Germline variant affecting p53β isoforms predisposes to familial cancer

Abstract

Germline and somatic TP53 variants play a crucial role during tumorigenesis. However, genetic variations that solely affect the alternatively spliced p53 isoforms, p53β and p53γ, are not fully considered in the molecular diagnosis of Li-Fraumeni syndrome and cancer. In our search for additional cancer predisposing variants, we identify a heterozygous stop-lost variant affecting the p53β isoforms (p.*342Serext*17) in four families suspected of an autosomal dominant cancer syndrome with colorectal, breast and papillary thyroid cancers. The stop-lost variant leads to the 17 amino-acid extension of the p53β isoforms, which increases oligomerization to canonical p53α and dysregulates the expression of p53's transcriptional targets. Our study reveals the capacity of p53β mutants to influence p53 signalling and contribute to the susceptibility of different cancer types. These findings underscore the significance of p53 isoforms and the necessity of comprehensive investigation into the entire TP53 gene in understanding cancer predisposition.

Fig. 1. Pedigrees of families carrying the TP53β-stop-lost variant.

The pedigree of (a) family 1, (b) family 2, (c) family 3, and (d) family 4 are shown. Genotypes of all tested individuals are indicated: Mut, mutation carrier; Mut*, obligate carrier; WT, wild-type. Symbols: filled quarters represent cancer patients; filled top left quarter, colorectal cancer; filled top right quarter, breast cancer; filled bottom right quarter, thyroid cancer; filled bottom left quarter, other cancers; question mark, unknown phenotype. Numbers in brackets indicate relatives merged in the pedigree for clarity. The abbreviations under each symbol indicate the diagnosis of a malignancy or carcinoma in situ (asterisk), followed by the age at diagnosis or age at death (d) reported in 5-year aggregates: B breast, Bl bladder, Br brain, C colorectum, Cx cervix, Ga gallbladder, Ki kidney, Lu lung, Me melanoma, NHL non-Hodgkin lymphoma, O ovary, SL serrated colonic lesions, St stomach, Thr throat, Thy thyroid. The abbreviations above each symbol indicate E, whole-exome sequencing analysis of leukocyte DNA; T, somatic hotspot analysis of neoplasia; B, patient-derived B-lymphocyte cell experiments. Clinicopathological characteristics of malignant and benign tumors are provided in Supplementary Table 2.

Fig. 2. Schematic representation of the TP53 gene and its isoforms.

a The TP53 gene locus is composed of 11 exons and two cryptic exons (9β and 9γ). Its two promoters (P1 and P2) result in four different translation initiation sites (full-length, Δ40, Δ133 and Δ160). b The TP53 gene encodes for at least 12 different p53 protein isoforms, where alternative splicing of intron-9 leads to three distinct C-terminal domains (α, β, and γ). p53α is composed of the transactivation domain, the DNA-binding domain, the hinge domain, the oligomerization domain, and the p53α regulatory domain. The p53β and p53γ isoforms only contain the first 7 amino acids of the oligomerization domain. The TP53β-stop-lost variant is predicted to elongate the p53β-domain (cyan) by 17 amino acids (red) for all p53β protein isoforms.

Fig. 3. The TP53β-stop-lost variant is expressed at the mRNA and protein level in a carrier-derived B-lymphocyte cells.

