A mutation affecting laminin alpha 5 polymerisation gives rise to a syndromic developmental disorder

Basement membranes (BM) are protein-rich sheets which typically separate epithelial or endothelial cells from underlying mesenchyme, thereby mediating the structural integrity of a tissue and communication between cell layers. As a consequence, they are central to normal embryonic development. Laminins constitute a major BM component and are composed of alpha (α), beta (β) and gamma (γ) chains which assemble into 16 different heterotrimeric isoforms (Durbeej, 2010). One of the most broadly and highly expressed of these proteins is laminin alpha 5 (LAMA5) (Miner et al., 1995) which is necessary for normal embryonic development (Miner et al., 1998). By associating with laminin β1, β2, γ1 and γ3, the LAMA5 chain contributes to three major BM trimers: LM-511 (α5/β1/γ1), which is found principally in the BM of epithelia, endothelia and smooth muscle; LM-521 (α5/β2/γ1), which contributes to the BM of epithelia, endothelia, smooth muscle, neuromuscular junctions and the glomerular basement membrane (GBM) of the kidney; and LM-523 (α5/β2/γ3), which has been detected in the retina and CNS. The crucial role that LAMA5 plays in dictating the normal development of these tissues is evident upon its germline inactivation in mice, which results in late gestation embryonic lethality associated with a failure in neural tube closure and placental insufficiency (Miner et al., 1998). In addition, a large number of other developmental defects including syndactyly, defective lung lobation, altered neural crest migration, aborted hair folliculogenesis and abnormal kidney development (amongst others) speak to the breadth of the role of the protein in directing the normal cellular interactions necessary for organogenesis.

LAMA5 is a large (∼400 kDa), evolutionarily conserved multidomain protein which forms the central arm of the cruciform structure typical of laminin trimers. The laminin N-terminal domain (LN) present in most other members of the β- and γ-subtypes is associated with polymerisation of laminin hetero-trimers. This process is necessary for the formation of the extensive mesh-like laminin networks that are a feature of BMs (Garbe et al., 2002; Behrens et al., 2012). The laminin complex is linked to (and integrated with) a second major BM protein network composed principally of collagen IV. Early investigations provided evidence that this is mediated through a shared association with nidogen (Fox et al., 1991); however, a more recent study in the skin suggests that perlecan (also known as HSPG2) may also arbitrate these linkages (Behrens et al., 2012). It appears to be likely that such proteoglycan linking is a common feature of BMs generally and that these may be dependent on the tissue in question. The laminin network also interacts extensively with underlying cells in a process mediated in large part through laminin α chains, particularly (though not exclusively) through the engagement of cellular receptors with the laminin globular (LG) domains at their C termini. In the case of LAMA5, potential binding of subsets of the five LG domains (LG1-5) have been identified with integrins α2β1, α3β1, α6β1, α5β4, α7β1 and αvβ3 (Kikkawa et al., 2000; Nishiuchi et al., 2006). In addition, LAMA5 binds to the plasma membrane/membrane-associated proteins α-dystroglycan and syndecan-4 through its LG4 domain (Nishiuchi et al., 2006; Lin and Kurpakus-Wheater, 2002). Uniquely among α chains, it also binds to basal cell adhesion molecule (BCAM; also known as the Lutheran blood group antigen) (Moulson et al., 2001; Kikkawa et al., 2002).

Many studies of the action of laminins have relied on the use of recombinant proteins, antibody blocking or heterologous in vitro expression systems; meaning that the relative contribution of these different interactions to normal embryonic development remain unclear. However, a series of in vivo experiments expressing full-length and chimeric LAMA5 in a Lama5^−/− background explored the specific physiological role of the LG domains (Kikkawa and Miner, 2006). On the basis of these experiments, it appears that LG1-2 harbours much of the functionality involved in embryonic development, as expression of these domains in a fusion transgene with a cDNA encoding LAMA1 LG3-5 rescues the exencephaly, placental vascularisation and lung lobe septation associated with gene deletion. However, these mice still develop progressive renal failure, suggesting that LG3-5 are important for establishing and maintaining the glomerular filtration barrier (Kikkawa and Miner, 2006). Although these findings are important, they may be limited by the universal expression of the transgene employed.

