Meckel–Gruber syndrome (MKS) is an autosomal recessive lethal malformation syndrome characterized by renal cystic dysplasia, central nervous system malformations (typically, posterior occipital encephalocele), and hepatic developmental defects. Two MKS genes, MKS1 and MKS3 , have been identified recently. The present study describes the cellular, sub-cellular and functional characterization of the novel proteins, MKS1 and meckelin, encoded by these genes. In situ hybridization studies for MKS3 in early human embryos showed transcript localizations in agreement with the tissue phenotype of MKS patients. Both MKS proteins predominantly localized to epithelial cells, including proximal renal tubules and biliary epithelial cells. MKS1 localized to basal bodies, while meckelin localized both to the primary cilium and to the plasma membrane in ciliated cell-lines and primary cells. Meckelin protein with the Q376P missense mutation was unable to localize at the cell membrane. siRNA-mediated reduction of Mks1 and Mks3 expression in a ciliated epithelial cell-line blocked centriole migration to the apical membrane and consequent formation of the primary cilium. Co-immunoprecipitation experiments show that wild-type meckelin and MKS1 interact and, in three-dimensional tissue culture assays, epithelial branching morphogenesis was severely impaired. These results suggest that MKS proteins mediate a fundamental developmental stage of ciliary formation and epithelial morphogenesis.
Johann Friedrich Meckel described the clinical entity that subsequently became known as Meckel syndrome in 1822 ( 1 ) in two newborn children presenting with occipital encephalocele, polydactyly, cleft palate and large cystic kidneys. The condition was called dysencephalia splanchnocystica by Gruber ( 2 ), John Marius Opitz further delineated the syndrome ( 3 ) and it has since been referred to as Meckel–Gruber syndrome (MKS). The classic triad of features associated with MKS are cystic kidneys (100% of cases), occipital encephalocele (90%) and polydactyly (<80%) ( 4 , 5 ). However, the clinical phenotype has been extended and altered since the disorder was first characterized and there is known to be a wide phenotypic variation. It has been suggested that at least two of the classic manifestations should be present to establish a definitive MKS diagnosis ( 6 ) or one classic feature and two other relevant anomalies ( 7 ). Central nervous system (CNS) abnormalities associated with MKS include Dandy-Walker malformation, agenesis of the corpus callosum, microcephaly, rhombic roof dysgenesis and prosencephalic dysgenesis. It has been suggested that a distinctive triad of CNS malformations (prosencephalic dysgenesis, occipital encephalocele and rhombic roof dysgenesis) should be included in the diagnostic criteria of MKS ( 8 ). Fibrocystic changes to the liver are possibly the only other constant finding, typically manifesting as impeded development of the intrahepatic biliary system and ductal plate malformation ( 4 , 7 ). Other associated MKS features include post-axial polydactyly, cleft lip/palate, laterality defects and congenital heart malformations including dextrocardia, shortening and bowing of the long tubular bones and abnormal development of the male genitalia. Survival beyond birth is unusual with a majority of cases dying in utero . Incidence is variable and high frequencies have been estimated in North Africa (1:3500) and Finland (1:9000) ( 5 ). In the USA, the incidence is reported to be 1:13 250 ( 9 ).
MKS displays locus heterogeneity and, to date, three loci have been identified. MKS1 was mapped to chromosome 17q21-q24 in an endogamous Finnish population ( 10 ) and, more recently, Kyttälä et al . ( 11 ) redefined the critical MKS1 locus to a region <100 kb through linkage disequilibrium and identified mutations in the novel gene FLJ20345/MKS1 . MKS1 encodes a protein, MKS1, of previously unknown function. MKS1 is 559 amino acids in length, lacks any predicted transmembrane helices but contains a B9 domain, the significance of which is currently unknown. Broad tissue expression of this gene was reported, although increased expression was identified in the brain, liver, kidney and the cartilage tissue of the developing digits of the upper limbs. MKS1 has been previously identified as a gene within the flagella and basal body proteome (FABB; see the ciliary proteome web server http://www.ciliaproteome.org/ ( 12 ), suggesting a link between MKS and ciliary or basal body function that is borne out by the clinical features ( 13 ).
Roume et al . ( 14 ) reported locus heterogeneity in eight families of North African and Middle Eastern ancestry. Subsequent autozygosity mapping in seven consanguineous families revealed a second MKS locus, MKS2 , on chromosome 11q13 ( 15 ). The region spans ∼8 cM and a causative gene is yet to be identified.
MKS3 was mapped to chromosome 8q24 ( 16 ) and, recently, we reported the identification of mutations in the novel gene, MKS3/TMEM67 , in a cohort of six MKS cases originating from Pakistan and Oman ( 17 ). In addition, a missense mutation was identified in the rat orthologue, LOC313067/Tmem67 , in the wpk rat model of MKS, which is characterized by polycystic kidney disease and CNS malformations, including hypoplasia to agenesis of the corpus callosum and severe hydrocephalus ( 18 ). At the RNA level, we found ubiquitous, moderate expression of MKS3 , with higher expression in embryonic adrenal gland, brain, kidney, lung and spinal cord. The predicted protein, meckelin, comprises 995 amino acids in humans, and although it is evolutionarily conserved with orthologues, for example, in Caenorhabditis elegans and Drosophila melanogaster , it has no known paralogues. The predicted structure of meckelin has some topological similarity with the Frizzled (FZD) family of receptor proteins, which are thought to have diverse roles in signal transduction during differentiation and development by the Wnt class of ligands [reviewed in ( 19 )]. A role for FZD4 in retinal angiogenesis has been suggested by the identification of mutations in a subset of patients with familial exudative vitreoretinopathy (EVR) ( 20 ). Frizzled receptors also have a role in establishing and maintaining epithelial planar cell polarity (PCP) during morphogenesis in both D. melanogaster and vertebrates [reviewed in ( 21 )]. In addition, the MKS3 promoter sequence contains an X-box motif, a sequence known to be involved in the regulation of primary ciliary genes in C. elegans ( 22 ), and therefore clearly implicates meckelin in mediating primary ciliary function.
