Lrig2 and Hpse2 , mutated in urofacial syndrome, pattern nerves in the urinary bladder

Mutations in leucine-rich-repeats and immunoglobulin-like-domains 2 (LRIG2) or in heparanase 2 (HPSE2) cause urofacial syndrome, a devastating autosomal recessive disease of functional bladder outlet obstruction. It has been speculated that urofacial syndrome has a neural basis, but it is unknown whether defects in urinary bladder innervation are present. We hypothesized that urofacial syndrome features a peripheral neuropathy of the bladder. Mice with homozygous targeted Lrig2 mutations had urinary defects resembling those found in urofacial syndrome. There was no anatomical blockage of the out ﬂ ow tract, consistent with a functional bladder outlet obstruction. Transcriptome analysis revealed differential expression of 12 known transcripts in addition to Lrig2 , including 8 with established roles in neurobiology. Mice with homozygous mutations in either Lrig2 or Hpse2 had increased nerve density within the body of the urinary bladder and decreased nerve density around the urinary out ﬂ ow tract. In a sample of 155 children with chronic kidney disease and urinary symptoms, we discovered novel homozygous missense LRIG2 variants that were predicted to be pathogenic in 2 individuals with non-syndromic bladder outlet obstruction. These observations provide evidence that a peripheral neuropathy is central to the pathobiology of functional bladder outlet obstruction in urofacial syndrome, and emphasize the importance of LRIG2 and heparanase 2 for nerve patterning in the urinary tract.

Mutations in leucine-rich-repeats and immunoglobulin-likedomains 2 (LRIG2) or in heparanase 2 (HPSE2) cause urofacial syndrome, a devastating autosomal recessive disease of functional bladder outlet obstruction. It has been speculated that urofacial syndrome has a neural basis, but it is unknown whether defects in urinary bladder innervation are present. We hypothesized that urofacial syndrome features a peripheral neuropathy of the bladder. Mice with homozygous targeted Lrig2 mutations had urinary defects resembling those found in urofacial syndrome. There was no anatomical blockage of the outflow tract, consistent with a functional bladder outlet obstruction. Transcriptome analysis revealed differential expression of 12 known transcripts in addition to Lrig2, including 8 with established roles in neurobiology. Mice with homozygous mutations in either Lrig2 or Hpse2 had increased nerve density within the body of the urinary bladder and decreased nerve density around the urinary outflow tract. In a sample of 155 children with chronic kidney disease and urinary symptoms, we discovered novel homozygous missense LRIG2 variants that were predicted to be pathogenic in 2 individuals with non-syndromic bladder outlet obstruction. These observations provide evidence that a peripheral neuropathy is central to the pathobiology of functional bladder outlet obstruction in urofacial syndrome, and emphasize the importance of LRIG2 and heparanase 2 for nerve patterning in the urinary tract.
B ladders undergo filling and voiding cycles controlled by autonomic nerves. 2,3 Preganglionic autonomic neurons have cell bodies in the spinal cord, and their axons synapse within ganglia with postganglionic neurons whose axons innervate bladder muscles. Sympathetic noradrenergic neurons elicit outflow tract contraction during urinary storage, whereas during voiding, parasympathetic cholinergic motor neurons drive detrusor smooth muscle contraction and neuronal nitric oxide synthase (nNOS)-expressing neurons dilate the outflow tract. 4 Urofacial, or Ochoa, syndrome (UFS) is an autosomal-recessive disease with around 150 families reported worldwide. 5,6 In persons with UFS, detrusor smooth muscle contracts against a poorly dilated outflow tract, causing functional bladder outlet obstruction. 5,6 This dyssynergia causes dribbling incontinence and residual urine, leading to a risk of ascending bacterial infection and kidney failure. The UFS bladder is a "nonneurogenic neurogenic bladder" resembling that caused by damaged bladder nerves, yet such gross lesions are absent in persons with UFS. 5,6 Persons with UFS typically have a facial expression "as if in pain or sadness when they tried to smile or laugh." 5 Ochoa,5 who first delineated the syndrome, and others 7 speculated that there is a neuropathic basis for UFS.