Insights into the functional importance of different laminins have also been provided by the identification of gene mutations in inherited disease. Mutations in LAMA3, LAMB3 or LAMC2, which abrogate LM-332 function, cause a skin disorder termed junctional epidermolysis bullosa Herlitz type that is characterised by diminished dermal-epidermal adhesion, skin fragility and blistering (Vidal et al., 1995; Pulkkinen et al., 1994a,b; Aberdam et al., 1994). LAMA2 loss-of-function mutations cause congenital merosin-deficient muscular dystrophy (MDC1A) (Helbling-Leclerc et al., 1995), whereas hypomorphic missense mutations underpin a milder but related disease limb girdle muscular dystrophy-23 (LGMDR23) (Allamand et al., 1997). Mutations in LAMB2 give rise to Pierson syndrome, a congenital nephrotic syndrome which rapidly progresses to end-stage renal disease within the first year of life (Zenker et al., 2004). Variants in LAMB1 are associated with brain malformations known as lissencephaly (Radmanesh et al., 2013) and mutations in LAMA1 cause cerebellar dysplasia and renal cysts (Aldinger et al., 2014). Despite its broad expression profile and requirement during embryonic development, the evidence for disease-causing mutations in LAMA5 is less compelling. A homozygous missense variant in domain II of the protein was found in a family with presynaptic congenital myasthenic syndrome (Maselli et al., 2017), but confirmation of causation was not established in vivo and interpretation of the patient phenotype was complicated by co-inheritance of a homozygous missense variant in LAMA1. Similarly, Sampaolo and colleagues identified heterozygous missense variants in the LAMA5 LG3 domain in a family affected by dominant cutaneous defects and a range of other conditions including variable night blindness, muscle weakness, osteoarthritis, malabsorption syndrome and hypothyroidism (Sampaolo et al., 2017). However, analogous knock-in mice did not display many of these features. Finally, homozygous missense LAMA5 variants (in different protein domains) have been reported in three families with recessive paediatric nephrotic syndrome (Braun et al., 2019) and heterozygous changes in three families with autosomal dominant adult-onset focal segmental glomerulosclerosis (Chatterjee et al., 2013; Gast et al., 2016). However, neither of these studies include functional characterisation to establish whether these variants are causative, correlative or modifiers of the disease phenotypes observed.

Polymerisation is a process which directs the assembly of α/β/γ laminin trimers to form sheet-like networks characteristic of most BMs (Behrens et al., 2012; Garbe et al., 2002). The relative importance of polymerisation to LAMA5 function during fetal development is poorly understood. However, mutations in LAMB2 LN domains cause Pierson syndrome, indicating that abrogated polymerisation is important for maintaining the in vivo integrity and functionality of the glomerular BM (McKee et al., 2018; Funk et al., 2018, 2020). Moreover, recent crystallographic studies have identified a protein interface within the LN domain in the region of a PLENGE sequence motif which appears to be important for this oligomerisation (Hussain et al., 2011). In this paper, we report the identification of a compelling disease-causing patient variant in the LN domain of LAMA5 associated with a complex developmental disorder affecting multiple organ systems and reminiscent of the Lama5^−/− mouse phenotype. This altered residue lies close to the PLENGE sequence thought to be necessary for laminin polymerisation. Using biochemical approaches, we provide evidence that this variant prevents laminin polymerisation. We then employ CRISPR-driven in vivo modelling to show that mice carrying the analogous change develop phenotypes which very strongly copy those of the patient. Comparison between the phenotypes of humans and mice with this mutation, and those apparent upon gene deletion, helps to define the functional importance of the LAMA5-LN domain and more broadly of laminin polymerisation to embryonic development.

The proband was the only child to consanguineous healthy parents of Italian descent (Fig. S1) born at 36 weeks of gestation following induced labour, initiated after ultrasound scans indicated advanced polyhydramnios. The pregnancy was also complicated by abnormal nuchal translucency and omphalocoele. Shortly after birth he was diagnosed with 2 large ‘cysts’ in a dysplastic non-functioning right kidney and with bilateral vesicoureteral reflux (VUR) (Fig. 1A). A dimercapto-succinic acid (DMSA) scan in infancy was suggestive of dysplasia in the left kidney. At 2 years he developed atypical focal segmental glomerulosclerosis which was unresponsive to therapy and progressed to end-stage kidney disease and renal transplant shortly afterwards. He then developed a lymphoproliferative disorder 3 years post-transplant, which was thought to be driven by Epstein-Barr virus. In addition, he presented with a complex syndromic condition affecting several other organs. This included short stature and craniofacial dysmorphism (Fig. 1B) including down slanting palpebral fissures, low set dysplastic ears, prominent eyes, micrognathia and mild tongue tie. He also had syndactyly of the 2nd and 3rd toes (Fig. 1C), pyloric web (which was surgically corrected 3 days after birth), atelectasis, undescended testes, bilateral inguinal hernia, absent left vas deferens, and mild and moderate hearing loss in the left and right ears, respectively. He had normal motor and language development and there were no vertebral anomalies.

Panel-based genomic testing for variants in known congenital anomalies of kidney and urinary tract (CAKUT) genes was negative and comparative genomic hybridisation, transferrin isoform profiling and urinary metabolic screening was unremarkable. He was recruited to the KidGen Research Genomics Program (Mallett et al., 2015) and was subjected to whole-genome sequencing (along with his mother), generating mean depth of coverage of 35× and 34×, respectively (with 96% and 93% of bases covered at >20×). Sequence analysis identified a homozygous missense variant [NM_005560.4:c.857G>T; p.(Arg286Leu)] in the LN of LAMA5. This residue is highly conserved across species (Fig. 1D) and, at a structural level, lies adjacent to the PLENGE sequence central to the polymerisation of multiple laminin α/β/γ heterotrimers (Fig. 1E) (Hussain et al., 2011). This variant is not present in the gnomAD database (Lek et al., 2016) and analysis using multiple damage prediction tools (SIFT, Polyphen, LRT, MutationTaster, MutationAssessor, FATHMN, PROVEAN and CADD) consistently predicted the variant as damaging/pathogenic. A summary of further homozygous candidate changes present in the patient is detailed in Table S1.

