Introduction

Arthrogryposis, renal dysfunction and cholestasis (ARC) syndrome (OMIM 208085) typically presents with neonatal cholestatic jaundice, renal tubular leak and hypotonia-related arthrogryposis. Other features variably reported include ichthyosis, mild dysmorphic signs, absent corpus callosum and recurrent infections resulting in severe metabolic acidosis, worsening nephrogenic diabetes insipidus and rarely liver failure. Significantly, a platelet storage pool defect similar to grey platelet syndrome is reported to occur in ∼ 25% of cases (Eastham et al. 2001). Previously, we mapped the ARC disease locus to 15q26.1 and identified 9 different germline VPS33B mutations in 14 of 15 kindred tested (Gissen et al. 2004). In order to further define the molecular pathology of ARC syndrome we undertook molecular genetic studies in a further 20 ARC syndrome kindreds and obtained detailed clinical information on 62 affected individuals from at least 14 different ethnic backgrounds providing the largest cohort of ARC patients yet analysed.

Patients and methods

Patients

Clinical data was recorded prospectively for living patients and information on deceased patients was obtained from the case notes and pathology records. Eleven of the 35 families in the current report have also been reported previously (Table 1). A clinical diagnosis of ARC syndrome was based on a triad of arthrogryposis, renal tubular dysfunction and cholestasis with a low γ-glutamyl transpeptidase (gGT) activity. Clinical details were available for 62 patients (35 males) of varied ethnic background (Table 1). Study protocols were approved by the South Birmingham research ethics committee.

Table 1 Clinical features in ARC patients
Full size table

Genotyping

DNA was extracted from blood by standard methods. To estimate the age of the most recent common ancestor for the p.Arg438X mutation we analysed DNA from eight affected individuals from eight unrelated consanguineous families from Mirpur region of Northern Pakistan. The extent of the common haplotype in the ARC locus was determined by genotyping fluorescently labelled microsatellite markers, which were identified using public genome databases and deCODE, Marshfield and Généthon genetic maps. Novel microsatellite markers from the ARC locus were identified using the current chromosome 15 draft genome sequence (NT_033276 and NT_033277). Patients’ DNA was amplified by PCR with novel primers flanking microsatellite markers as described previously (Gissen et al. 2004). PCR products were run on an ABI 3730 DNA analyser and then analysed with genemapper software package (ABI). In order to estimate population frequency of the individual marker alleles we genotyped DNA from 30 anonymised ethnically matched controls (UK Asians).

Mutation analysis

Mutation analysis was performed by direct sequencing of coding exons and flanking sequences (Gissen et al. 2004). All mutations were verified bidirectionally. Twenty distinct mutations (11 novel) were identified comprising 7 nonsense, 5 frameshift and 8 splice site mutations (Table 2). Segregation of the putative disease-causing mutations was analysed in the affected families. We used DNA from 100 anonymised controls to assess population frequencies of DNA variants. DNA from 100 anonymised healthy UK Asians was used as controls for the UK Pakistani patients. We used 100 anonymised samples from healthy UK Caucasians as controls for the patients with European origin. A mixture of 50 Caucasian and 50 Asian samples was used as controls for the Turkish and Tahitian patients with identified mutations. None of the presumed pathological DNA variants were found in the controls.

Table 2. VPS33B mutations in patients with ARC syndrome
Full size table

Estimating age of the most recent common ancestor

For estimating the age of the most recent common ancestor we applied a likelihood-based method, which uses multilocus marker data from a small number of patients suggested by Genin et al. (2004). We assume that all affected individuals in the sample descended from a common ancestor, who introduced the mutation n gen generation ago. The problem is to estimate n gen from the size of the haplotype shared by the individuals on each side of the disease locus D. The method is based on two functions: (a) the probability that no recombination occurred between D and marker M x located at recombination fraction θx from D, S(x) = (1−θ x )n and (b) the probability that one crossing-over occurred in the interval between marker M x−1 and M x , f(x) = S(x−1)−S(x). Allele frequencies of the markers and mutation rates are taken into account to allow for the possibility that recombination occurred in previous intervals but with haplotypes sharing the same alleles as the ancestral haplotype or to let haplotype diversity be due to mutations rather than recombinations.

