The Common PKD1 p.(Ile3167Phe) Variant Is Hypomorphic and Associated with Very Early Onset, Biallelic Polycystic Kidney Disease

and neonatal demise in one dizygotic twin was referred for clinical testing. Further individuals with the putative hypomorphic PKD1 variant, p.(Ile3167Phe), were identi ﬁ ed from the UK 100,000 genomes project (100K), UK Biobank (UKBB), and a review of the literature. We identi ﬁ ed a likely pathogenic PKD1 missense paternal variant and the putative hypomorphic PKD1 variant from the una ﬀ ected mother in the deceased twin but only the paternal PKD1 variant in the surviving dizygotic twin. Analysis of 100K cases identi ﬁ ed a second family with two siblings with similar biallelic inheritance who presented at birth with VEO-PKD and reached kidney failure in their teens unlike other a ﬀ ected relatives. Finally, a survey of 618 UKBB cases con ﬁ rmed that adult patients monoallelic for PKD1 p.(Ile3167Phe) had normal kidney function. Our data reveals that p.(Ile3167Phe) is the second most common PKD1 hypomorphic variant identi ﬁ ed and is neutral in heterozygosity but is associated with VEO-PKD when inherited in trans with a pathogenic PKD1 variant. Care should be taken to ensure that it is not automatically ﬁ ltered from sequence data for VEO cases.


Introduction
Autosomal dominant polycystic kidney disease (ADPKD), due to pathogenic variants in PKD1 or PKD2 genes, is generally an adult-onset disorder which commonly progresses to kidney failure. Very rarely, the presentation can be very early onset (VEO), presenting in utero or the neonatal period with a severe phenotype occasionally leading to neonatal demise.
Several case reports [1][2][3] and two recent series of VEO cases [4,5] have elucidated the underlying genetic mechanism of disease. In most cases, biallelic PKD1 variants are detected, more rarely biallelic PKD2 variants [5,6] and very rarely trans-heterozygous variants in PKD1/PKD2/PKHD1/ HNF1B [5,7]. Fully penetrant pathogenic biallelic PKD1 variants are assumed to be early gestational embryonic lethal. Hence, all VEO cases reported have at least one variant with likely partial protein function, also known as a hypomorphic variant. In such cases, the hypomorphic variant does not cause ADPKD in isolation in heterozygous parents. The variant classification for hypomorphic variants is, however, problematic as these alleles may be present at high frequency in population studies, are likely to be benign when monoallelic, and could be automatically filtered from sequence data [5].

VEO-PKD Pedigree.
A family with a history of VEO-PKD and neonatal demise was referred for NHS diagnostic testing. The couple's first pregnancy was a male foetus with enlarged cystic kidneys and anhydramnios, detected antenatally at 21 weeks. The neonate (II.1) was delivered at 33 weeks gestation but suffered neonatal demise shortly after birth. Array CGH and genetic testing for autosomal recessive polycystic kidney disease (ARPKD) did not identify a molecular diagnosis and DNA was not stored for further testing.
The couple's subsequent pregnancy was dizygotic twins. At 16 weeks gestation, the male foetus II.2 had enlarged bright cystic kidneys detected by ultrasound. In contrast, his twin sister (II.3) had normal kidney echogenicity and length at 16 weeks. At 32 weeks, II.2 had anhydramnios with polycystic kidneys measuring >99.6 th centile; II.3 had a few kidney cysts detectable antenatally, with normal kidney length. Both foetuses were delivered at 33 + 5 weeks gestation. II.2 subsequently died shortly after birth, but II.3 has remained clinically well postnatally. There was no known significant family history at the time of referral. The paternal grandfather had a kidney removed for an unknown indication, and the paternal mother had died from cancer aged 41.
A DNA sample extracted from an uncultured postmortem skin sample from II.2 underwent next-generation sequencing (NGS) and dosage analysis using a custom hybridisation capture 17-gene cystic disease diagnostic panel sequenced on a HiSeq 2000 [5]. Dosage analysis for whole exon deletions and duplications was performed using comparative depth of coverage of NGS data (DeCON software ([8]; local validation data for single and multiexon CNVs sensitivity > 0:999 and specificity 0.989).  [9]. We examined the UKBB cohort exome data for the presence of the c:9499A > T p.(Ile3167Phe) variant in the PKD1 gene and obtained demographic data from baseline assessment including age and sex. Clinically relevant phenotype data including CKD-EPI eGFR, systolic and diastolic blood pressure, and ACR from enrolment in UKBB were obtained along with HES data for CKD and cysts. Statistical differences in clinical data were determined by independent t-test for continuous data. The c:9499A > T p.(Ile3167Phe) and c:11957C > T p.(Ala3986Val) variants were shown to be inherited from the unaffected mother. Obstetric kidney ultrasound scans on the mother (age 29) did not detect any kidney cysts. Based on familial testing, the variant classification (https:// www.acgs.uk.com/media/11631/uk-practice-guidelines-forvariant-classification-v4-01-2020.pdf) was revised and two variants were classified as unlikely to be clinically relevant (Supplementary Table 1).