a Targeted transcriptome sequencing of the whole TP53 gene showed a relative increase in 9β exon expression in TP53β-stop-lost carrier-derived (NA50) compared to non-carrier-derived (NA68) B-cells. b Increased TP53β expression was confirmed in NA50 cells compared to p53 wild-type B-cells (NA68, NA16). Difference in TP53α expression were <2-fold between the samples. Expression was measured with isoform-specific qPCR primers. Mean and standard deviations are plotted of biological replicates (n = 3). P-values were calculated with an unpaired two-sided t test (NA16:NA68: TP53α p < 0.001, TP53β p = 0.12; NA50:NA68: TP53α p < 0.001, TP53β p = 0.01). c Immunoblotting using the p53β-specific KJC8 antibody and full-length p53-specific antibody DO-7 was performed on nine B-cell samples. DO-7 preferentially detects p53α. Wild-type p53β and p53β-stop-lost co-migrate with gel electrophoresis, making it impossible to distinguish them using generic p53 antibodies. All samples showed comparable protein expression of p53 isoforms, including p53α, p53β, Δ133p53β and Δ160p53β. d The p53β-stop-lost-specific (p53β-SL) antibody was generated against synthetic p53β-stop-lost (p53β-SL) peptide (CSREHENGSMTLPDTDAT) and showed specificity for p53β-stop-lost protein isoforms in p53-null cells (H1299) transiently transfected with different p53 isoform expression vectors. Proteins were analyzed by immunoblotting. DO-1 blot was cut. e The p53β-SL and Δ133p53β-SL proteins were detected in mutant B-cells (NA50) but not in non-carrier B-cells (NA16). Both samples showed comparable expression of other p53 isoforms, shown by re-probing with full-length specific DO-1 and pan-p53 antibody SAPU. f The mRNA expression of 21 p53-target genes was measured for biological replicates (n = 3) with a targeted qPCR panel, showing altered regulation in mutant NA50 compared to p53 wild-type (NA16, NA68) B-cells. 13 genes (in bold) are consistently altered in the same direction. Median fold changes and p-values, calculated with an unpaired two-sided t test are shown in Supplementary Table 5. g Protein expression of the p53-target protein p21 (encoded by CDKN1A) was assessed with immunoblotting. Mutant B-cells (NA50) showed higher basal expression of p21 and an altered response to cellular stress, tested by doxorubicin treatment (0.5 μM), at different timepoints. All immunoblotting was performed in two independent experiments. Source data are provided as a Source Data file.

Fig. 4. Altered p53 signaling in p53β-stop-lost-overexpressing cells.

a p53 wild-type cells (HCT-116 and MCF-7) were transduced with full-length p53β, p53β-stop-lost (p53β-SL) or empty vector. p53β-stop-lost overexpression resulted in increased basal expression of most p53-target genes measured by qPCR, compared to p53β and vector controls. Minor differences in expression between the two cell-lines can be attributed genetic background and/or cell-type. Gene expression was measured for biological duplicates and normalized to GAPDH expression. Median fold changes and p-values, calculated with an unpaired two-sided t test, are shown in Supplementary Table 6. b Immunoblots of p21 protein expression, encoded by CDKN1A, after induction of cellular stress by doxorubicin (1.0 µM continuous for 2 h, 4 h or 8 h). HCT-116 p53β-stop-lost overexpressing cells showed increased basal expression of p21, which did not increase upon doxorubicin treatment, as was seen in both control and wild-type p53β overexpressing cells. MCF-7 cells showed induction of p21 after 8 h, irrespective of the overexpression of p53β-stop-lost or p53β. Immunoblotting was performed in two independent experiments (see Supplementary Fig. 5). c Cell cycle analysis of transduced cancer cell lines and response to doxorubicin treatment. In the controls, doxorubicin treatment (1.0 µM continuous for 24 h) resulted in the accumulation in G2M in HCT-116 and S-phase cells in MCF-7, while treatment caused an increased replication blockade in S-phase in the p53β-stop-lost overexpressing HCT-116 cells, but its MCF-7 counterpart, where wild-type p53β overexpressing cells showed the most pronounced blockade in S-phase. Cell cycle analysis was assessed in independent experiments (n = 3) with the NucleoCounter (Chemometec). Data points, means, standard deviations and significance (* p < 0.05; ** p < 0.01) are plotted. Cell cycle distribution and p-values, calculated with a paired two-sided t test, are shown in Supplementary Table 7). d Unbiased gene expression analysis of bulk RNA-sequencing of the transduced cell lines. Thirteen genes were consistently downregulated (logFC < −0.6) in both p53β-stop-lost-overexpressing cell lines, while two genes were upregulated (logFC > 0.6).Source data are provided as a Source Data file.

