https://orcid.org/0000-0001-7983-3109
Abstract
Hereditary leiomyomatosis and renal cell carcinoma (HLRCC) is an autosomal dominant condition characterized by the development of cutaneous and uterine leiomyomas and risk for development of an aggressive form of papillary renal cell cancer. HLRCC is caused by germline inactivating pathogenic variants in the fumarate hydratase (FH) gene, which encodes the enzyme that catalyzes the interconversion of fumarate and L-malate. We utilized enzyme and protein mobility assays to evaluate the FH enzyme in a cohort of patients who showed clinical manifestations of HLRCC but were negative for known pathogenic FH gene variants. FH enzyme activity and protein levels were decreased by 50% or greater in three family members, despite normal FH mRNA expression levels as measured by quantitative PCR. Direct Nanopore RNA sequencing demonstrated 57 base pairs of retained intron sequence between exons 9 and 10 of polyadenylated FH mRNA in these patients, resulting in a truncated FH protein. Genomic sequencing revealed a heterozygous intronic alteration of the FH gene (chr1: 241498239 T/C) resulting in formation of a splice acceptor site near a polypyrimidine tract, and a uterine fibroid obtained from a patient showed loss of heterozygosity at this site. The same intronic FH variant was identified in an unrelated patient who also showed a clinical phenotype of HLRCC. These data demonstrate that careful clinical assessment as well as biochemical characterization of FH enzyme activity, protein expression, direct RNA sequencing, and genomic DNA sequencing of patient-derived cells can identify pathogenic variants outside of the protein coding regions of the FH gene.
Introduction
Patients with hereditary leiomyomatosis and renal cell carcinoma (HLRCC) are predisposed for the development of smooth muscle cutaneous and uterine leiomyomas. Individuals with HLRCC are also at risk for the development of an aggressive form of Type 2 papillary renal cell cancer (RCC) that has the potential to spread when the primary tumor is small (1). The estimated cumulative risk for development of renal cancer in HLRCC patients is 15%, and renal tumors have occurred in individuals as young as 10 years of age (2). The aggressive and infiltrative nature of HLRCC-related RCC tumors has led to the recommendation that individuals with genetically confirmed HLRCC should undergo annual abdominal imaging for surveillance of renal manifestations starting at 8 years of age (2). Additionally, early surgical intervention with wide surgical margins and consideration of retroperitoneal lymph node dissection is recommended for HLRCC patients manifesting with renal lesions (1).
HLRCC is caused by germline heterozygous inactivating mutations in the fumarate hydratase (FH) gene, which encodes the FH enzyme that catalyzes the interconversion of fumarate and L-malate in the Krebs cycle. HLRCC patients harboring FH mutations in their germline show diminished FH enzyme activity levels in cultured patient dermal fibroblasts and lymphocytes, typically at levels of 50% or less than controls (3,4). FH-deficient tumors exhibit complete loss of FH enzyme activity in the affected tumor cells and accumulate high levels of fumarate (5). This leads to a pseudohypoxia phenotype characterized by stabilization of HIF1α (6) and inhibition of α-ketoglutarate-dependent dioxygenases resulting in hypermethylation of DNA and histones (7–9). Accumulation of fumarate in FH-deficient tumors is also associated with the non-enzymatic covalent modification of cysteine residues in proteins leading to the formation of an S-(2-succinyl)-cysteine (2-SC) protein adduct. This modification known as succination inactivates multiple proteins resulting in modulation of tumor biology and metabolism (10,11). Additionally, succination inactivates the mitochondrial nucleoid components responsible for the maintenance and replication of mitochondrial DNA resulting in copy loss and mutation of mtDNA in FH-deficient tumors (12). The 2-SC protein adduct in FH-deficient cells and tumors can be detected by histochemical and proteomic methodologies (10,13,14), and positive 2-SC immunoreactivity in tumor cells is increasingly used as a tool to facilitate diagnosis of FH-deficient RCC (15–17).
Screening for HLRCC in patients and their families involves obtaining a detailed family history as well as careful clinical evaluation of patients for manifestations of uterine leiomyomata, cutaneous leiomyomas, and renal lesions. Clinical diagnosis includes germline DNA sequencing of the FH gene, performed by a qualified Clinical Laboratory Improvement Amendments (CLIA)-approved laboratory. The FH gene is evaluated for missense, frameshift, and nonsense mutations in the exon sequences, and intronic sequences very near the exon boundaries are evaluated for splice site mutations (3). In this manner, 31 pathogenic FH variants were identified in 56 families with manifestations of HLRCC, with an overall detection rate of over 90% (3). Next-generation techniques have facilitated the identification of germline heterozygous large-scale deletions of the FH gene in additional individuals presenting with clinical manifestations of HLRCC (18,19).