Some families with UFS have biallelic variants in heparanase 2 (HPSE2) 8-10 encoding heparanase 2, an inhibitor of the enzyme activity of the classical heparanase protein, 11 hereafter called "heparanase." Homozygous Hpse2 gene-trap mice have incomplete bladder emptying, 10,12 phenocopying UFS. Whether Hpse2 mutant mouse bladders have innervation defects is unknown. Other families with UFS lack HPSE2 mutations; some instead have biallelic variants of leucine-rich repeats and Ig-like domains 2 (LRIG2). 13,14 The most studied of the 3 mammalian Lrig proteins is Lrig1, a tumor suppressor that downregulates growth factor signaling. [15][16][17] Belonging to the family of leucine-rich repeat-containing proteins, LRIG2 comprises multiple leucine-rich repeats, 3 Ig-like domains, a transmembrane segment, and a long cytosolic portion. 18 Homozygous Lrig2 mutant mice are protected from growth factor-induced gliomas, but their bladders were not studied. 19 Because heparanase 2 and Lrig2 are immunodetected in fetal mouse 10 and human 13 bladder nerves, these proteins are well placed to affect nerve patterning. We hypothesized that UFS features a peripheral neuropathy of the bladder. We studied gene-targeted Lrig2 mice, demonstrating that they have defective urination and abnormal patterns of bladder nerves. We went on to show that Hpse2 mutant mice have similar defects. To date, only a few families with LRIG2 mutations have been reported. 13,14 Accordingly, we studied persons with nonsyndromic bladder outlet obstruction and report homozygous LRIG2 likely pathogenic missense variants in a subset.

Lrig2 mutant mice have abnormal urination
We used mice with a targeted deletion of Lrig2 exon 12, 19 which introduces a frameshift preceding the region encoding the transmembrane domain, generating multiple stop codons. Mice had been maintained on a C57BL6 background for more than 10 generations, and mating heterozygotes generated offspring for study. Upon analyzing the bladder and outflow tract unit, quantitative reverse transcription-polymerase chain reaction (qRT-PCR) detected Lrig2 in wild-type neonatal (first postnatal day) samples. In Lrig2 -/littermates, transcripts were undetectable using exon 12 primers and were significantly diminished (P ¼ 0.0163) to one third of wild-type levels using exon 17 primers (Figure 1a). Using an antibody to a cytosolic Lrig2 epitope absent in Lrig1 and Lrig3, 19 Lrig2 was detected as a 125-kDa band in Western blots of wild-type bladders (Figure 1b). Lrig2 was absent in Lrig2 -/littermates, with intermediate levels in Lrig2 þ/samples (Figure 1b and c). Neonatal mice of the 3 genotypes had similar weights. At 2 weeks, Lrig2 -/mice had gained less weight than the other genotypes, being on average 14% lighter than Lrig2 þ/þ or Lrig2 þ/littermates. Neonatal necropsy bladders, examined within 5 minutes of cervical dislocation, were classified as empty or containing urine by visual inspection. Approximately 90% of bladders contained urine in all 3 genotypes (Figure 1d). At 2 weeks (Figure 1e), 67% of Lrig2 -/mouse bladders contained urine, a significant (P ¼ 0.014) increase versus 21% of wild-type bladders. The frequency of heterozygous bladders containing urine was similar to wild types. These and adult mice were studied by the voided stain on paper (VSOP) technique 20 where they urinate onto filter paper, with numbers and sizes of spots measured. Wild-type and heterozygous Lrig2 mice voided once or twice per hour, whereas homozygous Lrig2 mutant mice voided up to 20 times per hour, producing smaller spots than Lrig2 þ/þ or Lrig2 þ/mice ( Figure 1f). Quantification (Figure 1g) showed that Lrig2 -/mice voided more frequently (P ¼ 0.001), and their average volume per void was lower (P ¼ 0.0124) compared with control subjects. Total urine voided per time was similar in control (Lrig2 þ/þ and Lrig2 þ/-) mice and Lrig2 -/mice. These aberrant urination patterns, observed from 14 days to adulthood, resemble those in UFS. 5,6 UFS uropathy is not sex limited, 5 and both male and female Lrig2 -/mice had abnormal urination. Ink was injected into the bladder lumen at autopsy, and upon gentle palpation, ink exited the urethra. Upon histologic examination (Supplementary Figure S1A-F), bladder outflows were patent in neonatal, juvenile, and adult Lrig2 -/mice. Histology of the bladder body was similar in wild types and mutants at 2 weeks (Supplementary Figure S1G-L). Adult Lrig2 -/bladders appeared larger than wild types (Figure 2a). Histology with picrosirius red, a collagen reactive dye, revealed prominent staining in mutant lamina propria and detrusor (Figure 2b and c). Lrig2 -/kidneys showed no scarring (Figure 2d and e). After draining urine from bladders, mutant organs were significantly heavier than those of control subjects (Figure 2f). Immunostaining adult bladders for Ki67 (Supplementary Figure S2) revealed similar proportions of proliferating nuclei in mutants and control subjects (Figure 2g-i). Detrusor nuclei number was similar (P ¼ 0. 36) in control mice (mean AE SEM, 431 AE 88/mm 2 , n ¼ 4) and mutants (453 AE 53/mm 2 , n ¼ 4).

