Activation of lymphoma-associated MyD88 mutations via allostery-induced TIR-domain oligomerization

Abstract

Myeloid differentiation 88 (MyD88) is the key signaling adapter of Toll-like and interleukin-1 receptors. Recurrent lymphoma-associated mutations, particularly Leu265Pro (L265P), within the MyD88 Toll/interleukin-1 receptor (TIR) domain sustain lymphoma cell survival due to constitutive nuclear factor κB signaling. We found that mutated TIR domains displayed an intrinsic propensity for augmented oligomerization and spontaneous formation of cytosolic Myddosome aggregates in lymphoma cell lines, mimicking the effect of dimerized TIR domains. Blocking of MyD88 oligomerization induced apoptosis. The L265P TIR domain can recruit the endogenous wild-type MyD88 for oligomer formation and hyperactivity. Molecular dynamics simulations and analysis of additional mutations suggest that constitutive activity is caused by allosteric oligomerization.

Figure 1

Lymphoma-associated MyD88 mutants and their isolated TIR domains constitutively hyperactivate NF-κB signaling. (A) Structure of a human MyD88 TIR monomer (Protein Data Bank accession number 2JS7) with lymphoma-associated mutations highlighted in red. The α helices are shown in beige and β sheets in light pink. The BB loop is shown in cyan. (B-F) HEK293 cells were transfected with plasmids for full-length (B-C) or TIR-domain–only (D-F) constructs of lymphoma-associated MyD88 mutants, in parallel with NF-κB–inducible firefly luciferase and constitutive Renilla luciferase reporters, and TLR4 and MD-2 plasmids (F only). Cells were harvested 24 hours later (B-E) or stimulated with LPS (100 ng/mL) for another 24 hours (F). Luciferase activity was measured by DLA (B,D,F) or separately transfected cell lysates loaded on SDS polyacrylamide gel electrophoresis (PAGE) for immunoblot (C,E). For panels B, D, and F, data presented are means + standard deviation (SD) of triplicate samples, and for panels B to F, data shown are representative of at least 3 independent experiments. EV, empty vector; IB, immunoblot.

Figure 2

Augmented heterodimerization of lymphoma-associated mutants and WT MyD88 TIR domain. (A-D) MyD88-deficient HEK293-I3A cells were transfected with plasmids for full-length (A-B) or TIR-domain–only (C-D) constructs of lymphoma-associated MyD88 mutants, in parallel with NF-κB–inducible firefly luciferase and a constitutive Renilla luciferase reporters. Cells were harvested 48 hours later; 1 lysate aliquot was used for luciferase activity measurement by DLA (A,C) and another for immunoblot upon 15% SDS-PAGE separation (B,D). Data presented are means + SD of triplicate samples, and data shown are representative of at least 3 independent experiments. (E-F) The L265P mutation leads to increased TIR-TIR oligomerization. LUMIER luciferase analysis from HEK293T cells transfected with Protein A–tagged and Renilla luciferase–tagged WT or L265P mutant full-length (E) or TIR-domain–only (amino acid sequence 155-294) (F) MyD88 constructs. Forty-eight hours posttransfection, cells were lysed and raw luciferase measured in 10% of the lysate sample. The remainder was used for Protein A purification on immunoglobulin G magnetic beads and subsequent measurement of bound luciferase. Data represent ratios of bound vs raw luciferase for each transfection upon subtraction of background (Protein A–only control bait) combined from 7 identical experiments. Means ± SDs are shown and differences tested using a Mann-Whitney U t test. EV, empty vector; IB, immunoblot.

Figure 3

Aggregation of constitutively active lymphoma-associated TIR mutants in lymphoma cell lines. (A-B) HEK293 cells were transfected with plasmids for mCitrine-TIR (yellow pseudocolor) or mutants and full-length MyD88-mCerulean (blue pseudocolor, B). Twenty-four hours later, cells were fixed and visualized under the confocal microscope at ×1000 magnification. Shown are representative confocal images of at least 3 independent experiments. (C-G) Augmented amount of MyD88 in aggregates in different DLBCL cell lines. Cell lysates were fractionated by centrifugation and lysate and pellet fraction analyzed by SDS-PAGE followed by anti-MyD88 (C-E) or IRAK1 (G) and anti-tubulin immunoblot (see blots in panels E and G). Relative MyD88 or IRAK1 and tubulin levels were quantified by charge-coupled device detection and plotted as the tubulin-normalized ratios of lysate (WCL) vs pellet MyD88 or IRAK1. (C-D) Data were combined from 9 experiments and are shown as min-to-max box-and-whisker plots. In panel C, WT MyD88 DLBCL cell lines (white, OCI-LY19, BJAB, U2392) were compared with MyD88-mutated cell lines (burgundy, OCI-LY10, TMD8, HBL1, OCI-LY3 with L265P; HLY1 with S119C, SUDHL2 with S122R). In panel D, cells were plotted according to L265P mutational status. White, WT MyD88 (OCI-LY19, BJAB, U2392); red, L265P heterozygous (OCI-LY10, TMD8, HBL1); burgundy, L265P homozygous (OCI-LY3). (E) Comparison of unmutated OCI-LY19 (white) vs L265P homozygous OCI-LY3 (red) lysate vs pellet ratio from 9 combined experiments and showing a representative immunoblot. (F) Differences in MyD88 lysate expression in L265P-mutated OCI-LY3 cells are not due to lower mRNA expression as assessed by quantitative PCR in OCI-LY19 (white) and OCI-LY3 (burgundy) cells. One representative of 3 independent experiments was done in triplicate (mean + upper limit). (G) MyD88-containing aggregates also contain IRAK1. Comparison of unmutated OCI-LY19 (white) vs L265P homozygous OCI-LY3 (red) lysate vs pellet ratio from 3 combined experiments and showing a representative immunoblot. IB, immunoblot; WCL, whole-cell lysate.

Figure 4

Lymphoma-associated TIR-domain mutants are located at stable communication hubs in the TIR domain. (A-B) HEK293 cells were transfected with plasmids for designed MyD88 mutants and NF-κB–inducible firefly luciferase and constitutive Renilla luciferase reporter plasmids. Twenty-four hours later, cells were lysed and luciferase activity measured (A) or the lysates were run on SDS-PAGE and analyzed by anti-MyD88 immunoblot (B). (C-D) The MyD88 TIR domain contains highly stable RMSF minima. The 40-nanosecond (C) or 100-nanosecond (D) molecular dynamics simulations were done on WT (black dotted line) and mutant (colored as shown) MyD88 NMR structure 2js7 ensemble (composed of 20 structures each, 20 simulations each); shown are averaged data (C) or the single most representative conformer (D). RMSF over the TIR-domain residues were plotted. RMSF profiles are highly similar and show RMSF minima (black arrows pointing to residues 177, 204, 231/232, 265, and 303/304). (E-F) RMSF minima are shown as green or orange (in case they coincide with lymphoma-associated mutations) spheres and map to a plane describing MyD88 dimer formation (see schematic representation, panel F) as proposed by Bovijn et al. Communication hubs are indicated by arrows and coincide with RMSF minima or locate in the interaction plane within the MyD88 dimer. Some lymphoma-associated mutations directly map to hub positions (orange spheres) or locate within this plane (red spheres), with the exception of S143 and T294 (red). Hub positions for which so far no lymphoma-associated mutations have been reported are shown as green spheres. The region of higher flexibility in WT (black dotted line) vs mutated (colored as shown) TIR domains in molecular dynamics simulation (F) is shown in dark blue in panel E. EV, empty vector; IB, immunoblot.