CLDN16 Genotype Predicts Renal Decline in Familial Hypomagnesemia with Hypercalciuria and Nephrocalcinosis

Abstract

Familial hypomagnesemia with hypercalciuria and nephrocalcinosis (FHHNC) is a rare autosomal recessive tubular disorder caused by CLDN16 mutations. CLDN16 encodes the renal tight junction protein claudin-16, which is important for the paracellular reabsorption of calcium and magnesium in the thick ascending limb of Henle's loop. That FHHNC is frequently associated with progressive renal failure suggests additional roles for claudin-16 in the maintenance of tight junction integrity. An investigation of 32 patients with FHHNC and 17 different mutations was previously reported; here, the analysis is expanded to 39 additional patients and 12 new mutations. Expression studies revealed that five of the 12 new mutations led to partial loss of claudin-16 function and the remaining seven led to complete loss of function. The 23 patients who had mutations resulting in complete loss of function of both alleles were significantly younger at the onset of symptoms than the 46 patients who had at least one mutant allele providing partial function (2.2 versus 5.6 years; P < 0.01). In addition, those with complete loss of function had a more rapid decline in GFR (7.3 versus 2.9 ml/min per 1.72 m2/y; P < 0.01), leading to 54% requiring renal replacement therapy by age 15 compared with 20% of those with residual function (P < 0.05). These data suggest that residual function of claudin-16 may delay the progression of renal failure in FHHNC. Familial hypomagnesemia with hypercalciuria and nephrocalcinosis (FHHNC; OMIM *248250) is a rare autosomal recessive tubular disorder. It is characterized by massive urinary losses of magnesium (Mg) and calcium (Ca) leading to hypomagnesemia and bilateral nephrocalcinosis. The disease usually presents with recurrent urinary tract infections and polyuria/polydipsia. Additional symptoms include rickets, nephrolithiasis, hematuria, muscular tetanies, seizures, failure to thrive, vomiting, and abdominal pain.1,2 Ocular abnormalities and hearing impairment have been described in a subset of patients with FHHNC.3,4 Additional biochemical abnormalities include signs of incomplete distal renal tubular acidosis, hypocitraturia, increased parathyroid hormone (PTH) levels (independent of GFR), and hyperuricemia.3,5–8 Unlike most other inherited tubular diseases affecting electrolyte homeostasis, FHHNC is generally complicated by progressive renal failure during childhood or adolescence, but the pathogenesis of chronic renal failure remains a matter of debate. Clearance studies in patients with FHHNC localized the site of disturbed handling of Mg and Ca to the thick ascending limb (TAL) of Henle's loop.6 The TAL plays an important role in the tubular reabsorption of Mg and Ca, which occurs by paracellular flux, a process driven by the lumen-positive transepithelial potential in this nephron segment. In 1999, Simon et al.9 identified a new gene (CLDN16, formerly PCLN1) and characterized mutations in this gene as the underlying molecular defect in FHHNC. Since then, approximately 30 additional families with FHHNC as a result of CLDN16 defects have been described.2,10–13 Hypercalciuria and nephrolithiasis have also been observed in heterozygous FHHNC mutation carriers.2,3 FHHNC is a genetically heterogeneous disease because recently CLDN19 mutations have been identified in a cohort of patients mainly originating from Spain.14 The renal phenotype of these patients is very similar to patients exhibiting CLDN16 mutations; however, patients with CLDN19 mutations also have severe ocular abnormalities in most cases. CLDN16 and CLDN19 encode the tight junction (TJ) proteins claudin-16 (paracellin-1) and claudin-19, both being members of the claudin multigene family. Claudins are important components of the TJ strands in various tissues.15 They are part of a complex protein network built up by a variety of different proteins, and there is clear evidence that claudins confer ion selectivity to the paracellular pathway.16–18 In the kidney, claudin-16 expression is restricted to the TAL of Henle's loop. It was speculated that claudin-16 constitutes the core of an intercellular pore, allowing paracellular reabsorption of Mg and Ca ions.9,19 Following the observation that a naturally occurring knockout model in Japanese black cattle shows early-onset renal failure with diffuse interstitial nephritis,20 it was recently speculated that claudin-16, like other claudins, may also be involved in the regulation of cell growth, proliferation, differentiation, and dedifferentiation.21 Hou et al.22 described the functional analysis of claudin-16 in polarized cell lines. They demonstrated that in LLC-PK1 cells, claudin-16 modulated the ion selectivity of the TJ by selectively