Transcriptome analyses of Lrig2 mutants
To investigate the molecular milieu before urination defects became established, we undertook RNA sequencing of neonatal bladders and their attached outflow tracts harvested  b a s i c r e s e a r c h from 5 Lrig2 þ/þ and Lrig2 -/littermate pairs. We used females because the male tract includes the prostatic rudiment, with a transcriptome that could complicate interpretation of resulting arrays. The data set is available from ArrayExpress (E-MTAB-6089). Unsupervised hierarchical clustering distinguished between homozygous Lrig2 mutants and wild types (Figure 3a). Thirteen significantly differentially regulated known transcripts were identified after adjusting for multiple comparisons. As expected, Lrig2 was one of the downregulated transcripts. Gene ontology analysis (Figure 3b) revealed enrichment of cell differentiation and system development terms. The only tissuespecific enriched term was nervous system development. Several of these transcripts encode proteins implicated in neurobiology, as detailed in the Discussion section.

UFS proteins in bladder nerves
The RNA sequencing results were generally consistent with the hypothesis that the UFS bladder has a neural component, and this led us to examine nerve patterns in Lrig2 -/bladders. In mice, the 2 pelvic ganglia flank the bladder outflow. 3, 21 We studied wild-type mice two thirds through gestation, when these ganglia are composed of loose aggregates 10 of neural crest-derived cells 22 and the bladder rudiment has acquired well-defined body and outflow tract. Lrig2 in situ hybridization demonstrated prominent signals at the junction of the body and outflow tract (Figure 4a), the sites of pelvic ganglia immunostaining for peripherin, a neural intermediate filament ( Figure 4b). Embryonic pelvic ganglia were explanted and cultured, adapting published protocols. 23,24 During the next day, processes and cell bodies emanated from cell aggregates ( Figure 4c). The processes immunostained for the neuronal microtubule protein b3-tubulin ( Figure 4d) and migrating cell bodies immunostained for S100B, a calciumbinding protein expressed by autonomic ganglia glia. 25 Lrig2 was immunodetected in cell bodies within explanted aggregates; it was immunodetected faintly in the emerged neurites and more prominently in emerged glia-like cells ( Figure 4e). Heparanase 2 was immunodetected in cell bodies, neurites, and glia-like cells ( Figure 4f). Heparanase, the classical heparanase, 11 was immunodetected in neurite-like structures ( Figure 4g). Upon histologic examination, pelvic ganglia of neonatal wild-type mice immunostained for Lrig2, heparanase 2, and heparanase ( Figure 4h). Immunohistochemistry for Wnt4 and Capza1, encoded by 2 transcripts highlighted in unsupervised hierarchical clustering (Figure 3), showed that they were detectable in neonatal pelvic ganglia (Supplementary Figure S3). In wild-type neonatal ganglia, 9% of nuclei were Ki67 þ . Pelvic ganglia in Lrig2 -/littermates had a similar (P ¼ 0.190) proportion of positive nuclei (n ¼ 4 mice, each genotype).
