Delineation of the Clinical, Molecular and Cellular Aspects of Novel JAM3 Mutations Underlying the Autosomal Recessive Hemorrhagic Destruction of the Brain, Subependymal Calcification, and Congenital Cataracts
Nadia A. Akawi,
Department of Pathology, Faculty of Medicine and Health Sciences, United Arab Emirates University, Al-Ain, United Arab Emirates
Contract grant sponsors: Dubai Harvard Foundation for Medical Research (DHFMR) (2008-04); United Arab Emirates PhD scholarships programme, National Institutes of Health (NINDS) (RO1 NS035129).
Correspondence to: Lihadh Al-Gazali, Department of Pediatrics, Faculty of Medicine and Health Sciences, United Arab Emirates University, P.O. Box 17666, Al-Ain, United Arab Emirates. E-mail: firstname.lastname@example.org or Bassam R. Ali, Department of Pathology, Faculty of Medicine and Health Sciences, United Arab Emirates University, P.O. Box 17666, Al-Ain, United Arab Emirates. E-Mail: email@example.com, firstname.lastname@example.org
We have recently shown that the hemorrhagic destruction of the brain, subependymal, calcification, and congenital cataracts is caused by biallelic mutations in the gene encoding junctional adhesion molecule 3 (JAM3) protein. Affected members from three new families underwent detailed clinical examination including imaging of the brain. Affected individuals presented with a distinctive phenotype comprising hemorrhagic destruction of the brain, subependymal calcification, and congenital cataracts. All patients had a catastrophic clinical course resulting in death. Sequencing the coding exons of JAM3 revealed three novel homozygous mutations: c.2T>G (p.M1R), c.346G>A (p.E116K), and c.656G>A (p.C219Y). The p.M1R mutation affects the start codon and therefore is predicted to impair protein synthesis. Cellular studies showed that the p.C219Y mutation resulted in a significant retention of the mutated protein in the endoplasmic reticulum, suggesting a trafficking defect. The p.E116K mutant traffics normally to the plasma membrane as the wild-type and may have lost its function due to the lack of interaction with an interacting partner. Our data further support the importance of JAM3 in the development and function of the vascular system and the brain.
Junctional adhesion molecule 3 (JAM3; MIM #613730) is a member of the JAM subfamily of proteins, which include F11R (JAM1, JAM-A), JAM2 (JAM-B), JAM3 (JAM-C), and IGSF5 (JAM4) [Ebnet, 2008; Weber et al., 2007]. Other related members of this subfamily of proteins include endothelial cell selective adhesion molecule and coxasackie virus and adenovirus receptor (CXADR) [Weber et al., 2007]. JAMs and related proteins are localized at intercellular contacts and participate in the assembly and maintenance of tight junctions (TJs) and the control of cellular permeability [Ebnet, 2008]. TJs are morphologically distinct subcellular structures that are located closer to the apical side (compared with other junctional complexes) and are highly regulated areas of close contact between the plasma membranes of adjacent epithelial and endothelial cells [Balda and Matter, 2008]. TJs are important for the formation of polarized epithelial and endothelial barriers by forming intermembrane diffusion barriers and by controlling diffusion along the paracellular pathway [Cereijido et al., 2008]. It has also been found that TJs are important for intracellular signaling [Matter and Balda, 2007; Steed et al., 2009; Terry et al., 2010]. JAMs have been found to regulate adhesion between leukocytes and endothelial cells and the paracellular transmigration of leukocytes across the endothelium [Bradfield et al., 2007; Weber et al., 2007]. Moreover, JAMs play a role in cell polarization during spermatogenesis [Gliki et al., 2004].
