-dependent in effector T cells (2). First, whereas naïve p62^−/− CD4⁺ T cells exhibit normal IKK phosphorylation in response to anti-CD3/28 antibodies, by 24 hours after stimulation (by which time naïve CD4⁺ T cells have differentiated into IL-2-producing effector T cells), p62^−/− CD4⁺ T cells fail to activate IKK (12). Second, the nuclear translocation of RelA in response to stimulation with anti-CD3 antibody is blocked in D10 cells in which p62 is knocked down (7). The implication of these observations is that differentiated T cells require one or more functions of p62 for the activation of IKK and NF-κB.

To establish a mechanistic relationship between p62 abundance and the activation of NF-κB in effector T cells, we assessed the extent of IKK activation in the presence and absence of p62. Western blotting analysis demonstrated minimal IKK phosphorylation in p62^−/− T_H2 cells compared to that in wild-type T_H2 cells in response to anti-CD3/CD28 antibodies (Fig. 4C). In addition, the stimulation of wild-type or p62^−/− T_H2 cells with APCs loaded with SEB (Staphylococcus enterotoxin B) resulted in no difference in the translocation of PKCθ to the plasma membrane (Fig. 4, D and E), indicating that membrane-proximal TCR–NF-κB signaling events were not affected by the loss of p62. However, cytosolic Bcl10 clustering was substantially impaired in p62^−/− T cells compared to that in wild-type T cells in response to stimulation with antigen-loaded APCs or anti-CD3/28 antibodies (Fig. 4, E to G), consistent with our previous observations of reduced formation of POLKADOTS in p62-silenced D10 cells (7). We speculate that the small population of cells with residual Bcl10 clustering may reflect the activity of Nbr1, a protein functionally similar to p62 (18), which co-localizes with p62-speckles in T effector cells (7). Alternatively, the Carma1-dependent aggregation of Bcl10 (11) may inefficiently lead to the production of large cytosolic Bcl10 clusters in the absence of p62.

Finally, the percentage of p62^−/− T_H2 cells that exhibited nuclear translocation of RelA was statistically significantly reduced compared to that of wild-type T_H2 cells in response to stimulation with anti-CD3/28 antibodies (Fig. 4, F and H), consistent with the observed reduction in IKK phosphorylation. Furthermore, those p62^−/− T cells that exhibited nuclear translocation of RelA were the same cells that formed Bcl10-containing POLKADOTS (Fig. 4I). Moreover, the small population of responding p62^−/− cells exhibited the same degree of RelA nuclear translocation as that of wild-type cells, as demonstrated by equivalent ratios of nuclear to cytosolic RelA (Fig. 4J). These data are consistent with our finding that the activation of NF-κB is digital; that is, that varying the intensity of TCR stimulation results in variations in the percentage of cells that respond, whereas the extent of activation of NF-κB on a per-cell basis is invariant in the responding population (10). These data suggest that p62 markedly enhances the efficiency of NF-κB signal transduction, increasing the percentage of effector T cells that successfully activate NF-κB in response to TCR stimulation. In the absence of p62, the activation of NF-κB is very inefficient, likely reflecting the near-absence of p62-dependent cytosolic clustering of Bcl10.

Mice and the differentiation of T_H2 cells

Tissues were harvested from 6- to 12-week-old C57BL/6 wild-type (WT) mice (National Cancer Institute) and p62^−/− mice. The p62^−/− strain (from ES cell clone EPD0162_1_G07) was obtained from the NIH–supported KOMP (knockout mouse project) Repository, generated by the CSD consortium for the NIH-funded KOMP, with established methodology (25). Harvesting of organs and the purification and in vitro differentiation of T_H2 cells were performed as described previously (7, 26). Briefly, CD4⁺ T cells were isolated from lymph nodes from C57BL/6 WT or p62^−/− mice by negative sorting with a CD4 isolation kit (Invitrogen). To differentiate naïve cells into T_H2 cells, naïve CD4⁺ T cells were stimulated with plate-bound anti-CD3 and anti-CD28 antibodies in the presence of IL-2, IL-4, anti-IFN-γ antibody, and anti-IL-12 antibody for 4 days, which was followed by 2 days of culture in IL-2–containing medium to increase cell numbers. Animal experiments were approved by the USU Animal Care and Use Committee.