The proximity of the variant residue to the PLENGE sequence previously proposed as a mediator of laminin trimer polymerisation (Hussain et al., 2011) raised the possibility that the formation of higher-order laminin networks may be specifically impacted by this alteration. The sequence of the affected residue and those upstream and downstream of this site are highly conserved in laminin α5, α1 and α2 subunits. Because of this, it was possible to study the impact of this change on laminin polymerisation using established protocols for studying this process in chimeric α1 protein. In this approach, a polymerisation-enabling chimeric fusion protein called αLNNd, which comprises the LN domain and a fragment of nidogen (McKee et al., 2018, 2007) can be used in conjunction with full-length γ and β chains and a truncated α chain to model the impact of LN domain variants on polymerisation (Fig. 2A). To this end, we generated wild-type αLNNd protein and a version of the protein carrying the LAMA1 variant equivalent to the LAMA5 mutation (α1R263L). These were then compared to examine whether they could ‘rescue’ the polymerising capacity of a nidogen-binding LAMA1 construct lacking the LN domain (Lmα1ΔLN-L4b). Laminin assembly was then assayed by the application of recombinant proteins to cultures of Schwann cells, which do not ordinarily generate their own laminin matrix. As expected, wild-type LM-111 was able to contribute to extensive matrix on the surface of these cells, but Lmα1ΔLN-L4b was not (Fig. 2B, upper panels). Although wild-type αLNNd was able to restore matrix assembly with Lmα1ΔLN-L4b, αLNNd bearing the analogous R263L mutation was not (Fig. 2B, lower panels). These findings were confirmed by quantitation of laminin fluorescence (Fig. 2C). The same mutation was also evaluated in a solution-based polymerisation assay with complexes of wild-type or mutated αLNNd+Lmα1ΔLN-L4b. In line with the Schwann cell laminin assembly assay, the incapacity of Lmα1ΔLN-L4b to polymerise (Fig. 2D) was restored by the addition of αLNNd (Fig. 2E) but not by αLNNd bearing the α1R363L substitution (Fig. 2F). These observations were confirmed by quantitation (Fig. 2G). Taken together, this work indicates that the variant in our patient impairs the capacity of LAMA5 to polymerise.

To further assess whether the LAMA5 variant in our proband is pathogenic, the change was introduced into the mouse genome using CRISPR-Cas9 genome editing. This was complicated by the fact that the mouse codon analogous to that affected in the patient utilises a different first base (AGG versus CGG). Therefore, the introduction of the same genetic change (G>T) at the second position would result in a different protein alteration (Arg-Met instead of Arg-Leu) (Fig. 3A). Further, the base in question lies near an intron-exon boundary. To address this, we generated two independent mouse lines – a control in which the sequence of the codon was altered to that found in human populations (Lama5^em1Smy; hereafter Lama5^hs for ‘humanised sequence’) and a second in which the genetic change in the proband was introduced to this sequence (Lama5^em2Smy; hereafter Lama5^pm for ‘point mutation’). This leads to an analogous amino acid change (R291L, Fig. 3A). Transmission of these alleles from founder mice occurred and inter-crosses were established to generate homozygous mice. Lama5^hs/hs mice were born at Mendelian ratios (n=60, χ² P=0.165) and were healthy and fertile. No phenotypes have been noted in these animals in any assay undertaken to date. By contrast, although Lama5^pm/pm pups were born at the expected ratio (n=7 from 43 pups, χ² P=0.073), five were found dead at or shortly after birth, one died at postnatal day (P)15 and the other survived until 4 months of age (P135). Both of the surviving mice were smaller than control littermates (Fig. 3B). Those that died at or around birth had empty stomachs, indicating they had not fed before death. Western blot analysis of extracts from embryonic kidneys at embryonic day (E)18.5 found no difference in the levels of the protein and no evidence of alternate splicing producing smaller forms of the protein (Fig. 3C). Analysis of Lama5 transcripts in kidney RNA by RT-PCR using primers between exons 4 and 10 found no evidence for alternate splicing caused by either the humanisation of the allele or the introduction of the change at the end of exon 5 (Fig. 3D). Although hair follicle growth and eruption appeared normal in embryos at E18.5 (Fig. 3E), in the two surviving mice there was a significant reduction in fat deposition in the hypodermis, suggestive of wasting (Fig. 3F).

