MYD88L265P has recently been discovered as an extraordinarily frequent somatic mutation in benign monoclonal IgM gammopathy, Waldenström’s macroglobulinemia, and diffuse large B cell lymphoma. In this study, we analyze the consequences for antigen-activated primary B cells of acquiring MYD88L265P. The mutation induced rapid B cell division in the absence of exogenous TLR ligands and was inhibited by Unc93b13d mutation and chloroquine or TLR9 deficiency, indicating continued dependence on upstream TLR9 activation. Proliferation and NF-κB activation induced by MYD88L265P were nevertheless rapidly countered by the induction of TNFAIP3, an NF-κB inhibitor frequently inactivated in MYD88L265P–bearing lymphomas, and extinguished by Bim-dependent apoptosis. MYD88L265P caused self-reactive B cells to accumulate in vivo only when apoptosis was opposed by Bcl2 overexpression. These results reveal checkpoints that fortify TLR responses against aberrant B cell proliferation in response to ubiquitous TLR and BCR self-ligands and suggest that tolerance failure requires the accumulation of multiple somatic mutations.
B cell lymphoproliferative diseases represent natural mutagenesis experiments that shed light on normal B cell regulatory mechanisms (Rui et al., 2011) in addition to being major causes of human morbidity and mortality. These take numerous forms, encompassing non-Hodgkin and Hodgkin lymphomas, chronic lymphocytic leukemia, Waldenström’s macroglobulinemia, myeloma, and clinical or subclinical monoclonal gammopathies (Shaffer et al., 2002). Learning about normal B cell regulation from malignant B cells is confounded, however, by the accumulation of 20 or more protein-altering somatic mutations in malignant B cell clones (Morin et al., 2011; Pasqualucci et al., 2011; Puente et al., 2011). The drive toward malignancy must begin with individual mutations, but aside from a few well-studied mutations like MYC and BCL2 translocations (ar-Rushdi et al., 1983; Tsujimoto et al., 1985; Vaux et al., 1988), little is known about the consequences of recurring lymphoma mutations individually or combinatorially for the behavior of otherwise normal mature B cells.
MYD88 mutations have emerged as one of the most frequently recurring mutations in mature B cell lymphoproliferative disease. Somatic missense mutations in MYD88 were discovered by Ngo et al. (2011) in 39% of cases of a common form of non-Hodgkin’s lymphoma, activated B cell type diffuse large B cell lymphoma (ABC-DLBCL), with a single L265P substitution accounting for 75% of the mutations. The L265P mutation occurs in almost 100% of cases of Waldenström’s macroglobulinemia (Treon et al., 2012; Xu et al., 2013), at least 47% of cases of IgM monoclonal gammopathy of undetermined significance (Xu et al., 2013), 3–10% of cases of chronic lymphocytic leukemia (Puente et al., 2011; Wang et al., 2011), and 13% of splenic marginal zone lymphoma (Trøen et al., 2013). Other TIR domain mutations, such as S219C, predominate in germinal center B cell type diffuse large B cell lymphoma (GCB-DLBCL; Ngo et al., 2011).
MYD88 is an important adaptor protein that bridges TLR and the IL-1 receptor to the activation of downstream IL receptor–activated kinases (IRAKs) and NF-κB transcription factor activation (Akira and Takeda, 2004). MYD88 has two distinct domains, the Toll/IL-1R like domain (TIR), via which MYD88 proteins homodimerize upon activation, and the death domain (DD), which recruits IRAKs to form the signaling complex (Akira and Takeda, 2004). Interestingly, all MYD88 lymphoma mutations are found in the TIR domain and result in uncontrolled formation of the MYD88–IRAK signaling complex (Ngo et al., 2011). An ABC-DLBCL cell line with the MYD88L265P mutation showed hyperphosphorylation of IRAK1 and elevated NF-κB activity, whereas shRNA studies established that the dysregulated MYD88 to NF-κB signaling was necessary for the survival of this cell line (Ngo et al., 2011). Similarly evidence for this mutation driving exaggerated NF-κB activity has been obtained in malignant cells from Waldenström’s macroglobulinemia (Treon et al., 2012) and CLL (Wang et al., 2011). However, it remains unclear whether MYD88L265P mutation actively drives the proliferation of these malignant B cells or only maintains their survival, and the consequences of MYD88L265P mutation in the precursors of malignant B cells that do not carry numerous other somatic mutations are unknown.
