All individuals reported in this study provided consent to participate in a study to define the genetic cause of their family diagnosis. Institutional ethical approval for the study was granted (United Kingdom; University of Manchester [06138] and National Research Ethics Service Northwest, Greater Manchester Central ethics committee [06/Q1406/52 and 11/NW/0021]). Where a clinical diagnosis of UFS was made prior to genetic testing a targeted approach of sequencing HPSE2 and LRIG2 was employed. Where there was clinical uncertainty, but UFS lay within the differential diagnosis a broader candidate gene or unbiased exome approach was employed. In Families 1 and 4, Sanger sequencing was undertaken for all coding exons of HPSE2 and LRIG2. Primers for amplification of exons and exon-intron boundaries of HPSE2 and LRIG2 were designed with Primer3Plus. For HPSE2 NM_021828.4 for exons 1 to 12 and NM_001166246.1 for transcript variant 4 alternative exon 12 (exon12b) was used (details available on request). For LRIG2 NM_014813 for exons 1–18 was sequenced (details available on request). Sanger sequencing was performed using the BigDye Terminator v3.1 kit (Life Technologies, CA, United States) according to manufacturer’s instructions and resolved on an ABI3730 sequencer (Life Technologies, CA, United States). Genotyping for HPSE2 and LRIG2 variants was undertaken by sequencing the relevant amplicons in other family members. In Family 2, genetic testing was performed by a TruSight One capture kit (Illumina) using Nextera rapid capture for 4813 genes that were considered clinically relevant at the time of the design. This was sequenced on a NextSeq550 (Illumina) with 2 × 150 bp paired-end reads. The alignment to GRCh37 was performed on NextGene (SoftGenetics, v2.4.1) with the in-built copy number variation (CNV) tool for CNV detection. Bioinformatic analysis of 57 genes (Supplementary Data) associated with urinary tract malformations was performed. In Family 2, confirmation of the CNV detected and cascade testing of family members were undertaken on an Agilent SurePrint G3 Human whole genome microarray. In Family 3, a trio analysis with a CytoScan HD single nucleotide polymorphism array was undertaken (ThermoFisher Scientific). Data were processed and analysed using NxClinical v5.1 software (BioDiscovery, CA, United States). Results were confirmed visually in whole exome sequencing data. In short, capture was performed using the SureSelect Human All Exon V7 capture kit (Agilent) according to manufacturer’s instructions and subsequent sequencing was performed on a NovaSeq 6000 sequencing system (Illumina). Mapping was performed using an in-house GATK-based pipeline and resulting data was visualized using the Integrative Genomics Viewer (IGV, Broad Institute, CA, United States) (Robinson et al., 2017).

FreestyleTM HEK293-F cells (Thermo Scientific) were transiently transfected in FreestyleTM 293 expression medium (Thermo Scientific) in duplicate with either wild-type pcDNA3: HPSE2c myc, or the myc-tagged p.Asn543Ile (Mahmood et al., 2012) or p.Pro140Arg (current paper) variant constructs. Three days after transfection at 37°C shaking at 130 rpm, the cultures were split and heparin (10 μg/ml) was added to one set and left to grow for a further 24 h with shaking at 37°C. Heparin is known to bind wild type heparanase-2 and in cultured cells adding heparin to the media will sequester heparanase-2 protein that was associated with the cell surface (Levy-Adam et al., 2010; McKenzie, 2020). The conditioned media was clarified and concentrated 10-fold using a vivaspin 5 KDa concentrator. Ten µg of cell lysate protein per lane was used for western blotting. β-actin was used as a cellular loading control. For the supernatant lane loadings equal amounts of cells were quantified using a haemocytometer and the clarified media was collected next day for blotting. All samples were concentrated to the same volume to standardise before SDS analysis and blotting. RIPA lysis buffer was added to the cell pellets on ice and left for 30 min. The lysates were sonicated (30% setting for 30 s on ice) and then centrifuged at 15,000 g for 30 min at 4°C. Samples were removed and mixed with 2 x SDS Laemmlli buffer containing 2-mercaptoethanol and heated for ^95oC for 5 min. Samples were resolved on a 4–20% SDS PAGE gel and blotted onto polyvinylidene fluoride membranes. Membranes were blocked in PBS-T, 5% skimmed milk for 1 h and then incubated overnight with anti-myc antibody (Sigma) at 4oC. Next day, membranes were washed with PBS-T, milk and incubated with secondary mouse anti-myc antibody for 1 h in PBS-T milk. Blots were finally washed in PBS-T and incubated with ECL reagents (GE). Chemiluminescence was detected using the Syngene western blot system.