increasing the permeability of Na+ with no effects on Cl−, resulting in a high permeability ratio of Na+ to Cl−. Instead, Mg flux across cell monolayers showed a far less pronounced change after claudin-16 expression. From these data, it might be concluded that claudin-16 defects lead to a loss of cation selectivity with a subsequent decrease in lumen-positive potential that is the driving force for the paracellular flux of cations. This hypothesis of a nonselective paracellular cation channel is supported by a mouse model using transgenic RNAi depletion of claudin-16. Loss of CLDN16 in this model caused TJ in TAL to lose the cation selectivity.23 Hou et al.22 also analyzed the consequences of most of the reported human CLDN16 mutations by heterologous expression in vitro. Whereas most mutations resulted in a complete loss of function, some mutations retained a substantial residual function. One of these mutations (L151F) is by far the most frequent FHHNC mutation, occurring in almost 50% of the patients described so far.2 Combining this information with additional functional analysis of new mutations using the same expression system, we present a genotype/phenotype correlation with a special focus on the progression of renal failure in a large cohort of patients with FHHNC. We provide clinical data indicating that homozygous or compound heterozygous patients who carry at least one CLDN16 mutation with residual function have a much more benign course of the disease than patients with a complete loss of function. RESULTS Mutation Analysis of CLDN16 In our initial study,2CLDN16 mutations were identified in 32 patients with FHHNC, 67% of the mutant alleles exhibited a missense mutation affecting the first extracellular loop of claudin-16, and 48% were affected by an L151F exchange. This mutational hot spot is due to a widespread founder effect (Germany and Eastern European countries). Mutation analysis in families F29 to F88 (with 39 affected individuals) revealed 17 different mutations (Table 1). In addition to the ones already described, 12 novel CLDN16 mutations were identified, including seven missense mutations, one nonsense mutation truncating the protein in the fourth transmembrane domain; one small internal deletion, one splice-site mutation in exon four, one frameshift mutation truncating the protein before the first extracellular domain, and one mutation resulting in the loss of the translation initiation start site (Figure 1A). Figure 1B depicts the amino acids affected by missense mutations in the claudin-16 protein. All mutations co-segregated with the FHHNC phenotype, and none of the mutations was observed in at least 100 control chromosomes. Both mutant alleles were detected in 34 affected individuals; however, in the remaining five patients from three families (F37, F61, and F73), only one heterozygous mutation could be identified. Families F37 and F61 both have two affected children. Therefore, we performed haplotype analysis to demonstrate identical haplotypes also for the second mutant allele. In both families, the affected siblings share the same haplotypes compatible with linkage to the CLDN16 locus. Thus, it is most likely that the second mutation was missed because it is not located in the coding sequence. Family F73 could not be studied by haplotype analysis because there is only one affected child. Functional Analysis of the New Mutations The functional consequences of the new CLDN16 missense mutations were analyzed after heterologous expression in MDCK and LLC-PK1 cells. In addition, we studied H141D and L151W, which have been described previously2 but not functionally characterized by Hou et al.22 The profile of expression and localization of the mutant proteins are summarized in Table 2. Most of the mutant proteins display a normal trafficking to the cell membrane. Only two mutations (M71T and C131R) were retained inside the cell; the M71T mutation is localized to lysosomes, whereas C131R is retained in the endoplasmic reticulum (Figure 2). The mutants that were properly targeted to TJ showed a significant loss of function compared with wild-type claudin-16 when expressed in LLC-PK1 cells. We observed a significant decrease of the dilution potential and the ratio of permeability of Na+ over Cl−. This indicates a disturbance of the cation selectivity of LLC-PK1 cells expressing wild-type claudin-16. As in the previous study, not all mutations showed a complete loss of function (CL; Figure 3). Several mutants retain a significant residual claudin-16 function. To compare the results obtained in this study with the data from the initial expression study,22 we calculated the residual function as percentage relative to wild-type claudin-16 function. Mutants with a residual function >40% were considered “partial” loss of function (PL); the remaining