Discrimination between chemical components of infecting microbes and self-tissues is the central problem for normal B cell regulation. B cells express multiple TLRs, each serving as a sensor for infection by binding evolutionarily conserved molecules that differ between microbes and self (Akira and Takeda, 2004; Beutler, 2004). TLR3, TLR7, and TLR9 bind features of RNA or DNA that are enriched in microbial as opposed to mammalian nucleic acids, such as unmethylated CpG-rich DNA sequences or double-stranded RNA (Krieg, 2002). Because these features are also present at lower abundance in self-nucleic acids, the nucleic acid–sensing TLRs must use additional mechanisms to ensure they tolerate and do not trigger immune responses to self-nucleic acids. The mechanisms for TLR self-tolerance are nevertheless not well understood. One important mechanism is restriction of the activity of TLR3, TLR7, and TLR9 to acidified endosomes, where microbes are frequently trafficked by endocytosis after being captured by cell surface immunoglobulin (B cell antigen receptors [BCRs]). Restriction is achieved by Unc93b1-mediated TLR3, TLR7, and TLR9 trafficking to endosomes (Tabeta et al., 2006; Kim et al., 2008), and by requirement for proteolytic activation of the TLR ectodomain by endosomal proteases active only at low pH (Ewald et al., 2008). Because self-binding BCRs are negatively selected through processes of central and peripheral B cell deletion, editing, and anergy (Goodnow and Ohashi, 2013), these BCR tolerance mechanisms assist self–nonself discrimination with nucleic acid–sensing TLRs in B cells.
TLR tolerance to self-nucleic acids breaks down when BCR tolerance mechanisms fail, allowing BCRs to capture self-nucleic acids directly or indirectly and deliver them to acidified endosomes (Leadbetter et al., 2002). TLR and BCR signaling converges on NF-κB activation through separate adapter proteins, MYD88 and CARD11, respectively. BCR tolerance involves an uncoupling of CARD11 signals to NF-κB, and is disrupted by a range of activating somatic mutations in CARD11 frequently found in non-Hodgkin lymphoma (Lenz et al., 2008; Jeelall et al., 2012). It is conceivable that BCR tolerance mechanisms might also be disrupted by uncontrolled activity of nucleic acid sensing TLRs or gain-of-function MYD88 mutations, potentially initiating a vicious cycle of cross-activation by self-sensing BCRs and TLRs, but this theoretical possibility has yet to be examined.
Here, we analyze the effect of MYD88 mutations on the behavior of normal antigen-activated B cells, including B cells bearing self-reactive BCRs, using a retroviral gene transfer system in mice. We show that MYD88L265P initially breaks tolerance to RNA- or DNA-sensing TLRs, resulting in Unc93b1 and TLR9-dependent B cell proliferation in the absence of foreign ligands for TLR9. However, activation of NF-κB signaling and proliferation by MYD88L265P is rapidly terminated through inhibition of NF-κB signaling by Tnfaip3 and by Bcl2/Bim-regulated apoptosis. These results reveal a multilayered mechanism for B cell tolerance to self-ligands of nucleic acid–sensing TLRs, whose progressive disruption by mutations often found in human B cells cross-interferes with tolerance to self-ligands of the BCR.
MYD88L265P mutation drives B cell proliferation independent of foreign TLR ligand in vitro
To investigate the consequences of MYD88 mutations arising in mature, antigen-activated B cells that lack other lymphoma mutations, we used retroviral gene transfer to introduce normal or mutated mouse Myd88 alleles into activated B cells from mice that bear a transgenic BCR to a known antigen, hen egg lysozyme (HEL; Goodnow et al., 1988; Fig. 1 A). B cells were activated by antigen, and then stimulated by CD40; retrovirally transduced with mutant Myd88, WT Myd88, or EGFP-only control reporter vectors; washed; and placed back into tissue culture in the absence of mitogen stimulation to compare the fate of EGFP+ cells over time (Fig. 1, A and B). We initially tested the effects of six MYD88 TIR domain mutations (Ngo et al., 2011): L265P, S219C, S222R, M232T, S243N, and T294P (Fig. 1 A). Four of the six, L265P (Mut1), S219C (Mut2), M232T (Mut4), and S243N (Mut5), promoted mitogen-independent B cell proliferation in culture that was not observed when wild-type MYD88 was expressed, whereas S222R (Mut3) and T294P (Mut6) showed no proliferation. The wild-type and mutant Myd88 proteins were expressed at comparable levels, demonstrating that proliferation resulted from specific mutations and not simply from overexpression of the protein (Fig. 1, B–D). B cells transduced with the MYD88L265P:EGFP vector spontaneously increased their number in the first 3 d of culture to 4–5 times the starting culture number, despite the absence of antigen or anti-CD40 antibody (Fig. 1 B). In contrast, over the same period the MYD88WT:EGFP and empty:EGFP vector-transduced B cells did not increase in number (Fig. 1 B). EGFP− nontransduced B cells served as an internal control and all declined in number over the course of 3 d, indicating that proliferation induced by MYD88L265P was limited to the B cells that expressed the mutation and not the result of a secreted or cell–cell interaction factor (Fig. 1 B). However, the MYD88L265P-induced proliferation was short lived and the numbers of live EGFP+ B cells dropped sharply after day 3 in culture (Fig. 1 B).
