Whole-exome sequencing in relapsing chronic lymphocytic leukemia: clinical impact of recurrent RPS15 mutations

Abstract

Fludarabine, cyclophosphamide, and rituximab (FCR) is first-line treatment of medically fit chronic lymphocytic leukemia (CLL) patients; however, despite good response rates, many patients eventually relapse. Although recent high-throughput studies have identified novel recurrent genetic lesions in adverse prognostic CLL, the mechanisms leading to relapse after FCR therapy are not completely understood. To gain insight into this issue, we performed whole-exome sequencing of sequential samples from 41 CLL patients who were uniformly treated with FCR but relapsed after a median of 2 years. In addition to mutations with known adverse-prognostic impact (TP53, NOTCH1, ATM, SF3B1, NFKBIE, and BIRC3), a large proportion of cases (19.5%) harbored mutations in RPS15, a gene encoding a component of the 40S ribosomal subunit. Extended screening, totaling 1119 patients, supported a role for RPS15 mutations in aggressive CLL, with one-third of RPS15-mutant cases also carrying TP53 aberrations. In most cases, selection of dominant, relapse-specific subclones was observed over time. However, RPS15 mutations were clonal before treatment and remained stable at relapse. Notably, all RPS15 mutations represented somatic missense variants and resided within a 7 amino-acid, evolutionarily conserved region. We confirmed the recently postulated direct interaction between RPS15 and MDM2/MDMX and transient expression of mutant RPS15 revealed defective regulation of endogenous p53 compared with wild-type RPS15. In summary, we provide novel insights into the heterogeneous genetic landscape of CLL relapsing after FCR treatment and highlight a novel mechanism underlying clinical aggressiveness involving a mutated ribosomal protein, potentially representing an early genetic lesion in CLL pathobiology.

Figure 1

Somatic mutation frequencies in CLL relapsing after FCR treatment. (A) Average number of nonsynonymous mutations in both the pretreatment and relapse samples and the number of shared mutations in the 28 samples with matched constitutional DNA. (B) Frequency of the 6 mutation classes for pretreatment and relapse-specific mutations in the 28 samples with matched constitutional DNA. Statistical significance was assessed using 796 pretreatment mutations and 425 relapse-specific mutations. (C) Recurrently mutated genes. Columns represent patients (n = 41) and rows genes or genetic lesions. Color-coding indicates the type of mutation or genomic alteration. Case names in gray were analyzed without matched normal DNA. The majority of cases with TP53 aberrations harbored a mutation without coexisting del(17p); this is explained by the fact that the TP53 mutation status was not known in most cases before the start of the FCR regime in contrast to fluorescence in situ hybridization detection of del(17p), which had been performed in all cases. All TP53 mutations were deemed damaging and have been reported previously. CNA, copy-number aberration; PR, partial relapse; UPD, uniparental disomy.

Figure 2

RPS15 mutations in CLL. (A) Localization of SNVs detected by either WES or with targeted resequencing of exon 4 in the extension cohorts (data for each cohort are provided in supplemental Table 10). With the exception of a single mutation in amino acid 33 (not shown in figure), all mutations clustered to the C-terminal of the RPS15 protein. Exons are marked with dashed lines and the region covered in the targeted resequencing is color coded. (B) Frequency of RPS15 mutations in the FCR relapse cohort (n = 41), the extended screening cohort (n = 605), the CLL4 trial cohort (n = 329), in RS cases (n = 30) and IGHV-mutated/stage A patients (n = 185). (C-D) Concurrent mutations and genetic lesions in RPS15-mutated cases in the extended screening cohort and the UK CLL4 cohort. (E-F) Overall survival for subgroups carrying recurrent cytogenetic and molecular aberrations. Pairwise log-rank test: RPS15^mut vs del(13q), P < .001; RPS15^mut vs TP53^abn, P = .42; RPS15^mut/TP53^abn vs TP53^abn, P = .15; RPS15^mut vs RPS15^mut/TP53^abn, P = .12. Ten-year survival rate for RPS15^mut/TP53^wt was similar to RPS15^mut/TP53^abn and RPS15^wt/TP53^abn patients (0%, 0%, and 22%, respectively), but lower than the remaining RPS15^wt/TP53^wt patients (59%, see supplemental Figure 7). NA, not available.

Figure 3

Temporal dynamics of somatic mutations. SciClone analysis of variant allele frequencies for case S42 revealed 6 predicted subclones (A). Although cluster 1 (harboring a mutation in RPS15), clusters 2 and 3 (carrying a frame-shift deletion in MGA, a gene previously found disrupted in high-risk CLL)^, remained stable over time; clusters 4 and 6 represent relapse-specific (sub)clones, with the latter harboring a TP53 mutation that affects 25% of tumor cells, whereas cluster 5 was eliminated. In case S17, a single stable population was detected; however, a drastic shift in (sub)clonal populations was also observed (B); clusters 4 and 6 were successfully eliminated by treatment. This is noteworthy because the former harbored the classical 2-bp deletion of NOTCH1 and the recurrent p.K700E SNV in SF3B1. On the other hand, clusters 2, 3, and 5 represent relapse-specific subclones with cluster 2, which was the dominant clone at relapse, harboring a stop-gain mutation in NOTCH1 and an SNV in EGR2. Results for the remaining samples are provided in supplemental Figure 4. (A-B) The mean variant allele frequency of each cluster at both time points (top right) and the fraction of each cluster in comparison with the major cluster (cluster 1; bottom right). (C-E) Variant allele frequencies for FCR relapsing cases with RPS15, EGR2, and NOTCH1 mutations before treatment and at relapse. *Denote 2 unique mutations in the same case (S17).

Figure 4

In vitro characterization of RPS15 mutations. (A-B) RPS15^G132A and RPS15^P131S interact with MDM2 and MDMX. Immunoblot analysis of whole-cell lysates (input) and immunoprecipitates (IP) of HCT116 cells cotransfected with 1 of the Myc-DDK-RPS15 vectors (wt, G132A or P131S) and either tGFP-MDM2 (A) or tGFP-MDMX (B). Twenty-four hours posttransfection 500 μg of cell lysates were subjected to immunoprecipitation using anti-tGFP antibody followed by immunoblotting with indicated antibodies. Immunoglobulin G heavy chain was used as loading control for IP fraction. (C-D) Transient expression of RPS15^wt, RPS15^P131S, and RPS15^G132A in HCT116 cells revealed an impaired ability to stabilize endogenous p53, in particular for the RPS15^G132A mutant. Western blot images from 1 representative experiment and quantification results from 3 independent experiments are shown. Images from 2 additional experiments are presented in supplemental Figure 7. (E) Ubiquitination experiments revealed increased ubiquitin-mediated p53 degradation in both RPS15 mutants (40.1% and 40.6% increase for RPS15^P131S and RPS15^G132A, respectively, measuring the intensity for the upper 4 bands) compared with wild-type RPS15.