Marker allele frequencies were obtained from a sample of 30 unrelated controls from the same ethnic origin as the patients. Because genetic distances may not be accurate between closely linked markers, recombination fractions between disease locus and the different markers were obtained from the physical distances between distant markers and their genetic distance in the Marshfield map. For the most distant markers of the map, D15S655 and D15S816, the physical distance was found to be approximately 7 Mb and the genetic distance (Marshfield) is 17.75 cM, giving 2.52 cM per Mb. However, when we considered separately each side of the mutation, very different results were obtained: from markers D15S655 to D15S127: we obtain 1.16 cM per Mb and from arc4 to D15S816, the correspondence is 4.20 cM per Mb. Therefore, analysis was performed using either the average 2.56 cM per Mb estimate or using the two side specific estimates and results were compared. The 95% confidence interval (95% CI) of the n gen estimate was computed using a Bayesian approach.

Results

Clinical features of ARC syndrome

The clinical features from 62 children with ARC syndrome are presented in Table 1. As expected the four major diagnostic features of ARC syndrome (arthrogryposis, renal tubular dysfunction and cholestasis with a low gGT activity) were present in the vast majority of cases, but in addition, all patients failed to thrive (struggled to maintain birth weight). Almost half of the patients were born small for gestational age (30/62, 48%) and in 10 pregnancies oligohydramnios was reported. Parenteral nutrition was attempted in one patient (Denecke et al. 2000, pedigree 17) and some weight gain was achieved before the child died of overwhelming sepsis age 7 months. Elemental feeds were attempted in one child (pedigree 34), which did not prevent failure to thrive (increase in 10% of birth weight at 8 months), but the child is alive at the age of 8 months and has not suffered severe infections (pedigree 34). 17 patients (27%) were reported to have intermittent episodes of diarrhoea but this was not persistent or severe. All patients had conjugated jaundice and the level of bilirubin in an individual patient fluctuated between extremely high (300 μmol/l) and normal levels. Liver disease was associated with normal gGT and normal/slightly elevated ALT and AST (less than twice the upper limit of normal). Fourteen out of 15 studied patients had non-excreting biliary isotope studies suggesting biliary obstruction or severe intrahepatic cholestasis. Liver biopsy features included bile duct paucity in 7/16 and lipofuscin deposition in 5/16. In 12 patients giant cell transformation of hepatocytes associated with neonatal hepatitis was reported. These changes were relatively mild and not associated with lobular disarray. One patient developed acute liver failure during an episode of sepsis at the age of 6 weeks (pedigree 11), characterised by raised bilirubin and hepatic transaminases, and deranged coagulation.

All patients had a degree of renal tubular dysfunction with aminoaciduria, occasional glycosuria and renal tubular acidosis. The severity of renal tubular acidosis and hypernatraemic dehydration worsened during the episodes of intercurrent illness. Renal ultrasound (performed in 14 patients) and biopsy (performed in 3 patients) findings included poor cortico-medullary differentiation in 6 patients, nephrocalcinosis in 6 patients and tubular atrophy in 2 patients. Renal ultrasound scan was reported as normal in four cases.

Thrombocytopaenia and/or abnormal platelet morphology on light microscopy were identified in 7/62 patients. Seven out of 16 patients who underwent organ biopsies suffered life threatening or fatal haemorrhage and in further 4 patients there was spontaneous severe bleeding (pulmonary haemorrhage in 2, gastrointestinal in 2). Nine out of the above 11 patients with bleeding episodes had normal platelet count and morphology reported. Abnormal platelet function studies were found in 4 out of 4 studied patients (pedigrees 1, 6, 22, 30, see Table 1), one of whom had normal platelet count and morphology on light microscopy. None of the patients with abnormal platelet function had episodes of spontaneous bleeding, but one underwent organ biopsy and suffered severe haemorrhage.

Four patients had severe anaemia requiring blood transfusions.

Neurological abnormalities

Arthrogryposis was present in all but two patients (pedigree 2, p.Arg438X mutation), but the severity ranged from isolated talipes to severe forms including congenital hip dysplasia. Corpus callosum dysgenesis and other intracranial abnormalities were found in nine patients.

Other abnormalities

Ichthyosis was noted in most patients. Dysmorphic features included large hands, proximally inserted thumbs, low set ears, sloping forehead and hirsutism. Congenital cardiac anomalies included atrial septal defect (2), ventricular septal defect (2) and patent foramen ovale (2). Deafness was detected in 4 patients. 3 patients were found to have hypothyroidism.

