VNtyper enables accurate alignment-free genotyping of MUC1 coding VNTR using short-read sequencing data in autosomal dominant tubulointerstitial kidney disease

Abstract

The human genome comprises approximately 3% of tandem repeats with variable length (VNTR), a few of which have been linked to human rare diseases. Autosomal dominant tubulointerstitial kidney disease-MUC1 (ADTKD-MUC1) is caused by specific frameshift variants in the coding VNTR of the MUC1 gene. Calling variants from VNTR using short-read sequencing (SRS) is challenging due to poor read mappability. We developed a computational pipeline, VNtyper, for reliable detection of MUC1 VNTR pathogenic variants and demonstrated its clinical utility in two distinct cohorts: (1) a historical cohort including 108 families with ADTKD and (2) a replication naive cohort comprising 2,910 patients previously tested on a panel of genes involved in monogenic renal diseases. In the historical cohort all cases known to carry pathogenic MUC1 variants were re-identified, and a new 25bp-frameshift insertion in an additional mislaid family was detected. In the replication cohort, we discovered and validated 30 new patients.

Keywords: Genetics; Genomics; Genotyping; Techniques in genetics.

Graphical abstract

Figure 1

Mucin-1 structural domains and recurrent MUC1 dupC variation hotspot (A) Mucin-1 is a transmembrane glycoprotein widely expressed in different segments of the nephron in the kidney. Protein domains of the full-length mucin-1 protein are shown: N-terminus signal sequence, VNTR regions (variable repetition of 20aa blocks), SEA cleavage domain (is a highly conserved domain that undergoes an autocatalytic cleavage during folding in the endoplasmic reticulum), transmembrane domain (TM), and cytoplasmic domain. (B and C) The frameshift caused by the insertion of a C in the VNTR motifs (blue boxes) with stretch of seven C (GCCCCCCCAGC) creates a new stop codon shortly (85aa) beyond the VNTR domain. This variation could be studied by the SNaPshot method using MwoI restriction enzyme. The 20 amino acid (aa) repeat blocks for wild-type (WT) and mutant (MT) protein are shown. The mutant protein harbors novel aa repeats (yellow boxes) and lacks the C-terminal domain.

Figure 2

Historical cohort description flowchart (A) The inclusion and exclusion criteria for MUC1-ADTKD are shown. In this cohort, 108 families with 237 individuals were studied by short-read sequencing. No pathogenic variant was found with standard pipeline in any gene related to ADTKD. (B) The SNaPshot assay was performed to detect the known and recurrent MUC1 dupC variation. This method identified the dupC pathogenic variant in 31 index cases and their 62 symptomatic relatives. In one family with three affected members, our modified SNaPshot approach detected a 5bp deletion in the MwoI site. (C) The VNtyper pipeline was applied to all individuals in the historical cohort. VNtyper re-identified all MUC1 dupC and 5bp deletion events. In one symptomatic case from MUC1-positive family (NTIH_140), linkage analysis validated the VNtyper results by confirming the segregation of the risk allele. In one family including six symptomatic members and negative SNaPshot, the pipeline detected a 25bp insertion.

Figure 3

Schematic overview of the VNtyper pipeline The pipeline of short-read sequencing (SRS)-based deleterious variant detection in the coding VNTR of the gene MUC1 in ADTKD is shown.

Figure 4

Modified SNaPshot method explaining new events at the Mwo1site (A–C) Examples of the SNaPshot PCR product migration from symptomatic and asymptomatic individuals are shown. (B) In case of positive dupC event (REN6122000742), a 7C + C and 8C + A signals are present. With our modified protocol direct sequencing of the PCR product is feasible, which confirms the SNaPshot results. (C) A very strong 7C + A signal could be observed in some SNaPshots if Mwo1 digestion failed at one of the sites (due to a Mwo1 restriction site variants). In this instance, sequencing the PCR product could aid in the identification of potentially pathogenic (NTI6120004559) or polymorphic variants (HYP2316, NTI195). (D and E) In one index patient with the clinics of ADTKD, a very strong 7C + A signal led to the detection of a 5bp deletion in the Mwo1 site and was confirmed by subcloning and sequencing the SNaPshot PCR product.

Figure 5

Historical cohort characterization VNtyper analysis of the historical cohort identified MUC1 pathogenic variants in 33/108 families (30.6%). (A) The VNtyper results of the historical cohort based on the depth score (log10 ratio) and the estimated depth of the alternate variant (AltDepth) are shown. Independent clustering of the MUC1 SNaPshot positive and negatives cases allowed the definition of a threshold to filter out variants with low support. All SNaPshot negative cases except (NTIH_140, marked with $) clustered together below the threshold (0.00469, in red), one patient was localized in the 10% above the threshold, and all SNaPshot positive cases (in green) as well as six cases with 25bp insertion (marked with #, in lightblue) were located above the 10% of the threshold (0.00515). (B) In this section (magnification of the framed zone in (A)), the y axis scale was adjusted, and two plots were merged. Variants with depth scores below the threshold were filtered out and classified as false positive. Any variant with a depth score between the threshold and 10% above [0.00469–0.00515] or with AltDepth less than 20 is deemed low-confidence variants (L shape in gray), whereas any variant with a depth score above 0.00515 and AltDepth above 20 is considered to be high-confidence results. (C) The analysis of the pipeline’s sensitivity and specificity with different depth score-based thresholds identified 0.00469 as the optimal threshold for distinguishing true positives from false positives. (D) Among MUC1-positive families in this cohort, 31 had MUC1 dupC (94%), one had a 5bp deletion, and another had a 25bp insertion event (both 3%).

Figure 6

Characterization of the renome cohort. VNtyper analysis of the renome cohort identified 30 MUC1-positive patients (A and B) The depth score-adapted threshold revealed from the historical cohort was applied to the naive renome cohort. Forty cases were reported above the threshold (dashed line), and 662 instances were below the threshold. To confirm the reliability of the threshold, we analyzed all cases with Mwo1 site variations (dupC or dupA) reported above (n = 35) the threshold. We extended the SNaPshot analysis to the 11 individuals that clustered just below the threshold (i.e., between the threshold and 10% below [0.004176–0.00469]) to ensure that no true positive were missed. All cases tested below the threshold and 9/12 of the cases reported with low confidence were SNaPshot negative (red points). SNaPshot verified MUC1 dupC events in 26 cases (25 high-confidence and 1 from low-confidence cases) as well as a case with MUC1 dupA event (green points). With a second high depth NGS, we verified a patient with low confidence single base insertion as true positive. Variations identified in the early conserved motifs were validated using IGV, see Figure S5. In section B (magnification of the framed zone in A), the y axis scale was adjusted, and the plots were merged. (C) Patients from this cohort were assigned to different groups based on the initial clinical diagnosis. (D) MUC1 pathogenic variants were discovered in various patient groups, particularly NTI. Several cases were also identified in other cohort groups like NPH, REN, NCR, and CAKUT. (E) The description flowchart of this cohort is shown. Note: ^$Individuals with variants that cannot be confirmed by SNaPshot. ∗High confidence: patients with AltDepth above 20 or depth score above 0.00515. #Low confidence: patients with AltDepth below 20 or depth score between 0.00469 and 0.00515.