In addition to JAMs, TJs contain other proteins including claudins, occludins, and tricellulins [Ebnet et al., 2004; Ebnet, 2008]. JAMs are single-span transmembrane proteins, whereas claudins, occludins, and tricellulins are tetraspan transmembrane proteins [Balda and Matter, 2008; Ebnet, 2008]. The tetraspan proteins form the paracellular permeability barrier and therefore determine the selectivity and extent of paracellular diffusion. JAMs on the contrary are postulated to mediate homotypic cell–cell adhesion [Bazzoni et al., 2000; Weber et al., 2007]. Several inherited conditions have been shown to be caused by mutations in genes encoding members of TJs protein complexes [O'Driscoll et al., 2010]. Hypomagnesemia has been shown to be caused by mutations in genes encoding claudins 16 and 19 [Konrad et al., 2006; Simon et al., 1999] and familial hypercholanemia to be caused by mutations in TJP2 and BAAT genes [Carlton et al., 2003]. Similarly, mutations in claudin 14 and tricellulin have been found to cause hereditary deafness [Riazuddin et al., 2006; Wilcox et al., 2001] and mutations in NHS to cause Nance-Horan syndrome [Burdon et al., 2003]. The variability in the clinical presentations reflects the variable functions of TJs in various tissues and organs.
We have recently identified a homozygous mutation in JAM3 in a consanguineous family from the United Arab Emirates (UAE) with a rare recessive disorder that is characterized by hemorrhagic destruction of the brain, subependymal calcification, and congenital cataracts (MIM #613730) [Mochida et al., 2010]. In this manuscript, we describe three additional novel mutations in JAM3 in patients from different ethnic backgrounds with similar phenotypes and elucidate the cellular mechanisms underlying some of those mutations.
Materials and Methods
All individuals participating in this study or their guardians provided written informed consents. Clinical assessments were performed by experienced clinical geneticists in Turkey, Spain, Australia, and the UAE. Full medical history was obtained. Participating members underwent careful clinical examination including imaging of the brain by magnetic resonance imaging (MRI).
Genomic DNA was extracted from peripheral leukocytes using the Flexigene DNA kit (Qiagen, Santa Clara, CA). Intronic primer pairs were designed to amplify individual JAM3 exons and its exon/intron junctions using Primer3 (http://frodo.wi.mit.edu/). Following PCR amplification, the PCR products were cleaned using the ExoSAP-IT reagent (USB Co., Cleveland, OH) followed by DNA sequencing using the BigDye Terminator v3.1 Cycle Sequencing kit and run on an ABI PRISM 3130xl DNA Analyzer (Applied Biosystems, Foster City, CA). Sequencing data were analyzed using clustalw2 referencing the NCBI GeneBank NM_032801.4 for JAM3 gene.
A transfection-ready JAM3-pCMV6 plasmid tagged C-terminally with Myc and Flag was purchased from OriGene Inc. (OriGene, Rockville, MD). The identified missense mutations (p.E116K and p.C219Y) were introduced by site-directed mutagenesis using Pfu Ultra high-fidelity DNA polymerase (Stratagene, La Jolla, CA) following the manufacturer's protocol. The primers used for site-directed mutagenesis are the following with the mutagenic nucleotide in bold:
For generating c.346G>A (p.E116K) we used
Forward: 5′ CCCTTTATCGCTGTAAGGTCGTTGCTCG 3′
Reverse: 5′ CGAGCAACGACCTTACAGCGATAAAGGG 3′
For generating c.656G>A (p.C219Y) we used
Forward: 5′ CTGGGCAGTACTACTACATTGCTTCCAATG 3′
Reverse: 5′ CATTGGAAGCAATGTAGTAGTACTGCCCAG 3′
All constructs were confirmed by DNA sequencing.
Human cervical cancer (HeLa) cell lines were cultured in DMEM (Invitrogen, Carlsbad, CA) supplemented with 10% heat-inactivated fetal bovine serum at 37°C with 5% CO2. Transient transfection of cells was achieved with FuGENE HD transfection reagent (Promega, Madison, WI) using 0.5 μg plasmid cDNA for confocal microscopy experiments. Cotransfection with hRas-GFP plasmid (0.2 μg cDNA) served as a transfection indicator and a plasma-membrane marker. All experiments were performed 24–48 hr following transfection.