Aberrant patterns of nerves in bladders of mice mutant for Lrig2 or Hpse2 Bladders harvested from mice in the second week after birth were cut open, generating butterfly shaped sheets, and wholemount immunostained for peripherin. In Lrig2 mutant and wild-type littermate bladders, pelvic ganglia expressed peripherin, as did nerves at the junction of the outflow and the bladder base and nerves in the body ( Figure 5). Representative bladders are shown in increasing magnifications in Figure 5a, b, and c. We noted 2 characteristic changes in mutants versus wild types. First, large nerve bundles in the bladder base (black asterisks, Figure 5a) continuous with the pelvic ganglia appeared to be less compactly bundled. Second, there appeared to be an increased density of smaller nerves in the bladder body ( Figure 5b and c). Upon quantification by observers blinded to genotypes, a significant (P ¼ 0.048) increase of peripherinþ structures was confirmed in Lrig2 mutant bladders ( Figure 5d). Using VSOP analyses (Figure 6a), Hpse2 homozygous gene-trap mice 10 produced significantly more voids per unit time (P ¼ 0.003) than did control (wild-type and heterozygous) littermates ( Figure 6b). Mutant void volumes were significantly (P ¼ 0.012) smaller than those of control mice (Figure 6c), but with no significant difference (P ¼ 0.423) in total volumes urinated per unit time ( Figure 6d). These aberrant urination patterns resemble those reported in another Hpse2 gene trap mouse. 12 Bladder sheet immunostaining demonstrated bundles of peripherinþ nerves in bladder body that appeared more prominent in Hpse2 mutants than in control bladders ( Figure 6e). Quantification showed an increase (P ¼ 0.004) of peripherinþ structures in Hpse2 -/versus Hpse2 þ/þ littermates ( Figure 6f). Hpse2 and Lrig2 lines were interbred, generating transheterozygotes-that is, mice carrying one mutant allele at each locus. Upon being studied by VSOP (n ¼ 6 adults, a cohort containing males and females), their urination patterns appeared similar to wild-type or single heterozygous littermates.
Nerves around the outflow tract were analyzed in the second week after birth ( Figure 7a). In wild types, peripherin immunostaining revealed ganglia flanking the outflow tract, with prominent nerve bundles between them, lying in the = Figure 1 | (continued) homozygous mice nearly always contained urine (P ¼ 0.100, Fisher exact test). (e) Frequency of bladders containing urine at necropsy at 2 weeks after birth. Lrig2 -/bladders contained urine significantly more often than did wild types (P ¼ 0.014, Fisher exact test), whereas heterozygous bladders were similar to wild types (P ¼ 0.462, Fisher exact test). (f) Urination patterns in weaned mice as assessed by voided stain on paper. The wild-type and heterozygous mice produced a few large voids in the observation period, whereas the Lrig2 -/mouse produced many smaller voids in the same period. (g) Left panel: numbers of voids per unit time, with a significant increase in homozygous mutants versus control subjects (P ¼ 0.001). Middle panel: volumes of urine per void, with a significant decrease in homozygous mutants versus control subjects (P ¼ 0.012). Right panel: total volumes of urine produced per unit time, with no significant difference (P ¼ 0.832) between control (comprising wild-type and heterozygous mice) and Lrig2 -/littermates. Data are mean AE SEM, with values factored for weights of individual mice, with comparisons using 2-tailed unpaired Student t-test. Note that increased frequency of urination was observed in both male (boxes) and female (circles) homozygous mutant mice. To optimize viewing of this image, please see the online version of this article at www.kidney-international.org.  bladders, weights of mutant organs (n ¼ 6, age between 2 and 12 months), shown here factored for body weight, were significantly (P ¼ 0.013) greater than control subjects (n ¼ 9, age 2-12 months). (g-i) Immunostaining adult bladders for Ki67 revealed no significant differences in percent proportions of positive nuclei in urothelial (g; P ¼ 0.855), lamina propria (h; P ¼ 0.500), and detrusor (i; P ¼ 0.804) layers of Lrig2 -/homozygous mutants (n ¼ 4) and control subjects (n ¼ 4). Ranges of total nuclei counted for each layer: urothelium, 2000-8000; lamina propria, 3500-10,000; and detrusor, 8000-20,000. Bars in (f-i) are means AE SEMs. To optimize viewing of this image, please see the online version of this article at www.kidney-international.org. b a s i c r e s e a r c h dorsal side of the outflow. In Lrig2 or Hpse2 homozygous mutants, a spectrum of aberrations was observed, from disorganized to almost absent peripherinþ nerves. Next, we used markers of specific neural subtypes: anti-nNOS immunostaining to detect nitrergic parasympathetic nerves and anti-tyrosine hydroxylase immunostaining to mark sympathetic nerves. In control subjects, a subset of cell bodies in pelvic ganglia were nNOS þ , and fine bundles connecting the pelvic ganglia were also nNOS þ . nNOS þ cell bodies also were observed in ganglia of Lrig2 -/or Hpse2 -/bladders. In contrast, patterns of nNOS þ connecting nerves appeared aberrant in either mutant, sometimes with barely any positive processes detected. In wild-type outflow tracts, anti-tyrosine hydroxylase immunostaining was detected in a subset of cell bodies in ganglia and in nerve bundles spanning pelvic ganglia. In mutant outflow tracts, anti-tyrosine hydroxylase immunostainingþ bundles appeared disorganized or reduced. Numbers of peripherin-positive nerves around outflow tracts ( Figure 7b) were significantly less in either Lrig2 or Hpse2 homozygous mutants compared with control mice. In histologic sections of neonatal bladders, nNOS was immunodetected in pelvic ganglia (Supplementary Figure S3), with a small but statistically significant reduction of signal intensity in Lrig2 -/ganglia. qRT-PCR for Nos1 in neonatal bladders and outflow tracts showed significantly reduced levels in Lrig2 and Hpse2 mutants (Supplementary Figure S4). This finding was consistent with the RNA sequencing, which revealed that Lrig -/mutant bladders had average Nos1 levels of 48% of wild-type values (P ¼ 0.001, when multiple comparison testing was not used). Numbers of processes emanated from explanted pelvic ganglia fragments were similar in mutants and control subjects (Supplementary Figure S5).