Results
Testing of the clinically well twin sister II.3 confirmed that she had inherited the c:2534 T > C p.(Leu845Ser) likely pathogenic missense variant from her father but not the c: 9499A > T p.(Ile3167Phe) variant from her mother. Pedigree, ultrasound scans, and antenatal kidney length from II.1, II.2, and II.3 are shown in Figure 1. Both had presented at birth with bilateral renal cysts, enlarged kidneys, and hypertension. Their kidney function was noted to be mildly reduced at ages 3 and 5, respectively, but the exact values were not recorded. Both siblings required dialysis and transplantation aged 17 and 22, respectively. There was a paternal history of ADPKD, although of a very different severity, with the affected father starting dialysis followed by kidney transplantation aged 66. DNA sequence analysis in 2 affected paternal cousins identified the c.10071dup p.(Thr3358Hisfs * 32) pathogenic variant with no evidence of the p.(Ile3167Phe) variant; therefore, the variants are highly likely to be in trans ( Figure 2). Nonetheless, since no DNA was available from the affected father, the rare possibility that the p.(Ile3167Phe) variant was a de novo event on the same allele as c.10071dup p.(Thr3358Hisfs * 32) could not be excluded.
MinION nanopore long-read sequencing was performed on 100 K.1a and 1b to confirm phase (Supplementary Table 2). Parental haplotypes based on the reference/ nonreference nucleotide at position chr16:2100465 (c.9499) were established based on the proportion of reads at the downstream duplication site (chr16:2097964, c.10071) that contained an insertion was recorded (Supplementary Figure 1). While the identification of insertion-deletion variants from long-read nanopore datasets remains challenging, our data is consistent with the c:9499A > T p.(Ile3167Phe) and c.10071dup variants being arranged in trans, i.e., on different parental haplotypes. The presence of~6% of c.9499 T (chr16:2100465A) reads containing apparent insertions is likely due to a combination of (i) chimeric read formation (strand switching) during long-range PCR enrichment of the target locus and (ii) a reduction in base-calling accuracy caused by the duplicated base occurring at the end of a string of four G nucleotides. Clinical details from her clinician stated very severe PKD at referral with an eGFR of 22 ml/ min/1.73m 2 , diagnosis at age 12, and approaching kidney failure at age 32. There was, however, insufficient information on her relatives outside the UK, and no DNA was available to confirm the phase.

3.2.3.
Other Individuals with the p.(Ile3167Phe) Variant in 100 K. A second search of the 100 K data, where the PKD1 gene was not included in the relevant clinical panel analysis, identified a total of 99 unaffected heterozygotes and 1 homozygote. The homozygote was recruited as the unaffected parent of a child with multiple congenital anomalies including macrocephaly, hearing impairment, and dysmorphism but with no kidney disease. The unaffected homozygote, aged 49, has no recorded history of renal disease, and a kidney ultrasound scan did not identify any cysts (personal communication).  Table 3). There was no evidence of chronic kidney disease (CKD) in the heterozygotes based on baseline measurements of eGFR, BP, and ACR. However, two individuals had a diagnosis of polycystic kidneys, unspecified on HES data. Both were diagnosed in the seventh decade of life or later. Neither individual had CKD based on GFR criteria alone. Without information on family history or imaging, it was not possible to determine whether this represents a true diagnosis of ADPKD or an incidental finding of acquired cysts.