Despite the relatively high detection rate for HLRCC using the standard genetic screening procedures outlined above, pathogenic alterations in the FH gene may remain undetected in some patients and their families since large-scale chromosomal rearrangements, promoter alterations, and deep intronic variants are not typically evaluated using the standard screening techniques. In this study, we utilized enzymatic activity assays to evaluate FH activity in material from patients with clinical manifestations of HLRCC who lacked standard CLIA-identifiable variants in the FH gene. By comparing enzymatic activity, protein expression and mobility, and mRNA expression of the FH gene in cultured dermal fibroblast and lymphoblast cell lines obtained from these individuals, we identified a discrepancy between FH protein and mRNA levels. Direct RNA sequencing of patient-derived mRNA transcripts identified retained intronic sequences, consistent with aberrantly spliced FH mRNA. Finally, genomic DNA sequencing confirmed the presence of a novel deep intronic point mutation resulting in the creation of a novel splice acceptor site near a polypyrimidine tract. Together, these data demonstrate the clinical utility of evaluating FH activity and mRNA expression in cells obtained from patients suspected of having HLRCC but lacking a detectable sequence variant in the FH coding region.
Materials and methods
Patients and patient-derived primary cells
Patients were evaluated and treated at the Clinical Center of the National Institutes of Health on Urologic Oncology Branch, National Cancer Institute protocols (NCI-97-C-0147 and NCI-03-C-0066) approved by the NCI Institutional Review Board, and written informed consent was given for participation in this study. Patents were tested for germline mutations in the FH gene by CLIA-certified genetic testing. Dermal fibroblast cultures were obtained from dermal biopsy explants maintained on tissue culture dishes in a humidified incubator at 37 °C in the presence of 5% O2 and 5% CO2 and propagated in DMEM medium containing 5 mM D-glucose, 4 mM glutamine, 1 mM pyruvate, 25 mM HEPES, 10% fetal bovine serum (FBS), 100 units/ml of penicillin and streptomycin, and 0.25 μg/ml amphotericin-B. All experiments were performed on fibroblast cultures at passages 3–6. Immortalized lymphoblastoid B cell lines were generated by EBV transformation of blood lymphocytes and were cultured in RPMI 1640 medium supplemented with 1 mM pyruvate, 4 mM glutamine, 10% FBS and 100 units/ml of penicillin and streptomycin, 0.25 μg/ml amphotericin-B (all reagents from Gibco, Thermo Fisher Scientific Inc., MA, USA).
In-gel enzyme activity assays
FH and malic NAD + -reductase enzyme activities were separated by native gel electrophoresis using pre-cast 1 mm 7% Tris-acetate polyacrylamide gels (Invitrogen, CA, USA). Briefly, whole cell lysates were prepared on ice by sonication in lysis buffer consisting of 10 mM Tris, pH 8.0, 0.1% Triton X-100, 3 mM KCl, 3 mM MgCl2, 3 mM sodium citrate, and protease inhibitors (Complete™ EDTA-free, Roche), followed by centrifugation at 21 000× g for 30′ at 4 °C. Protein concentration of the recovered supernatants was measured by BCA assay and 10 μg of total cellular protein was loaded into the gel, and the samples were subjected to electrophoresis at 4 °C for 2 h at 125 V, with running buffer consisting of 25 mM Tris, 192 mM glycine. For measurement of FH activity, the gels were immersed in 10 ml of a solution pre-warmed to 37 °C, consisting of 100 mM Tris, pH 7.4, 1 mg/ml 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT), 1 mM NAD+, 0.184 mg/ml phenazine methosulfate (PMS), 150 units of porcine heart malic dehydrogenase (Sigma-Aldrich), and 5 mM Na fumarate. The gels were gently rocked at 37 °C in the dark for 5–10 min, until FH enzyme activity bands were developed, at which point the gels were de-stained in the dark with sequential incubations in de-ionized water until the background was clear, and the gels were scanned immediately using a transmitted light flatbed scanner. NAD-malate dehydrogenase enzyme activity was assessed by incubation of duplicate gels with 100 mM Tris, pH 7.4, 1 mg/ml iodonitrotetrazolium bromide (INT), 1 mM NAD+, 0.184 mg/ml phenazine methosulfate (PMS), and 5 mM Na L-malate.
Fluorogenic FH activity assay
Whole cell proteomes were extracted from patient samples by sonication. Briefly, 100 μl of lysis buffer containing ice-cold PBS, protease inhibitor cocktail (1X, EDTA-free, Cell signaling #5871S) and phenylmethylsulfonyl fluoride (1 mM Sigma #78830), was added to each sample. Samples were then incubated on ice in preparation for sonication. Each sample was sonicated on ice using a 100 W QSonica Q700 sonicator (10 × 3-s pulses, amplitude 1, 30 s resting in between pulses) equipped with microtip #4417. Following sonication, lysates were pelleted by centrifugation (20 000 relative centrifugal force × 30 min, 4 °C) to discard insoluble cellular debris and the supernatant were recovered for protein concentration measurement on a Qubit 2.0 Fluorometer using a Qubit Protein Assay Kit. Lysates were diluted to a 2 mg/ml concentration and stored at −80 °C for FH enzyme activity assay.