Confocal Fluorescence Microscopy
Transfected cells were grown on coverslips (Ibidi, Martinsried, Germany) and fixed (5 min) with methanol (Sigma–Aldrich, St. Louis, MO) at −20°C. After three washes (5 min each) with PBS, cells were blocked with 1% BSA (Sigma–Aldrich) for 30 min. Membranes were incubated with primary antibodies at room temperature for 1 hr. Mouse antiflag (Sigma–Aldrich) and rabbit anticalnexin (Santa Cruz Biotechnology, Santa Cruz, CA) were used at 1:1,000 and 1:100 dilutions, respectively. After the primary antibodies incubations, the slides were washed three times (5 min each) then incubated for 1 hr at room temperature with secondary antibodies (fluorescence-labeled “Alexa-Fluor-568” [antimouse] and “Alexa-Fluor-488” [antirabbit]; Santa Cruz Biotechnology). Confocal microscopy was performed with a Nikon Eclipse system (Nikon Instruments Inc., Melville, NY). Control experiments without the primary antibodies or with nontransfected cells were carried out and revealed very low-level background staining.
Three families from different ethnic backgrounds (one Turkish, one Afghani, and one Moroccan) have been recruited to participate in this study (Fig. 1). Consanguinity is evident in two of the families and the clinical features of all the affected children are listed in Supp. Table S1.
Family 1: The parents of the affected child in this family are second cousins of Turkish origin. The parents are normal. Their first child (II1) was a preterm baby girl who had intracranial bleeding and died in the neonatal period. One of the mother's sisters died with similar phenotype but no clinical data were available. The second child (II2) of the couple was a female who was the product of normal pregnancy and term delivery. Her birth weight was 2,480 gm (<10th centile). She was admitted to the neonatal intensive care unit on day 5 because of the tonic seizures, which were resistant to treatment. Physical examination revealed a head circumference of 30 cm, slopping forehead, prominent, bulging anterior fontanelle, bitemporal grooving, and bilateral nuclear cataracts (Fig. 2A-1). Cranial ultrasound showed a prominent expansive bleeding. Brain computerized tomography showed calcification of the subependymal region and multifocal intraparenchymal hemorrhage (Fig. 2A-2). Serological tests were negative for toxoplasmosis, rubella, and cytomegalovirus. Cerebrospinal fluid PCR for cytomegalovirus and toxoplasma were also negative. Platelets levels were normal. The infant developed posthemorrhagic hydrocephalus and died at 2 months of age.
Family 2: The parents are nonconsanguineous and healthy of Afghani origin. Patient 1 (II1) was the first child of the family born at term by emergency caesarean section. His birth weight was 2,480 gm (<10th centile). He was noted to have bilateral cataracts and had feeding problems from birth. He was discharged home on day 4. At 2 weeks of age, he presented with vomiting, irritability, and a high-pitched cry. Brain MRI at the age of 15 days showed Grade IV intraventricular hemorrhage, severe ventricular dilatation, and white matter abnormalities in both cerebral hemispheres and multiple subependymal cysts (Fig. 2B). Renal ultrasound showed horseshoe kidneys with echogenic cortex and altered corticomedullary differentiation. He had progressive neurological deterioration and died at 2 weeks of age. Investigations were all normal and included respiratory chain enzymes on muscle and liver, 7 DHC, very long chain fatty acids, transferrin isoforms, blood karyotype, Affy 250K array, and coagulation study including platelets. Mitochondrial point mutations (MELAS, NARP, Leigh syndrome) were all negative.
Patient 2 (II3) was the third child of the family, with the family's second child being a healthy male. Patient 2's pregnancy was normal and delivery was by elective caesarean section at term. His birth weight was 3,315 gm (25th centile), length was 52 cm (50–90th centile), and head circumference was 35 cm (50th centile). At birth, he was noted to have bilateral cataracts. His tone and reflexes were normal and there was no clinical evidence of seizures. Cranial ultrasound showed periventricular echogenicity but there was no bleed. However, he deteriorated on day 5 and his hemoglobin dropped from 180 g/l to 96 d/l. He required ventilation. EEG on the third day of life showed bilateral epileptiform activity. Cranial ultrasound revealed extensive bilateral intracerebral hemorrhage. He died on the fifth day of life. Abdominal ultrasound showed pelvicalyceal dilatation of the right kidney. Full metabolic work up was normal.