Human embryonic tissues, collected after maternal consent and with ethical approval (REC18/NE/0290), were sourced from the Medical Research Council and WellcomeTrust Human Developmental Biology Resource (https://www.hdbr.org/). Seven week embryonic tissues were fixed, paraffin embedded, and sectioned as described (Lopes et al., 2019) and serial sections were immunostained with one of the following primary antibodies: rabbit anti-heparanase-2 (1:200; custom made and raised against an epitope starting at amino acid 82) (Roberts et al., 2014); rabbit anti-LRIG2 (1:200; AP13821b; Abgent); or chicken anti-β3-tubulin (1:400; AB9354; Millipore). The primary antibodies were detected with secondary antibody and signals generated with a peroxidase-based system, as described (Lopes et al., 2019).

The proband (II:1) is a 20 year old male from a consanguineous Turkish family (Figure 1). His clinical course featured recurrent urinary tract infections (UTIs), and VUR was diagnosed when he was 7 years old. He was initially treated with intermittent bladder catheterization and anticholinergic medication. His small capacity bladder was surgically augmented when he was 12 years old. Subsequently, he reached end-stage renal failure and received a kidney transplant at the age 17 years. UTIs persisted and a video-urodynamic study demonstrated VUR into both his native kidneys and into the transplanted kidney. He has a striking grimace upon attempting to smile (Supplementary Video S1). The proband’s parents are clinically unaffected. Sequencing of LRIG2 revealed no significant variants. Sequencing of HPSE2 in the proband revealed a homozygous pathogenic variant, as defined under ACMG guidelines (PVS1, PM2_Mod, PS4_Supp) (Richards et al., 2015) c.429T>A, p.(Tyr143*), which has previously been reported in a family affected by UFS (Stuart et al., 2015). Segregation analysis revealed that his parents, who were each clinically unaffected, were each heterozygous for the variant, as were two of the proband’s clinically unaffected brothers (II:2 and II:3) where DNA samples were available. A sample from the other unaffected brother (II:4) was not available.

The proband (II:2), is the middle of three sisters born to a non-consanguineous couple (Figure 2) from the Torres Strait Islands, Australia. The pregnancy leading to her birth, and the birth itself, were uneventful. She has a grimace typical of UFS. She presented aged 3 years with recurrent UTIs. Investigations revealed that she had a thickened bladder wall, VUR and hydronephrosis. She had a bladder stoma fashioned at that point. She had an ileocystoplasty bladder augmentation in her early teenage years and she currently self-catheterises. Investigations in her early teenage years revealed persistent thickening of the bladder wall, with bladder diverticulae and a large residual volume after micturition, together with VUR, bilateral hydronephrosis and kidney cortical scarring. Her blood creatinine was elevated at 106 μmol/L (upper normal for age 82 μmol/L). II:3 is her younger sister who presented with a UTI aged 4 years. Investigations revealed bilateral VUR with a thickened neurogenic bladder. She currently self-catheterises. Like II:2, II:3 has a grimace typical of UFS, and she also had a bladder that failed to void completely, with a residual of 117 ml, with a thickened wall and bilateral VUR. II:3’s blood creatinine was elevated at 63 μmol/L (upper normal for age 58 μmol/L). The eldest sister, II:1, had had several UTIs as a child but did not report other urinary symptoms. Ultrasonography revealed a normal bladder capacity (205 ml) and a minimal residual volume (2 ml) after micturition, and the upper urinary tract also appeared normal. She did not have an overt grimace on attempting to smile. In the proband in Family 2 next generation sequencing of a panel of genes associated with urinary tract malformations identified a potential homozygous triplication of exons 8 and 9 in HPSE2, p.(Val367_Pro [3]). This variant is defined as variant of uncertain significance (4F, 5D) using the ACMG and ClinGen guidelines (Rooney Riggs et al., 2020). This was confirmed by a high resolution microarray (arr [GRCh37] 10q24.2 (100365741_100390995)x6. Segregation analysis by microarray revealed that the two sisters were also homozygous for the triplication and the parents were heterozygous. Sequencing of LRIG2 revealed no significant variants in the proband.