Here, we provide the first evidence of the localization and function of the novel proteins MKS1 and meckelin. To determine the cellular and sub-cellular localization of the novel proteins, MKS1 and meckelin, we raised specific polyclonal antibodies for use in a series of western blotting, immunohistochemical staining and immunofluorescence microscopy studies. Furthermore, we used siRNA-mediated reduction of Mks1 and Mks3 expression in a ciliated cell-line to characterize the roles of these proteins in ciliogenesis.
In situ hybridization studies of MKS3
The expression of MKS3 mRNA was determined by a series of in situ hybridization experiments on human embryo sections (Fig. 1 ). A sense RNA control probe, transcribed from nucleotides 2289 to 3070, which contains much of the 3′-UTR of MKS3 , gave a negligible background signal (Fig. 1 A and C; data not shown). An antisense probe for SHH was used as a positive control for staining, with specific and restricted expression in the gut, notochord and floor plate of the CNS (data not shown). In the embryonic brain, an antisense probe for MKS3 revealed moderate to high levels of expression restricted to ventral midline regions of the posterior forebrain (Fig. 1 B), particularly the hindbrain floor (Fig. 1 E) and the developing sphenoid bone (Fig. 1 F). MKS3 was also expressed at moderate levels in post-mitotic neurons in the midbrain and posterior forebrain (Fig. 1 E). Intense expression of MKS3 was detected in the cartilage of developing limbs, particularly in the digits (Fig. 1 J'), the cartilaginous cervical vertebrae (Fig. 1 D, E, H, I and J), the bronchioles of the lung (Fig. 1 H), intestine (Fig. 1 D and I), stomach (Fig. 1 D, G and I) and kidney (Fig. 1 D, G and I). Expression was also noted in large intrahepatic bile ducts (Fig. 1 G and I) and in the outer layer of the developing retina (Fig. 1 K and K').
Immunohistochemical detection of meckelin and MKS1
To determine the cellular and sub-cellular localization of meckelin and MKS1, we developed polyclonal antisera against synthetic peptides. The specificities of the antisera were characterized by ELISA assays (data not shown) and western blotting, which showed that they detected single bands of expected size 70 kDa (for anti-MKS1) and 120 kDa (for meckelin) (Fig. 2 A). We also determined the absolute levels of expression of the Mks1 and Mks3 genes using quantitative real-time PCR with Hprt as a normalizing control. Mks1 and Mks3 were expressed at 41 and 46% of the level of Hprt , respectively. Hprt is between 0.01 and 0.1% of total cellular protein ( 23 ).
We visualized the location of meckelin and MKS1 in sections from fixed, paraffin-embedded liver and kidney tissue blocks (Fig. 2 B and C) obtained between weeks 18 and 20, at the time of termination of pregnancy, for two gestationally age-matched normal controls (Fig. 2 B) and three MKS cases (Fig. 2 C). Two of the MKS cases had MKS3 mutations that have been reported previously ( 17 ); the remaining case had a novel MKS1 frameshift mutation. In normal controls and in agreement with our in situ hybridization experiments for MKS3 (Fig. 1 ) and those reported previously for MKS1 ( 11 ), moderate to high levels of meckelin and MKS1 were localized at the proximal renal tubule epithelia, but not at the glomeruli (Fig. 2 B), and at the biliary epithelium of larger bile ducts (data not shown). Diffuse background staining of hepatocytes was also detected, but since this was also present in pre-immune negative controls we ascribe this to endogenous peroxidase activity. For MKS cases, MKS1 was not detected in the epithelial cell layer of the renal cysts for case 102, which had a homozygous MKS1 c.1448_1451dupCAGG duplication causing a frameshift mutation (p.T485fsX591) and is therefore a null allele. Complete loss of meckelin expression was also seen for case 67F, with a homozygous frameshift mutation c.647delA (p.E216fsX221), for which we have already confirmed the complete absence of MKS3 transcript in fetal liver tissue ( 17 ). Meckelin immunostaining was also absent for case 40T with a homozygous missense mutation c.1127A > C (p.Q376P), but we ascribe this to the accumulation of the mutant protein in the endoplasmic reticulum (ER) (see below). Expression of protein from the wild-type MKS gene was not affected in any of the three cases, with strong staining in the epithelial cell layer of renal cysts (Fig. 2 C).
Aberrant localization of mutant meckelin with missense mutation Q376P
As MKS3 encodes meckelin, a putative transmembrane receptor ( 17 ), we assessed the localization of HA-tagged wild-type meckelin and an HA-tagged mutant protein with the Q376P missense change in RCC4 and HEK293 cells, which are derived from proximal renal tubular epithelium and are therefore likely to process nascent proteins in a physiologically relevant fashion. Cell surface localization was apparent in permeabilized RCC4 cells that expressed wild-type meckelin, but not cells that expressed the Q376P mutant (Fig. 3 A). We quantified the number of HEK293 cells transfected with either wild-type meckelin ( n = 194) or meckelin Q376P ( n = 154), and also the number of transfected cells with either clear cell surface localization of HA-tagged protein, or only intracellular localization (Fig. 3 C). Protein with the missense mutation displayed statistically significant intracellular accumulation (unpaired t test; P < 0.001) that predominantly co-localized with anti-calreticulin antibody (Fig. 3 D, arrows), suggesting retention in the ER. We also determined some co-localization with EEA1, a marker of early endosomes (data not shown). Thus, cell-surface accumulation of the wild-type protein is consistent with its functional role as a transmembrane receptor, with a possible role for the extracellular domain of meckelin in mediating intracellular transport from the ER.
We labelled transfected cells with the membrane-impermeant biotin compound sulfo-NHS-SS-biotin to analyse cell surface processing of meckelin. While both HA-tagged wild-type meckelin protein and the Q376P mutant protein were expressed at high levels (Fig. 3 E), pull-down of biotinylated proteins with streptavidin-agarose demonstrated that only wild-type meckelin was biotinylated and therefore found at the cell surface (Fig. 3 E).