Cell division was measured by labeling the transduced B cells with cell trace violet (CTV), a dye that is progressively diluted with each cell division. MYD88L265P:EGFP+ B cells diluted the CTV dye over the course of 3 d as a result of multiple cell divisions (Fig. 1 E). Most B cells expressing comparable levels of WT MYD88wildtype:EGFP or empty:EGFP vector did not divide over this time course, although a subset of MYD88WT–transduced B cells expressing very high levels of EGFP did persist and divide (Fig. 1 E and not depicted). In contrast, the accumulating B cells with MYD88L265P expressed the bicistronic MYD88L265P:EGFP vector over a broad range that mirrored the starting distribution and the distribution in cells transduced with empty:EGFP.
We next looked at the downstream pathways that drove the proliferation of MYD88L265P B cells. Since MYD88 and CARD11 are adaptor proteins that connect TLRs and the B cell antigen receptor, respectively, to NF-κB activation, we compared the effects of activated mutant MYD88 and CARD11 alleles on NF-κB activity in primary B cells, alongside a mutant IKBKB allele (K171E; Rossi et al., 2011) that directly activates NF-κB (Mercurio et al., 1997). As described previously (Jeelall et al., 2012), B cells transduced with CARD11L232LI:EGFP vector displayed increased NF-κB p65 phosphorylation (Fig. 1 F). CARD11L232LI or IKKK171E increased expression of CD25, CD95, and CD69, whereas the latter also induced CD23, each encoded by an NF-κB–induced gene (Fig. 1 G). Paradoxically, MYD88L265P-transduced B cells had markedly diminished NF-κB p65 phosphorylation and decreased CD25, CD95, and CD23 expression relative to control B cells transduced with empty vector (Fig. 1, F and G).
TNFAIP3 mediates MYD88L265P-induced shutdown of proliferation and NF-κB activation
The aforementioned experiments showed that B cell dysregulation caused by MYD88L265P differs from other NF-κB–activating lymphoma mutations in multiple respects: MYD88L265P-induced cell division in vitro was rapidly self-limiting, and MYD88L265P triggered a decrease instead of induction of NF-κB–activated genes encoding CD23, CD25, and CD95. These results, together with the frequent co-occurrence of MYD88 and TNFAIP3 mutations in ABC-DLBCL (Ngo et al., 2011), raised the possibility that MYD88L265P might have an intrinsically self-limiting effect on B cells by inducing a negative feedback loop through induction of TNFAIP3 (also designated A20). TNFAIP3 mRNA and protein are rapidly induced by NF-κB and form a critical negative feedback loop to diminish NF-κB activity by adding K48-linked ubiquitin chains, removing K63-linked ubiquitin chains, and inhibiting the addition of linear ubiquitin chains to key molecules in the NF-κB signaling axis (Harhaj and Dixit, 2012; Tokunaga et al., 2012; Verhelst et al., 2012).
To test the involvement of TNFAIP3, we first measured the mRNA level of TNFAIP3 in B cells transduced with MYD88L265P and found it was increased 300% compared with B cells transduced with empty:EGFP vector (Fig. 2 A). We then took advantage of a recently discovered ENU mutant mouse strain, lasvegas, with a partial loss of function Tnfaip3 mutation in the OTU domain that abolishes deubiquitinating activity (Tnfaip3lsv; unpublished data). When B cells with homozygous Tnfaip3lsv mutation were transduced with MYD88L265P and cultured without mitogens, the number of EGFP+ B cells continued to increase between days 3 and 5, in contrast to the rapid drop that occurred in WT B cells expressing MYD88L265P (Fig. 2, B and E). WT and Tnfaip3lsv mutant B cells had diluted the cell division tracking dye, CTV, to comparable levels on day 3 but by day 5 the mutant B cells had greater dilution (Fig. 2 F). The increased division of mutant B cells between days 3 and 5 corresponded to less than one additional division (Fig. 2 F), consistent with less than a doubling of the numbers of mutant cells over this time (Fig. 2 E), and is insufficient to account for the large difference in numbers of mutant and wild-type B cells at day 5. The main effect of Tnfaip3 mutation in B cells expressing MYD88L265P appears to be a delayed induction of apoptosis.
EGFP+ WT or Tnfaip3lsv B cells expressing MYD88L265P or control vectors were sorted on day 1 of culture without mitogens, and cell lysates were analyzed by Western blotting for phosphorylation of NF-κB p65. MYD88L265P decreased p65 phosphorylation in WT cells, as observed above, but this effect was negated in Tnfaip3lsv B cells (Fig. 2 C). Likewise, in WT B cells MYD88L265P decreased the expression of CD23, CD25, CD95 and CD86 on EGFP+ cells relative to EGFP− internal control B cells, but this was abolished in Tnfaip3lsv B cells (Fig. 2 D). These results established that the rapid shut down of MYD88L265P-driven B cell proliferation and NF-κB activity is partially Tnfaip3 dependent.