Owing to the postnatal mortality observed in the Lama5^pm/pm mice, we generated cohorts of embryos at E14.5, E16.5 and E18.5 to examine the impact of the mutation on fetal development. At the latter time point, homozygous embryos were present in the expected Mendelian ratios (χ² P=0.277, 69 embryos; Fig. 3G), which, combined with postnatal testing, confirms that, unlike Lama5 knockout mice, in utero death is not associated with this variant. Indeed, the exencephaly associated with gene deletion (Miner et al., 1998) was never observed in Lama5^pm/pm animals. Embryos from Lama5^pm/pm at E14.5 and E16.5 were smaller than wild-type littermates, but this difference was no longer evident at E18.5 (Fig. 3H), indicating that in utero growth restriction was unlikely to contribute to failure to thrive and survive after birth. There was no evidence of sexual dimorphism in the context of survival (χ² P=0.138) or with respect to embryo weight (P=0.395, Welch's t-test) at this developmental time point. Germline Lama5 deletion results in significant failure in the attachment of placental trophoblasts to associated BMs (Miner et al., 1998). On this basis, we examined the structure of the placenta of Lama5^pm/pm in an attempt to identify analogous defects. Immunohistochemical staining for trophoblasts (pan-cytokeratin), endothelial cells (CD31; also known as Pecam1) and basement membrane (LAMA5) revealed no significant differences in the zonation of the placenta in Lama5^pm/pm mice versus controls (Fig. 3I,J), but did show restricted areas of subtle trophoblast detachment (Fig. 3J). Notably, staining with an anti-LAMA5 antibody was equivalent in both intensity and localisation to wild-type controls, indicating that the variant protein is incorporated into BMs.

As was the case with our patient, syndactyly was a fully penetrant phenotype observed in Lama5^pm/pm mice. In all instances, the forelimbs were more severely affected than hindlimbs and usually had a club-shaped appearance in which individual digits were not delineated (Fig. 4A). Staining of the axial skeleton with Alizarin Red/Alcian Blue found no defects in the underlying patterning of bone and cartilage or fusion of these elements, findings which parallel the soft tissue syndactyly evident in our patient (see Fig. 1C). Comparison of skeletal elements at E16.5 suggested that osteogenesis may be delayed as a consequence of the Lama5 mutation, although these phenotypes were not evident at E18.5 (Fig. 4A). Syndactyly in Lama5^−/− mice has previously been attributed to a breakdown in inter-phalangeal BM and extrusion of underlying mesenchyme (Miner et al., 1998), and similar defects were observed in Lama5^pm/pm embryos at E14.5 (Fig. 4B). This analysis also highlighted the development of duplicated BM layers, which were a feature of the knockout mice. Eye abnormalities were recorded in 26% (n=6 of 23) of E18.5 Lama5^pm/pm mice, manifested unilaterally or bilaterally, were variable in their severity (Fig. 4C) and extended to anophthalmia. In some cases, these were associated with malformation of the frontonasal processes (Fig. 4C, right). Sectioning of skulls at E15.5 (Fig. 4D) and E18.5 (Fig. 4E) revealed malformation of the lens and retina in Lama5^pm/pm mice, identifying a role for the protein in regulating the normal differentiation of these structures. Craniofacial malformations extended to the nasal cavity and included reduced trabeculation of the nasal epithelia (Fig. 4F). Lastly, omphalocoele was observed in one (4%; n=23) of the E18.5 Lama5^pm/pm embryos (Fig. 4G), a variably penetrant feature which mirrors the patient phenotype. None of the defects in the limb, face, eye or body wall were observed in Lama5^+/+ littermates (n=28) or Lama5^hs/hs controls (n=29).

Our patient presented with recurring atelectasis and Lama5 knockout mice have been reported to show defects in pulmonary development (Nguyen et al., 2002). To examine whether lung involvement was also a feature of Lama5^pm/pm mice we examined embryos at E18.5. In wild-type mice, the right lung was always fully septated into the cranial, medial, caudal and accessory lobes (Fig. 5A). However, fusion of the right lung lobes was evident in 27% (n=9 of 33) of Lama5^pm/pm embryos at E18.5, as was smoothing of the normally defined lobe margins (Fig. 5B). When fusion occurred, the cranial and medial lobes were always affected, although the degree of fusion varied considerably in its extent. There was no evidence of sexual dimorphism in the penetrance of this phenotype (χ² P=0.062). Sections of lung showed that fusions were often incomplete, with areas of separation and fusion along the same inter-lobar margin (arrowhead, Fig. 5B). Staining for both LAMA5 and the basement membrane protein collagen IV at E16.5 found that both proteins were still deposited into the BMs of both airways and pleura (Fig. 5C), an observation which differs from those made of Lama5^−/− mice, which lack a BM at this stage.

One of the most prominent clinical features of our patient was the development of multiple defects in renal development and function, some of which appeared functional (focal segmental glomerulosclerosis) and others structural (VUR) in origin. To examine the former, we examined sections of late fetal kidneys but found no gross defects in organ morphology. Immunostaining for collecting ducts and nephron segments also indicated that the development of the organ was relatively normal (Fig. S2). However, upon closer examination of Lama5^pm/pm kidneys, multiple glomeruli were observed which showed ectatic congested capillaries with fewer vascular profiles than control animals (Fig. 6A). Staining with endothelial (CD31) and podocyte (P57; Coro1a) markers identified a subset of glomeruli that displayed obvious vascular abnormalities, including an enlargement of the vascular space and erosion of the thin vascular walls that create distinct capillaries (Fig. 6A, lower panels). Despite these changes, the cellular organisation of the podocytes around the periphery of the differentiating endothelium/mesangium was indistinguishable from control embryos. These findings are significantly different from the changes observed in Lama5^−/− mice, in which endothelial and mesangial cells are extruded from the developing glomerulus and the remaining podocytes stratify (Miner and Li, 2000). LAMA5 is thought to contribute significantly to the formation of LM-521 and consequently the integrity of the GBM. Indeed, deletion of the gene results in the absence of this structure (Miner and Li, 2000). To investigate whether the LN domain mutation similarly affected the integrity of the GBM, sections were stained with antibodies to collagen IV. This demonstrated that the GBM does form in LAMA5-LN mutant mice but also that some glomeruli had reduced GBM complexity, irregular deposits of the collagen IV protein and constricted vascular profiles not present in the controls (Fig. 6B). To examine these changes in more detail, we performed transmission electron microscopy on kidney sections at E18.5. Compared with the thin, continuous GBMs present in control glomeruli, Lama5^pm/pm kidneys displayed intermittent areas of disorganisation (Fig. 6C). These alterations in GBM ultrastructure correlated with defective morphology of podocyte foot processes, the broader outlines of which indicate mild effacement compared to the thin and evenly spaced structures present in control embryos (Fig. 6C).

