PRDM10 directs FLCN expression in a novel disorder overlapping with Birt-Hogg-Dubé syndrome and familial lipomatosis

Abstract

Birt-Hogg-Dubé syndrome (BHD) is an autosomal dominant disorder characterized by fibrofolliculomas, pulmonary cysts, pneumothoraces and renal cell carcinomas. Here, we reveal a novel hereditary disorder in a family with skin and mucosal lesions, extensive lipomatosis and renal cell carcinomas. The proband was initially diagnosed with BHD based on the presence of fibrofolliculomas, but no pathogenic germline variant was detected in FLCN, the gene associated with BHD. By whole exome sequencing we identified a heterozygous missense variant (p.(Cys677Tyr)) in a zinc-finger encoding domain of the PRDM10 gene which co-segregated with the phenotype in the family. We show that PRDM10Cys677Tyr loses affinity for a regulatory binding motif in the FLCN promoter, abrogating cellular FLCN mRNA and protein levels. Overexpressing inducible PRDM10Cys677Tyr in renal epithelial cells altered the transcription of multiple genes, showing overlap but also differences with the effects of knocking out FLCN. We propose that PRDM10 controls an extensive gene program and acts as a critical regulator of FLCN gene transcription in human cells. The germline variant PRDM10Cys677Tyr curtails cellular folliculin expression and underlies a distinguishable syndrome characterized by extensive lipomatosis, fibrofolliculomas and renal cell carcinomas.

Figure 1

Pedigree and photos of skin and mucosal phenotype. (A) Pedigree of the family. Pr; prostate cancer, uRCC; unclassified renal cell carcinoma, B/M; bilateral and multifocal, ccRCC; clear cell renal cell carcinoma, Pap2 RCC; papillary type 2 renal cell carcinoma, Th folll; follicular thyroid carcinoma, Lu; lung carcinoma, NHL; non-Hodgkin lymphoma. The proband is indicated by an arrow. (B) Photos of affected patients. I (patient IV-1): partly confluent papules on the flank, skin tags in the armpit, intraoral papules, papules on the lip and multiple skin-colored papules on the trunk. II (patient IV-4): papules in the neck. III (patient IV-3): papules on the nipple, intraoral papules. IV (patient III-2): partly confluent papules on the flank and intraoral papules.

Figure 2

Histology of skin and mucosal phenotype and PRDM10 sequence analyses. (A) Histology of skin lesions from patient IV-3: panel I is H&E staining of skin biopsy of the cheek showing strands of epithelium surrounded by stromal cells in loose connective tissue with mucin. This histological picture is consistent with a diagnosis of a fibrofolliculoma (squared red). Panel II is H&E staining of scrotum biopsy showing a skin tag of fibrous stroma covered by squamous cell epithelium. Scale bar is 200 μm. (B) Sanger sequencing on DNA isolated from tissues derived from family members (III-4) and (IV-3) confirms the PRDM10 c.2030G > A mutation without loss of the second PRDM10 allele.

Figure 3

PRDM10^Cys677Tyr phenocopies cellular effects of FLCN loss in human embryonic kidney cells. (A) Schematic presentation of PRDM10 with known domains. Variant Cys677Tyr is located in seventh zinc finger and indicated in red. There are seven isoforms of PRDM10 described at Uniprot, the longest is 1160. (B) Creation of endogenous mutants in 293 T cells. Sanger sequence chromatogram of DNA derived from CRISPR prime edited cells shows a homozygous endogenous PRDM10^Cys677Tyr c.2030G > A mutation (indicated in orange). To prevent further genome editing, a silent mutation (indicated by asterisk) was introduced to disrupt the PAM site. (C) qPCR (n = 4) shows downregulation of FLCN expression upon PRDM10^Cys677Tyr. Expression levels of EIF3B and BCCIP are increased. (D) Western blot (n = 3) shows downregulation of FLCN protein upon PRDM10^Cys677Tyr. We detected higher PRDM10 protein levels in the PRDM10^Cys677Tyr mutant as compared to the wild type (WT). GAPDH was used as a loading control. (E) qPCR showed that PRDM10^Cys677Tyr resulted in upregulation of specific genes, which are also induced upon FLCN loss (FLCN^KO) in 293 T cells. Results are representative for three independent experiments with at least two technical replicates. To determine quantitative gene expression levels, data were normalized to the geometric mean of two housekeeping genes. Note that FLCN mRNA is still detectable in the FLCN^KO cell line, possibly due to incomplete nonsense-mediated mRNA decay or a transcriptional feedback mechanism. (F) Western blots (n = 2) of 293 T mutant cell lines. FLCN protein is absent in both PRDM10^Cys677Tyr and FLCN^KO 293 T. Induction of GPNMB in PRDM10^Cys677Tyr was similar to induction levels observed in FLCN^KO cells. Actin was used as loading control. (G) Both PRDM10^Cys677Tyr and FLCN^KO cells grew slower when compared to wild type. Cells were seeded in equal densities and total cell number was counted for six consecutive days (n = 2).

Figure 4

FLCN is a transcriptional target of PRDM10. (A) Schematic overview of primer locations used for X-ChIP qPCR experiments as described in (B). (B) FLCN promoter binding by PRDM10 was assessed by PRDM10 X-ChIP of 293 T PRDM10 wildtype and Cys677Tyr (C677Y) mutant cells. Fold enrichment of binding capacities between wild-type and C677Y mutant PRDM10 was determined by qPCR, showing that the mutation strongly diminished promoter binding. Bar graphs are representative of differences observed in three independent ChIP experiments, with two technical replicates per qPCR, and normalized to input and a predicted non-binding region ~20 kB upstream (left) or downstream (right) of the FLCN promoter. Two different primer sets surrounding the predicted PRDM10 motif (GGTGGTACGGCTCA) in the FLCN promoter were used. (C) qPCR (n = 2) shows that inducible overexpression (OE) of PRDM10^Cys677Tyr slightly repressed FLCN expression within 72 h, while this effect did not occur upon overexpression of PRDM10^WT in 293 T cells. (D) Volcano plot showing gene expression changes upon five days induction of PRDM10^Cys677Tyr in RPTEC/TERT1. 6087 genes are differentially expressed (P-value ≤ 0.05). Top 20 of most significant differential genes are indicated, plus, as a reference, these marker genes: PRDM10, FLCN, GPNMB, RRAGD, PTEN, CCND1, CDH1, CDH3, ACP5, PDGFRB and CCL2. Asterisks indicate genes of potential interest overlapping between FLCN^KO and PRDM10^Cys677Tyr cell lines. (E) Volcano plot showing gene expression changes upon FLCN^KO in RPTEC/TERT1. A total of 3703 genes are differentially expressed (P-value ≤ 0.05). Top 20 of most significant differential genes are indicated, plus, as a reference, these marker genes: PRDM10, FLCN, GPNMB, RRAGD, PTEN, CCND1, CDH1, CDH3, ACP5, PDGFRB and CCL2. Asterisks indicate genes of potential interest overlapping between FLCN^KO and PRDM10^Cys677Tyr cell lines. (F) Overlap of gene expression changes upon FLCN loss or PRDM10^Cys677Tyr induction in RPTECs. A total of 446 genes are significantly upregulated in both conditions and 473 genes are significantly downregulated in both conditions. Gene lists are provided as Supplementary Material, Table S3.