Meckelin localizes to primary cilia and MKS1 to centrosomes
We investigated the sub-cellular localization of endogenous MKS1 and meckelin in HEK293 cells and biliary epithelial cells (BECs). As expected, meckelin had a widespread distribution at the border of HEK293 cells (Fig. 4 A, arrowheads), and very high expression in BECs (data not shown, and Fig. 4 B), with co-localization at primary cilia indicated by intense staining with anti-acetylated α-tubulin (Fig. 4 A and B, white arrows). In contrast, MKS1, a putative cytoplasmic protein ( 11 ), had a broad intracellular localization with occasional distribution at the border of HEK293 cells (white arrowhead in Fig. 4 A), but was absent from primary cilia (marked by white arrows) in both HEK293 cells and BECs. However, MKS1 clearly displayed a punctuate doublet staining pattern characteristic of a centrosomal localization in BECs (green arrows in Fig. 4 B). To confirm this, we transiently transfected a number of cell-lines, including IMCD-3 cells, which are derived from mouse inner medullary collecting duct, and HEK293 cells, with Myc-tagged MKS1 and showed co-localization with anti-γ-tubulin (Fig. 4 C, white arrows), a marker of centrosomes and basal bodies (in post-mitotic cells).
MKS1 and meckelin are required for ciliogenesis
We tested the effect of silencing Mks1 or Mks3 on ciliogenesis. We used a pool of three to four siRNA duplexes targeted against either Mks1 or Mks3 . Each pool was transiently transfected into IMCD-3 cells, and the presence of cilia assessed 96 h later using anti-acetylated α-tubulin antibody as a ciliary marker. We used quantitative real-time PCR assays to determine that RNAi had been successful, with total transcript levels for Mks1 at 15 ± 2.3% that of negative controls, and Mks3 at 33.7 ± 2%. In contrast to wild-type (data not shown) and negative control siRNA-transfected (Fig. 5 A) cells, ciliary formation was markedly reduced in Mks1 - (Fig. 5 B) and Mks3 - (Fig. 5 C) silenced cells, although centrioles/basal bodies could still be observed with anti-γ-tubulin antibody. Sixty-five percentage of negative control-transfected cells showed ciliary staining, while Mks1 - and Mks3 -silenced cells displayed only 18 and 25% ciliary staining, respectively (Fig. 5 D). Silencing of the intraflagellar transport protein IFT-88 ( 24 ) was used as a positive control for loss of cilia; <10% of IFT-88 -silenced cells exhibited ciliary staining. Cilia in Mks1 - and Mks3 -silenced cells, when present, were significantly shorter than control cells (Student's t -test, P < 0.001), with an average length of 2.5 ± 0.16 µm for Mks1 -silenced cells and 4.3 ± 0.20 µm for Mks3 -silenced cells when compared with 7.4 ± 0.27 µm for negative control-transfected cells. Simultaneous silencing of Mks1 and Mks3 yielded the same phenotype as the individual silencing experiments.
We asked whether this effect on cilium formation could be caused by a general perturbation of apical–basal polarity. The establishment of cell–cell contacts is essential for epithelial polarity and tissue morphogenesis ( 25 , 26 ). We tested the effect of Mks1 - or Mks3 -silencing on the subcellular localization and distribution of the tight-junction marker ZO-1 and the adherens junction marker E-cadherin and found no difference in the localization or appearance of either marker compared with negative control-transfected cells (data not shown). We measured the transepithelial resistance across the monolayer of Mks1 -, Mks3 - and negative control-silenced cells. No significant differences were observed after 5 days of culture on Transwells (data not shown). Finally, we assessed whether Mks1 - and Mks3 -silenced cells retained the ability to differentiate their apical membrane and form microvilli. Scanning electron microscopy revealed the expected loss of cilia (Fig. 5 E–G. However, Mks1 or Mks3 siRNA-transfected cells additionally had fewer microvilli (Fig. 5 E'–G'). We quantified the number of microvilli present in a 25 µm 2 region at the centre of each cell and found significant differences between negative control-transfected cells and Mks1 - or Mks3 -silenced cells (Student's t -test result: P < 0.001 compared with control cells for Mks1 -silenced cells and P < 0.005 compared with control cells for Mks3 -silenced cells). An average of 71 ± 3.8 microvilli were present within the defined region in the negative control-transfected cells, while Mks1 - and Mks3 -silenced cells had an average of 39 ± 2.6 and 45 ± 3.1 microvilli, respectively, suggesting a possible polarization defect independent of the ability to form functional junctions.
All cilia are extended from a basal body ( 27 , 28 ). The primary cilium is extended from a single basal body analogous to the pre-existing mature centriole in the cell. In dividing IMCD-3 cells, the centrosome (two centrioles surrounded by a pericentriolar matrix) is localized in a juxtanuclear position closely associated with the nuclear envelope (data not shown). To form a primary cilium, the centrioles must migrate to and dock with the apical membrane by an as yet unknown mechanism. We therefore asked whether the inability to form a cilium was due to incorrect positioning of the centriole/basal body. Anti-γ-tubulin was used to mark the position of the centrosome. Z-stacks were taken every 0.5 µm and cells were scored according to whether the γ-tubulin labelling was observed in an apical position (within the first 2.5 µm of the cell, above the nucleus), in the mid-cell region or in the basal region of the cell (below the nucleus, in the last 1 µm of the cell). γ-Tubulin staining was present in the apical region of the cell in 58% of negative control-transfected cells (Fig. 6 A, arrow); however, the number of cells with an apically positioned centrosome was reduced to 28 and 24% in Mks1 - and Mks3 -silenced cells, respectively (Fig. 6 C). Instead, the γ-tubulin label was positioned in the mid-cell region (Fig. 6 B, arrow). As a control experiment, we assessed centrosome positioning in IFT-88-silenced cells as these do not form cilia (see above), but centriole migration should not be impaired. γ-Tubulin staining was apparent in the apical region of the cell in 63% of IFT-88 -silenced cells, indicating that intraflagellar transport is not required for centriole migration.