To determine FH activity, proteomes (5 μg) were incubated with neutralized L-malic acid (25 mM) and sodium phosphate buffer (100 mM, pH 7) in a final reaction volume of 100 μl in a UV-Star 96-well plate (Greiner Bio-One #655801). Plates were covered with sealing film (Eppendorf #0030127781) and incubated at 37 °C for 2 h. Following incubation, 100 μl of tetrazole 1 (300 μM in acetonitrile, final reaction concentration of 150 μM) was added to each well and mixed by pipetting then irradiated with 302-nm light for 2 min. Samples were then transferred to a black clear-bottom 96-well plate (Greiner Bio-One #655090), and fluorescence measurements were taken on a BioTek Cytation5 plate reader controlled by Gen5 detection software (version 3.09). Measurements were taken by using an excitation wavelength of 410 nm, and a spectral scan was performed to monitor emission wavelengths between 430 and 630 nm by 5-nm increments. To determine FH activity in patient samples, a fumarate standard curve was generated using 0, 25, 50, 100, 200, 400, and 800 μM concentrations in 100-mM sodium phosphate buffer (pH 7) in a final volume of 100 μl. Following addition of tetrazole (300 μM in acetonitrile, final reaction concentration of 150 μM) to the fumarate standards, samples were subjected to irradiation and fluorescence measurements were taken as described above. Spectral measurements were plotted using GraphPad Prism 9 software (version 9.1) and a smooth of curve function was used to generate area under the curve (AUC) values for all samples. A simple linear regression function was fitted to the AUC values of the fumarate standards to determine the production of fumarate in the patient samples. Patient sample measurements were taken in quadruplicate technical replicates, while standard curve samples were performed in duplicate. Tetrazole 1 was synthesized as previously described (20).
Nanopore library preparation and bioinformatic processing
Cultured dermal fibroblasts from an affected patient and negative-control fibroblasts obtained from a patient affected with von Hippel Lindau disease were collected in Trizol, and total RNA was isolated following the manufacturer’s protocol (Invitrogen, CA, USA). Total RNA underwent two rounds of polyA selection using NEB oligodT beads and protocol (New England Biolabs S1419S). PolyA quality was assessed by Agilent High Sensitivity RNA ScreenTape Assay. PolyA RNA (750 ng) was prepared for nanopore direct RNA sequencing following the Oxford Nanopore Technology (ONT) SQK-RNA002 kit protocol, including the optional reverse transcription step recommended by ONT. RNA sequencing on the MinION was performed using ONT R9.4 flow cells and the standard MinIT protocol script for 60 h. Data quality control was performed using MinION QC (v. 1.4.1) (21). Raw fast5 files were basecalled using Guppy (v3.5.4) and mapped to Human GRCh38.p12. genome using minimap2 (v 1.7) in splice-aware RNA mapping mode with a kmer size of 14 (minimap2 -ax splice -uf -k14). SAM and BAM file manipulations were performed with samtools (v 1.9), and mapped data was visualized in the Integrated Genome Viewer (IGV-2.8.0) (Broad Institute) (22). The code for nanopore processing and figure generation is available at https://github.com/BatistaLab/Exon9A_Nanopore.
Whole genome sequencing
Germline variant analysis was performed as described in Ricketts et al. (23), using the hg38 reference genome.
Immunohistochemistry
Five-micron thick formalin-fixed paraffin-embedded sections were cut from surgical material obtained from a myomectomy as well as from a biopsy of a dermal leiomyoma lesion. IHC staining was performed by VitroVivo Biotech (Rockville, MD) using standardized methodologies, using either anti-FH (Santa Cruz Biotechnology; 1:1000) or anti-2-succinocysteine (2SC) (Cambridge Research Biochemicals; 1:5000). Images were captured using an AxioScan.Z1 Slide Scanner (Zeiss, Oberkochen, DE).
Data access
Sequencing data will be made available by request on dbGaP (https://www.ncbi.nlm.nih.gov/gap/).