To investigate how mutation of LAMA5 might contribute to these GBM changes we stained sections to determine whether protein deposition was altered. These studies revealed no differences in either the level or localisation of the mutant LAMA5 protein to the GBM (Fig. 6D). Co-staining for LM-111, which is present in early embryonic GBMs and under normal circumstances gradually disappears during embryonic development to be replaced by LAMA5 (Miner and Li, 2000), also detected no significant differences (Fig. 6D). As with our previous studies of LAMA5 deposition in the placenta (Fig. 3J) these studies indicate that LAMA5^pm is appropriately translated, transported and incorporated into the GBM. The LM-111 staining also indicates that the transition of laminin subtypes in this structure proceeds normally despite the alterations in GBM ultrastructure and vascular differentiation.

Although the neonatal lethality associated with introduction of the Lama5^pm mutation limited some investigations of postnatal disease, we were able to study whether other phenotypes emerged in the pups which did survive after birth. At the same time as developing what appeared to be functional glomerular impairment associated with glomerular sclerosis, our patient's renal health was significantly and progressively impaired by VUR. We therefore examined whether the surviving mice at P15 and P135 had evidence of the hydronephrosis typically associated with this condition. Upon dissection, kidneys from both animals exhibited medullary ablation and the formation of voids consistent with the development of this phenotype (Fig. 7A). To confirm that this was caused by hydronephrosis and not by some other mechanism (e.g. the formation of large isolated cysts) we stained sections of kidneys with antibodies to uroplakin IIIa. This protein is expressed in the lining of the renal pelvis, the structure which expands in hydronephrosis. The lining of the voids from these mice stained positive, confirming their origin (Fig. 7B). We also examined the kidneys from surviving mice to determine whether there was evidence of progressive glomerular disease. Surprisingly, given the often-severe phenotypes evident in mutant embryos at E18.5 (see Fig. 6A), the nephrons of these mice were relatively normal, with only occasional vascular malformations noted (Fig. 7C). Variable tubule dilation was observed in the older animals (Fig. 7D) but such phenotypes may be secondary to high grade reflux affecting the animals. The high rate of neonatal death and the poor overall health of these mice precluded more detailed studies of their renal architecture.

LAMA5 is one of the most broadly expressed of the laminin subunits during embryonic development but, to date, there is limited evidence for mutations in the gene giving rise to congenital disease. The severity of the phenotypes observed upon germline Lama5 deletion in mice (Miner et al., 1998) could explain why this might be the case, particularly as exencephaly and placental insufficiency would likely contribute to embryonic death in the event of loss-of-function mutations. However, in the current study we identify a missense mutation in the gene in a patient with a complex syndromic condition whose phenotype correlates with known sites of LAMA5 expression. Analysis of this change in mouse models profiled phenotypes that extensively overlap with those of our patient. These include malformations of the kidney, face, limb, lung and body wall. Importantly, phenotypes such as exencephaly and extensive placental malformation that have been associated with fetal death in Lama5^−/− mice are absent or ameliorated in our mouse model and in the patient. Taken together with biochemical studies of the variant protein showing that its capacity to polymerise is severely abrogated, this strongly supports the identification of a new syndromic condition specifically associated with hypomorphic mutations in LAMA5. Notably though, the mutant LAMA5 protein is normally localised in BMs in multiple tissues. This finding suggests that trimerisation and secretion of the protein is unaffected in our mouse model but also that incorporation of LAMA5 into BM is not wholly dependent on its capacity to form higher order polymerised networks on its own. In this respect, the localisation of laminins and their stable integration into the BM is likely to also be mediated by interactions with a range of cellular and extracellular matrix components including cell-surface receptors, sulphated glycolipids in the cell membrane and through interactions mediated by intermediary proteins such as agrin and nidogen. From our mouse studies it appears that BM integrity is only compromised in tissues which we propose are under greater mechanical stress as a consequence of rapid changes in tissue morphology (limb, lung) or acquisition of function (glomerulus).