Genotype-phenotype correlation in ARC

No clear genotype-phenotype correlations were identified (Table 2). Most children died in the first 6 months (median survival 5 months, range 1 week–20 months), but the two patients who survived more than 1 year had novel VPS33B mutations. Both were from non-consanguineous families. A homozygous frameshift mutation c.556_557delCT (p.Leu175fsX219) was detected in a non-consanguineous Italian patient (pedigree 18). Although this child had the classical triad of renal tubular dysfunction, cholestasis and arthrogryposis, he had no infections until the age of 14 months when he developed fatal pneumonia. In addition to a hypoplastic corpus callosum, he had other intracerebral abnormalities noted on the MRI of the brain such as an increased T1 weighted signal in the basal ganglia, capsular, corticospinal and Rolandic areas bilaterally. In view of the apparently mild phenotype in association with an exon eight frameshift mutation, we considered the possibility that his mutation might affect splicing and have a less-than-predicted effect on protein function. We used “GENSCAN” software program to look for the splicing signal in the genomic DNA (http://www.genes:mit.edu/GENSCAN.html) and this predicted that the two base pair deletion would create a new splice site earlier in the exon that would alter protein sequence encoded by exon 8 of the new protein from Start>>>>HLLSTLYGPFPNCYGIGRCAKMAYEL>>>Stop to Start>>>>>HLLSTLWTLSKLLWNWQMAYEL>>>Stop, but thereafter the protein sequence would be unchanged.

A heterozygous splice donor site mutation (c.498+1 G>A) was identified in a child (pedigree 28) of non-consanguineous Swedish parents, who was born with severe renal tubular acidosis and cholestasis in the neonatal period. The patient had other typical features of ARC and also deafness and a hypoplastic corpus callosum. This child had severe failure to thrive despite nutritional supplementation and died at the age of 20 months due to pulmonary haemorrhage.

Evidence for a second ARC locus

Direct sequencing of all 23 exons and exon–intron boundaries of VPS33B gene did not identify any pathogenic changes in 7 (pedigrees 15, 17, 19, 23, 25, 26, 31) out of 35 families (20%). Patients from 3 of these families (15, 19 and 25), had typical features of ARC and came from non-consanguineous families that were unsuitable for linkage studies. In one family genotyping was consistent with linkage to VPS33B (pedigree 23), but three Turkish patients from consanguineous families (pedigrees 17, 26, 31) were heterozygous for VPS33B intragenic SNPs providing evidence for a possible second ARC locus.

Age of R438X mutation

The p.Arg438X mutation was identified in 12 Pakistani kindreds and we then genotyped 12 markers around the VPS33B gene (six markers on each side) in 8 patients who were not knowingly related (pedigrees in the families were traced at least 4 generations back, Table 3). The results were consistent with a common founder mutation and we then used the method of Genin et al. (2004) to estimate mutation age. Thus we performed two age estimations using either the average correspondence (2.52 cM per Mb) or using the two different correspondences on each side. (1) With 2.52 cM per Mb the age estimate is 39 generations (95% CI 23–70), when assuming a negligible mutation rate at the markers. If the mutation rate is 10−4 per marker per generation, the age estimate is 38 (95% CI 22–69).

Table 3 Haplotype analysis for the ARC region in patients with R438X mutation in VPS33B
Full size table

(2) With the two different distance correspondences on each side of the mutation (from markers D15S655 and D15S127: 1.16 cM per Mb and on the other side: 4.20 cM per Mb), the age estimate is 35 generations (95% CI of 20–62). As before, these results are not significantly affected by mutation rates. If we assume a mutation rate of 10−4, the age estimate is 34 generations (95% CI 20–61). This shows that estimates are concordant with the two sets of recombination fractions and if we assume that one generation is 25 years, then the common ancestor should be about 900–1,000 years old (95% CI 500–1,525).

Discussion

VPS33B encodes a 617 amino acid protein that is a homologue of yeast Vps33p, a class C vacuolar protein sorting (vps) protein known to be involved in multiple stages of a vesicular trafficking pathway. VPS33B belongs to the Sec1/Munc18 (SM) family of proteins that are known to be involved in SNARE-dependent vesicle targeting and fusion (Gissen et al. 2005). Immunostaining of renal and liver biopsy material from ARC patients identified mislocalisation of several apical membrane proteins (Gissen et al. 2004; Bull et al. 2006). Interestingly, mislocalisation of the MDR1 protein was also found in hepatocytes of zebrafish Vps18 mutant. Vps18 protein is also a class C vps protein, whose function is closely associated with Vps33 (Sadler et al. 2005). These findings suggest that class C vps proteins are involved in apical protein transport. Liver disease in ARC patients leads to cholestasis and bile duct paucity and knockdown of the vps33b ortholog in zebrafish embryo also leads to the bile duct paucity (Matthews et al. 2005). Thus VPS33B appears to be important in both function and development of the biliary tracts. Interestingly, embryonic expression of the zebrafish vps33b is particularly prominent in the liver and intestine but not in the kidneys, which is reflected in the fact that the vps33b knockdown zebrafish had no renal functional or morphological defects (Matthews et al. 2005).