MYD88L265P-driven B cell proliferation is chloroquine sensitive and Unc93b1 and IRAK dependent
To test if normal B cell proliferation induced by MYD88L265P required upstream signaling by the RNA/DNA-sensing receptors TLR3, TLR7, and TLR9, as has been observed in malignant ABC-DLBCL cell lines (unpublished data), we repeated the retroviral transduction experiments in B cells where RNA/DNA-sensing TLRs were inhibited. Trafficking of these TLRs to endosomal compartments needed for active signaling was blocked by transducing B cells homozygous for the Unc93b13d mutation (Tabeta et al., 2006; Kim et al., 2008). Endosomal acidification needed for proteolytic activation of the TLR7 and TLR9 ectodomains was inhibited by treating the B cells with chloroquine (Ewald et al., 2008), and TLR9 was selectively eliminated by transducing B cells homozygous for a Tlr9-null mutation (Hemmi et al., 2000). These genetic (Fig. 3, A–F) or pharmacological (Fig. 3, C and D) interventions specifically inhibited proliferation of MYD88L265P:EGFP-expressing B cells, but had no effect on proliferation of CARD11L232LI:EGFP-expressing B cells tested in parallel. Slow residual division nevertheless consistently occurred in Unc93b13d or Tlr9 mutant B cells bearing MYD88L265P, indicating either weak activation by other Unc93b1-independent receptors or low-level, receptor-independent signaling by the MYD88 mutant protein. Thus, mutations in MYD88 induce transient B cell proliferation in the absence of normal B cell mitogens that nevertheless depends for the most part on an intact DNA-sensing TLR9 signaling apparatus.
We tested whether downstream IRAK activity was also required for proliferation by treating transduced B cells with an IRAK1/4 inhibitor, N-acyl 2-aminobenzimidazole (Powers et al., 2006). MYD88L265P transduced B cells incubated with 5 µM IRAK1/4 inhibitor showed an 80% reduction in proliferation (Fig. 3 G). This inhibitory effect was specific for B cells carrying mutant MYD88, as the IRAK inhibitor had little inhibitory effect on proliferation of B cells transduced with vectors encoding gain-of-function alleles of CARD11 or IKK analyzed in parallel cultures (Fig. 3 G).
MYD88L265P induces T cell–independent B cell proliferation in vivo
To examine the consequences of MYD88L265P mutation for normal B cells in vivo, we tracked its effect on mature retrovirally transduced B cells transplanted into Rag1−/− recipient mice. After in vitro transduction with MYD88L265P:EGFP, MYD88WT:EGFP, or empty:EGFP vector, as above, the activated B cells were loaded with the cell division dye CTV and transplanted by intravenous injection into syngeneic Rag1−/−-recipient mice (Fig. 4 A). Groups of recipients were analyzed 5, 7, 11 and 14 d after transplantation for the number of EGFP+ donor B cells in the spleen. Flow cytometry was used to identify donor B cells by their surface expression of B220 and IgM, and to enumerate the percentage and absolute number that expressed EGFP before and after transplantation (Fig. 4, A and B). While similar percentages and numbers of EGFP+ cells expressing each vector were injected into the mice (Fig. 4, C and G), the percentage and total number of EGFP+ cells expressing MYD88L265P increased 8 to 12 fold by day 5 after transplantation whereas no increase occurred in control B cells expressing MYD88WT (Fig. 4, D–H). The increase in MYD88L265P-transduced B cells was accompanied by extensive dilution of CTV, demonstrating that the mutation induced at least eight rounds of cell division in many of the EGFP+ B cells in vivo (Fig. 4, B and F). When later times after transplantation were analyzed, there only a slight additional increase or decrease in the numbers of MYD88L265P-expressing B cells (Fig. 4 I). Thus, MYD88L265P was sufficient to initiate spontaneous proliferation of mature B cells both in vitro and in vivo, although in both cases the aberrant clonal growth was rapidly self-limiting.
