The primary cilium is an antenna-like structure extending from the surface of most vertebrate cells. Loss or mutation of ciliary proteins can lead to polycystic kidney disease and other developmental abnormalities. inv mutant mice develop multiple renal cysts and are a model for human nephronophthisis type 2. The mouse Inv gene encodes a 1062-amino-acid protein that is localized in primary cilia. In this study, we show that the Inv protein (also known as inversin) is localized at a distinctive proximal segment of the primary cilium, using GFP-tagged Inv constructs and anti-Inv antibody. We named this segment the Inv compartment of the cilium. Further investigation of the Inv protein showed that 60 amino acids at its C-terminal, which contains ninein homologous sequences, are crucial for its localization to the Inv compartment. Fluorescence recovery after photobleaching analysis revealed that the Inv protein was dynamic within this compartment. These results suggest that localization of the Inv protein to the Inv compartment is actively regulated. The present study revealed that the primary cilium has a distinct molecular compartment in the body of the primary cilium with a specific confining and trafficking machinery that has not been detected previously by morphological examination.
A primary cilium is an antenna-like structure extending from the surface of most vertebrate cells. Loss of cilia or mutation of ciliary proteins leads to developmental defects such as situs inversus, skeletal abnormalities, blindness, obesity and polycystic kidney disease, as well as hepatic and pancreatic cysts (Fliegauf et al., 2007). Two types of primary cilia are known: motile and non-motile. It is now clear that motile primary cilia in the primitive node control left-right asymmetry of the body, and a defect in motile primary cilia results in situs inversus (Hirokawa et al., 2006). By contrast, the role of non-motile primary cilia is not well understood. They are proposed to act as chemical or mechanical sensors (Marshall and Nonaka, 2006; Singla and Reiter, 2006). Primary cilia are also proposed to be involved in intracellular signaling pathways such as the Shh and Wnt pathways (Attanasio et al., 2007; Corbit et al., 2005; Corbit et al., 2008; Thoma et al., 2007). Among the primary-cilia-associated abnormalities, renal cystic disease is the most commonly observed. Renal cysts are characterized by an expansion of renal tubules. Renal epithelial cells have a non-motile primary cilium that is thought to sense urinary flow and transmit a signal to the cytoplasm (Nauli et al., 2003; Praetorius and Spring, 2001; Schwartz et al., 1997). Mutations of ciliary proteins are thought to affect structure and function of the cilia, and can cause renal cystic disease (Hildebrandt and Zhou, 2007; Yoder, 2007). However, how ciliary proteins function, where they are localized and how they interact with each other in the primary cilium are largely unknown.
The primary cilium is structurally divided along its vertical axis into sub-compartments that include the ciliary tip, the body (or shaft), the ciliary necklace segment, the transitional zone and the basal body (Gilula and Satir, 1972; Wheatley, 1967; Wheatley, 1995). It is covered by a membrane and inside has the axoneme. Intraciliary proteins that participate in cilia signaling are thought to be localized at a sub-compartment of the primary cilium, where they exert their function. A first step in the understanding of how ciliary proteins are involved in cilia signaling is to determine where these intraciliary proteins are localized in this complex structure.
The inv mouse mutant shows situs inversus associated with multiple renal cysts (Mochizuki et al., 1998; Morgan et al., 1998; Yokoyama et al., 1993). Mutation in the INVS gene in human was later found to cause nephronophthisis type 2 (NPHP2) (Otto et al., 2003). No abnormality is observed in primary cilia in the kidney of inv mice (Phillips et al., 2004). The length of primary cilia and the bending mechanics of primary cilia in response to physiological fluid flow are almost identical for renal epithelial cells from wild-type and inv mutant mice (Shiba et al., 2005). Therefore, loss of Inv does not affect primary cilium structure, suggesting that the Inv protein has a functional, rather than a structural, role in primary cilia. Although Inv was localized in primary cilia (Otto et al., 2003; Watanabe et al., 2003), detailed intraciliary localization and dynamics have not been examined so far.
The present study used Inv-GFP fusion constructs as well as high-resolution immunocytochemistry and immunoelectron microscopy to examine Inv localization in the primary cilium. Our study showed that Inv was confined to a proximal segment of the primary cilium. We defined this segment as the Inv compartment of the cilium. We then generated a series of truncated Inv-GFP fusion constructs and identified sequences essential for targeting and confining this protein to the Inv compartment. We identified 60 C-terminal amino acids that are essential for localization to the Inv compartment. Finally, fluorescence recovery after photobleaching (FRAP) analysis revealed that Inv is dynamic within the Inv compartment. Our study proposes that the primary cilium has molecule-based compartments in addition to morphological compartments.
The Inv-GFP signal is confined to a proximal segment of the body of the primary cilium
Inv-GFP mice are inv mutant mice carrying a GFP-tagged full-length Inv transgene, which rescues all the associated inv phenotypes (Watanabe et al., 2003). Therefore, the Inv-GFP construct is functional and localization studies should thus reveal its functional site. First, we examined the in vivo localization of the Inv-GFP protein. Renal tubules isolated from Inv-GFP mice were immunohistochemically stained with anti-acetylated α-tubulin antibody. Acetylated tubulin staining and the Inv-GFP fluorescent signal overlapped partially, showing that the Inv-GFP protein is localized to a specific part of the primary cilia in vivo (Fig. 1A). Localization of GFP-tagged Inv was further analyzed using renal epithelial cells derived from Inv-GFP mice (supplementary material Fig. S1). In living cultured Inv-GFP cells, the primary cilia extended perpendicular to the apical membrane and were seen as a dot when examined through the microscope (Fig. 1B). The Inv-GFP fluorescent signal was detected at the base of primary cilia (Fig. 1B) and not in the cytoplasm or plasma membrane. Confocal laser scanning z-axis images of primary cilia and Inv-GFP fluorescence are available as a supplemental movie (supplementary material Movie 1). The side view also showed Inv-GFP signal at the base of primary cilia (Fig. 1B). Anti-γ-tubulin immunohistochemistry stained the centrosome and the basal body, and showed that the Inv-GFP signal did not overlap with the γ-tubulin signal, indicating that Inv-GFP is not localized in the centrosome or basal body (Fig. 1C). Line-scan analysis confirmed that the Inv-GFP signal was detected at the base of primary cilia, but not in the basal body (Fig. 1C). The localization pattern of polycystin-2 (the gene product of pkd2), the causal gene product of autosomal dominant polycystic kidney disease, was different from the Inv-GFP signal (Fig. 1C).