MKS1 and meckelin are required for cell branching morphogenesis
Under normal circumstances, sheets of epithelial cells within the kidney form an elaborate network of tubules. Given that one of the classic clinical manifestations of MKS is the formation of large cystic kidneys suggestive of a gross impairment of tubulogenesis, we asked whether silencing of MKS1 or MKS3 affected branching morphogenesis and tubulogenesis. Unlike other epithelial cell lines such as MDCK, IMCD-3 cells spontaneously form highly branched structures and tubules in three-dimensional culture ( 29 , 30 ). Branched structures are formed within 72 h of three-dimensional culture (Fig. 7 A). Tubules are just beginning to form at this stage, although apical–basal polarity is not yet fully determined (Fig. 7 B); they can be clearly seen after 10 days of collagen culture (data not shown). To test whether silencing MKS1 or meckelin would induce abnormal branching morphology, we transfected cell monolayers with the pooled siRNAs and 24 h later placed the cells into a three-dimensional collagen I matrix culture. Cultures transfected with either siRNA pool showed a decrease in the number of elongated and branched structures visible within the gel compared with the negative control-transfected cultures. Instead, cells aggregated into clumps with few small branches (Fig. 7 C and D). At the ultrastructural level, clusters of cells were visible adjacent to hollow cysts (Fig. 7 E, arrow marks the cyst). Elongated branched structures failed to form in 75–80% of MKS1 - or MKS3 -silenced cell colonies (Fig. 7 F) as opposed to around one-third of negative control-transfected cell colonies. This significant decrease in the frequency of branched structures ( χ2 test: P < 0.001, n = 150–300 cell colonies) suggests that reduction of MKS1 or meckelin expression prevents tubule formation by perturbing early branching morphogenesis in three-dimensional cultures.
Meckelin and MKS1 interact
The similar siRNA phenotypes observed for MKS1 and meckelin suggest that they may act in a common pathway. We determined that endogenous meckelin had a widespread distribution at the cell surface of HEK293 cells, where it co-localizes with a proportion of Myc-MKS1 (arrows in Fig. 8 A; top panel). In addition, although endogenous MKS1 was predominantly cytoplasmic, some could also co-localize with wild-type HA-meckelin at the cell surface (arrows in Fig. 4 A; third panel). We noted a statistically significant increase ( t -test: P < 0.05) in multinucleated cells over-expressing Myc-MKS1: over 70% of cells transfected and over-expressing Myc-MKS1 after 48 h were multinucleated (Fig. 8 A and B), whereas <20% of cells over-expressing MKS3 were multinucleated (Fig. 8 B). As MKS1 can localize at the cell surface (Fig. 8 A), we evaluated the possibility that MKS1, a soluble protein, can interact with meckelin, a putative receptor. We confirmed a biochemical interaction between meckelin and MKS1 in a series of co-immunoprecipitation experiments (Fig. 8 C). These clearly indicated that purified antibodies against meckelin, but not pre-immune serum or antibodies blocked with the cognate peptide, could immunoprecipitate epitope-tagged Myc-MKS1. We also confirmed the reciprocal interaction (immunoprecipitation of epitope-tagged HA-meckelin with anti-MKS1; Fig. 8 C).
Many cystic diseases of the kidney have been shown to result from defects in primary cilia, a rapidly expanding area of research. Watnick and Germino ( 31 ) originally suggested a unifying pathogenic mechanism for such disorders that, although clinically and pathologically heterogeneous, have a similar outcome. Hildebrandt and Otto ( 32 ) reviewed the role of defective primary cilia and basal body proteins in autosomal dominant and recessive polycystic kidney disease (ADPKD1/ADPKD2/ARPKD), nephronophthisis (NPHP), Alström syndrome (ALMS1) and Bardet–Biedl syndrome (BBS). BBS is an autosomal recessive disorder with a phenotypic overlap with MKS. Patients typically present with obesity, retinal degeneration, polydactyly, renal and gonadal malformation and behavioural and developmental problems ( 33 ). To date, 12 BBS genes have been identified ( BBS 1–12 ). The phenotypes are so similar at the prenatal stage that they may even be indistinguishable and mutations in BBS genes have been found in patients prenatally diagnosed with an MKS-like phenotype: kidney anomalies, polydactyly and brain anomalies, but not encephalocele ( 34 ). Recently, PCP signalling has been shown to be disrupted in BBS ( 35 ).
Our findings suggest that MKS results from defects in primary cilia and centrosome/basal body function, as expected from the ciliopathy phenotype ( 36 ). In situ hybridization revealed expression of MKS3 in the kidneys, liver and digits, which is in keeping with the typical MKS phenotype of cystic kidneys, liver dysplasia and polydactyly. Expression in the retina is consistent with other ciliopathies such as BBS. Retinitis pigmentosa is a common feature of BBS and can result from defective cilia in the rods and cones of the retina. The expression pattern in the brain is especially interesting, with high levels of expression at the midline, in addition to other locations. Defects of the midline, such as agenesis of the corpus callosum and cerebellar vermis, are frequent features of MKS and Joubert syndrome (JBTS). However, it is occipital encephalocele, a defect of the hindbrain, which is considered a diagnostic requirement of MKS. It is significant that in situ hybridization detected intense expression of MKS3 in both the hindbrain (Fig. 1 E) and the developing sphenoid bone (Fig. 1 F), which forms the base of the skull and joins the basal occipital bone. In contrast, arhinencephaly, a defect of the forebrain is common in MKS, but significant MKS3 expression was not detected at this location. As MKS3 expression is also intense in cartilage tissue, it seems more likely that the encephalocele of MKS arises due to defects in the formation of the occipital bone which, unlike the skull vault, develops via a cartilaginous phase ( 37 ). Therefore, as suggested previously ( 38 ), encephalocele develops as a result of herniation of the meninges and brain through a bony defect, and not from a neural tube closure defect per se . Expression of MKS3 at the midline reiterates the often-noted phenotypic overlap of CNS and other anomalies between MKS and JBTS. Thus, it is probable that these are allelic conditions, with MKS and ‘cerebello-oculo-renal syndrome’ ( 39 , 40 ) at the severe end of the phenotypic spectrum.