Results
Families evaluated in this study
Family 1
Patient #1 presented with biopsy-proven cutaneous leiomyomas consisting of light tan-to-brown papules present on the left posterior shoulder, right lumbar back, and central chest. This patient also had symptomatic uterine leiomyomas and underwent a total hysterectomy at an outside institution at age 33 (Fig. 1, Family #1). Patient #2, the sister of Patient #1, also presented with biopsy-proven cutaneous leiomyomas as well as symptomatic uterine leiomyomas for which she underwent a total hysterectomy at an outside institution at age 30. Patient #3, the daughter of Patient #1, presented with biopsy-proven cutaneous leiomyomas and underwent a total hysterectomy at an outside institution at age 35. All three patients were tested for pathogenic germline variants in the FH gene using CLIA-certified Sanger sequencing platforms; Patient #1 underwent additional deletion/duplication testing of the FH gene as well as karyotype analysis to rule out germline translocations involving chromosome 1. None of these tests identified a known pathogenic alteration of the FH gene. Several other family members in the paternal lineage were reported to have cutaneous and/or uterine leiomyomas, and a paternal uncle was reported to have RCC of undefined histology (Fig. 1, Family #1).
Pedigree of affected family members. Clinical phenotypes are denoted in colors; Red: cutaneous leiomyomata. Blue: renal cell carcinoma. Yellow: uterine leiomyomata. Asterisk (*) indicates that CLIA-certified genetic testing for FH mutation was performed and yielded negative results for Patients #1–4.
Family 2
Patient #4 presented with a history of uterine leiomyomas and underwent a myomectomy at an outside institution at age 30 for the removal of fibroids, the largest of which was 10 cm in diameter. Histological evaluation at the NIH Clinical Center revealed high cellularity and nuclear atypia. The patient reported dermal lesions on the right lower leg that were sensitive to touch and cold, which were confirmed to be leiomyomas upon biopsy at the NIH Clinical Center. Sequencing and deletion/duplication analysis did not reveal germline alterations in the FH gene, and the patient’s karyotype was normal. The patient reported that two paternal aunts had a history of uterine fibroids (Fig. 1, Family #2).
In-gel activity measurement of FH enzyme activity and protein mobility in HLRCC patient-derived primary cells
In order to evaluate the functional effect of FH mutations on the activity and assembly of FH protein, we developed a tetrazolium-based in-gel electrophoretic FH activity assay using tris-acetate polyacrylamide gels ((12); see Materials and Methods). Concomitantly, we ran a replicate tris-acetate native gel loaded identically and electroblotted the resolved proteins onto a nitrocellulose membrane, followed by immunoblotting using a monoclonal antibody raised against FH. We evaluated FH enzyme activity and native gel mobility in cultured dermal fibroblasts obtained from an unaffected control patient as well as genetically confirmed HLRCC patients harboring FH variants R233H, H318Y, M454X, and L244R or complete FH deletion (Supplementary Material, Fig. S1A and B). All patient samples harboring FH missense variants exhibited a shift in the mobility of their FH enzyme activity as well as altered FH protein mobility as assessed by native FH immunoblot (Supplementary Material, Fig. S1A and B). In contrast, the patient sample harboring the FH deletion showed unaltered mobility but decreased activity and expression of the FH protein (Supplementary Material, Fig. S1A and B). Together, these data demonstrate that evaluation of native FH activity and protein expression can reveal diminished FH activity and altered FH protein mobility in HLRCC patient-derived primary cells.
Analysis of FH enzyme activity in HLRCC-phenotypic patients
Using the methodologies outlined above, we next investigated FH enzyme activity and protein expression in dermal fibroblasts and immortalized B lymphoblasts derived from our HLRCC-phenotypic patients. Using the in-gel FH activity assay, we found that FH activity was diminished in cultured fibroblasts obtained from Patients #1 and #2 as well as in fibroblasts obtained from a known HLRCC patient harboring a R343X FH variant, relative to fibroblasts obtained from a patient unaffected by HLRCC (Fig. 2A, upper panel). Native immunoblot of FH demonstrated decreased FH protein expression in the absence of a notable shift in FH protein mobility (Fig. 2A, middle panel). In contrast, malate dehydrogenase (MDH) enzyme activity, assessed as a “housekeeping” enzyme in these analyses, was similar across all four samples (Fig. 2A, lower panel).
Analysis of FH enzyme activity in HLRCC-phenotypic patients. (A) Fumarate hydratase and malate dehydrogenase in-gel enzyme activity and FH protein mobility were assessed in cultured primary dermal fibroblasts obtained from a patient unaffected with HLRCC, a known HLRCC patient with a truncating R343X FH mutation, and Patients #1 and #2. (B) FH and MDH in-gel enzyme activity and FH protein mobility were assessed in cultured lymphoblasts derived from two unaffected patients, three HLRCC-phenotypic patients, and an HLRCC patient harboring a K447X FH mutation. (C) Quantitation of FH and MDH enzyme activities was performed on triplicate fibroblast cultures obtained from three unaffected individuals (FH+/+), a known HLRCC patient (FH+/−) with a truncating R343X FH mutation, and Patients #1 and #2. (D) Fluorogenic fumarase assay was performed on fibroblast cultures obtained from two unaffected individuals (FH+/+), a known HLRCC patient (FH+/−) with a Q439P FH point mutation, and Patients #1 and #2 (n = 4 cell pellets per patient). *P < .05, **P < .01.