LRIG2 variants in nonsyndromic bladder outlet obstruction
Putative loss-of-function LRIG2 variants have been described in UFS. 13 We also reported compound heterozygous missense LRIG2 variants (p.Arg550Cys in exon 13 and p.Ile852Phe in exon 16) in a patient with a non-neurogenic neurogenic bladder but with a normal smile. 13 To seek LRIG2 variants in other uropathies, we used Sanger sequencing for LRIG2 in children with chronic kidney failure from the 4C Study (http://www.4cstudy.org/). We focused on a subset of 155 patients with urinary tract symptoms (e.g., incontinence of urine and poor stream) and/or signs (e.g., abnormal urinary tract radiology). Two individuals were found to have homozygous missense variants in LRIG2. One harbored c.1337T>C; p.(Leu446Pro) in exon 12, absent in control databases including >120,000 persons in the Genome Aggregation Database (http://gnomad.broadinstitute. org) and predicted to be deleterious by in silico tools including Sorting Intolerant From Tolerant, Polymorphism Phenotyping version 2), and MutationTaster. This individual had a nonneurogenic neurogenic bladder but lacked UFS facial features. Another patient harbored c.1948C>T; p.(Arg650Cys) in exon 14 of LRIG2. This characteristic has been reported in 8 of 277158 alleles in the Genome Aggregation Database but never in the homozygous state and is predicted to be deleterious by Sorting Intolerant From Tolerant, Polymorphism Phenotyping version 2, and MutationTaster. The patient had congenital bladder outlet obstruction but lacked UFS facial features. It was not possible to obtain to access relatives. The 2 cases were Turkish, so we screened 85 healthy Turkish control subjects but failed to find either Leu446Pro or Arg650Cys variants in 170 chromosomes. Figure 8a shows that both Leu446 and Arg650 are conserved between human, mouse, chicken, xenopus, and zebrafish protein sequences. As detailed in the Supplementary Results text and depicted in Figure 8b-d, these missense LRIG2 variants may have an impact on the structural integrity of the LRIG2 extracellular region by affecting disulphide bonds for Leu446Pro and a salt bridge for Arg650Cys. LRIG2 exons 12, 13, 14, and 16, known to harbor missense variants in non-neurogenic neurogenic bladder, were sequenced in 192 index cases from the United Kingdom Vesicoureteric Reflux (VUR) DNA Collection. 26 They had familial primary nonsyndromic VUR, ureter malformations without anatomic bladder outlet obstruction, non-neurogenic neurogenic bladder, or extrarenal malformations. Five rare heterozygous LRIG2 variants were identified in index cases, but no biallelic variants were detected. Only 2 variants segregated with disease and neither was uniformly predicted.