Antibodies

Primary antibodies used in this study were as follows: rabbit anti-Carma1 (Enzo, ALX-210-903), mouse anti-Bcl10 (Santa Cruz Biotechnology, sc-5273), rabbit anti-p62 (Sigma, P0067), mouse anti-tubulin (Santa Cruz Biotechnology, sc-5286), rabbit anti-PKCθ (Cell Signaling Technology), mouse anti-PKCθ (Enzo), anti-pERK1/2 (Cell Signaling Technology, 4370), mouse anti-GAPDH (Santa Cruz Biotechnology, sc-32233), mouse anti-tubulin (Santa Cruz Biotechnology, sc-5286), rabbit anti-pIKKα/β (Cell Signaling Technology, 2694), mouse anti-IKKβ (Imgenex, clone 10A9B6 (used in Fig. 1D) and clone 10AG2 (used in fig. S1B), mouse anti-TRAF6 (Santa Cruz Biotechnology, sc-8409), mouse anti-pIκBα (Cell Signaling Technology, 9246), and rabbit anti-RelA (Santa Cruz Biotechnology, sc-372). Secondary antibodies included anti-rabbit and anti-mouse immunoglobulin G1 (IgG1) labeled with Alexa Fluor 488, Alexa Fluor 555, or Alexa Fluor 647 (Molecular Probes). Alexa Fluor 647–conjugated streptavidin (Molecular Probes) was used to crosslink anti-CD3 antibodies to induce TCR capping. DRAQ5 (Cell Signaling Technology) or DAPI [which is present in Prolong Gold mounting media (Invitrogen)], were used to mark the nucleus.

Cells, microscopy, and Western blotting

The D10 T_H2 cell clone and B cell lines (CH12 and CHb) were maintained in Eagle’s Ham’s Amino Acids (EHAA) medium with and without IL-2, respectively, as previously described (9). D10 cell lines expressing PKCθ, Bcl10, and Malt1 proteins fused to fluorescent proteins were previously described (8, 9). Formation of T cell–B cell conjugates was performed as previously described (7, 9). For experiments with SEB (a gift from C. Ventura and A. O’Brien), CHb cells were loaded with SEB (10 μg/ml) overnight, and then were centrifuged at 2000g for 30 s with an equal number of primary T_H2 cells. Stimulation of cells with anti-CD3 antibody alone or together with anti-CD28 antibody was performed as previously described (7). Capping of TCR on naïve T cells involved the use of biotinylated anti-CD3 antibody and Alexa Fluor 647–conjugated streptavidin, as previously described (3). For inhibitor experiments, T cells were incubated with 2 μM 5Z-7-oxozeaenol (Enzo) or DMSO (vehicle) 30 min before being stimulated, or with BAY 11-7082 (5 μg/ml, Calbiochem) 5 min after initiation of stimulation. For confocal microscopic analysis, cells were fixed and incubated with antibodies as previously described (8), and were visualized with a 40× 1.4 NA oil objective, a Zeiss 710 NLO microscope, and Zen software. Cell stimulation, harvesting, and Western blotting analysis were performed as previously described (7, 8).

Quantification of microscopy data

Raw data exported from Zeiss Zen software were analyzed with ImageJ software (US National Institutes of Health). For quantification of RelA nuclear occupancy, a region of interest (ROI) was drawn around the nucleus and cytoplasm of individual cells, using DRAQ5 or DAPI staining to define the nuclear periphery and the DIC image to define the plasma membrane (cytoplasm periphery). The mean pixel intensity of anti-RelA fluorescence within the nuclear and cytosolic ROIs was obtained for each cell and used to derive the nuclear/cytosolic RelA ratio. At least 20 cells were examined in each group. For quantifying POLKADOTS, cells with ≥2 Bcl10 clusters were scored as positive for POLKADOTS formation. Cells were scored as positive for nuclear RelA based on the presence of overlap of between RelA and DRAQ5 (or DAPI) fluorescence. At least 20 cells were counted in each group. GraphPad Prism 6.0 software was used to plot data and calculate errors and statistics.

Statistical analysis

Where indicated in the figure legends, P values for the differences in means were calculated with an unpaired, two-tailed Students t test.