Discussion
The major finding of this paper is genetic evidence that the common c:9499A > T p.(Ile3167Phe) PKD1 variant manifests as a hypomorphic variant. Data from gnomAD v2.2.1 (https:// gnomad.broadinstitute.org/variant/16-2150466-T-A?dataset= gnomad_r2_1) shows the c:9499A > T variant is present in 340/280802 alleles from multiple ethnicities, including 2 homozygotes, with the highest allele frequency of 0.21% in the European Non-Finnish population. It is found on 181/152192 alleles including 1 homozygote on gnomAD 3.1.1 (https://gnomad.broadinstitute.org/variant/16-2100465-T-A?dataset=gnomad_r3) with the highest allele frequency 0.23% (NFE). This variant fulfils the ACMG criteria BS1 (allele frequency is greater than expected for the disorder) and BS2 (observed in a healthy adult individual for a recessive (homozygous), dominant (heterozygous), or X-linked (hemizygous) disorder with full penetrance expected at an early age) and, therefore, could be classified as class 1 benign without further investigation. Based on this high population frequency, it is likely to be automatically filtered by bioinformatic pipelines designed to select rare variants for analysis. It may also be filtered from WES or WGS trio analysis where the inheritance pattern is selected as autosomal dominant or complete penetrance where it has been inherited from an unaffected parent, or it could be manually discarded by analysts without specialist knowledge of VEO-PKD and hypomorphic variants.