We have characterised the clinical and molecular features of a cohort of 62 patients with a clinical diagnosis of ARC syndrome from 14 different ethnic backgrounds. All patients had severe failure to thrive (difficulty to maintain birth weight) and most died in the first 6 months of life. The severity of the failure to thrive observed exceeded that expected for the degree of liver and intestinal dysfunction and poses the question about the cause for such growth restriction. The identification of mislocalised apical membrane proteins suggests a possibility of a global abnormality in intestinal absorption and renal tubular reabsorption. Further research may identify the functional defect and collaborative studies could provide objective evidence for the use of special feeds which may overcome the reduced absorption of nutrients.

Most patients with ARC are reported to have skin ichthyosis. Interestingly some of the clinical features found in ARC patients overlap with those in the recently described syndrome cerebral dysgenesis, neuropathy, ichthyosis and keratoderma (CEDNIK) associated with a mutation in SNAP29 gene, coding for a SNARE protein (Sprecher et al. 2005). Patients with CEDNIK have severe ichthyosis associated with abnormal maturation and fusion of the lamellar bodies. Secretion of lamellar bodies is critical for epidermal cohesion and waterproofing (Elias et al. 2003). Lamellar bodies represent a type of secretory vesicles, which belong to a class of lysosome-related organelles analogous to the lamellar bodies of alveolar type 2 cells and also platelet alpha granules, found to be deficient in the ARC patients (Lo et al. 2005). Abnormal synthesis and function of lamellar bodies could be the cause of ichthyosis in ARC as VPS33B is involved in regulation of vesicular membrane fusion by interacting with SNARE proteins. Alternatively, abnormal intestinal absorption leading to nutritional deficiencies could also result in ichthyosis. Other similarities between ARC and CEDNIK include brain MRI findings and neurogenic atrophy found on muscle biopsy. Linkage to the CEDNIK locus was excluded in some ARC patients unlinked to the VPS33B.

Only a small number of patients survived beyond 1 year and it is possible that this might reflect partial retention of VPS33B function. The severity of arthrogryposis was variable even amongst those with the common p.Arg438X mutation. A small number of patients with ARC without clinically detected arthrogryposis have been reported elsewhere (Coleman et al. 1997; Bull et al. 2006). However, mutation analysis for patients reported by Coleman et al. (1997) was not available. Bull et al. (2006) reports a patient with classical ARC features, who had rocker-bottom feet but no arthrogryposis, homozygous for a mutation in VPS33B gene. We would have classified rocker-bottom feet as part of the arthrogryposis spectrum. Patients in pedigree 2 (see Table 1) had no clinically detected arthrogryposis unlike their two siblings and three cousins. Although arthrogryposis in ARC syndrome may be partially neurogenic in origin, the degree of arthrogryposis may depend on the fetal position and severity of oligohydramnios (reported in ten pregnancies). As polyuria is one of the cardinal signs of ARC, the oligohydramnios could not be accounted for by a fetal renal abnormality, thus the cause of this phenomenon may be a placental insufficiency resulting in oligohydramnios and also intrauterine growth retardation (reported in 48% of cases) secondary to a protein trafficking defect (Sanseverino et al. 2006).

We note that approximately 10% of ARC patients had a structural cardiac defect implicating VPS33B-related vesicular transport in normal cardiac development. In view of the susceptibility of the ARC patients to infection, the diagnosis of congenital cardiac anomaly should be sought in all ARC patients. Finding such abnormality would make physicians aware of the possibility of bacterial endocarditis.

Grey platelet syndrome results from an abnormal biosynthesis and function of platelet alpha granules and there was a spectrum of severity in this disorder. There were frequent episodes of severe bleeding leading to premature death in this cohort. These episodes occurred in patients without any morphological platelet abnormalities and so normal routine platelet analysis will not identify the risk of severe bleeding in suspected ARC syndrome patients. We recommend that organ biopsies should not be performed as a diagnostic procedure for ARC syndrome but should be replaced by VPS33B mutation analysis. If this is negative, then organ biopsy should only be considered after detailed platelet function studies and support (Lo et al. 2005).

The identification of specific mutations which occur at higher frequency in different ethnic groups (e.g., p.Arg438X in Pakistanis, p.Arg507X in Portuguese) will facilitate mutation detection. Interestingly we dated the p.Arg438X mutation to 900–1,000 years ago and this may correspond to when the original Arabic and Muslim settlers came to Kashmir (Wolpert 2000). Identification of the further locus for ARC syndrome will facilitate molecular genetic diagnosis and may offer insights into VPS33B function.