Novel somatic mutations in UBA1 as a cause of VEXAS syndrome

TO THE EDITOR:

Systemic autoinflammatory disorders encompass a heterogeneous group of monogenic disorders that are characterized by recurrent episodes of systemic and organ-specific inflammation.¹ 

Using a genotype-first approach, Beck et al recently described VEXAS (vacuoles, E1 enzyme, X-linked, autoinflammatory, somatic) syndrome, a new late-onset treatment-refractory inflammatory syndrome with associated hematological abnormalities.²  VEXAS is caused by acquired somatic mutations at methionine 41 (p.Met41) of UBA1, the major E1 enzyme responsible for initiating ubiquitylation. The mutations were predominantly found in myeloid lineages and were absent in lymphoid lineages. Functional analysis identified loss of the cytoplasmic isoform UBA1b, initiated from p.Met41, and the subsequent gain of a new isoform, UBA1c, as the underlying disease mechanisms. No additional variants, other than those at p.Met41, were identified, suggesting that mutation of this residue alone causes VEXAS.

Here, we report 10 additional cases of VEXAS and describe 2 new mutations as a cause of this disease.

We investigated 18 patients for a severe systemic inflammatory disorder, cytopenias, and with bone marrow dysplastic features; all were men older than 40 years of age. Four of these patients, who were already deceased, were identified from a database based on their phenotype. Sanger sequencing was undertaken, resulting in the identification of pathogenic or likely pathogenic variants in 10 of 18 patients (P1-P10). Of the 10 patients with a mutation, 8 (P1-P8) had mutations at p.Met41, and 2 had novel variants. To determine whether samples negative for mutations following Sanger sequencing harbor somatic mutations at a low allele frequency, we performed deep amplicon sequencing of exon 3 to an average depth of 2900 reads per sample. No variant was identified in any of the sequenced samples. Clinical features of all patients are shown in Table 1.

The first novel variant identified was a somatic variant, NM_153280:c.167C>T, p.Ser56Phe in P9, with an approximately equal ratio of reference and variant alleles in peripheral blood. This variant is not present in the Genome Aggregation Database (gnomAD)³  and is predicted to be deleterious by all tested bioinformatics predictions (supplemental Table 1, available on the Blood Web site). To determine whether the c.167C>T variant was preferentially expressed in myeloid cell lineages, we used magnetic-activated cell sorting to sort peripheral blood into myeloid (CD14⁺/CD15⁺) and lymphoid (CD3⁺/CD19⁺) lineages. Sanger sequencing of the variant in extracted genomic DNA revealed that the myeloid lineage populations had predominantly mutant alleles, whereas B- and T-cell lymphoid lineages were predominantly wild-type (Figure 1A). Similar to canonical VEXAS mutations, the p.Ser56Phe variant did not affect the cellular localization of UBA1 (supplemental Figure 1). Surprisingly, however, it did not result in the cytoplasmic isoform swap from UBA1b to UBA1c observed in p.Met41 variants (Figure 1B). Rather, this variant resulted in a temperature-dependent impairment in UBA1 catalytic activity compared with wild-type enzyme (Figure 1C-D). The second novel variant identified was at the splice acceptor site (NM_153280:c.118-1G>C) of exon 3 (Figure 1E) in P10. This variant is not present in gnomAD and is predicted to alter splicing by SpliceAI (supplemental Table 1). Analysis of patient-derived RNA revealed a reduction in properly spliced transcript and the creation of multiple incorrectly spliced products (Figure 1F).

The remaining 8 patients (P1-8) had a somatic variant that substituted the Met41 residue. Five of the patients had c.122T>C; p.Met41Thr substitution, whereas the remaining 3 patients had the c.121A>G; p.Met41Val substitution (Table 1). Bone marrow histopathology was consistent with previously reported patients, with vacuolated promyelocytes, increased cellularity and granulopoiesis, and decreased erythropoiesis (supplemental Figure 2). P9 (p.Ser56Phe) also showed increased cellularity; however, in contrast with the other VEXAS cases, he had increased erythropoiesis and reduced granulopoiesis (Figure 1F-G). Eight patients who tested negative for the pathogenic UBA1 variant had many clinical features similar to VEXAS patients. All presented with fever, a high degree of skin involvement, and chondritis. Although all had anemia, the macrocytic type was only seen in 50% of the UBA1⁻ individuals (Table 1).