MYD88L265P cooperates with dysregulated Bcl2 or Bim deficiency
Bcl2 mRNA and protein are frequently elevated in malignant B cells and correlate with poor clinical outcome (Shen et al., 2004; Iqbal et al., 2006). Because Bcl2 inhibits B cell apoptosis induced by the Bim, Bak, and Bax proteins (Cory, 1995), we tested the possibility that the short-lived proliferative burst induced by MYD88L265P was curtailed by activation of this apoptotic pathway. Myd88WT:EGFP-, Myd88L265P:EGFP-, or empty:EGFP-transduced B cells washed off CD40 were placed in culture for either 1 or 3 d and stained intracellularly for Bim protein levels. Compared with empty:EGFP vector, B cells transduced with Myd88L265P showed elevated Bim protein on day 1 and day 3 (Fig. 5 A). The increased Bim in MYD88L265P-transduced cells contrasts with our previous finding that Bim is decreased in B cells expressing activated Card11 or Ikk (Jeelall et al., 2012). To test whether Bcl2 overexpression could overcome the effects of Bim, activated B cells were obtained from Vav-Bcl2 transgenic mice that constitutively express Bcl2 in hematopoietic cells (Egle et al., 2004) and from WT control mice. These were transduced with MYD88L265P:EGFP or empty:EGFP vector, and cultured without mitogenic stimuli to enumerate EGFP+ cells. The number of WT control B cells expressing MYD88L265P increased until day 3, and then declined sharply by day 5 as observed above, whereas the corresponding Vav-Bcl2 B cells increased to greater numbers on day 3 and continued to increase by day 5 (Fig. 5 B). Vav-Bcl2 B cells transduced with the empty:EGFP-only vector did not proliferate, but survived better in culture than corresponding WT cells, as expected, because of resistance to apoptosis (Fig. 5 B). To distinguish if the increased accumulation of Vav-Bcl2 B cells expressing MYD88L265P was caused by enhanced cell division or better survival of the progeny, the transduced B cells were labeled with CTV. Despite much higher accumulation of Vav-Bcl2 than WT control B cells, there was no difference in the dilution of CTV within the MYD88L265P:EGFP+ population on days 3 or 5 (Fig. 5 C). Thus, the Vav-Bcl2 transgene did not alter the rate of cell division induced by MYD88L265P but enhanced the accumulation of progeny at each division. Despite this, the modal CTV fluorescence only decreased from 1/8 to 1/16 of the nondividing empty-vector control cells between days 3 and 5 (Fig. 5 C), indicating that MYD88L265P-induced B cell proliferation arrested after 4 divisions independent of apoptotic loss.
We then asked whether mutations that inhibit B cell apoptosis either through Bcl2 overexpression or inactivation of the proapoptotic Bim protein (Bcl2l11) also enhanced the accumulation of MYD88L265P-expressing B cells in vivo, where the prosurvival cytokine BAFF/Blys is available (Woodland et al., 2006). Transduced B cells from Vav-Bcl2 transgenic, Bim-deficient (BimKO; Bouillet et al., 1999), or WT control mice were injected into Rag1−/− recipient mice and spleen cells analyzed 11 d after injection (Fig. 5 D). In control groups that received B cells transduced with empty vector, three to four times more B220+IgM+ B cells remained in recipients that received Vav-Bcl2 or BimKO B cells compared with WT B cells, which is consistent with enhanced survival of nontransduced and transduced B cells in vivo (Fig. 5 D). However the percentage of transferred B cells expressing the empty EGFP vector did not increase between the time of transfer and 11 d after transfer, regardless of Bcl2 or Bim genotype (black lines in Fig. 5 E). In contrast, the percentage of B cells expressing MYD88L265P:EGFP increased 4-fold by day 11 in the case of wild-type B cells, and increased 7-fold and 11-fold in Vav-Bcl2 transgenic or Bim-deficient B cells, respectively (red lines in Fig. 5 E). Measured as total number of EGFP+ B cells, half as many EGFP+ cells were injected in the groups receiving Vav-Bcl2 or BimKO B cells (Fig. 5 D) due to lower transduced percentages on day 0 (Fig. 5 E). There was nevertheless a multiplicative effect of combining MYD88L265P with either Vav-Bcl2 or BimKO on the numbers 11 d later: alone, each mutation increased the numbers of EGFP+ B cells by ∼6-fold relative to the number of empty-vector–expressing WT B cell controls, whereas combined they increased the number of EGFP+ B cells 25–35-fold (Fig. 5, E and G).
MYD88L265P blocks self-antigen–induced elimination only when combined with dysregulated Bcl2
All lymphoma CARD11 mutations tested so far have the property of switching B cells from self-antigen–induced death into self-antigen–induced proliferation and plasma cell differentiation (Jeelall et al., 2012). To test if lymphoma mutations in MYD88 have the same activity, HEL antigen–specific B cells transduced with MYD88L265P:EGFP or EGFP-only empty vectors were transplanted into Rag1−/− mice that express HEL in their circulation as a self-antigen (Fig. 6 A). When analyzed 13 d after transfer, EGFP+ B cells expressing MYD88L265P proliferated and persisted in the absence of antigen in nontransgenic Rag1−/− control mice as observed previously, but were eliminated in HEL-transgenic recipients (Fig. 6, A–C). The unimpeded deletion of MYD88L265P-expressing B cells contrasts with the protection from elimination and induction of plasma cell differentiation induced by CARD11 mutations under the same experimental conditions (Jeelall et al., 2012).