As expected, null alleles for both MKS3 and MKS1 , gives rise to a loss of meckelin and MKS1 expression in epithelial cells of the kidney. This is presumably due to nonsense-mediated decay, although we were unable to test this directly because the tissues did not provide useable total RNA. However, a missense mutation (Q376P) in the N-terminal extracellular domain of meckelin ( 17 ) results in an inability to localize the mutant protein at the cell surface, predominantly by retention in the ER. Possible mechanisms include the formation of an amino acid sequence that promotes ER retention ( 41 ), the occlusion of ER exit codes ( 42 ) or aggregation caused by local misfolding.
Our sub-cellular localization studies indicate that MKS1 is located at the centrosomes in dividing cells, and that meckelin can localize at primary cilia. However, meckelin also has a widespread distribution at the cell surface, as demonstrated by both immunofluorescence microscopy and biochemical assays of cell surface processing, and meckelin can interact directly with MKS1. We noted that cells transfected with MKS1 were multinucleated, suggesting that over-expression of this protein may mediate a defect in cell division through a dominant-negative effect and may implicate MKS1 in cell cycle processes. These observations suggest that both proteins have a number of functions, as yet not clearly defined, but with meckelin likely implicated as a membrane receptor involved in signalling events during development.
siRNA-mediated silencing of MKS1 and MKS3 revealed that both MKS1 and meckelin play important roles in ciliogenesis. We do not yet know whether there is a direct relationship between cilia and branching in these cells, or if the ciliation defect is a consequence of poor branching. While the topology of meckelin makes it is likely that meckelin mediates signalling events that regulate cilium formation, MKS1 could be a structural protein required for ciliogenesis, or it could be involved in signalling. Elucidation of the binding partners of MKS1 and meckelin, and determination of the function of the B9 domain found in MKS1 should help in resolving this issue.
The N-terminal extracellular domain of meckelin is particularly rich in cysteine residues, and is similar to the cysteine-rich domain (CRD) that is found in mammalian orthologues of the Drosophila frizzled ( fz ) receptor ( 43 ). In Drosophila , members of the frizzled family of tissue-polarity genes encode cell-surface proteins of the seven transmembrane-domain receptor (7TMR) class, and the CRD in the extracellular region has been implicated as a binding domain for Wnts, cysteine-rich glycoproteins that are homologues of Drosophila wingless ( 19 , 44 ). In mammals, the FZD family of WNT receptors have a pivotal role in various cellular processes including determination of cell fate, control of cell polarity and malignant transformation ( 45 ). Interestingly, PCP signalling has been shown to be disrupted in BBS ( 35 ) as well as other ciliopathies. Furthermore, mutations in the human frizzled-4 gene ( FZD4 ) cause an autosomal dominant form of familial exudative vitreoretinopathy (FEVR), which is a hereditary ocular disorder characterized by a failure of peripheral retinal vascularization ( 20 ), with loss of visual acuity due to hyperpermeable blood vessels and neovascularization. It remains to be determined if this pathogenic mechanism occurs during renal cytogenesis in MKS.
Centriole movements to the apical membrane are a critical early stage in ciliogenesis. The pathways that control centriole movements during ciliogenesis of the primary cilium are unknown; however, a large body of evidence implicates the actin cytoskeleton in centriole migration in multi-ciliated epithelial cells ( 46–50 ) and in the ciliated ctenophore Beroe ( 51 ). While the signalling events that regulate these processes remain obscure, recent evidence has implicated the PCP effectors inturned ( in ) and fuzzy ( fy ) in the formation of an actin array essential for basal body docking ( 52 ). The cytoskeleton is also likely involved in moving the mature centriole during ciliogenesis of the primary cilium, as microtubules position the interphase centrosome ( 53 ), while the Rho-associated coiled-coil containing protein kinase p160ROCK is required to maintain the position of the mature centriole in interphase cells ( 54 ). Here, we provide the first evidence of factors that control centriole positioning during primary cilium formation. The upstream and downstream pathways which MKS proteins impinge upon to direct redistribution of the centrioles to the apical membrane remain unknown and require further study, but may involve cell polarity pathways. We speculate that meckelin may act as a non-canonical Wnt receptor that links PCP signalling with actin or tubulin cytoskeleton rearrangments. During Drosophila development, the Drosophila Rho-associated kinase, Drok, connects frizzled-mediated PCP signalling to the actin cytoskeleton ( 55 ), and we suggest that ROCK signalling may be required during ciliogenesis. p160ROCK activity is found at the centrosome ( 54 ) and it is possible that MKS1 may provide the link between meckelin at the membrane and ROCK at the centrosome, as it can localize to both regions. Whatever the mechanism, it is likely that MKS1 and meckelin play similar roles in tissues containing primary cilia as well as multi-ciliated epithelia, as both types of tissue can be affected in MKS.
Our results demonstrate a fundamental role for both meckelin and MKS1 in primary cilia and basal body function. MKS therefore joins the increasing list of genetic diseases linked to cilia or basal body dysfunction. Further studies will help to decipher the mechanisms of pathogenicity of MKS, and the role of these proteins in the signalling pathways that mediate development.
