Translation of cytoplasmic UBA1 contributes to VEXAS syndrome pathogenesis

Abstract

Somatic mutations in UBA1 cause vacuoles, E1 ubiquitin-activating enzyme, X-linked, autoinflammatory somatic (VEXAS) syndrome, an adult-onset inflammatory disease with an overlap of hematologic manifestations. VEXAS syndrome is characterized by a high mortality rate and significant clinical heterogeneity. We sought to determine independent predictors of survival in VEXAS and to understand the mechanistic basis for these factors. We analyzed 83 patients with somatic pathogenic variants in UBA1 at p.Met41 (p.Met41Leu/Thr/Val), the start codon for translation of the cytoplasmic isoform of UBA1 (UBA1b). Patients with the p.Met41Val genotype were most likely to have an undifferentiated inflammatory syndrome. Multivariate analysis showed ear chondritis was associated with increased survival, whereas transfusion dependence and the p.Met41Val variant were independently associated with decreased survival. Using in vitro models and patient-derived cells, we demonstrate that p.Met41Val variant supports less UBA1b translation than either p.Met41Leu or p.Met41Thr, providing a molecular rationale for decreased survival. In addition, we show that these 3 canonical VEXAS variants produce more UBA1b than any of the 6 other possible single-nucleotide variants within this codon. Finally, we report a patient, clinically diagnosed with VEXAS syndrome, with 2 novel mutations in UBA1 occurring in cis on the same allele. One mutation (c.121 A>T; p.Met41Leu) caused severely reduced translation of UBA1b in a reporter assay, but coexpression with the second mutation (c.119 G>C; p.Gly40Ala) rescued UBA1b levels to those of canonical mutations. We conclude that regulation of residual UBA1b translation is fundamental to the pathogenesis of VEXAS syndrome and contributes to disease prognosis.

Graphical abstract

Figure 1.

Survival analysis in VEXAS syndrome. (A) Frequencies of common clinical diagnoses assigned to patients with VEXAS syndrome are shown stratified by genotype. (B) Kaplan-Meier curves demonstrate survival in the entire cohort and (C) are stratified by genotype. (D) Forest plot showing the results of a multivariable Cox proportional hazards model where the Val variant (c.121 A>G, p.Met41Val), transfusion dependence (time-varying predictor), and ear chondritis were significantly associated with mortality in VEXAS syndrome. *, P < .05; #, P < .01 for survival difference between all groups.

Figure 2.

p.Met41Val produces less UBA1b as compared with other patient variants. (A) p.Met41Val supports less UBA1b translation than p.Met41Thr and p.Met41Leu. HEK293T cells were transfected with C-terminally FLAG-tagged WT UBA1 and indicated VEXAS variants, lysed, subjected to anti-FLAG immunoprecipitation and analyzed by anti-UBA1a/b/c or anti-UBA1a/b immunoblotting. Shorter and longer exposures of the respective immunoblots are shown. (B) Quantification of the experiment shown in panel A. Error bars denote standard error of the mean, n = 3 technical replicates, *P < .05; **P < .01, ordinary 1-way ANOVA. (C) p.Met41Val carriers express less UBA1b than p.Met41Thr and p.Met41Leu carriers. CD3-depleted peripheral blood mononuclear cells (PBMCs) of healthy volunteers (C1, C2) and VEXAS patients with similar VAFs were subjected to anti-UBA1a/b/c or anti-UBA1a/b immunoblotting. (D) Quantification of anti-UBA1a/b immunoblots of CD3-depleted PBMCs or CD14⁺ cells of healthy volunteers (control) or VEXAS patients (p.Met41Val/Thr/Leu). Error bars denote standard deviation, n = 8 (4 biological replicates of independently processed patient samples per genotype with 2 technical replicates each; technical replicates of each biological replicate are depicted in the same color and symbol), *P < .05, ordinary 1-way ANOVA.

Figure 3.

Residual UBA1b translation defines a threshold for VEXAS syndrome pathogenesis. (A) Schematic overview of the translation reporter system used to determine UBA1b translation. The first cistron of these reporters is translated in a CAP-dependent mechanism and encodes for C-terminally HA-tagged UBA1 constructs varied in p.Met41 and mutated at p.Met1 and p.Met67 (to suppress UBA1a and UBA1c translation and thus allowing for sensitive quantification of UBA1b by anti-HA immunoblotting). The second cistron is translated using an IRES and encodes for a C-terminally HA-tagged MBP, which is used for normalization across different p.Met41 variants. (B) p.Met41Val supports less UBA1b translation than p.Met41Thr and p.Met41Leu. Anti-HA immunoblot analysis of HEK293T lysates expressing indicated UBA1b translation reporters. (C) Quantification of the experiment shown in panel B. Error bars denote SD, n = 12 (4 biological with 3 technical replicates each; technical replicates of each biological replicate are depicted in the same color and symbol), **P < .01; ****P < .0001, ordinary 1-way ANOVA. (D) The canonical VEXAS variants (p.Met41Val/Thr/Leu) support more UBA1b translation than any of the other possible single-nucleotide variants at the Met41 start codon. Anti-HA immunoblot analysis of HEK293T lysates expressing indicated UBA1b translation reporters. (E) Quantification of the experiment shown in panel D. Error bars denote SD, n = 4 (2 biological with 2 technical replicates each; technical replicates of each biological replicate are depicted in the same color and symbol), *P < .05; ****P < .0001, ordinary 1-way ANOVA. (F) Two novel VEXAS variants (c.121 A>T; p.Met41Leu and c.119 G>C; p.Gly40Ala) are present in a patient in cis on the same allele. Overview of the genome sequencing reads of the p.Met41Leu^TTG/p.Gly40Ala patient. (G) The p.Gly40Ala variant rescues residual UBA1b translation of the p.Met41Leu^TTG to the level of the canonical VEXAS variant p.Met41Val. Anti-HA immunoblot analysis of HEK293T lysates expressing indicated UBA1b translation reporters. (H) Quantification of the experiment shown in panel G. Error bars denote SD, n = 10 (2 biological with 5 technical replicates each; technical replicates of each biological replicate are depicted in the same color and symbol), **P < .01; ****P < .0001, ordinary 1-way ANOVA. Shaded area represents minimum threshold below p.Met41Val expression. SD, standard deviation.