Although previous studies of LAMA5 function have focussed on the role of the C-terminal LG domains that are important for engagement with cellular receptors such as integrins, syndecans, dystroglycan and BCAM (reviewed by Spenlé et al., 2013), in vivo roles for the LN protein domain are less well understood. The mutation in our patient lies within this part of the protein, adjacent to a sequence considered important for the formation of laminin networks through polymerisation (Hussain et al., 2011). In vitro assessment of the LN mutation provides evidence that this change alters the capacity of LAMA5 to polymerise with other LN domains – an essential facet of BM laminin network formation. Consequently, comparing the phenotypes of knockout and Lama5^pm/pm mice provides an in vivo opportunity to assess the function of the LN domain during development. Some aspects of the knockout phenotypes are shared with Lama5^pm/pm mice, including syndactyly and lung lobe fusion. Both phenotypes have been suggested to arise because of breakdown in BM integrity. In the case of the limb, this is thought to result from the extrusion of underlying digit mesenchyme as a result of a partially competent forming BM (Miner et al., 1998) and in the lung there is evidence that the pleural BM does not form without LAMA5 (Nguyen et al., 2002). Although we observed BM formation in the lung in Lama5^pm/pm mice, one possibility is that insufficiencies in laminin network formation still result in structural failures of the BM. Notably, the fusion we did observe was variable in its penetrance and often focal in nature. Other phenotypes differ considerably between the two mouse genotypes, including the development of exencephaly, defective hair follicle and tooth morphogenesis, renal agenesis and significant defects in placental vascularisation. All of these features are evident in the knockout mouse but are absent or considerably reduced in Lama5^pm/pm animals. One interpretation of these results is that engagement with cellular receptors through LAMA5 LG domains and other interacting motifs are still possible in our mutant. Given the normal expression levels and localisation of the LAMA5-PM protein, we propose that such interactions are functionally relevant to the phenotypes which do not develop in Lama5^pm/pm mice but do develop in knockouts.

Our findings in the kidney represent an example where there exist both phenotypic similarities and differences between knockouts and LN mutants. Germline Lama5 inactivation results in a failure in GBM formation, an incapacity of the endothelial cells to form capillaries and a failure in podocyte maturation. Introduction of the LN domain mutation leads to very different outcomes. The defects we observed in mutant glomeruli are largely vascular in nature, i.e. the GBM forms but not completely, and the maturation and arrangement of podocytes within the glomerulus are largely normal. In knockout mice the mesangial and endothelial components of the structure are extruded across the course of development, but we do not see similar changes in the LN mutant mice. These findings correlate well with the patient, in whom sclerotic phenotypes parallel the focal breakdown in GBM observed in our mice and with transgenic experiments which suggest that the LG domains still present in our mutant mice are important for glomerulogenesis (Kikkawa and Miner, 2006). The absence of compensatory overexpression of LAMA1, which has previously been observed in Lama5 knockouts (Goldberg et al., 2010), is not evident. Interestingly the glomeruli of the two Lama5^pm/pm mice which survived beyond birth were largely normal – with only minor defects in structure. One possibility, given the variability in the appearance of glomerular vascular defects in the Lama5^pm/pm mice, is that those that are severely affected degenerate as the animals age, but that the aspects of LAMA5 function affected by the mutation are not required for homeostasis after development has ceased.

To the extent that we can tell from the limited number of surviving mice, it appears that changes in the LN domain are not associated with immediate breakdowns in GBM or renal function after birth. However, the extended survival of these mice revealed hydronephrosis as a novel developmental phenotype associated with defects in LAMA5 function and one which may reflect the VUR present in the patient carrying the same mutation. Hydronephrosis can be caused by obstruction of the urinary tract or by urinary reflux associated with changes in the ureter entering the bladder. Although the limited number of surviving Lama5^pm/pm mice restricted our capacity to assess this functionally, the patient carrying the same variant underwent a micturating cystourethrogram, which showed reflux of urine from the bladder. Given the extensive clinical correlates with our mouse model, this suggests that the development of hydronephrotic phenotypes in LN mutant mice does not arise because of blockage of the urinary tract. However, we cannot definitively exclude the possibility that the same overall phenotype (hydronephrosis) arises through completely different mechanisms.

The hypomorphic nature of the mutation, its proximity to the polymerisation domain of LAMA5 and the specific effect that it has on protein function likely contributes to the rarity of this condition. In general terms it is likely that a proportion of inherited human disease is represented by such ‘n=1’ families and it may well be that the individual in our study is one of only a handful of people with this type of complex developmental syndrome with causative LAMA5 mutations. Nevertheless, the evidence that we present to implicate LAMA5 in disease satisfies the three broad categories outlined by McArthur and colleagues for experimentally investigating causality (MacArthur et al., 2014). Namely, the normal function of LAMA5 in stabilising BMs is consistent with disease aetiology in our patient, the mutation in question functionally disrupts the protein and specifically its capacity to polymerise, and recapitulation of the patient mutation in a model organism results in an analogous phenotype. As such, we feel that studying further rare conditions using CRISPR-based mouse modelling will be clinically and diagnostically valuable. Moreover, our study highlights how such subtle hypomorphic changes can provide deeper insights into protein function. In this case they have helped to define the in vivo functions of the LAMA5 LN domain and the relative importance of laminin polymerisation in shaping the embryonic development of tissues and organs.