The proband was the son of consanguineous first cousin Turkish parents (Figure 3) who had been referred with a tentative diagnosis of facial nerve paresis. Otherwise, he was fit and well with no developmental problems. He was toilet trained at 3 years of age, with only occasional episodes of enuresis. There was no history of UTIs and a bladder ultrasound at 3 years of age was normal, with no residual urine volume after micturition. Later in childhood, however, there was a history of mild dribbling urinary incontinence during the day, and there was a significant volume (70–158 ml) of urine after micturition. He defecates small amounts, four to five times a day. SNP microarray of the proband revealed a potential homozygous deletion of exon 4 in HPSE2, arr [GRCh37] 10q24.2 (100501035_100514963)x0. with a minimal size of 13.9 kb. Subsequent exome sequencing confirmed this finding. His parents had no urinary tract symptoms or facial signs, and they were both heterozygous for the deletion. The deletion of exon 4 is in frame, leading to the predicted loss of 58 amino acids and formation of a truncated protein p.(Ala_Asn261del). This variant is defined as a variant of uncertain significance (2E, 5H) using the ACMG and ClinGen guidelines (Rooney Riggs et al., 2020). Upon further assessment, the proband’s abnormal smile, with downturned corners of the mouth upon smiling, was considered consistent with UFS. Sequencing of LRIG2 revealed no significant variants in the proband.

The proband (II:1) was female and the first child of healthy non-consanguineous German parents (Figure 4). She presented in early childhood with recurrent UTIs associated with poor bladder emptying and she was initially treated with intermittent bladder catheterization and anticholinergics. Magnetic resonance imaging revealed normal spinal anatomy. She underwent a vesicostomy aged 4 years and assessment in her seventh year led to a diagnosis of neurogenic bladder with detrusor sphincter dyssynergia. Her course was complicated by VUR and damage to her left kidney so that, as assessed by isotope scanning when 12 years old, it contributed only 22% of total kidney function. The next two siblings (II-2 and II-3) were also female. They had no urinary symptoms, but II-2 died at 18 years from epilepsy. The next sibling (II:4), another girl, presented in utero with a thickened bladder wall reported on an anomaly screening ultrasound scan. After birth, a micturating cystourethrogram (MCU) was abnormal, consistent with a neurogenic bladder. When assessed aged 17 years, urodynamics revealed a low compliance bladder with high intravesical pressures of 70 cm H₂O after filling, rising to 180 cm H₂O during micturition, the latter over three times the upper limit of normal (Lemack et al., 2002). MCU revealed incomplete bladder emptying with 130 ml residual urine. Ultrasonography revealed bilateral hydronephrosis and her overall renal function was impaired with a blood creatinine of 114.4 μmol/L (upper normal 105.6 μmol/L). The fifth sibling (II:5) was a healthy female. The sixth and final sibling (II:6) was a boy. He was investigated for nocturnal enuresis aged 10 years when urodynamics revealed a low compliance and low-capacity bladder with staccato micturition. Ultrasonography showed normal kidneys and blood tests showed normal renal function. He commenced intermittent self-catheterization and anticholinergic therapy. A grimace when smiling was noted and the diagnosis of UFS was considered; in further inspection the two other siblings with urinary tract disease were also noted to have a grimace. Sequencing of LRIG2 revealed no significant variants. Sanger sequencing of HPSE2 identified two variants, c.419C>G, p.Pro140Arg and c.1099-1G>A in the three affected children (II:1, II:4, and II:6). Two sisters unaffected by urinary disease (II:2 and II:5) were carriers for a single variant, and the third clinically unaffected sister (II:3) was wild type. The parents were heterozygous carriers, confirming that the variants were on separate alleles, consistent with autosomal recessive inheritance. The c.1099-1G>A variant has previously been reported in a patient with UFS (Stuart et al., 2015) and is predicted to result in the loss of a splice acceptor within exon 8, so introducing a premature stop codon. This variant was classified as pathogenic (PVS1, PM2_Mod, PS4_Supp, PP1_Supp, PM3_Mod) by ACMG guidelines (Richards et al., 2015; Ellard et al., 2020). Before this report, however, the c.419C>G variant has not been associated with UFS. This variant occurs at a minor allele frequency of 0.00001 (i.e., 2 in 152,996 alleles in the gnomAD database) (Karczewski et al., 2020). The c.419C>G variant is a missense change, predicted to result in the substitution of a proline for an arginine residue, p.Pro140Arg. This proline residue is conserved in heparanase-2 between humans to zebrafish. In silico tools of pathogenicity gave conflicting predictions: the variant was disease-causing as assessed with Mutation Taster (http://www.mutationtaster.org/); benign on Polyphen-2 (http://genetics.bwh.harvard.edu/pph2/); but tolerated using the Sorting Intolerant From Tolerant (SIFT) tool (http://sift.bii.a-star.edu.sg/); and with a Combined Annotation Dependent Depletion (CADD) score of 22.4 (a score of >20 predicts that this variant in the top 1% of most likely deleterious variants) (Rentzsch et al., 2019). This variant is classified as a VUS (PM2_Mod, PP3, PM3_Mod, PP1_Supp) according to ACMG guidelines (Richards et al., 2015; Ellard et al., 2020).