DISCUSSION
In rats, neuromuscular functional maturation of detrusor smooth muscle occurs in the few first weeks after birth. 27 Our results show that bladders in newborn mice, whether wild type or Lrig2 mutant, are nearly always full at necropsy. In contrast, by 2 postnatal weeks, wild-type and Lrig2 heterozygous necropsy bladders are usually empty, whereas bladders of Lrig2 -/mice usually contain urine. Therefore, a bladder emptying mechanism is becoming established in wild-type and heterozygous mice between birth and 2 weeks of age,  but this is defective in Lrig2 -/mice. In older mice, Lrig2 -/mutants urinated more frequently than did wild types but their individual voids were of lower volume. These dysfunctional patterns resemble those in people with UFS. 5 Lrig2 -/mouse urination defects were not caused by anatomic blockage of the outflow tract, consistent with functional bladder outflow obstruction. Urination patterns of Lrig2 heterozygous mice appeared normal. Again, this finding is similar to that in human families affected by UFS because members who carry only one variant LRIG2 allele are phenotypically normal. 13 The fact that Lrig2 heterozygous mice had normal urination despite containing markedly less Lrig2 protein compared with wild-type bladders suggests that future treatments for UFS-for example, using gene therapy 28 -may be effective if they produce only a modest amount of Lrig2 protein. A previous report stated that another Lrig2 homozygous mutant line 29 had normal bladder pressure as assessed by cystometrography. 12 This mouse was created by random insertion of a gene trap into exon 11, resulting in fused Lrig2 and b-galactosidase and neomycin resistance (bgeo), with a low level of Lrig2 detectable in homozygous tissues. 29 Perhaps the different mode of targeting Lrig2, or background strain differences, account for the apparently discrepant bladder findings between this and the current study.
Mice homozygous for either Lrig2 or Hpse2 mutations had increased nerve density within the bladder body and decreased density around the outflow tract. LRIG2, heparanase 2, and heparanase itself were present in neural cells emanating from explanted embryonic pelvic ganglia and are present in intact pelvic ganglia. Our observations provide strong evidence that a peripheral neuropathy is part of the pathobiology of UFS bladder disease, whether caused by HPSE2 or LRIG2 variants. Figure 9 summarizes aberrations in the UFS bladder, linking established clinical observations with the new data.
Considering the UFS pathology, Lrig2 and heparanase 2 also may be implicated peripherally in the facial nerve biology because persons with UFS have a characteristic grimace upon smiling. 5,6 It had been hypothesized that a lesion affecting the pontine micturition center and the facial nerve nuclei, which control urination and facial expression, respectively, could account for the disparate phenotypes observed in the disease. 5 However, the fact that the micturition center and the facial nerve nucleus are discrete structures makes a single lesion theory unlikely. 30 Further studies may resolve whether molecules mutated in UFS also play roles in the facial nerve and brainstem.
Lrig2 is implicated in vestibular and auditory nerve function, 29 and it modulates optic nerve regeneration. 31 Another member of the Lrig family, Lrig1, acts as a break on primary dendrite formation and dendritic branching of hippocampal neurons. 32 Heparanase, the enzyme inhibited by heparanase 2, 11 modulates neurite outgrowth from pheochromocytoma cells, 33 and heparanase protects against axonal degeneration after sciatic nerve injury. 34 Moreover, embryonic frogs depleted of heparanase 2 have disorganized peripheral nerves. 35 The current data showing aberrant nerves in Hpse2 or Lrig2 mutant bladders therefore add to a growing body of literature that the heparanase 2 and Lrig axis modulates neural biology. A biological explanation for the abnormal patterns of nerves in mutant and wild types is not yet clear. Our current study, however, detected no increased proliferation in cells within neonatal pelvic ganglia of Lrig2 homozygous mutant mice versus wild types. We also found no significant differences in outgrowth of processes from explanted pelvic ganglia, comparing either Hpse2 or Lrig2 tissues with wild-type explants. One caveat is that it is not possible to know which neurites would have been destined to populate the bladder body or the outflow. The same applies to the nature of individual neuronal cell bodies visualized in pelvic ganglia histology. These considerations could be important because we counted more peripherin-positive nerves in the bladder body but fewer around the outlet. Bladders from adult Lrig -/mutants were bulkier than those of control subjects, with prominent picrosirius red staining upon histologic examination. Indeed, prominent lamina propria collagen was reported in Hpse2 mutant bladders in their third postnatal week. 12 These changes may be secondary to bladder outlet obstruction.