Human studies were approved as part of the KidGen Research and Functional Genomics project protocols: HREC/14/QRBW/34 and HREC/15/QRCH/126. All participants and their families provided written informed consent for data collection and to undergo clinically relevant genetic/genomic testing. Whole genome sequencing was undertaken on an Illumina X instrument at the Garvan Institute (Australia). Reads were aligned to the reference human genome (GRCh37) using BWA-mem. Single nucleotide variants (SNVs) and indels were called using GATK v3.6 (McKenna et al., 2010), and copy number variants (CNVs) were identified with Canvas and Manta (Roller et al., 2016; Chen et al., 2016). All variants were annotated using SnpEff v3.4 (Cingolani et al., 2012). Analysis and identification of candidate variants were performed with an in-house workflow incorporating the annotated variant data and pedigree information. Parallel analysis was undertaken by Broad Research Genomics. Measures included in silico damage prediction (SIFT, Polyphen, LRT, MutationTaster/Assessor, FATHMM, PROVEAN, CADD), assessment of population allele frequencies (gnomAD; Karczewski et al., 2017), read quality analysis and determination of pedigree concordance for all hypothesised inheritance patterns.

Lama5^hs and Lama5^pm mouse alleles were generated by the Monash Genome Modification Platform (including the Monash University node of the Australian Phenomics Network) by direct injection of CRISPR-Cas9 reagents [guide RNA, recombinant Cas9 protein, HDR repair template (all IDT; 30 ng of each)] into the pronucleus of fertilised mouse oocytes. Mutant alleles were identified in founder mice and backcrossed to wild-type controls for >2 generations before the analysis of any phenotypes. All animals in the study were on a pure C57BL/6J background. Procedures complied with standards set under Australian guidelines for animal welfare and experiments were subject to Monash University animal welfare ethics review (Approval #MARP/2019/13606).

Immunohistochemistry was performed using antibodies to the following proteins: Dolichos biflorus agglutinin (DBA) (Vector Laboratories, B-1035, 1:250), Lotus tetragonolobus lectin (LTL) (Vector Laboratories, B-1325, 1.250), CD31 (R&D Systems, AF3628, 1:40), cytokeratin (Abcam, Ab11595, 1:100), uroplakin 3a (Progen, 610108, 1:100), collagen IV (Abcam, Ab6586, 1:200), LAMA5 (1:800; Miner et al., 1997), p57 (Santa Cruz Biotechnology, Sc8298, 1:100), Tamm–Horsfall glycoprotein (THP) H-135 (Santa Cruz Biotechnology, Sc20631, 1:100), laminin 111 [gift from Prof. Dale Abrahamson (Abrahamson et al., 1989), 1:1600]. All species-specific Alexa conjugated secondary antibodies were purchased from Invitrogen and used at a dilution of 1:400 [Alexa Fluor Streptavaidin 488, S-11223; Alexa Fluor donkey anti-goat IgG (H+L) 488, A-11055; Alexa Fluor donkey anti-rat IgG (H+L) 488, A-21208; Alexa Fluor donkey anti-rabbit IgG (H+L) 488, A-21206; Alexa Fluor donkey anti-rabbit IgG (H+L) 555, A-31572; and Alexa Fluor donkey anti-mouse IgG (H+L) 555, A-31570]. With the exception of the antisera to LAMA5, all staining was performed on 5 µm paraffin sections following citrate-based antigen retrieval. LAMA5 staining was performed on 10 µm frozen sections of OCT-embedded tissues. Stained sections were mounted in ProLong^® Gold Antifade Mountant (Thermo Fisher Scientific) and imaging was performed using a Nikon C1 Upright Confocal microscope. Western blotting was performed using the LAMA5 antibody described above (dilution 1:10,000) and proteins were separated on a precast gradient gel (Bio-Rad). For fetal samples, staining and imaging was performed on a minimum of three different samples from three different animals of each genotype. As indicated, owing to high rates of neonatal death, analysis of surviving Lama5^pm/pm animals at P15 and P135 was undertaken on limited sample numbers but included n≥3 for control genotypes.

RNA from E18.5 snap-frozen kidneys was extracted by crushing organs in RLT buffer using Eppy-pestles and using QiaShredders and RNAeasy minikits (Qiagen) with on-column DNase treatment according to the manufacturer's protocols. cDNA synthesis was performed using Life Technologies VILO on 1 μg RNA. Endpoint PCR was undertaken with 25 ng of cDNA using Green Go Taq mastermix (Promega) with 30 cycles of amplification at 95°C, 30 s at 54°C, 1 min at 72°C. Exon 4 forward primer sequence was 5′-ggggccttgaacttctccta-3′ and exon 8 reverse primer sequence was 5′-gaagaagccaggcagacaac-3′.

Skeletal preparations were performed as per Rigueur and Lyons (2014).

After fixation in glutaraldehyde fixative [16% (v/v) paraformaldehyde, 25% (v/v) glutaraldehyde, 0.2 M sodium cacodylate buffer (pH 7.4), 5 mM CaCl₂, 10 mM MgCl₂] tissues were submitted to the Ramaciotti Centre for Cryo-Electron Microscopy at Clayton for osmication followed by epon resin embedding and sectioning. Imaging was performed using a Jeol 1400Plus microscope.