mutants were considered CL. Theses categories were also used for the genotype/phenotype analysis (see next section). Genotype/Phenotype Correlation Assignment of Mutation Category To analyze a possible genotype/phenotype correlation in FHHNC, we assigned CLDN16 mutations to three different categories: (1) Complete loss of function (CL) mutations including missense mutations resulting in a complete loss of function after heterologous expression (Figure 3). In addition, nonsense, truncating and splice-site mutations were attributed to this category. Bioinformatic analysis of the three splice site mutations clearly indicated significant disturbance of CLDN16 mRNA splicing. (2) Missense mutations that displayed substantial residual function >40% as compared with wild-type claudin-16 function were considered as mutations with partial loss of function (PL) (Figure 3). (3) Missense mutations that could not be assigned in categories CL or PL are given as X. This assignment was done separately for both mutant alleles. Following these criteria, 23 patients had a CL/CL genotype, 29 patients had PL/PL, 12 patients had PL/CL, five patients had PL/X, one patient had X/X, and one patient had CL/X. Because of the lack of information in X/X and CL/X patients, these two patients were completely excluded from the genotype/phenotype analysis (Table 3). Confirmation of the Recessive Nature of FHHNC In a first step of the genotype/phenotype analysis, only the groups with classified mutations on both mutant alleles were analyzed (CL/CL, PL/CL, and PL/PL). PL/X patients were excluded from this first statistical analysis because no information is available with respect to the severity of the second allele (for comparison, the data are shown in Figure 4). In theory, given the recessive nature of the disease, one might expect that the effect of a PL mutation can also be observed if this mutation occurs in compound heterozygosity with a severe mutation (CL). This has been previously shown for other recessive diseases, for example, with respect to the preservation of pancreatic sufficiency in cystic fibrosis24: as long as a loss of function mutant was in compound heterozygosity with a mutant with residual function, no severe pancreatic phenotype was observed. Analysis of the three groups with categorized mutations on both alleles (CL/CL, PL/CL, and PL/PL) revealed a significant correlation between the genotype and age at onset (P < 0.05) and decline of renal function (P < 0.01). CL/CL patients were significantly younger at presentation of first symptoms (2.2 yr; 95% confidence interval [CI] 1.1 to 3.3 yr) compared with PL/PL patients (5.7 yr; 95% CI 3.3 to 8.2 yr), whereas age at onset in PL/CL patients (4.9 yr; 95% CI 2.3 to 7.6 yr) was similar to that of PL/PL patients, nevertheless not reaching statistical significance (Table 4, Figure 4A). The progression of renal failure expressed as loss of GFR/yr was significantly faster in the CL/CL patients (7.3 ml/min per 1.73 m2/yr; 95% CI 5.0 to 9.6) as compared with either PL/PL (3.3 ml/yr; 95% CI 1.6 to 4.9) or PL/CL patients (3.0 ml/yr; 95% CI 1.1 to 4.9), whereas no difference between PL/PL and PL/CL patients could be observed (Table 4, Figure 4B). The decline of GFR in the various groups during the study period is shown in Figure 5. The similar clinical courses of the PL/PL and PL/CL groups, which is in sharp contrast to the CL/CL indicates that in FHHNC, the nature of the second allele in with a PL mutation is not important for the clinical The PL/X which was excluded from this first analysis, these (Table 4, and all criteria, the PL/X is very similar to either the PL/PL or the PL/CL We PL/PL, PL/CL, but also PL/X patients one large This was compared with the of CL/CL at and Loss of CL and PL at onset in the CL loss of function of both was significantly than in the PL loss of function of at least one with CI 1.1 to 3.3 yr) versus 5.6 CI to yr; P < Table 4). CL patients showed a faster decline of GFR than PL patients, as expressed by a loss of GFR of versus 2.9 ml/min per 1.73 (P < 0.01). As a at the of the the renal function significantly between both groups (P < CL patients had a GFR ml/min per 1.73 compared with PL patients versus and more CL than PL patients had a GFR ml/min per 1.73 versus or renal replacement therapy versus Table CL patients also in than PL the age of 15 of CL patients renal replacement therapy versus 20% of PL patients (P < these increased to in CL and in PL by the age of We also analyzed the Mg and Ca but it has to be in that of with Mg and Ca at different of chronic renal failure may these The Mg and Ca levels were in both groups but not significantly between the CL and PL Mg was high in both groups but a significant In the urinary Ca was significantly in the CL versus the PL both in and in spot levels were high in both