Family 3: The parents are first cousins of Moroccan origin, who had a total of five pregnancies. The family history is significant in that the mother's brother (I3) died on the seventh day of life without any known diagnosis. This couple's first child was a male born at 36 weeks gestation who had cerebral hemorrhage, hydrocephalus, and died at 10 days of age. They then had two healthy children. The fourth pregnancy ended with a stillborn child who had hydrocephalus and the cerebral tissues showed necrotic degeneration. The index case (II5) is 28 days old female, the product of normal pregnancy. Fetal tachycardia was detected at 37 weeks gestation and therefore the baby was delivered by caesarean section. Her birth weight was 2,520 gm (10–25th centile), length 46.5 cm (25th centile), and head circumference 34.5 cm (50th centile). Bilateral cataracts were detected at birth both central and nuclear. On the second day of life, she presented with irritability, crying, and a bulging anterior fontanelle. Brain ultrasound showed dilated lateral ventricles, intraventricular hemorrhage in the right ventricle, and frontal periventricular cysts. On the eighth day of life, she was referred to another hospital for neurosurgical evaluation. Head CT revealed multifocal intraparenchymal hemorrhages, subependymal calcification, and diffusely hypodense brain parenchyma (Fig. 2C-1). MRI of the brain revealed intraventricular and intraparenchymal hemorrhage, edema, and cystic changes of the brain parenchyma, dilated ventricles, and thin corpus callosum with hemohrragic damage (Fig. 2C-2). There was normal circulation in the circle of Willis arteries. The EEG showed electrical status epilepticus, however, there were no seizures clinically. Metabolic screening was normal. Platelets levels were normal. It was decided that the child was not a candidate for ventricular shunt. She had severe anemia. Brain CT scan demonstrated bilateral subependymal calcification in the lateral ventricles. On the eighteenth day of life, her head circumference was 41 cm (+4 SD) and she was discharged for palliative care at home. She died at home the age of 39 days.
DNA sequencing revealed three different homozygous missense mutations in JAM3 gene in the three studied families (Table 1; Supp. Fig. S1). In family 1, we detected a nonsynonomous mutation (c.656G>A) in exon 6 of JAM3 gene, which led to an amino acid substitution at position 219 (p.C219Y). In family 2, we identified a single nucleotide substituation (c.346G>A) in exon 4 of JAM3 leading to a missense mutation at position 116 (p.E116K). The third mutation (c.2T>G) changed the initiator codon (methionine) of JAM3 leading to amino acid substitution to arginine (p.M1R). The three mutations were not reported in dbSNP or the 1000 genome project. All the amino acids affected by these mutations were highly conserved across species (Table 1) and were predicted to be disease causing by Polyphen2 (http://genetics.bwh.harvard.edu/pph2/), SIFT (http://sift.jcvi.org/), and Mutation taster (http://www.mutationtaster.org/). Furthermore, the detected mutations segregated with disease status in the three families; the parents were heterozygous, consistent with carrier status, and the available unaffected siblings were homozygous for the wild-type (WT) or heterozygous for the mutation. All the detected mutations were submitted to the LOVD database (www.lovd.nl/JAM3).
Table 1. Mutations Detected in JAM3
Likelihood of pathogenicity
aMutation nomenclature is based on NM_032801.4 and NP_116190.3 starting from A of the ATG as 1.
bThe nomenclature used by Mochida et al. (2010) based on NM_032801.3.
cConservation alignments for all the missense mutations are based on homoloGene results for JAM3 (ID: 83700), but the splice site mutation alignments were adopted from Mutationtaster (http://www.mutationtaster.org/) results.
C219Y Mutation Results in JAM3 Trafficking Defect
To further understand the pathophysiological significance of the two detected missense mutations (p.E116K and p.C219Y), we employed confocal fluorescence microscopy to establish their subcellular localization relative to the WT-JAM3 protein. HeLa cells, known to lack endogenous expression of JAM3 [Betanzos et al., 2009], were either transfected with individual JAM3 constructs (WT, p.E116K or p.C219Y) or cotransfected with GFP-hRas. The cells were then stained with antiflag antibodies to visualize JAM3 constructs. The cells transfected with JAM3 only constructs were costained with anticalnexin antibodies to visualize the endoplasmic reticulum (ER) network. As shown in panels A–F of Figure 3, the WT JAM3 is predominantly localized to the plasma membrane and colocalized largely with the GFP-hRas. Similarly, the p.E116K is predominantly localized to the plasma membrane as shown in panels G–L of Figure 3. On the contrary, the p.C219Y mutant colocalized predominantly with the ER marker (calnexin) and not with GFP-hRas indicating its retention in the ER (Fig. 3, panels M–R, Supp. Fig. S2).