Humans who have UFS associated with mutations of either HPSE2 or LRIG2 have similar bladder phenotypes, 6 which is consistent with the hypotheses that the proteins encoded by these 2 genes maintain neural health by independent pathways or that the proteins interact, directly or indirectly, during their normal functions. Transheterozygote mice carrying one mutant allele at both the Lrig2 and Hpse2 loci did not have abnormal urination patterns, perhaps arguing against the idea that the encoded proteins directly interact. However, this observation also may be explained by a lack of haploinsufficient effects. It also should be noted that heparanase 2 may play biological roles in certain diseases unrelated to UFS, for example, in Alzheimer's disease 36 and head and neck cancer growth. 37 The same applies to Lrig2 that has been implicated in tumor biology. 15,16,19  nNOS is a well-established mediator of smooth muscle relaxation, including in the lower urinary tract. 38 Indeed, Nos1 homozygous mutant mice have incomplete bladder emptying caused by defective neurogenic relaxation of the outflow tract. 4 We observed that the outflow tracts of mutant Lrig2 or Hpse2 bladders had a reduced number of nitrergic nerves compared with control subjects; if this finding translates to humans, the mouse observations could explain why the UFS bladder outflow fails to dilate, a key defect in the syndrome. 5 We also documented an increased density of nerves in the bodies of bladders of mice with homozygous mutations of either Lrig2 or Hpse2. The major nerve trunks within the normal murine bladder body contain cholinergic parasympathetic neurons that stimulate detrusor smooth muscle contraction. 3 This feature may correlate with the detrusor overactivity typical of patients with UFS 4 and that also is reported to be present in Hpse2 homozygous mutant mice. 12 As with the long-term changes in the walls of Lrig2 -/bladders, increased nerve density in the bladder body may be secondary to chronic outlet obstruction. On the other hand, loss of Lrig2 can, in some circumstances, directly cause neuronal overgrowth, and Lrig2 knockdown causes overexuberant regeneration after optic nerve injury. 31 We note that molecules involved in nerve biology are dysregulated in bladders of newborn Lrig2 mutant mice. Upregulated transcripts included Wnt family member 4, encoding a growth factor implicated in establishing motor neurons 39 and neuromuscular junctions 40,41 ; period circadian regulator 2, encoding period circadian clock 2, implicated in micturition rhythm and expressed in bladders 42 ; and shroom family member 1, a member of a family implicated in central nervous system morphogenesis. 43 Apart from Lrig2, downregulated transcripts included G protein-coupled receptor 17, encoding a leukotriene-responsive G protein-coupled receptor implicated in neural damage 44 and glial maturation 45 ; capping actin protein of muscle Z-line subunit alpha 1, encoding an F-actin-capping protein involved in neurite extension 46 ; Wnt family member 2B that maintains a neural progenitor state in retinal progenitors during embryogenesis 47 ; contactin associated protein like 2, encoding a presynaptic protein implicated in neuronal migration 48 ; and SRY-box 10, encoding a transcription factor expressed in neural crest progenitors and glia in autonomic ganglia. 21,25,49 Further experiments are needed to define whether these molecules are functionally implicated in UFS. Lrig2 binds neogenin, a receptor for neuronal guidance molecules, and prevents neogenin shedding mediated by ADAM metallopeptidase domain 17-mediated proteolysis. 31 Transcripts encoding neogenin and ADAM metallopeptidase domain 17 were unchanged in the Lrig2 -/neonatal bladder transcriptome.
Biallelic putative null variants of LRIG2 have been reported in a subset of families with UFS. 13 Further emphasizing the role for Lrig2 in bladder biology, this study discovered novel homozygous missense LRIG2 variants in people with nonsyndromic bladder outlet obstruction. The collective evidence we have presented suggests that each variant is likely to be pathogenic. In the future, proof of pathogenicity could be provided by finding dysfunctional bladders in mutant mice engineered to be homozygous for either variant. Neither patient had facial features of UFS, something also noted in a case report of a child 13 with bladder disease harboring a homozygous LRIG2 variant (c.2125C>T) resulting in a stop codon. Therefore, evidence exists that LRIG2 variants occur in some persons with nonsyndromic bladder outlet obstruction. Conversely, we found no strong evidence that LRIG2 variants are implicated in primary nonsyndromic VUR, and we reached a similar conclusions regarding HPSE2 in this condition. 10 Congenital bladder voiding dysfunction can be also caused by variants of acta alpha 2, smooth muscle encoding alpha smooth muscle actin, 50 acta gamma 2, smooth muscle encoding g2actin, 51 cholinergic receptor muscarinic 3 encoding an acetylcholine receptor, 52 myosin heavy chain 11, encoding myosin heavy chain 11, 53 and myosin light chain kinase, encoding a kinase involved in myosin activation. 54 Whether any of these are implicated in primary nonsyndromic VUR awaits investigation.