Whereas MYD88L265P alone was insufficient to block antigen-induced death, we next asked if it could do so in cooperation with dysregulated Bcl2, given the cooperation observed above in the absence of antigenic stimulation and the capacity of dysregulated Bcl2 alone to delay elimination of mature self-reactive B cells (Cyster et al., 1994). HEL-specific B cells transduced with MYD88L265P or empty vector were transferred into Rag1−/− HEL-transgenic mice as above, except that half of the mice received B cells from Vav-Bcl2-Tg donors (Fig. 6 D). WT B cells transduced with the empty:EGFP or MYD88L265P:EGFP vectors were eliminated by day 13, whereas small numbers of Vav-Bcl2-Tg cells carrying empty vector survived; however, their frequency did not increase relative to the percentage EGFP+ on day 0 (Fig. 6, D–G). However the combination of MYD88L265P and Vav-Bcl2 resulted in 10 times more EGFP+ B cells accumulating in HEL-transgenic recipients compared with Vav-Bcl2 alone and 200 times more EGPF+ B cells than empty vector alone (Fig. 6, D–F). In contrast to CARD11 mutations (Jeelall et al., 2012), the combination of MYD88L265P and Vav-Bcl2 mutations remained insufficient to drive spontaneous differentiation of self-antigen–stimulated B cells into CD19low plasmablasts, as the majority of accumulating EGFP+ B cells remained CD19high (Fig. 6 D). Thus MYD88L265P was insufficient to interfere with BCR-induced deletion on its own, but exhibited a highly cooperative interaction with dysregulated Bcl2 to block elimination of self-reactive B cells and allow their accumulation.
B cells must normally tolerate self-ligands of their pathogen-receptor systems (BCRs and TLRs) and only proliferate in response to foreign ligands for these receptors. The aforementioned experiments reveal that tolerance to nucleic acid-sensing TLRs is disrupted by a somatic mutation in MYD88 that is very frequently found in the benign disorder, IgM monoclonal gammopathy of undetermined significance, and in a range of B cell malignancies. In the absence of foreign TLR ligands, MYD88L265P was sufficient to drive multiple rounds of B cell division provided the Unc93b1-dependent and chloroquine-sensitive steps in TLR9 activation were intact, and provided the B cells were not constantly binding self-antigen. This disruption of normal tolerance to TLR9 has parallels with the effects of weakly activating lymphoma CARD11 mutations, which break normal B cell tolerance to self-ligands of the BCR (Jeelall et al., 2012). However breakdown of TLR tolerance is fortified by more checkpoint mechanisms: MYD88L265P-induced proliferation was rapidly curtailed by Tnfaip3-mediated shutdown of NF-κB and by Bcl2-inhibited, Bim-dependent apoptosis. When the apoptotic checkpoint was corrupted by a second mutation, MYD88L265P promoted much greater B cell accumulation and blocked B cell elimination by self-antigen.
What drives the initial proliferation of MYD88L265P-bearing primary B-cells? The inhibition of proliferation by the Unc93b13d mutation, chloroquine, and Tlr9 deficiency indicates that activation of TLR9 is necessary, and hence the MYD88 mutation does not simply result in constitutively activated IRAK signaling. This result, in otherwise normal B cells, complements findings (unpublished data) that show malignant ABC-DLBCL cells bearing the MYD88L265P mutation remain dependent on active TLR9 for their survival or proliferation. A linkage between dysregulated inflammatory responses and cancer has long been proposed, starting with Virchow back in 1863 (Balkwill and Mantovani, 2001). Since then, many studies have linked chronic infection and inflammation with B cell malignancy, most notably in gastric MALT lymphoma associated with Helicobacter pylori infection (Wotherspoon et al., 1993). The induction of proliferation and resistance to self-antigen–induced elimination in B cells bearing the MYD88L265P mutation may represent an unbridled response to endogenous TLR9 ligands, with parallels to the B cell dysregulation observed in mice inheriting either a duplicated Tlr7 locus (Y-linked autoimmunity mutation; Yaa) or multiple copies of a Tlr7 transgene (Pisitkun et al., 2006; Subramanian et al., 2006; Deane et al., 2007). It also parallels the autoimmune B cell proliferation caused when B cell antigen receptors bind and deliver excessive amounts of self-DNA to TLR9 (Leadbetter et al., 2002). Hence the findings here provide a rationale for future studies to test for inherited or somatic MYD88 mutations in systemic lupus and other autoimmune diseases.
The results also indicate that overexpression of wild-type MYD88 causes B cell dysregulation. B cells that expressed the highest levels of the MYD88WT:EGFP vector were indeed stimulated to divide in culture. In contrast, EGFPmid B cells expressing the different vectors were only stimulated to divide if the vector encoded MYD88L265P:EGFP. These results provide experimental evidence to explain the observation that MYD88 overexpression without L265P mutation is seen in 36% of DLBCL cases and is associated with tumor recurrence and shortened disease-free survival (Choi et al., 2013).