We also evaluated FH and MDH enzyme activities in lymphoblasts obtained from each of three patients from Family #1 and compared these with a known HLRCC patient harboring a heterozygous truncating K447X variant in the FH protein, as well as with two unaffected control patients (Fig. 2B). Lymphoblasts from all three patients from Family #1 as well as the known HLRCC patient exhibited reduced FH enzyme activity (Fig. 2B, upper panel), whereas MDH activity was similar across all samples (Fig. 2B, lower panel). Notably, native FH immunoblots also revealed diminished FH levels and an absence of change in FH protein mobility in the lymphoblasts from the three unknown patients, while the cells harboring the K477X FH variant showed evidence for an upward shift in mobility suggesting altered characteristics of the FH multimer in these cells (Fig. 2B; middle panel).
Next, FH and MDH in-gel enzyme assays were performed in triplicate dermal fibroblast samples and analyzed by densitometry to obtain a quantitative assessment of the reduction in FH enzyme activity in our patients. As compared with three patients confirmed to be unaffected by HLRCC, fibroblast FH activity was significantly reduced in a known HLRCC patient and was similarly and significantly decreased in Patients #1 and #2 (Fig. 2C). In contrast, MDH activity was not significantly different between any of the fibroblast samples evaluated (Fig. 2C). To obtain an orthogonal assessment of FH enzyme activity in our patient fibroblasts, we employed a recently described fluorogenic FH activity assay that employs the conjugation of hydrazonyl chlorides and diaryl tetrazole to create a nitrileimine intermediate that reacts with fumarate to form a fluorescent pyrazolidine cycloadduct (20,24). This assay, which measures the quantity of fumarate produced from malate during a fixed incubation period, is inherently more quantitative than the densitometry-based native gel activity assay (Supplementary Material, Fig. S2). Using the fluorogenic assay, we observed a >50% decrease in the amount of fumarate produced in fibroblast lysates obtained from Patients #1 and #2 (Fig. 2D). Fumarate production was similarly diminished in fibroblasts from a known HLRCC patient harboring a Q439P FH variant (Fig. 2D). In summary, two complementary FH enzyme assays demonstrated diminished levels of FH enzyme activity in our patients showing clinical manifestations of HLRCC in the absence of detectable FH gene mutations.
Analysis of FH protein levels in HLRCC-phenotypic patients
Given the unknown reason for diminished FH enzyme activity in our cohort of HLRCC-phenotypic patients, we next set out to investigate FH protein expression levels in patient-derived cells. We performed SDS-PAGE western blots to evaluate FH protein expression and found that FH protein levels were diminished in cultured fibroblasts obtained from Patients #1 and #2 as well as in an HLRCC patient harboring the R343X FH variant, relative to fibroblasts obtained from a patient unaffected by HLRCC (green bands; Fig. 3A). Immunoblotting of MDH1 protein revealed comparable MDH1 protein levels in all four samples. Similar results were obtained in cultured lymphoblasts obtained from Patients #1–3, where FH protein levels were diminished in the three unknown patients as well as in an HLRCC patient harboring a K477X FH variant, as compared with two non-HLRCC controls (Fig. 3B). We repeated our SDS-PAGE protein analyses of FH and MDH1 protein levels in independent triplicate fibroblast cultures and found that FH protein levels were significantly decreased by ~50% in Patients #1 and #2 as well as in a patient harboring a truncating R343X FH variant, as compared with three control patients unaffected by HLRCC (Fig. 3C).
Analysis of FH protein expression in HLRCC-phenotypic patients. (A) Fumarate hydratase and malate dehydrogenase 1 protein expression was assessed in cultured primary dermal fibroblasts obtained from a patient unaffected with HLRCC, a known HLRCC patient with a truncating R343X FH mutation, and Patients #1 and #2. (B) Fumarate hydratase and malate dehydrogenase 1 protein expression was assessed in cultured lymphoblasts derived from two unaffected patients, the three HLRCC-phenotypic patients, and an HLRCC patient harboring a K447X FH mutation. (C) FH protein bands were quantified and normalized to MDH1 bands in triplicate fibroblast cultures obtained from three unaffected individuals (FH+/+), a known HLRCC patient (FH+/+) with a truncating R343X FH mutation, and Patients #1 and #2. *P < .05, **P < .01.