groups when compared with control patients with chronic renal failure of other (Figure In contrast to most other tubular affected by FHHNC are at high to progressive renal one of the patients already during adolescence, requiring renal replacement The pathogenesis of chronic renal failure in FHHNC remains a matter of debate. The progressive in FHHNC has been attributed to the hypercalciuria and however, other have this because other disease associated with early-onset nephrocalcinosis are not by a severe of renal All patients with FHHNC are affected by but of renal is to a The for this has been genotype/phenotype correlation could be in the patient that had first been analyzed for CLDN16 however, the of et in two different large mutations of CLDN16 in Japanese cattle already to a function of claudin-16 for tubular These present with chronic interstitial diffuse and with renal studies in these cattle characterized an nephron with and This knockout model suggests that the complete loss of claudin-16 is associated with a severe of renal function, whereas the heterozygous remains In no large CLDN16 gene have been however, the recently of Hou et demonstrated for the first functional in the human CLDN16 missense mutations identified so human missense mutations were in a system, and TJ localization and ion were of the mutants showed a CL TJ mutants showed a partial loss of function localization but ion and one mutant displayed a normal function to the wild-type protein not identified in our patient Mutants with residual function were expressed and targeted to TJ but showed a significant not loss of function compared with wild-type claudin-16. In this study, we performed mutational analysis of CLDN16 in 39 new patients with FHHNC and analyzed the course of renal in this cohort with the 32 patients of our initial the expression studies by functional data for the new mutations, all patients were to mutation for genotype/phenotype The clinical course in patients two mutations with CL also and splice-site is significantly compared with patients with at least one mutation with residual function with CL present symptoms in and a faster decline of As a significantly more patients with two CL mutations had a GFR ml/min per 1.73 at the of the study or renal replacement and was in than in the patients with PL of at least one allele. Whereas in the CL most of the families a CLDN16 mutation, in the PL the of patients carry the L151F founder mutation on at least one mutant allele. Only families have other mutations with PL on both mutant alleles. This the that the less severe phenotype in the PL on the L151F if this be it because the expression studies not functional for the courses of loss of GFR in the families L151F in the PL are to the patients with L151F mutations with the two patients a normal GFR at age and The previously observation of a with respect to the clinical course in the of the families the important role of CLDN16 for the progression of renal The handling of Mg and Ca in FHHNC was also analyzed in this As we hypomagnesemia and a a that is in FHHNC. We also could previous results increased levels already before the onset of chronic renal however, we were to between patients with CL and levels before chronic renal failure may be by the loss of Ca with subsequent from In addition, hypomagnesemia is to in of very severe in which it to have an This is observed in patients with hypomagnesemia with In these patients, is frequently very but the Mg levels in hypomagnesemia with are far the levels in FHHNC. In contrast to Ca, and urinary Ca in FHHNC to on the CLDN16 CL patients more Ca in the than PL this is a effect of the residual claudin-16 function is a residual claudin-16 function, as shown in the expression system, also in affected by these PL mutations, one expect reabsorption for both Ca and however, Mg not between the groups, but we have to that either Mg or in the of of renal failure are we could not for The increased Ca in CL patients also may be to which has been demonstrated to Ca in patients with It might be possible that patients with FHHNC and PL mutations have renal more than patients with renal which is more frequent in the CL The of tubular handling of Ca and Mg to be in much more in progression of renal could be between patients two mutations with residual function and patients affected by one mutation with residual function and one mutation with CL on the second allele. These results are in with the recessive nature of the One mutation with residual function is to a clinical we completely that mutations may different effects on the genotype of the second allele a can we in patients affected by mutations in characterized by interstitial and tubular is the to for patients with FHHNC of our initial demonstrated Ca tubular and interstitial to a In interstitial