MATERIALS AND METHODS
Affected individuals and families
All three MKS cases (67F, 40T and 102; see below) had a characteristic MKS phenotype. These comprised the presence of occipital encephalocele, renal cystic dysplasia and evidence of hepatic developmental defects (hepatic fibrosis, bile duct proliferation, ductal plate malformation). Informed consent was obtained from all participating families and the study was approved by the relevant local research Ethics Committees. DNA was extracted from blood, fibroblast cell lines and liver biopsy paraffin blocks using standard methods ( 56 ). We performed direct sequencing using the dideoxy chain termination method (ABI ‘BigDye 3.0’ system) on an ABI 3730 DNA Sequencer, and analysed sequences using Chromas v2.0 software. All mutations were verified bidirectionally.
Full-length MKS1 and MKS3 ORF clones for FLJ20345 and MGC26979 (IMAGE numbers 3952532 and 4825770) were obtained from Geneservice Ltd., Cambridge, UK for use as templates in PCR. For full-length inserts, amplimers were generated with forward primers that included the start codon, and reverse primers with the native stop codon. For MKS1 , the forward primer contained an Eco RI restriction site (5′-NNNNGAATTCTGGCGGAGACCGTCTGGAGC-3′-), and the reverse primer contained a Not I site (5′-NNNNGCGGCCGCCTAGGAGACCAGGGTTCC-3′) for sub-cloning into the pCMV-Myc vector (BD Biosciences Clontech, Palo Alto, CA, USA), giving the pCMV-Myc + MKS1 construct. For MKS3 , the forward primer contained an Xho RI restriction site (5′-NNNNCTCGAGCGGTTTGGTCCCTCTTA-3′-), and the reverse primer contained a Not I site (5′-NNNNGCGGCCGCTTAAATCAAAAATCTTTG-3′-) to allow sub-cloning into the pCMV-HA vector (BD Biosciences Clontech), to give the pCMV-HA + MKS3[wt; wild-type] construct. To mimic the effect of the MKS3 missense mutation c.1127A > C (p.Q376P), we generated a PCR amplimer, using the forward primer as above, and a reverse primer that contained an endogenous Bgl II site and a T > G mismatch (5′-ATTAAGATCTTAGAGATAGGAATCTCACAATTT G GTTGG-3′), that was sub-cloned into the full-length wild-type insert to give the pCMV-HA + MKS3[mt; Q376P mutant] construct. For in situ hybridization studies, a MKS3 cDNA fragment (nucleotides 2289–3070) was amplified by PCR using the following primers: 5′-TCGACAGTTCGTTGATTTATGC-3′ and 5′-AAAAATATGGCAAACTAACCTGA-3′. The PCR amplimer was sub-cloned into the pGEM-T Easy vector (Promega Ltd., Southampton, UK). All constructs were verified by bidirectional sequencing.
In situ hybridization
Normal human embryos were collected from terminations of pregnancy under the auspices of the Human Developmental Biology Resource ( http://www.hdbr.org/ ) with informed consent of the donors and Ethics Committee approval. In situ hybridization was performed on human embryo paraffin sections as described ( 57 ). Of a number of MKS3 cDNA fragments tested as probes for in situ hybridization, MKS3 (nucleotides 2289–3070), which contains much of the 3′-UTR of the gene, gave optimal results. Antisense and sense control RNA probes were produced by in vitro transcription from the T7 and SP6 RNA polymerase sites, respectively, of pGEM-T Easy vector using the ‘Riboprobe’ system (Promega). Expression patterns were analysed using the Axioplan 2 imaging system (Zeiss).
Cells and antibodies
Human renal clear cell (RCC4), human embryonic kidney (HEK293) and mouse inner medullary collecting duct (IMCD-3) cells were grown in Dulbecco's minimal essential medium (DMEM) or in DMEM/Ham's F12 supplemented with 5–10% fetal calf serum (Sigma–Aldrich Co. Ltd., Poole, UK). Biliary epithelial cells (BECs) were isolated and cultured in vitro as described previously ( 58 ). For functional analyses, IMCD-3 cells were cultured on Transwell permeable supports (#3413, Costar). The following primary antibodies were used: mouse anti-HA and anti-c myc, and rabbit anti-γ-tubulin (all Sigma–Aldrich); mouse anti-β-actin and rabbit anti-calreticulin (Abcam Ltd., Cambridge, UK); goat anti-EEA1 (Santa Cruz Technology, CA, USA); mouse anti E-cadherin (BD Transduction Laboratories), rat anti-ZO-1 (Zymed) and anti-acetylated alpha tubulin clone C3B9 ( 59 ). Secondary antibodies were: fluorescein isothiocyanate (FITC)-, tetramethyl rhodamine isothiocyanate (TRITC)- and horseradish peroxidase (HRP)-conjugated goat anti-mouse and anti-rabbit immunoglobulins, and rabbit anti-goat conjugates (Sigma–Aldrich), Alexa-fluor 488- and Alexa-fluor 594-conjugated goat anti-mouse and goat anti-rabbit (Molecular Probes) and rhodamine-conjugated goat anti-rat (Jackson Immunoresearch). CM-DiI was from Molecular Probes.
Polyclonal antibody production
Polyclonal antisera to MKS1 and meckelin were raised in rabbits by immunization with synthetic peptides: ETEGEKQELWKYTI-(C) and (C)-NLASKTLVDQRFLI, which correspond to internal sequence in MKS1 (residues 109–122) and the C-terminus of meckelin (residues 982–995). Cysteine residues (C) were included to enable covalent cross-linking. Antisera were raised commercially (Sigma–Aldrich Corp. and CovalAb UK Ltd., Cambridge, UK) and were characterized by ELISA and western blotting using standard methodologies ( 60 ). For co-immunoprecipitation experiments, we purified small volumes (2 ml) of antisera by passing them through a column containing protein A-sepharose beads (GE Healthcare UK Ltd., Little Chalfont, UK), essentially as described previously ( 61 ).