Analysis of FH mRNA in HLRCC-phenotypic patients
The enzyme activity and protein expression analyses described above indicate that somatic cells derived from normal tissues of the HLRCC-phenotypic patients exhibit diminished FH enzyme activity and express only ~ 50% FH protein levels relative to controls. This suggested that an as-yet undetected alteration in the FH gene sequence results in loss of FH protein expression. One possibility is that an intronic mutation could alter splicing and render mRNA transcripts produced from the mutant FH allele susceptible to nonsense-mediated decay. To test this hypothesis, we performed quantitative PCR (qPCR) to assess total cellular FH mRNA levels. We found that fibroblasts derived from Patients #1 and #2 showed levels of FH mRNA that were comparable to two control patients unaffected by HLRCC (Fig. 4A). In contrast, mRNA levels in two HLRCC patients harboring point mutations in the FH gene resulting in premature stop codons (R343X and A101X) showed FH mRNA levels that were ~50% of control (Fig. 4A), suggesting that these truncating FH mutations result in NMD of mRNA transcripts produced from the affected FH alleles. Finally, UOK348 cells that harbor both a point mutation resulting in a premature stop codon (K371X) as well as an FH exon 5 frameshift mutation (F191LfsX11) possessed nearly undetectable FH mRNA levels, suggesting nearly complete NMD of all mutant cellular FH mRNAs. Together, these data suggested that normal cells obtained from the HLRCC-phenotypic patients produce stable FH mRNA transcripts from both FH alleles, but that the FH mRNA produced from one of the FH alleles must not produce stable or functional FH protein.
Analysis of FH mRNA in HLRCC-phenotypic patients. (A) FH mRNA expression levels were assessed by TaqMan qPCR and normalized to MDH1 mRNA expression levels in cultured primary dermal fibroblasts obtained from two unaffected patients (FH+/+), two known HLRCC patients harboring R343X and A101X FH mutations, respectively (FH+/−), UOK348 tumor cells harboring compound heterozygous K371X and F191LX11 FH mutations (FH−/−), and Patients #1 and #2. (B) FH mRNA sequences obtained by direct Nanopore total mRNA sequencing of Patient #2 and control unaffected patient fibroblast RNA were aligned to the human genome (see Results). (C) RT-PCR using primers flanking FH exons 9 and 10 was performed on fibroblast mRNA obtained from all three affected patients as well as a fourth unrelated patient. (D) Schematic diagram demonstrating the normal splicing between exons 9 and 10 of FH mRNA. (E) Impact of FH intron 9 chr1:241498239 T/C mutation on splicing and protein translation of FH mRNA. The resulting A/G base substitution in FH mRNA creates a strong splice acceptor site leading to retention of intron 9 sequence, and an alternative splice donor site 83 bp downstream from the mutation facilitates splicing with exon 10. The result is altered C-terminal amino acid incorporation and premature termination of the FH protein.
To focus on the altered FH mRNA allele present in the HLRCC-phenotypic patient cells, we performed long-read direct mRNA sequencing of poly-A-selected total cellular mRNA using Nanopore technology. In addition to the canonical FH transcript, Nanopore analysis of the FH mRNA sequences revealed the presence of a novel FH mRNA isoform containing additional sequence derived from intron 9 (Fig. 4B). Nanopore direct mRNA sequencing detected 96 and 74 FH mRNA reads in two independent runs in RNA obtained from an unaffected patient, all showing proper splicing between FH exons 9 and 10 and no FH mRNA reads with inclusion of cryptic exon 9A. For two independent Nanopore runs with mRNA obtained from Patient #2, we observed 27 and 28 reads with proper splicing between exons 9 and 10, and 20 and 28 FH mRNA reads showing inclusion of exon 9A (Fig. 4B), suggesting that this alternatively spliced FH allele is expressed at comparable levels to the wild-type FH allele. This observation was further confirmed by RT-PCR with primers flanking exon 9 and 10, performed on cDNA prepared from all three patients from Family #1 as well as from Patient #4 from Family #2 (Fig. 4C), consistent with the presence of a new exon in the FH mRNA in these patients’ cells.
Analysis of whole genome data obtained from blood DNA of the HLRCC-phenotypic patients revealed a single point mutation, 1:241498239 T/C, present within FH intron 9 (Supplementary Material, Fig. S3). Cosegregation analysis revealed that this intronic 1:241498239 T/C FH variant was not present in the 1000 Genomes database, despite its presence in members of both families (Supplementary Material, Fig. S3). Sanger sequencing of cDNA generated from patient cells confirmed that splicing occurs at the site of mutation. Furthermore, sequence analysis identified an AGAG motif at a new putative splice donor site that delineates the 3′ end of the new FH exon 9A in these patients (Fig. 4D and E). The insertion of this new exon disrupts the protein reading frame and, when included, results in a truncated version of FH protein that is 37 residues shorter than the wild-type FH protein and creates a new c-terminus (Fig. 4D and E). Thus, in these patients an intronic mutation creates a new splice acceptor site, and a splice donor site present downstream is utilized to produce a retained exon in the mutant FH mRNA, termed exon 9A.