Fig. 5. Increased oligomerization properties of p53β-stop-lost.

a Crystal structure of the oligomerization domains of two p53α proteins (left panel), orientated anti-parallel for dimer formation (PDB 1SAE). Composite model of one p53α oligomerization domain and the predicted structure of the C-terminal domains of p53β (middle panel) and p53β-stop-lost (p53β-SL; right panel) showing the oligomerization domain sequence (dark gray), the β-specific protein sequence (cyan) and the p53-stop-lost peptide sequence (red). The TP53β-stop-lost variant is predicted to extend the α-helical structure. Visualization was performed in PyMOL. b Co-immunoprecipitation was performed using p53α-domain-specific antibodies (421/BP53.10) or p53β-domain-specific antibody (KJC8) on protein lysates of p53-null cells (H1299) transiently co-transfected with p53α/p53β or p53α/p53β-stop-lost, and compared to IgG control and whole cell lysates (WCL). p53β-stop-lost, but not wild-type p53β, co-immuno-precipitated with p53α, and vice versa. c Immunofluorescence staining using the p53β-stop-lost-specific (p53β-SL; green) antibody and DAPI (blue) of p53β-stop-lost expressing H1299 cells shows nuclear and cytoplasmic localization of this protein. d Immunoblot of expression levels of p53α, p53β and p53β-stop-lost protein in luciferase gene reporter assays using polyclonal p53 antibody (SAPU). H1299 cells were co-transfected with luciferase reporter plasmid driven by p53-target gene promoter, SV-renilla and empty vector or p53α; or p53α/p53β; or p53α/p53β-SL. Protein extracts were normalized according to the Renilla activity. e For the promoter sequences of p53-target genes CDKN1A, BAX and IGFBP3, the effect of p53β-stop-lost on p53α transcriptional activity was promoter specific (inducive at CDNK1A (p = 0.032) and inhibitory at BAX (p = 0.001) and IGFBP3 (p < 0.001) promoters), while p53β co-expression did not affect p53α transcriptional activity, at comparable protein levels. Expression of p53β-stop-lost alone, in the absence of p53α, did not transactivate these promoters. Luciferase activity was normalized to the basal activity in the absence of p53, and data points, means, standard deviations and p-values (* p < 0.05, ** p < 0.01, *** p < 0.001) are plotted for independent experiments; p53β and p53β-stop-lost (n = 3), other CDKN1A and BAX conditions (n = 6) and other IGFBP3 conditions (n = 7). Normalized transduction and p-values, calculated with an unpaired two-sided t test are shown in Supplementary Table 8. Source data are provided as a Source Data file.

Fig. 6. Clinical phenotype of carriers of germline TP53 variants.

Tumor distribution, median age of breast cancer diagnosis, and estimated penetrance’s are shown for carriers of pathogenic p53 DNA-binding domain variants (left) and the variant encoding p53α p.Arg337His (middle) compared to TP53-stop-lost (right). Germline TP53 variant data was obtained from the International Agency for Research on Cancer (IARC) TP53 database (Version R19) for the six most common DNA-binding domain variants (encoding p.Arg175His, p.Gly245Ser, p.Arg248Gln, p.Arg248Trp, p.Arg273His, and p.Arg282Trp), accounting for ~25% of all Li-Fraumeni syndrome families, and the Brazilian founder mutation encoding p53α p.Arg337His (6%). Penetrance of these two groups was calculated previously^,. The clinical data of TP53β-stop-lost carriers is based on 14 cancer patients, 11 unaffected individuals at the end of follow-up, and 2 relatives lacking clinical data, however, calculations of accurate cancer risk estimates require additional carriers.