Whole cell extract preparation and western blotting
Whole cell extracts (WCEs) were prepared from HEK293 or IMCD-3 cells. Approximately, 5 × 10 7 cells were scraped into PBS, washed and resuspended in lysis buffer [50 m m Tris, pH 7.5, 150 m m NaCl, 0.5 m m EDTA, 0.5 m m EGTA, 0.02% (w/v) NaN 3 , 1.0% (v/v) Igepal C630 (Sigma–Aldrich), 10% (v/v) glycerol, 0.1% (v/v) protease inhibitor cocktail for mammalian cell extracts in DMSO, Sigma–Aldrich]. The cell suspension was incubated on ice for 30 min with gentle stirring, sonicated on ice and clarified by centrifuging at 13 000 g for 5 min. WCE was also prepared from transiently transfected cells for use in co-immunoprecipitation experiments (see below).
For western blotting, 10 µg WCE total protein was analysed by 8% SDS–PAGE and transferred to activated ‘Invitrolon’ PVDF membrane (Invitrogen) using 1 × CAPS transfer buffer, pH 11 ( 56 ). Proteins were immunodetected by standard methods ( 60 ). Rabbit polyclonal antisera or mouse monoclonal antibodies (MAbs) were used at final dilutions of 200 × to 1000 × for polyclonals and 1000 × to 5000 × for MAbs. Appropriate HRP-conjugated secondary antibodies (Sigma–Aldrich) were used (at 5000 × to 10 000 × dilutions) for detection by the enhanced chemiluminescence (ECL) plus western blotting detection system (GE Healthcare UK Ltd., Little Chalfont, UK).
Human fetal (weeks 18–20) tissues for immunohistochemical detection of MKS1 and meckelin were obtained at autopsy from either three MKS cases or two gestationally age-matched controls that had undergone therapeutic termination for intrauterine infection. MKS3 mutations have been described previously ( 17 ) in two of the MKS cases in the present study: 67F with frameshift mutation c.647delA (p.E216fsX221) and 40T the missense mutation c.1127A > C (p.Q376P). The third case, 102, with typical features of MKS, has a novel MKS1 c.1448_1451dupCAGG duplication causing a frameshift mutation (p.T485fsX591). Tissues were fixed in 10% formaldehyde and embedded in paraffin. Thin sections (4 µm) of liver and kidney tissues were cut onto slides, deparaffinized in two 5 min incubations of clean xylene and rehydrated by a 100, 95, 70 and 50% ethanol series. Epitope recovery was through boiling in 1 m m EDTA pH 8.0, by microwaving at reduced power, for 15 min followed by 30 min cooling. Endogenous peroxidase activity was inactivated by incubation in 3% H 2 O 2 /50% methanol for 30 min, followed by washing in TBST [50 m m Tris, pH 7.5, 135 m m NaCl, 2.7 m m KCl, 0.05% (v/v) Tween 20]. Sections were blocked in diluent buffer supplemented with 5% normal goat serum for 1 h at room temperature. Primary antibodies (rabbit polyclonal antisera at 250 × to 500 × dilutions in diluent buffer, matched to appropriate pre-immune negative control sera) were applied to the slides in a moist chamber for 18 h at 4°C. Slides were washed thoroughly with TBST, and the secondary antibody (goat anti-rabbit HRP-IgG conjugate; Sigma–Aldrich) was applied at a dilution of 250 × . Sections were developed in ‘Sigma Fast’ 3,3′-diaminobenzidine (DAB) with CoCl 2 enhancer and counterstained with Meyer's hematoxylin (Sigma–Aldrich). Sections were dehydrated in an ethanol series, cleared in xylene and mounted with ‘CV Mount’ (Leica Instruments GmbH, Nussloch, Germany).
For transient transfection experiments, cells were grown on glass coverslips (22 mm × 22 mm, placed in a six-well plate) to ∼80% confluence, transfected with 0.4 µg of plasmid DNA (pCMV-HA or pCMV-Myc empty vector, pCMV-HA + MKS3[wt], pCMV-HA + MKS3[mt] or pCMV-Myc + MKS1 constructs) using ‘Effectene’ reagent (Qiagen GmbH, Hilden, Germany) for 24 h. Transfection complexes were removed and cells were allowed to grow for a further 24 h in normal medium prior to fixation.
siRNA of MKS1 and MKS3
siRNA duplexes were designed against different regions of the Mks1 or Mks3 sequence using Dharmacon's custom SMARTpool siRNA service (mouse Tmem67; NCBI accession no. NM_177861): 5′-UAAAUCGGAAAAUCUCUUGUU-3′, 5′-AAAUACGUUCUCCAUAUGCUU-3′, 5′-UUUCUCGCCCACUUACUGCUU-3′, 5′-CAAUCUGAAACGUCUGCCCUU-3′ or Invitrogen's Stealth siRNA service (mouse Mks1 ; NCBI accession no. AK190930): 5′-AUUCUUUGCAGAUGAACUCGGAGGC-3′, 5′-GCCUCCGAGUUCAUCUGCAAAGAAU-3′, 5′-GUCAGUGUAAGUGAAGAUUCGCCGG-3′. The medium GC non-targeting negative control (Invitrogen) was used as a negative control, and SMARTpool siRNA against IFT-88 (mouse Tg737; NCBI accession no. NP_033402, Dharmacon) was used as a positive control for loss of cilia. siRNA was pooled and 100 pmol of each siRNA was simultaneously transfected into IMCD-3 cells at 50–70% confluency using Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions. Transfected cells were identified using a fluorescent oligo transfection marker (Invitrogen). Assays were carried out at 96 h after transfection. Quantitative real-time PCR was used to measure the decrease of Mks1 or Mks3 transcript levels following siRNA treatment, essentially as described previously ( 17 ). Two sets of independent assays were each run in triplicate. PCR analysis of cDNA was performed using denaturing HPLC-purified primers specific that spanned at least one exon–intron boundary for Mks1 (exons 6–7: forward 5′-GAGGGCAGCAAACTCAAGTC-3′ and reverse 5′-TTCACTGTGATCACGCCATT-3′), Mks3 (exons 22–24: forward 5′- GTTTGTTGACCTGTGCTCCA-3′ and reverse 5′-ATCTGAAACGTCTGCCCATC-3′) and Hprt as a normalizing control (exons 1–3: forward 5′-CCAGCGTCGTGATTAGCGATG-3′ and reverse 5′-ATAGCCCCCCTTGAGCACACAGAG-3′).