Laminin-111 (LM-111) and non-polymerising laminin lacking the α1 short arm (Lmα1ΔLN-L4b) were prepared and characterised as described previously (McKee et al., 2007). The chimeric linker protein αLNNd was prepared as previously described (McKee et al., 2009). αLNNd bearing the LN domain mutation mouse α1R263L was prepared as follows. Overlapping PCR generated from 1895nt and 290nt products (1F-5′-gactcacggggatttccaagt-3′ and 2R-5′-gccaatagtaataaagtcttgtgacaatggg-3′; 2F-5′-cccattgtcacaagactttattactattcg-3′ and 1R-5′-gatggctggcaactagaaggc-3′) which were sewn together with the 1F and 1R primers. The resulting 2152nt insert was digested with NheI and XhoI (Invitrogen) and ligated with T4 ligase (Invitrogen) into the α1LNNd pcDNA3.1 zeocin vector (ThermoFisher 7648 nt). The α1R263L plasmid was transfected into HEK293 cells and media collected after 5 days of confluent culture. The protein was purified on a flag matrix according to the manufacturer's instructions (Sigma-Aldrich, A2220).

Schwann cells isolated from sciatic nerves from new-born Sprague-Dawley rats were maintained at high passage, cultured and used to evaluate laminin assembly on cell surfaces as previously described (McKee et al., 2018, 2009). Cells at passage 50 were plated onto 24-well dishes at 50,000 cells/well and treated with the indicated proteins (28 nM laminins, 28 nM αLNNd proteins) for 1 h at 37°C followed by washing, fixation and staining with a primary monoclonal antibody specific for laminin-γ1 as previously described (McKee et al., 2017). Detection of bound primary antibody was accomplished with Alexa Fluor 647 (far red) goat anti-rabbit IgG secondary antibody (A27040, Molecular Probes, 1/400) and nuclear counterstaining with 4′,6-diamidino-2-phenylindole (DAPI). Images were recorded (5-12 fields, each 1300×1030 pixels) using a ×10 objective, with the same exposure time within an experimental set. Fluorescence intensity levels were estimated from the digital images using ImageJ software (National Institutes of Health), with calculations performed in Microsoft Excel. Laminin staining intensities were calculated based on a protocol described for the measurement of laminin on Schwann cells (McKee et al., 2009). Briefly, a single segmentation cut-off value was set to exclude non-cellular regions for all images being compared. The summed pixel intensities overlying the treated cells in each field were then divided by the cell number as determined from DAPI-stained nuclei counts. Values were expressed as the mean±s.d. of summed intensities from different fields normalised to control, with plotting in SigmaPlot 12.5 (Systat Software).

Aliquots (50 μl) of wild-type LM-111, or Lmα1ΔLN-L4b without or with αLNNd, in polymerisation buffer were incubated at 37°C in a series of concentrations. Eppendorf tubes containing the aliquots were then centrifuged to separate polymerised protein (15,000 g). Supernatants and pellets were analyzed by SDS-PAGE with Coomassie Blue staining, scanned with GE Image Scanner III, and quantitated as previously described (McKee et al., 2009, 2007). Background-subtracted summed pixel values fitted by linear regression were plotted in SigmaPlot 12.5.

The studies were performed using the expertise, equipment and reagents of several Monash University Research Platforms including the Monash Genome Modification Platform (incorporating the Australian Phenomics Network), Monash Animal Research Platform, Monash Microimaging, the Ramaciotti Centre for Cryo-Electron Microscopy and the Monash Histology Platform. We are indebted to the patients and families involved and to the clinical, diagnostic and research staff of the KidGen Collaborative. The KidGen Collaborative Flagship is supported by the Melbourne Genomics Health Alliance (Melbourne Genomics), grants from the Royal Children's Hospital Foundation and Royal Brisbane and Women's Hospital Foundation and by the National Health and Medical Research Council Genomics TCR Grant 1113531, entitled ‘Preparing Australia for Genomic Medicine: A proposal by the Australian Genomics Health Alliance’. Melbourne Genomics is funded by the 10 members and the State Government of Victoria (Department of Health and Humans Services). In-kind funding support was provided for renal genetics clinic operation and genomic test provision by Queensland Health (Metro North Hospital and Health Service; Children's Health Queensland Hospital and Health Service), New South Wales Health, South Australia Health, Western Australia Department of Health, Northern Territory Department of Health and Tasmanian Department of Health and Human Services. The production of mouse lines in this study were subsidised by the Australian Phenomics Network which is supported by the Australian Government through the National Collaborative Research Infrastructure Strategy (NCRIS) Program. Funding for sequencing was provided by CostCo to the Sydney Children's Hospital Network and the Centre for Kidney Research. The research conducted at the Murdoch Children's Research Institute was supported by the Victorian Government's Operational Infrastructure Support Program. Sequencing and analysis were provided by the Broad Institute of MIT and Harvard Center for Mendelian Genomics (Broad CMG) and was funded by the National Human Genome Research Institute, the National Eye Institute, and the National Heart, Lung and Blood Institute (UM1 HG008900) and in part by the National Human Genome Research Institute grant R01 HG009141.

The peer review history is available online at https://dev.biologists.org/lookup/doi/10.1242/dev.189183.reviewer-comments.pdf