LOH and immunohistochemical analyses of patient-derived tumors
Given our new understanding of the molecular nature of the FH variant present in this cohort of HLRCC patients, we undertook sequencing and histological analyses to test for signs of FH deficiency in patient-derived uterine and dermal leiomyoma specimens. Loss of heterozygosity (LOH) of the wild-type FH allele is frequently observed in FH-deficient leiomyomas and renal tumors (25–27). We sequenced the region of FH intron 9 containing the 1:241498239 T/C alteration in genomic DNA obtained from histological sections of two uterine leiomyoma specimens obtained from Patient #3. Our results revealed that both tumor specimens exhibited loss of the wild-type FH allele (Fig. 5A). Next, we performed immunohistochemistry to examine FH protein levels in a uterine leiomyoma from Patient #3 and found that FH staining was negative in the tumor, whereas endothelial cells (white arrowhead) present within the tumor retained FH expression (Fig. 5B and C). These data suggest that the truncated FH protein produced by FH mRNA transcripts containing FH exon 9A was unstable in vivo. Proteins in FH-deficient tumors undergo covalent modification of cysteine residues forming an S-(2-succinyl)-cysteine (2-SC) protein adduct (10). Tumor cells in Patient #3’s leiomyoma were strongly positive for the 2SC modification (Fig. 5D and E), while endothelial cells were 2SC-negative (Fig. 5E; white arrowhead). We also evaluated a dermal leiomyoma obtained from Patient #4, who is unrelated to Patients #1–3. We found that, whereas normal cells associated with a hair follicle in the specimen retained FH staining (Fig. 5F; white arrowhead), tumor cells in the section were FH-negative (Fig. 5F; black arrowheads). In contrast, normal cells in the skin biopsy were negative for 2SC staining (Fig. 5G; white arrowhead) while tumor cells were 2SC-positive (Fig. 5G; black arrowheads). These data confirmed that the leiomyoma specimens from patients in both families evaluated lost functional FH protein expression and accumulated fumarate, leading to positive 2SC immunoreactivity.
Genomic DNA sequencing and immunohistochemistry of uterine and dermal leiomyomata. (A) Heterozygous chr1:241498239 T/C mutation in FH intron 9 was verified by Sanger sequencing of the FH genomic negative strand in germline DNA obtained from Patient #3 (top chromatogram). Uterine leiomyomata obtained from Patient #3 during hysterectomy showed loss of the wild-type FH intron 9 allele (middle chromatograms). Unaffected germline DNA was included as a control (bottom chromatogram). (B) Immunohistochemical analysis of FH protein expression in a uterine leiomyoma obtained from Patient #3 revealed that FH protein staining was absent in the tumor cells. (C) Higher magnification images showed absent FH protein staining in tumor cells whereas blood vessel cells were positive for FH protein. (D) Positive immunoreactivity for 2SC was observed in the same uterine leiomyoma specimen. (E) Strong 2SC staining was observed in tumor cells, whereas blood vessel cells did not show 2SC immunoreactivity. (F) Dermal leiomyoma obtained from Patient #4 (Family #2) revealed absent FH protein expression in tumor regions (black arrowhead), whereas the same regions were positive for 2SC (G). In contrast, hair follicle cells in the middle of the specimen stained positive for FH and negative for 2SC (white arrowhead). Scale bar in B, D = 200 μm, C, E–G = 50 μm.
Discussion
While relatively rare, the prevalence of undiagnosed HLRCC cases is of concern. The need for timely germline FH genetic testing and interval screening of individuals who are at risk for developing FH-deficient RCC is clear (28–31). Sensitive and specific clinical genetic testing is crucial for accurate diagnosis of HLRCC and for testing of additional family members of affected individuals. Intronic variations are more difficult to detect because most current CLIA-certified genetic tests focus on the exons of disease genes, and intronic sequences are inherently less conserved than exon sequences, making it more difficult to predict the pathogenicity of intronic point mutations (32).
In this study, we evaluated patients from two independent families with clinical manifestations of HLRCC including uterine and dermal leiomyomas, but for whom standard clinical genetic testing revealed no phenotypic variations in the FH gene. We undertook functional biochemical and molecular analyses in normal tissues derived from these patients, starting with assessment of FH enzymatic activity, FH protein levels, FH mRNA levels, FH mRNA sequence analysis, and finally whole genome analysis to conclude that a single deep intronic point mutation is responsible for manifestations of HLRCC syndrome in these patients. This study underscores the utility of enzymatic assay in assessing FH function in individuals suspected to harbor germline alterations in the FH gene and highlights the usefulness of direct unbiased nanopore mRNA sequencing in revealing functional alterations in cellular mRNA splicing.