Human Mutation
The most commonly reported hypomorphic variant identified in more than ten published VEO-PKD families is p.(Arg3277Cys) [2][3][4][5]10], present on 44 alleles and no homozygotes in gnomAD. PKD1 p.(Arg3277Cys) is the only PKD1 hypomorphic variant with experimentally proven reduced function. Studies using an engineered mouse model showed it to be a temperature-sensitive folding/trafficking mutant with approximately 20% retained activity in a compound heterozygous Arg3277Cys/null mouse mutant, consistent with a hypomorphic effect [11]. PKD1 p.(Arg3277Cys) homozygotes have been reported to develop adult-onset PKD with kidney failure or transplantation at ages 62, 75, and 50 consistent with this dosage-dependent model [2,4]. Based on functional studies, we would assume homozygotes have~40% functional polycystin-1 (PC1) protein. However, since the reported homozygotes do not have a more severe presentation than classic ADPKD with a heterozygous null allele, the level of functional PC1 protein may be closer to 50% [12].
For p.(Ile3167Phe), two homozygotes have been reported in gnomAD (1 exome aged 55-60, 1 genome aged 40-45) although their phenotype is unknown. A search of 100 K genomes and UK Biobank identified a single p.(Ile3167Phe) homozygous individual who was recruited as the parent of a child affected with multiple congenital anomalies but with no HPO terms for cystic kidneys or kidney disease. Kidney ultrasound did not detect any cysts in this individual at the age of 49 years though no kidney function tests were available. Assuming this is correct, the absence of cysts in this homozygote could imply that the retained activity of PC1 in the p.(Ile3167Phe) homozygote is sufficient to avoid cystogenesis, i.e., >55-60%. This would imply that a compound heterozygote for a null allele and p.(Ile3167Phe) has approximately 30% retained PC1 activity ( Figure 3). Unlike p. (Arg3277Cys), however, our phenotypic information is currently limited to a single individual. In addition, more sensitive kidney imaging (e.g., MR or CT) in more individuals will be needed to exclude the presence of tiny microscopic cysts below the sensitivity of ultrasound detection.
There is currently no direct functional evidence of the pathogenicity of PKD1 p.(Ile3167Phe). At codon 3167, both isoleucine and valine amino acids are commonly found in mammals, fish, and reptiles, and this region of the signature PLAT domain is conserved down to Cioana intestinalis. In silico tools have provided inconclusive data. The p.Ile3167Phe variant has an intermediate predicted pathogenicity score of 0.459 from REVEL, likely due to the presence of Ile and Val. The Ile→Phe variant is predicted as deleterious/damaging by Provean/SIFT, respectively, and probably damaging by PolyPhen-2 HumDiv and HumVar. Ile3167 is located in the PLAT (polycystin-1, lipoxygenase, and alpha toxin) domain, which regulates PC1 trafficking to the plasma membrane. The structure is a β-sandwich, with four strands in each sheet. The p.Ser3164 residue is critical for phosphorylation and function of this domain, and several likely pathogenic variants are reported in this region including the neighbouring 3168 amino acids at codon 3168 (p.(Arg3162Leu), p.(Phe3168Leu), p.(Phe3168del), and p.(Ala3171Pro)) (https://pkdb.mayo.edu/variants). Interestingly, p.(Arg3162Cys) has previously been reported as a hypomorphic variant, inherited in trans with a pathogenic truncating HNF1β mutation in a child with early-onset ADPKD and normal parents [7]. Functional studies showed loss of in vitro Ser3164 phosphorylation and reduction of   Human Mutation surface delivery to the plasma membrane and primary cilia predicted to be due to protein misfolding and impaired surface delivery by approximately 30%-50% [13]. In silico protein modelling by Missense3D for the p.(Ile3167Phe) missense variant predicts that it could affect a phosphatidylserine (PS) binding pocket in PLAT and, therefore, affect membrane association (Figure 1(c)) [5,13]. We previously reported the p.(Ile3167Phe) variant in two VEO-PKD neonates with severe disease [5]. Two additional VEO-PKD families (one with 2 affected siblings) have since been reported in the literature, both with p.(Ile3167Phe) inherited from the unaffected parent and a truncating variant inherited from the affected parent [10,14]. All 10 cases from 7 families are summarised in Table 1. Notably, while all cases show paediatric onset and a severe phenotype compared to classic adult-onset ADPKD, there is a significant variation in disease severity with cases ranging from neonatal demise to kidney failure aged 32. This variation could be due to unknown in cis or in trans genetic modifiers.

Human Mutation
Variant classification for hypomorphic variants is challenging with no guidelines currently in place and several ACMG codes not being applicable or difficult to apply [5]. Not surprisingly, p.(Ile3167Phe) has been reported on Clin-Var as likely benign (×2) or as a variant of uncertain significance (VUS × 4) (ClinVar variant ID: 440135). Guidelines for low penetrance and risk alleles using a modified version of the ACMG classes recommend the terms uncertain risk allele, likely risk allele, and established risk allele depending on the number of established case-control studies or a metaanalysis confirming a significant odds ratio [15]. Functional studies are not required for established risk alleles due to the small effect that may not be detectable in many assays. We propose the use of equivalent terminology for hypomorphic variants: uncertain hypomorphic variant, likely hypomorphic variant, and established hypomorphic variants depending on the number of reported cases and weight of evidence. Casecontrol studies for rare hypomorphic variants are not feasible. Functional studies may be challenging for some hypomorphic variants depending on the sensitivity of the assay to detect partial defects. Based on this terminology, we suggest that p.(Arg3277Cys) is an established hypomorphic variant, and p.(Ile3167Phe) is a likely hypomorphic variant.
PKD1 p.(Ile3167Phe) is now the second most common hypomorphic variant reported in VEO-PKD. Based on gno-mAD frequency alone, it is likely to be excluded from analysis by automated variant interpretation pipelines. Care should be taken to ensure this variant is analysed in all VEO cases due to the high recurrence risk in the affected families.

Data Availability
The genetic data used to support the findings of this study are included within the article.

Ethical Approval
Individual patient consent was not sought due to the retrospective nature of this study. All patient data was deidenti-fied. Access to deidentified patient data in the 100,000 Genomes Project and UK Biobank was approved by the project funders.