tubular cells were reported as with loss of both and to the membrane. between and was and the tubular was used to the of the cells of the renal These are with a process of tubular cells new that are considered to a role in the progression of renal is characterized by the of junction and the loss of cell cells cells. These suggest that TJ could be an in et demonstrated that of in is associated with of claudins and proteins of the TJ Therefore, one might that claudin-16 function could be an in of the tubular TJ by of cell and an of tubular who have FHHNC and are affected by two mutations have an onset of chronic renal failure as demonstrated in this patients with at least one mutation with residual function localization of the protein to the TJ are from rapid loss of renal function. a second study to the consequences of human CLDN16 mutations was That study on and the with the results of Hou et al.22 in however, with respect to there were some Most two of the mutations with residual function and could not be detected at the cell by et one mutation which was characterized as a trafficking mutant by Hou et was to be targeted to TJ by et respect to the two studies are not because different cell were It that the residual function described by Hou et as an phenotype between control and wild-type claudin-16 with respect to the permeability ratio of Na+ over and potential can be only in LLC-PK1 cells and not in cells, which were used by et These are to different transepithelial In LLC-PK1 cells, paracellular as a result of the transepithelial We also analyzed the of this study at onset and loss of in our patients with respect to the various of mutations by et attributed the patients either to expression at the cell as Mg by et or to at onset and loss of GFR were identical in both groups versus yr; loss of GFR versus ml/min per 1.73 that the on trafficking evidence of residual claudin-16 function is not for a genotype/phenotype This is supported by the observation of additional mutations that are targeted to TJ but display a CL (Table 2). our in data suggest that residual function of claudin-16 may delay the progression of renal failure in in patients with FHHNC the of the clinical course to some and of affected and Families patients who had FHHNC from families were in this study and and molecular results of families to families with affected individuals) were reported The remaining patients to 39 affected from were the same as in our initial hypercalciuria in the and bilateral and the of data of and were reported was in 15 17 families were with two or more affected The study was by the and was obtained from the patients and levels of Ca, and were analyzed using Additional urinary of Ca and Ca of per was considered for who were than The for and in spot in younger and were on the by et The control for levels with respect to GFR of to yr) with chronic renal failure tubular disorder. The GFR was calculated using the The loss of GFR was calculated from first and loss of only patients with a were Mutation Analysis CLDN16 mutation analysis was performed in families F29 by analysis and as described In all affected in analysis not both mutant CLDN16 was completely (F37, F61, and all new mutations, at least 100 control were In families F37 and F61, haplotype analysis was performed as described Expression The human claudin-16 was the The was performed with a for mutant were by were by of the cell with as described The were used for of MDCK and LLC-PK1 cells. The expression and localization of claudin-16 mutant proteins were by and with studies were performed on cell monolayers on as described The significance of difference between groups was by with and by as of different groups was compared by and not results are expressed as with 95% were used of Mutations are the presentation of CLDN16 mutations in protein model from affected by novel missense mutations are CL mutations are in PL mutations are in CLDN16 mutations resulting in trafficking localization of claudin-16 mutants in LLC-PK1 and MDCK cells. M71T is in the with a of the C131R shows a and indicating in the endoplasmic potential of CLDN16 missense mutations after heterologous expression in LLC-PK1 cells. are given as percentage relative to wild-type claudin-16 function. New mutations are the remaining mutations were from Hou et at of first clinical of renal failure expressed as loss of GFR per 1.73 are was performed only between the groups with categorized mutations on both alleles. < < of renal failure over in FHHNC to the mutation category. are levels with GFR in patients with FHHNC and control is indicated in analysis of the new patients with Functional analysis of CLDN16 mutations in LLC-PK1 Functional consequences of the CLDN16 correlation in FHHNC, correlation in are to the patients and families for We and for We also the and