MATERIALS AND METHODS Animals
Experiments were approved by the University of Manchester Biological Services Facility Committee and the United Kingdom Home Office (PPL 40/3550 and PAFFC1ffF), along with the Regional Ethics Committee of Umeå University (A193-12 and A1-2016). Experiments were conducted using Animal Research: Reporting of In Vivo Experiments best practice guidelines. Hpse2 mutant mice studied here were created by gene-trap insertion into intron 6. 10       26,55 The 4C study of children with chronic renal failure has been described (http://www.4c-study.org/). LRIG2 was analyzed using Sanger sequencing.  Figure S1. Neonatal and 2-week urinary bladder histology. (A-F) Bladder outflow: 5-mm paraffin sections cut transversely across the outflow comparing wild-type with leucine-rich repeats and Ig-like domains 2 (Lrig2) -/mutants at neonatal, 2-week, and adult stages. All had patent outflow tracts. Bars ¼ 100 mm. (G-L) Bladder body: 2-week bladder sections were reacted with Masson's trichrome (G and H; muscle and urothelium ¼ red and collagen ¼ light blue), or muscle immunostained (brown) for a-smooth muscle actin (I,J), or urothelium immunostained (brown) for uroplakin II (K,L). Note that staining patterns were similar in Lrig2 -/and Lrig2 þ/þ bladders. Bars ¼ 50 mm. Figure S2. Ki67 immunostaining of adult bladders. All sections were counterstained with hematoxylin. Panels show high-power views of detrusor, lamina propria, and urothelial layers in wild types and mutants. Selections of nuclei positive for Ki67 are indicated by arrows.
Bars are 50 ¼ mm. Figure S3. Immunostaining of neonatal pelvic ganglia. (A) Immunohistochemistry for Wnt family member 4 (Wnt4) and capping actin protein of muscle Z-line subunit alpha 1 (Capza1), encoded by 2 transcripts highlighted in unsupervised hierarchical clustering ( Figure 3) and neuronal nitric oxide synthase (nNOS), showed that all 3 were detectable in wild-type neonatal pelvic ganglia. The bottom left frame is a negative control with no primary antibody. All sections were counterstained with hematoxylin, with positive immunohistochemical signal in brown. (B) Quantification of nNOS immunohistochemical signals in neonatal ganglia showed a small but statistically significant reduction (Student t-test, P ¼ 0.014) in leucinerich repeats and Ig-like domains 2 (Lrig2) -/tissues (mean AE SEM in 5 control subjects ¼ 48.4 AE 10.7 arbitrary units and in 5 mutants ¼ 40.5 AE 10.4). nNOS immunostaining was semi-quantified by taking the average pixel intensity reading for the pelvic ganglia region only in ImageJ. Figure S4. Quantitative reverse transcription-polymerase chain reaction for nitric oxide synthase 1 (Nos1) in neonatal bladders and attached outflow tracts. For both leucine-rich repeats and Ig-like domains 2 (Lrig2) homozygous mutant tissues and the heparanase 2 (Hpse2) mutant tissues, Nos1 transcripts levels were reduced compared with littermate control subjects. Values are shown as mean AE SEM, with the value for the mean of each wild-type group defined as "1." Figure S5. Quantification of processes emanated from pelvic ganglia explants. Fragments of embryonic pelvic ganglia were explanted and the numbers of processes that had emanated from them were counted after 1 day in culture. Results are shown as median and ranges, with values factored for the length of the perimeter in each explant. The wild-type median was defined as "1." There were no significant differences (all P > 0.05) between wild-type explants and either leucine-rich repeats and Ig-like domains 2 (Lrig2) or heparanase 2 (Hpse2) mutants as assessed by Kruskal-Wallis tests. The number of ganglia assessed are shown in the figure. Figure S6. Western blot quantification of Lrig2 protein levels in bladder lysates. Total protein was extracted from wild-type, Lrig2 heterozygous, and Lrig2 homozygous mutant mouse bladders and 35 mg resolved on a polyacrylamide gel. Each lane represents a separate bladder sample, with genotypes loaded at random. Wild-type Lrig2 is indicated by the black arrow, at 125 KDa. The red box indicates an example of wild-type, heterozygous, and homozygous bladder samples, with strong, low, and absent Lrig2 bands, respectively. Supplementary material is linked to the online version of the paper at www.kidney-international.org.