The findings here reveal surprising differences between lymphoma MYD88L265P and CARD11 mutations. Both break the tuning of intracellular signals from their respective receptors (TLRs, BCRs) that normally prevents proliferative responses to self. However, based on the findings here TLR signaling appears more heavily fortified against this contingency, and has the potential to cross-interfere with normal tolerance to BCR self-ligands. MYD88L265P was sufficient to initiate B cell proliferation much like activated alleles of CARD11, but MYD88L265P differed from CARD11 because within 1–2 d, it induced a decrease in NF-κB p65 RelA phosphorylation, a decrease in down-regulation of NF-κB-induced cell surface proteins, and an increase in Bim protein. Cell division was prematurely terminated by apoptosis between 3 and 4 d. These self-extinguishing effects of MYD88L265P were negated by a Tnfaip3 partial loss of function mutation. Tnfaip3 is induced by NF-κB signaling as an early response gene and encodes the A20 protein, a ubiquitin modifier that inhibits formation of linear and K63-linked ubiquitin chains on key proteins in the NF-κB signaling pathway to provide a negative feedback that terminates further activation of NF-κB (Harhaj and Dixit, 2012; Tokunaga et al., 2012; Verhelst et al., 2012). Tnfaip3 inactivation accompanies MYD88L265P in 24% of ABC-DLBCL (Ngo et al., 2011) and in 38% of Waldenström’s macroglobulinemia (Braggio et al., 2009). The fact that many MYD88L265P-bearing lymphomas have not acquired an inactivating Tnfaip3 mutation raises the possibility that other recurring lymphoma mutations might represent alternative mechanisms to circumvent MYD88’s self-extinguishing property. CARD11 mutations may evade this feedback mechanism because CARD11 activates the paracaspase, MALT1, which cleaves and inactivates A20 (Coornaert et al., 2008; Rebeaud et al., 2008). MYD88 and CARD11 mutations also differed in their ability to counter cell death in the context of self-antigen stimulation, and this may reflect the capacity of activated CARD11 to diminish Bim protein levels (Jeelall et al., 2012). In support of that interpretation, we found that deletion could only be blocked by MYD88L265P when combined with a mutation that enforced expression of the Bim-inhibitor, Bcl-2.
BCL2 t(14:18) translocations that cause dysregulated BCL2 expression occur in almost half of apparently healthy adults and increase with age (Schüler et al., 2009). Many lymphomas bearing MYD88L265P lack a BCL2 translocation. In these cases, perhaps other recurring mutations provide alternative mechanisms to counteract Bim-induced apoptosis; e.g., by increasing BCL2 expression from its normal locus. For example, the delayed apoptosis observed when MYD88L265P was combined with Tnfaip3 mutation may result from increased NF-κB induction of Bcl2. The MYD88L265P and Vav-Bcl2 double mutation experiments indicate that additional layers of control exist to counter dysregulation of TLR responses even when MYD88L265P becomes paired with Bim inactivation via overexpressed Bcl-2. Although the two mutations cooperate to prevent elimination of self-reactive B cells, few of these rogue self-antigen–binding B cells differentiated into CD19low plasmablasts. This contrasts with lymphoma-associated CARD11 mutations, where even the weakest alleles caused self-antigen to drive not only proliferation but also extensive differentiation into plasmablasts (Jeelall et al., 2012).
Collectively, the findings in this study reveal checkpoints that normally prevent a single activating mutation in MYD88 from triggering dysregulated B cell proliferation in response to self-protein, RNA, or DNA. These checkpoints provide insights into the clinical associations observed between MYD88 somatic mutations and benign monoclonal IgM gammopathy (Xu et al., 2013); between somatic TNFAIP3 inactivation, MYD88 activation, and malignant DLBCL (Ngo et al., 2011); between inherited TNFAIP3 polymorphisms and rheumatoid arthritis or systemic lupus (Plenge et al., 2007; Thomson et al., 2007; Graham et al., 2008). They also provide insight into the association of these autoimmune diseases with development of B cell lymphoma (Goodnow, 2007). Although self-ligands for TLRs and BCRs are a ubiquitous potential stimulus for B cell proliferation and antibody secretion, the TLR signaling pathway appears fortified to ensure this possibility can only arise through accumulation of multiple somatic or inherited mutations.
materials and methods
Splenic mature B cells were obtained from (a) MD4 transgenic mice (Goodnow et al., 1988) bearing rearranged transgenic Ig encoding for HEL-specific antibodies (IgHEL transgenic); (b) Vav-Bcl2 transgenic mice where Bcl2 expression was driven by the pan-hematopoietic Vav1 promoter provided by S. Cory (Walter and Eliza Hall Institute of Mediacal Research, Parkville, VIC, Australia; Egle et al., 2004), either alone or as double-transgenic mice with IgHEL; (c) Bim (Bcl2l11) knockout mice, provided by A. Strasser (Walter and Eliza Hall Institute of Mediacal Research; Bouillet et al., 1999); (d) Unc93b3D, an ENU-induced loss of function mutation generated in C57BL/6 mice (Tabeta et al., 2006); and (e) TLR9−/− (Hemmi et al., 2000). All mice were either generated on a C57BL/6 background or backcrossed to that background for >10 generations, and were housed in specific pathogen–free environment at the Australian Phenomics Facility, ANU. Donor mice were 8–16 wk of age. Rag1−/− mice and Rag1−/− ML5-transgenic mice that express soluble HEL (Goodnow et al., 1988) were used as recipients at 8–14 wk of age.