The Impact of JAM3 Mutations on the Protein Product
Arrate et al. (2001) and Santoso et al. (2002) cloned and characterized JAM3 as an approximately 43 kDa transmembrane glycoprotein. This type I integral membrane glycoprotein consists of two immunoglobulin (Ig)-like domains with intrachain disulfide bonds at the extracellular region. By sequence analysis, Arrate et al. (2001) concluded that the N-terminal domain is a V-type, whereas the membrane proximal domain is a C2-type. Also they predicted that the intrachain disulfide bonds that stabilize each Ig domain are between C53–C115 and C160–C219. Ig-like domains are implicated in a range of vital functions, including cell–cell recognition, cell-surface receptors, muscle structure, and the immune system [Teichmann and Chothia, 2000]. The C-terminal region of the protein is composed of a short transmembrane domain followed by an intracellular domain containing a PDZ binding motif (Supp. Fig. S1A).
The p.C219Y mutation detected in family 1 alters one of the four highly conserved cysteine residues of JAM3 disrupting one out of two disulfide bonds known to stabilize the Ig-folds of the protein [Arrate et al., 2001; Santoso et al., 2002]. Several functional studies demonstrated that cysteine-substitutions disrupt the structure of cell surface proteins causing their misfolding, instability, and ER retention [Fukuda et al., 2011; Rudarakanchana et al., 2002]. In consistent, comparison of the subcellular localization of flag-tagged C219Y mutants transiently expressed in HeLa (Fig. 3; Supp. Fig. S2A) and COS7 (Supp. Fig. S2C) clearly showed that most of the mutated protein failed to reach the cell membrane. ER retention is a well established disease causing mechanism in many human syndromes caused by loss-of-function mutations [Aridor, 2007; Ali et al., 2010; 2011; Chen et al., 2005].
The p.E116K mutation found in family 2 did not seem to affect the subcellular localization or trafficking of the protein (Fig. 3). However, its position in the Ig-like V-type domain suggests an interfering role of this mutation with the dimerization or with the interaction(s) of this protein with one or more of its interacting partners. JAM3 controls TJ maintenance by engaging in homophilic and heterophilic interactions with neighboring JAM molecules [Ebnet et al., 2008].
Arrate et al. (2001) and Santoso et al. (2002) predicted the first ATG by homology alignments between JAM3, JAM1, JAM2, and mouse Jam3. Arrate et al. (2001) found using SignalP V1.1 the signal sequence of JAM3, which should be cleaved after glycine 30 from the putative ATG. The mutated methionine in family 3 is predicted to be the first methionine of JAM3 protein and such mutations almost invariably interfere with the translation of the protein. However, there are two in-frame ATG's upstream of this start codon and the improbable use of one of them or for the next in-frame ATG would encode a non-functional or unstable protein.
Clinical Consequences of JAM3 Deficiency
Mouse Jam3 is expressed at the apical junctions of endothelial cells, smooth muscle cells, fibroblasts, Schwann cells, spermatids, and hematopoietic stem cells. More recently, Jam3 expression was detected in the TJs of neural stem cells in the embryonic ventricular zone and the adult ependymal cell layer of mouse brain [Stelzer et al., 2012]. More than half of mice with homozygous mutation in Jam3 exhibit postnatal lethality. Jam3 deficient mice exhibit growth retardation, muscle weakness, abnormal spermatogenesis, granulocytes homeostasis defects, electrophysiological defects, and hypersensitivity to mechanical stimuli [Colom et al., 2012; Scheiermann et al., 2007]. Furthermore, mouse Jam3 apical expression was found on the embryonic retinal neuroepithelia, and its deficiency caused nuclear congenital cataracts [Daniele et al., 2007]. The brains of mice lacking Jam3 are abnormal in some strains [Wyss et al., 2012] but relatively unaffected in others [Stelzer et al., 2012].