Cellular pre-mRNAs undergo splicing during and after transcription in the nucleus of the cell by the ribonucleoprotein spliceosome complex. DNA mutations resulting in the retention of cryptic exons in mature mRNAs most frequently arise from formation of a novel intronic donor splice site and activation of a pre-existing acceptor splice site (32), as was the case for a deep intronic mutation in the β-globin gene in a cohort of β-thalassemia patients (33). Less frequently, cryptic exon retention can result from formation of a novel splice acceptor site, as is the case with our novel HLRCC patients described in the present work. In either case, cryptic exon retention most frequently results in creation of a premature stop codon in the mature mRNA, resulting in rapid nonsense-mediated mRNA decay (NMD) (34). We did not observe evidence for NMD in our patients, as total levels of FH mRNA were not decreased relative to controls, and ~50% of cellular FH mRNAs in the patients contained the retained exon 9A. The absence of NMD in our patients is consistent with the fact that the protein-truncating stop codon in the mutant FH exon 9A is located just 25 nucleotides upstream of the junction with the final exon of the mRNA (35).
At least two classes of pathogenic FH variants are known to exist in humans. One class of FH variants causes autosomal dominant inheritance of HLRCC, including point mutations, truncating frameshift mutations, and partial or complete deletions of the FH gene, all of which are thought to completely abrogate the enzymatic function of the affected FH allele. In contrast, certain germline homozygous or compound heterozygous FH mutations in humans can result in the autosomal recessive metabolic disorder fumarase deficiency, in which residual FH enzymatic activity in patient tissues is extremely low but not always undetectable (36,37), and there is evidence that not all fumarase deficiency FH allele carriers are at risk for development of HLRCC-related neoplasms (38,39). Thus, while bi-allelic deletion of the FH gene results in early embryonic lethality in mice (40), inheritance of either two low functioning FH alleles or inheritance of one low-functioning FH allele and one HLRCC-causing FH allele in humans can cause fumarase deficiency (39,41). We conclude that the 1:241498239 T/C alteration in the DNA of the patients evaluated in our study results in complete alternative mRNA splicing to form the stable, non-functional FH mRNA from the mutant allele, with no amount of normally spliced FH mRNA arising from this allele. Interestingly, the FH mutation discovered to cause HLRCC in our study was reported in one patient affected by fumarase deficiency (42), suggesting that the other affected FH allele in this patient (1078C > T; H402Y) must retain some amount of enzyme activity and thus be restricted to recessive inheritance of fumarase deficiency. Finally, given the genomic, biochemical, and histological data presented in this work, we believe that the 1:241498239 T/C FH alteration can be classified as a PVS1 null-variant FH allele according to the ACMG criteria (43,44), and that this allele is causative of both HLRCC and fumarase deficiency.
In addition to representing a new FH variant for genetic screening of patients suspected to be affected by HLRCC, the novel pathogenic FH variant identified in this work should also be considered in the context of screening for sporadic FH-deficient tumors. Disease-causing intronic mutations have been described in numerous inherited human disorders (45), including somatic occurrences in cancers (46). Recent advances have led to the introduction of clinically active, FDA-approved mRNA-based antisense oligonucleotide therapies for several human disorders (47). In contrast to protein-coding point mutations or deletions in the FH gene, the disease-causing 1:241498239 T/C FH alteration relies on the cellular splicing machinery to render a non-functional FH mRNA and protein. Thus, the concept of a targeted antisense oligonucleotide approach to restore even trace amounts of normal FH mRNA splicing from this mutant FH allele could constitute a patient-specific disease prevention strategy that does not rely on alteration of the cellular genome. Moreover, given that restoration of FH activity in cultured FH-deficient tumor cells reduced their invasiveness and abrogated tumor xenograft formation (48–50), restoration of FH activity in FH-deficient tumors harboring the 1:241498239 T/C FH alteration may provide a mutation-specific antiproliferative therapeutic strategy.
Conflict of interest statement: None declared.
Funding
This work was supported by the Intramural Research Program of NCI-CCR (ZIA BC011028, ZIC BC011932, and ZIA BC011766-01) and the Pamela Anne Cafritz Renal Cell Carcinoma Award. This project was also funded in part with Federal funds from the National Cancer Institute, National Institutes of Health under Contract No. HHSN261201500003I. Sanger sequencing was conducted at the CCR Genomics Core at the National Cancer Institute, NIH, Bethesda, MD 20892. C.M.F. was supported by a Postdoctoral Fellowship from the American Cancer Society (PF-19-157-01-RMC). The content of this publication does not necessarily reflect the views or policies of the Department of Health and Human Services, nor does mention of trade names, commercial products, or organizations imply endorsement by the US Government.