Retroviral vectors and packaging.
Myd88 was amplified by Platinum Pfx DNA polymerase (Invitrogen) from mouse spleen cDNA and cloned into pcDNA 3.1 (+) vector. PCR-based site-directed mutagenesis was used to introduce Myd88 mutations and the PCR mutagenesis products were sequenced on an AB 3730xl DNA Analyzer. WT and mutant Myd88 cDNAs were then subcloned in pMXs-IRES-GFP vector (provided by T. Kitamura, University of Tokyo, Tokyo, Japan). Retroviral vectors containing WT or mutant Myd88 were transfected into Phoenix ecotropic packaging cells (American Type Culture Collection) using calcium phosphate precipitation. Supernatants containing retroviral particles were collected 48 and 72 h after transfection and frozen at −80°C until use.
Retroviral transduction and culture of B cells.
B cells from MD4 transgenic mice were activated by in vivo HEL stimulation followed by in vitro anti-CD40 stimulation for 24 h, and then transduced with retroviral vectors as previously described (Jeelall et al., 2012). B cells from nontransgenic mice were activated by in vitro stimulation with anti-IgM (10 µg/ml) and anti-CD40 (10 µg/ml) for 24 h before transduction with the retroviral vectors, as above. The transduced cells were then washed three times with fresh complete RPMI by centrifugation at 300 g for 5 min at 8°C, and any remaining media containing trace amount of anti-CD40 antibodies was carefully removed. The number of live cells was determined by hemocytometer counting of Trypan blue–negative cells and the percentage of EGFP+ 7AAD− cells was determined by flow cytometry. Washed, transduced cells were then cultured in 24-well plates at 106cells/ml in fresh complete RPMI without any mitogen supplement, with the start of the mitogen-free cultures designated “day 0.” The number of EGFP+ cells after different days in culture was determined by harvesting all the cells in a culture well, counting Trypan blue–negative cells in a hemocytometer, and doing flow cytometric analysis of the same cells to measure the percentage 7AAD− cells that were B220+ EGFP+. Flow cytometric analysis was performed as described in (Jeelall et al., 2012). For cell division analysis, transduced cells at day 0 were loaded with 20 µM CTV (Invitrogen) for 30 min before culture.
Retrovirally transduced B cells were cultured in media containing anti-CD40 for 36 h after spin-infection, washed, and then analyzed by hemocytometer and flow cytometry as above. 5 × 106 to 10 × 106 cultured viable B cells were injected into the lateral tail vein of each recipient mouse.
EGFP+ transduced splenic B cells were sorted and analyzed by Western blotting as previously described (Jeelall et al., 2012). Myd88 and phospho-p65 NF-κB primary antibodies obtained from Cell Signaling Technology were used at 1:1,000 dilutions. Horseradish peroxidase–conjugated anti–rabbit IgG secondary antibody purchased from Cell Signaling Technology was used at 1:2,500 dilution. Membranes were reprobed with antibody to αβ tubulin obtained from Cell Signaling Technology at 1:5,000 dilution as a loading control.
Gene expression analysis.
EGFP+ B cells were harvested on day 1 of culture without anti-CD40, sorted, and resuspended in TRIzol. Phase separation was performed by the addition of chloroform and centrifugation at 12,000 g for 15 min, followed by isopropanol RNA precipitation. The air-dried RNA pellet was dissolved in RNase-free water, and mRNA expression was analyzed on Affymetrix mouse ST 1.0 arrays as per the manufacturer’s instructions.
All experiments were analyzed using Prism version 5 (GraphPad). Statistically significant p-values <0.05, <0.01, and <0.001, respectively are indicated and were determined using a two-tailed unpaired Student’s t test.
We thank S. Cory and A. Strasser for mouse strains, and the Australian Phenomics Facility for expert care and genotyping of animals.
J.Q. Wang was supported by an Australian Postgraduate Award. C. Goodnow was supported by NHMRC Australia Fellowship 585490. The work was supported by NHMRC Program Grants 1016953 and 427620 and NIAID National Institutes of Health grants U19 AI100627 and U54 AI054523.
The authors declare no competing financial interest.
Author contributions: J.Q. Wang, K. Horikawa, and C. Goodnow designed and analyzed the experiments and wrote the paper. J.Q. Wang performed most of the experiments. Y.S. Jeelall assisted with the experiments. B. Beutler provided the Unc93b3d mice and feedback on data analysis.
activated B cell type diffuse large B cell lymphoma
B cell receptor
chronic lymphocytic leukemia
cell trace violet
germinal center B cell type diffuse large B cell lymphoma
interleukin receptor activated kinase
myeloid differentiation primary response 88
tumour necrotic factor alpha-induced protein 3
K. Horikawa and C.C. Goodnow contributed equally to this paper.