For transient transfection experiments and localization of MKS1 and meckelin, cells were fixed in 4% paraformaldehyde in PBS for 30 min and blocked in PBST buffer [1 × PBS, 0.1% (v/v) Tween 20] supplemented with 5% (v/v) normal goat serum. Epitope-tagged and/or endogenous proteins were detected by application of antibodies diluted 100 × in diluent buffer. Primary antibodies were detected by application of FITC- and/or TRITC-conjugated goat anti-rabbit or anti-mouse IgG antibodies diluted 250 × in diluent buffer. For immunolocalization studies on siRNA cells, cells were extracted in 0.75% Triton X-100 in 100 m m PIPES, pH 6.9, 2 m m EGTA, 1 m m MgSO 4 , 0.1 m m EDTA for 30 s and fixed in methanol at −20°C for at least 10 min. Samples were rehydrated in PBS for 10 min prior to incubation with primary and secondary antibodies. Anti-fade (Vectashield, Vector Laboratories Ltd., Peterborough, UK) containing DAPI was applied, and samples were examined with a Zeiss Axioplan or Axioplan 2 epifluorescence microscope equipped with a motorized stage and using a 100 × , 1.4NA oil immersion lens. Images were captured on a CCD camera controlled by either SmartCapture 2 software (Digital Scientific Ltd., Cambridge, UK) or Metamorph software (Universal Imaging), and processed in Metamorph and Adobe Photoshop (Adobe).
Immunoprecipitation of immunocomplexes
WCEs were prepared from HEK293 cells that had been transiently transfected with 2.0 µg of plasmid DNA [empty vectors as controls, and the pCMV-HA + MKS3(wt), pCMV-HA + MKS3(mt) or pCMV-Myc + MKS1 constructs] in 90 m m tissue culture dishes. WCE supernatants were processed for immunoprecipitation experiments using 5 µg purified anti-HA or anti-c-myc MAb, or 5 µg purified IgG fractions from rabbit polyclonal antisera, coupled to protein G- and/or protein A-sepharose beads (GE Healthcare UK Ltd.) essentially as described previously ( 61 ).
Biotinylation of cell surface proteins
Transiently transfected HEK293 cells in 90 mm dishes were grown to confluence, washed thoroughly with ice-cold 1 × PBS and incubated with 0.5 mg/ml sulfo-NHS-SS-biotin (Perbio Science UK Ltd., Cramlington, UK) for 30 min at 4°C. The reaction was quenched with 50 m m ammonium chloride. Cells were lysed and WCEs prepared as described above. Biotinylated cell surface proteins were recovered by binding to streptavidin-agarose beads (Sigma–Aldrich) for 2 h at 4°C. After washing, the beads were resuspended in SDS–PAGE loading buffer (containing fresh 2-mercaptoethanol), incubated at 37°C for 20 min and loaded on to 8% SDS–PAGE gels. Epitope-tagged meckelin was immunodetected by western blotting.
Measurement of transepithelial resistance
IMCD-3 cells were grown and transfected as above and cultured on Transwell filters for a further 4 days. Transepithelial resistance measurements were made using a Millicell-ERS meter (Millipore).
IMCD-3 cells were grown on 13 mm glass coverslips, transfected as above and cultured for a further 4 days. Cells for transmission electron microscopy were fixed by addition of 2.5% gluteraldehyde, 2% paraformaldehyde and 0.1% picric acid in 100 m m phosphate (pH 6.5) to the culture medium followed by post-fixation in 1% osmium tetraoxide in 100 m m phosphate buffer (pH 6.5) for 1 h at 4°C. The fixed material was stained en bloc with 2% aqueous uranyl acetate for 2 h at 4°C. Following dehydration through a graded series of acetone and propylene oxide, the material was embedded in epon resin for sectioning. Samples fixed as above were prepared for scanning electron microscopy essentially as described ( 62 ).
Culture of IMCD-3 cells in type I collagen gels
IMCD-3 cells were cultured in three-dimensional collagen gels as previously described ( 63 ). To assess branching morphogenesis in MKS1- or meckelin-depleted cells, IMCD-3 cells were grown on 30 mm tissue culture dishes (Falcon) to 30–50% confluency, transfected with appropriate siRNAs as above and left for 24 h before trypsinization and plating into collagen in 96-well tissue culture plates. Two hundred microlitres of tissue culture medium containing 10% fetal calf serum was added to the top of each solidified gel. After 72 h, gels were fixed in 2.5% gluteraldehyde, 2% paraformaldehyde and 0.1% picric acid in 100 m m phosphate (pH 6.5) for 2 h at room temperature, examined using 10 × phase contrast optics on a Zeiss Axioskop microscope and captured on a CCD camera controlled by IPLab software (Scanalytics Inc., BD Biosciences).
We would like to thank Mike Shaw (University of Oxford) for assistance with E.M., and the Queen Elizabeth University Hospital Liver and Hepatobiliary Surgical Unit in procurement of liver tissue. We thank Dr Esther Maina for advice on quantitative real-time PCR and Dr Paloma Garcia for the gift of mouse Hprt primers. Work in our laboratories was funded by the BBSRC, the Wellcome Trust, University of Birmingham Medical School ‘Scientific Projects’, Children's Liver Disease Foundation, Birmingham Children's Hospital Research Foundation, Birmingham Women's Hospital R&D Fund and the E.P. Abraham Trust. J.L.B. is supported by a grant from the Polycystic Kidney Disease Foundation. The Human Developmental Biology Resource (D.G. and A.J.C.) is supported by the MRC and Wellcome Trust. H.R.D. is a Beit Memorial Research Fellow; K.G. is a Wellcome Trust Principal Research Fellow.
Conflict of Interest statement . None declared.