Nonstop mutations cause loss of renal tumor suppressor proteins VHL and BAP1 and affect multiple stages of protein translation

Abstract

Nonstop extension or stop-loss mutations lead to the extension of a protein at its carboxyl terminus. Recently, nonstop mutations in the tumor suppressor SMAD Family Member 4 (SMAD4) have been discovered to lead to proteasomal SMAD4 degradation. However, this mutation type has not been studied in other cancer genes. Here, we explore somatic nonstop mutations in the tumor suppressor genes BRCA1 Associated Protein 1 (BAP1) and Von Hippel-Lindau (VHL) enriched in renal cell carcinoma. For BAP1, nonstop mutations generate an extremely long extension. Instead of proteasomal degradation, the extension decreases translation and depletes BAP1 messenger RNA from heavy polysomes. For VHL, the short extension leads to proteasomal degradation. Unexpectedly, the mutation alters the selection of the translational start site shifting VHL isoforms. We identify germline VHL nonstop mutations in patients leading to the early onset of severe disease manifestations. In summary, nonstop extension mutations inhibit the expression of renal tumor suppressor genes with pleiotropic effects on translation and protein stability.

Fig. 1.. Recurrent nonstop mutations in tumor suppressor genes BAP1 and VHL in kidney cancer.

(A) Schematic view of nonstop (NSExt) or stop-loss mutations. Created in BioRender. Pal (2024); https://BioRender.com/z76o007. (B) Overlap of genes harboring nonstop mutations (COSMIC v98) with genes frequently mutated across all tumor entities (PANcancer) from The Cancer Genome Atlas (TCGA). (C) Number of nonstop mutations (blue) versus frequency of mutation in TCGA PANcancer data (orange) for the genes selected from (B). The blue line represents the threshold of five nonstop mutations and the orange line represents the threshold of 2% mutation frequency in TCGA PANcancer data. (D) Frequency distribution according to the primary site of the tumor for all point mutations in all Cancer Gene Census coding sequences in COSMIC v98, in VHL or in BAP1 in comparison to the entities with nonstop extension mutations in these two genes. (E) Representation of the four different nonstop mutations in the VHL gene occurring eight times in COSMIC. (F) Representation of the four different nonstop mutations in the BAP1 gene occurring five times in COSMIC. Created in BioRender. Pal (2024); https://BioRender.com/o34m945 (E and F).

Fig. 2.. Nonstop mutations in BAP1 lead to loss of the protein via reduced translation.

(A) BAP1 mRNA levels (N = 3) in HEK293 cells with endogenously either wild-type (WT, gray) or nonstop mutant (MUT, blue) BAP1 created by precision genome editing. G, L, R, and S refer to the change from stop codon to Gly, Leu, Arg, and Ser, respectively. WT clone refers to HEK293 clones generated from CRISPR-Cas experiments. The numbers identify the different homozygous clones for each genotype to exclude clone-specific differences. (B) BAP1 protein levels were reduced in MUT versus WT clones, while the ubiquitination of its downstream target histone H2A at K119 was increased (N = 3). (C) Treatment of WT and NSExt (MUT G1) HEK293 cells showed no rescue of the mutant protein with neither ubiquitin-proteasomal [MG132 and bortezomib (Bort.)] nor endo-lysosomal [chloroquine (Chlor.) and bafilomycin A1 (BafA1)] inhibitors (N = 3). MCL1 served as positive control for proteasomal inhibition. Untr. refers to untreated cells, i.e., 0.8% DMSO, which is the highest DMSO concentration used for the drugs. (D) Schematic workflow of polysome fractionation and RT-qPCR for BAP1. Created in BioRender. Pal (2024); https://BioRender.com/a30l718. (E) RT-qPCR quantification of the mRNA expression of BAP1 and PPIA as control mRNA in WT and mutant cells after polysome fractionation. The fold changes were calculated relative to the free RNA fractions. Here, error bars for individual values are not shown for ease of visualization (N = 3). (F) Comparison of the RNA abundance in the heavy polysome fractions for BAP1 and PPIA mRNAs normalized to their abundance in free RNA. For (A), (B), and (F), an F test followed by an unpaired t test was performed where ns and *** refer to P values of >0.05 and <0.001, respectively.

Fig. 3.. Nonstop mutations in VHL lead to loss of the protein via the ubiquitin-proteasome pathway.