In order to learn more about missense variant discovered in Family 4, we undertook further analyses. First, we expressed the p.Pro140Arg variant in cultured cells, and compared it with the p.(Asn543Ile) variant previously reported in a family with UFS (Mahmood et al., 2012). Each protein was expressed as a myc-tagged protein in HEK293 cells (Figure 5). Each variant protein was detected in cell lysates, as for cells transfected with a wild-type HPSE2 construct, also myc-tagged. As expected for a secreted protein, wild-type heparanase-2 was also detected in the conditioned media. In this context, however, there was a contrast between the two missense proteins, with only p.Pro140Arg being detected in the media. In parallel experiments, heparin, a molecule that binds wild-type heparanase-2 (McKenzie, 2020), was added to the media. Again, the myc-tagged p.Asn543Ile variant was not detected in the supernatant. In blots of both the wild-type and the p.Pro140Arg variant, the addition of heparin to the media appeared to increase the intensities of the detected protein bands in the supernatant compared with the intensities of the bands that were cell associated.

Second, we undertook protein modelling. No heparanase-2 structural information exists. The closest structural homologue is human heparanase-1 (encoded by HPSE) with approximately 50% sequence identity at the protein level (Wu et al., 2015). The heparanase-1 precursor is arranged into signal peptide-small subunit-proteolytic linker-large subunit. In the mature protein, the linker is cleaved off to form a ‘heterodimer’ of small and large subunits. The residue Asn543 in heparanase-2 (Mahmood et al., 2012) corresponds to Asn496 in heparanase-1 (using precursor numbering). This Asn position is strictly invariant in the heparanase-1/heparanase-2 orthologues and is located in the large subunit as part of the beta-sandwich domain. Asn496 in heparanase-1 forms main-chain hydrogen bonds with a nearby beta-strand to maintain the sandwich domain. These hydrogen bonds are likely conserved in heparanase-2, as suggested by the Alphafold predicted model (Tunyasuvunakool et al., 2021) and the p.Asn543Ile substitution in heparanase-2 is predicted to interfere with these hydrogen bonds. The effect of altering the residue Pro140 in heparanase-2 is more complex to rationalise (Supplementary Figure S1). Pro140 (underlined and cyan shaded in Supplementary Figure S1) and the surrounding residues are not well conserved with heparanase-1 in terms of sequence and structure. The equivalent region in the latter is part of the proteolytic linker (red letters) that gets cleaved off during activation to form the mature protein. It is therefore difficult to predict a possible effect of this variant in the absence of the heparanase-2 structure.

We studied histology sections of two human embryos, each of 7 weeks gestation. At this stage, the hindgut has separated from the urogenital sinus, the latter being the precursor of the urinary bladder (Jenkins et al., 2007). The immunostaining patterns were similar in each embryo. Heparanase-2 was immunodetected in loosely aggregated collections of cells flanking the hindgut (Figure 6), and also in the in the primitive urethra and in the genital tubercule. In transverse sections more cranial to these, heparanase-2 was detected in cords on cells flanking the hindgut and near the embryonic ureter. Serial sections revealed that these structures immunostained for the neural marker β3-tubulin as well as for LRIG2. The latter protein was also detected in other cells in this region, including loosely packed stromal-like cells.