Because VEXAS mutations are acquired and the disease is progressive, we hypothesized that its progressive nature could be due to the mutant clone expanding over time. Therefore, we identified a case from the original study²  and 1 from this study (P6) with the p.Met41Thr substitution for whom bone marrow biopsies spanning 5 years were available. Sequencing of genomic DNA extracted from each biopsy did not reveal any difference in the proportion of mutant/wild-type alleles, with the predominant allele in each biopsy being the mutant T allele (supplemental Figure 3).

By sequencing a cohort of patients suspected of having VEXAS, we identified 10 patients with UBA1 mutations. Although 8 of these substituted p.Met41, the remaining 2 patients harbored novel mutations that provide new insights into the disease mechanism of VEXAS. First, the c.118-1G>C variant altered splicing in vitro. We hypothesize that this mutation results in aberrant UBA1 messenger RNAs that lack regions around p.Met41 required for translation of the cytoplasmic UBA1b isoform. Second, analysis of the p.Ser56Phe variant led to the identification of a distinct disease mechanism whereby a temperature-dependent impairment in catalytic activity of UBA1, and not loss of the UBA1b isoform, results in VEXAS. Although we do not see a differential impact on UBA1 isoforms at the chosen conditions of our in vitro assays, we hypothesize that the p.Ser56Phe variant might lead to a preferential inactivation of cytoplasmic UBA1b isoform in cells (eg, through binding of activity-stabilizing proteins to the N terminus of nuclear UBA1a that is not present in cytoplasmic UBA1b); alternatively, cells may be more sensitive to reductions in activity to UBA1b. This novel disease mechanism may account for the difference in hematological pathology observed for P9 (ie, increased erythropoiesis and reduced granulopoiesis) compared with the rest of the cohort. Thus, our study suggests that a phenotypic spectrum may exist for VEXAS that is governed by the underlying mutation, the disease mechanism, and proportion of mutation-containing cells.

Previous analyses of sorted blood from patients harboring p.Met41 variants showed that the variant is predominant in myeloid cell lineages but not in lymphoid cell lineages.²  We observed the same restriction of mutations to myeloid lineages for the p.Ser56Phe patient, suggesting that this restriction is not specific to Met41 mutations. This may imply that loss-of-function mutations in cytoplasmic UBA1 invokes a survival advantage in myeloid lineages that does not exist in lymphoid lineages or, simply, that myeloid cells can survive the loss of cytoplasmic UBA1, which other cell types cannot. Interestingly we did not find any difference in the proportion of mutant/wild-type alleles in bone marrow over a 5-year period during which the disease progressed. Further studies will be required to better understand the clonal nature of this disease and how this associates with disease severity; however, the predominance of the mutant allele from the first biopsy onward suggests that, even at the initial presentation of symptoms, the mutation is already the prevalent allele.

Several studies have linked myelodysplastic syndrome (MDS) with an increased incidence of autoimmune and autoinflammatory complications.⁴  Previous studies have identified particular autoimmune manifestations associated with distinct karyotypes in MDS.⁵  A more recent study found that somatic mutation of transcription factor pathways and abnormal karyotypes are associated with autoinflammatory complications.⁶  The identification of somatic UBA1 mutations in VEXAS, in particular, helps to explain why some MDS patients also have autoinflammatory complications.⁷  Future studies of VEXAS-like UBA1-negative cases using the cell lineage–specific whole-exome sequencing approach might lead to the discovery of additional somatic mutations that link inflammatory and hematological abnormalities. This will also result in a better understanding of the underlying genetic causes of MDS and facilitate stratification and development of targeted treatments. Although VEXAS patients do not appear to have additional somatic mutations that are typically associated with high-risk MDS, the condition is associated with poor outcome and is refractory to standard therapies. A more radical treatment approach, such as bone marrow transplant, might need to be considered early in the disease course.