(A) VHL mRNA levels in HEK293 cells with wild-type (WT, gray) or nonstop mutant (MUT, blue) VHL created by precision genome editing (N = 3). WT par refers to parental HEK293 cells, while WT clone refers to a WT HEK293 clone obtained during the CRISPR-Cas–mediated mutagenesis of the VHL locus. C, W, and L refer to the stop codon change to Cys, Trp, and Leu, respectively. The numbers identify the different homozygous clones for each genotype to exclude clone-specific differences. (B and C) Reduced VHL and increased downstream target HIF-1α protein levels in MUT versus WT cells [quantification in (C), N = 3]. (D) Mutant cells (NSExt and MUT C2) displayed an increase in mRNA levels of HIF-1α downstream targets VEGFA, VEGFC, and FLT-1 (N = 3). (E) Mutant cell lines (NSExt) showed increased migration capacity in an IncuCyte Wound Healing assay (N = 4). h, hours. (F and G) Treatment of WT versus NSExt (MUT C2) HEK293 cells with ubiquitin-proteasomal pathway inhibitors [MG132 and bortezomib (Bort.)] led to a rescue of the mutant VHL protein, but not with endo-lysosomal pathway inhibitors [chloroquine (Chlor.) and bafilomycin A1 (BafA1)] [quantification in (G), N = 3] with MCL1 serving as a positive control for proteasomal inhibition. Untr. refers to untreated cells, i.e., 0.8% DMSO, which is the highest DMSO concentration used for the drugs. (H to J) Cycloheximide chase experiments unraveled reduced half-life of mutant VHL (NSExt and MUT C2). Both WT and NSExt cells were treated with 0.1 μM bortezomib for 8 hours followed by cycloheximide treatment. After adding the cycloheximide, cells were collected at 0, 6, 12, 24, and 36 hours (N = 3). For (A), (C), (D), (E), (G), and (J), an F test followed by an unpaired t test was performed. Here, ns, *, **, and *** refer to P values of >0.05, <0.05, <0.01, and <0.001, respectively.

Fig. 4.. Nonstop mutations in VHL alter translational start codon selection.

(A and B) Bortezomib (Bort.) treatments of mutant cells resulted in an increased expression specifically of the larger isoform of VHL in mutants (MUT) compared to wild types (WT) [quantification in (B), N = 3]. VHL²¹³ represents the 213–amino acid–long VHL isoform translated from the first translational start codon and VHL¹⁶⁰ refers to the 160–amino acid–long VHL isoform translated from the second translational start codon. (C) Schematic workflow of RNA affinity purification followed by mass spectrometry (RAP-MS). Cells were UV–cross-linked to fix the RNA-bound proteins directly on the RNAs. VHL mRNAs were pulled down using an raPOOL of 30 different biotinylated probes binding to VHL compared to a non-targeting control raPOOL. VHL-bound proteins were identified using mass spectrometry. Created in BioRender. Pal (2024); https://BioRender.com/o16w486. (D) Schematic workflow of the data analysis steps to identify VHL mRNA-bound proteins from the RAP-MS differentially binding to VHL WT versus mutant mRNAs (NSExt and MUT C2). (E) Heatmap of the differentially bound proteins to VHL WT and NSExt mRNA from RAP-MS including their propensity to bind to RNA as indicated by the RBP2GO score (N = 3). (F and G) siPOOL-mediated knockdown of ribosomal proteins RPSA and RPS3; translation initiation factors EIF3D, EIF4A1, and EIF4G2; and the poly-A binding protein PABPC1 in VHL NSExt (MUT C2) cells altered the VHL protein isoform ratio with an increase in VHL²¹³ abundance [quantification in (G) N = 3]. siN.C. refers to the non-targeting siPOOL control. For (B) and (G), F tests followed by unpaired t tests were performed. Here, ns, *, and *** refer to P values of >0.05, <0.05, and <0.001, respectively.

Fig. 5.. Clinical implication of germline VHL nonstop mutations.

(A) PCR-mediated sequencing of blood samples from an individual carrying the heterozygous germline nonstop variant VHL c.640T>G (p.Ter214Gly). (B) A Kaplan-Meier estimator assessing the age-related VHL disease penetrance (first diagnosis of retinal angioma, CNS hemangioblastoma, clear cell renal cell carcinoma, pancreatic neuroendocrine tumor, or pheochromocytoma) in participants of the Freiburg VHL Registry carrying the VHL nonstop variant c.640T>G (p.Ter214Gly) in comparison to the disease penetrance of carriers of the pathogenic variant c.499C>T (p.Arg167Trp), which revealed no significant difference (log-rank test). (C and D) Magnetic resonance images displaying clear cell renal cell carcinoma (C, right panel: 5× magnification) and pheochromocytoma (D, right panel: 5× magnification) in carriers of VHL nonstop variants. Scale bars represent 20 mm. (E and F) VHL *>G and *>R nonstop mutations also led to protein loss and HIF-1α stabilization [quantification in (F), N = 3]. *>G and *>R refer to the stop codon change to Gly and Arg, respectively. The numbers identify the different homozygous clones for each genotype to exclude clone-specific differences. (G and H) The mutant protein expression was rescued by bortezomib (BORT.) treatment. Also, the VHL²¹³ isoform was significantly more stabilized for the mutant VHL compared to WT [quantification in (H), N = 3]. For (F) and (H), an F test followed by an unpaired t test was performed where *** refers to a P value of <0.001.