In humans, Mochida et al. (2010) demonstrated that a homozygous mutation in JAM3 is the cause of a severe autosomal recessive syndrome segregating in a consanguineous family. The affected members in this family were born with congenital cataracts and developed severe brain abnormalities from progressive hemorrhagic destruction of the brain tissue, including the cerebral white matter and basal ganglia. Three out of six affected individuals died early in infancy. Brain imaging in all the affected showed calcification in the subependymal region of the brain, multifocal intraparenchymal hemorrhage with associated liquefaction, and massive cystic degeneration resulting in large ventricles [Al-Gazali et al., 1999; Mochida et al., 2010]. The phenotype of the children in this report is very similar to the phenotype of the affected children in this previously reported family. All the affected children in this report had bilateral cataracts and develop progressive neurological impairment, associated with seizures in some individuals. In addition neuroimaging in all patients showed intraparanchymal hemorrhage with evidence of brain destruction and subependymal, basal ganglia, and white matter calcification. These changes are very similar to those found in the original family. Some of the other features described in the original family, such as renal anomalies and liver enlargement, were more variable in this patient group. All the affected children in this report died. Therefore, of the total 10 cases reported so far the disorder was lethal in 7 cases. Death occurred in the first few weeks of life. Those who survived (three in total) are severely retarded with severe spasticity and seizures.
Other conditions that could be considered in the differential diagnosis include the syndrome of band-like intracranial calcification, simplified cerebral gyrai, and polymicrogyria [Abdel-Salam et al., 2008; O'Driscoll et al., 2010]. This syndrome is caused by mutations in another TJ gene, occludin (OCLN; MIM #602876). However, patients with this syndrome do not have intracranial hemorrhage or congenital cataract [Mochida et al., 2010]. Furthermore, our patients did not show any evidence of developmental malformation of the cerebral cortex and polymicrogyria. Similarly mutations in collagen type IV alpha (COL4A1; MIM #120130) can cause overlapping clinical presentations, which include antenatal intracranial hemorrhage [de Vries et al., 2009], brain small vessel disease with hemorrhage or vascular leukoencephalopathy, porencephaly [Gould et al., 2006; Sibon et al., 2007], and hereditary angiopathy with nephropathy, aneurysms, and muscle cramps [Plaisier et al., 2007]. All these disorders are inherited as autosomal dominant and the clinical presentations is different from the cases in this report. Proliferative vasculopathy and hydranencephaly–hydrocephaly syndrome (Fowler syndrome) is an autosomal recessive perinatally lethal disorder characterized by hydrocephalus associated with progressive destruction of the central nervous system (CNS) tissue as a result of an unusual and characteristic proliferative vasculopathy [Fowler et al., 1973]. The hallmark of the syndrome is the microvascular proliferation, which is associated with extensive necrosis and calcification of the CNS tissue. There are no visceral malformations [Lalonde et al., 2010]. It is caused by mutations in the FLVCR2 gene (MIM #610865). Clinically affected fetuses present with fetal akinesia deformation sequence with muscular neurogenic atrophy [Fowler et al., 1973; Meyer et al., 2010], which is different from the clinical presentation of the cases in this report.
Our study revealed several missense mutations scattered throughout the coding sequence of JAM3 affecting different domains yet the affected children shared similar phenotype indicating homogeneity of this disorder at the clinical and molecular level. This is likely due to the total loss of the protein function as a result of those mutations.
Our study confirms the importance of JAM3 as a component of the junctional complexes and its deficiency leading to a distinctive and catastrophic neonatal presentation of cataracts and hemorrhagic destruction of the brain.
We are grateful to the participating families and clinicians. We thank Davinia VAZQUEZ SANCHEZ, Ramon VIDAL SANAHUJA in Hospital de Tarrasa (Barcelona) for their help. We thank Dr. A. James Barkovich for his help in interpreting brain imaging studies. C.A.W. is an Investigator of the Howard Hughes Medical Institute.
Disclosure statement: The authors declare no conflict of interest.