Inv immunocytochemistry and immunoelectron microscopy
We next examined endogenous Inv localization using the anti-Inv antibody in renal epithelial cells derived from wild-type and inv mutant mice (supplementary material Fig. S1). Renal cell lines derived from inv mice were used as a negative control. Inv immunostaining was detected at the base of primary cilia (Fig. 2A). The staining partially overlapped with acetylated tubulin immunostaining, but did not overlap with γ-tubulin immunostaining, confirming the Inv-GFP findings (Fig. 1). Line-scan analysis also indicated that the endogenous Inv staining was detected at the base of the primary cilia but not in the basal body. By contrast, there was no positive immunostaining in primary cilia from inv cells, excluding the possibility that this ciliary signal is a false-positive reaction (Fig. 2A).
Immunoelectron microscopy confirmed the positive Inv labeling in the segment of the primary cilia body, and labeling was observed between the ciliary membrane and the axoneme (Fig. 2B). No signal was detected in the basal body (Fig. 2A,B). The area between the Inv-positive segment and the basal body is probably the ciliary necklace segment and transitional zone. Thus, Inv is specifically localized to a proximal segment of the primary cilia, but not in the ciliary necklace segment, transitional zone or basal body. We named the proximal Inv-positive segment in the primary cilium the Inv compartment.
The Inv C-terminus contains a compartment-targeting signal
Inv is composed of 1062 amino acids and has N-terminal ankyrin repeats and two IQ domains (Fig. 3A). To search sequences to target the Inv protein to the Inv compartment, we generated three GFP fusion constructs: a full-length Inv-GFP fusion construct [Inv (1-1062)], an N-terminal Inv-GFP fusion construct [Inv (1-743)] and a GFP-fused C-terminal Inv construct [Inv (742-1062)]. Inv (1-743) and Inv (742-1062) contained the regions coding for amino acids 1-743 and 742-1062 of the Inv protein, respectively. In cells transfected with the GFP-tagged full-length Inv fusion construct [Inv (1-1062)], the GFP signal was observed at the base of primary cilia (Fig. 3B) in all ciliated cells examined (n=101). By contrast, cells transfected with the N-terminal Inv-GFP fusion construct [Inv (1-743)] showed a weak or no GFP signal at the primary cilium, and only 27.2% of ciliated cells showed a GFP signal at the primary cilium (n=55) (Fig. 3B). We generated an Inv (1-418) construct that contains a region coding for amino acids 1-418 of Inv fused to GFP. The construct did not show any GFP signal within the primary cilium, even with increased exposure during image acquisition (data not shown). The Inv (742-1062) construct showed a strong GFP signal at the primary cilium in all ciliated cells examined (n=66) (Fig. 3B). These results suggest that Inv has at least two cilia-targeting signals: the first one between amino acids 1-743, probably between amino acids 418-743, and the second one between amino acids 742-1062.
Although the GFP signal in cells transfected with both the Inv (1-743) and Inv (742-1062) constructs was observed in cilia, the patterns were quite different for each construct. The GFP signal in cells transfected with Inv (1-743) was observed over the entire length of the cilia with a punctate appearance (Fig. 3C). Inv (742-1062) showed a GFP signal resembling that of the full-length Inv-GFP protein: a uniform signal at the base of the cilia (Fig. 3C). The GFP signal in cells transfected with Inv (742-1062) showed the same pattern as endogenous Inv immunostaining, suggesting that Inv (742-1062) is localized to the Inv compartment. Therefore, information for targeting the Inv protein to the Inv compartment should lie between amino acids 742-1062.
Two regions in the C-terminal Inv fragment are required for ciliary localization
To investigate which part of the sequence determines the targeting of Inv to the Inv compartment, we generated a series of truncated C-terminal Inv constructs (Table 1). The Inv protein has two IQ domains, one at amino acids 556-575 (IQ1) and the other at amino acids 915-935 (IQ2) (Fig. 3A). The IQ2 domain is present in the Inv (742-1062) construct. Inv (742-1062: ΔIQ2) encodes the same region as Inv (742-1062), but lacks the IQ2 domain. The Inv (742-843) construct has GFP fused to the region coding amino acids 742-843 of Inv. Deletion of the IQ2 domain (amino acids 915-935) resulted in the disappearance of the ciliary targeting in all ciliated cells examined (n=30), indicating that the IQ2 domain is required for ciliary localization of the Inv (742-1062) protein (Fig. 4A). We next generated five additional GFP-tagged C-terminal Inv constructs: Inv (842-1002), Inv (742-1002), Inv (842-1031), Inv (915-1062) and Inv (842-1062), with the encoded amino acids in parentheses. All five constructs carried the IQ2 domain. Positive GFP signals were detected at primary cilia of cells transfected with the Inv (915-1062) and Inv (842-1062) constructs, and the percentages of cells that showed the GFP signals at the primary cilia were 59.5% (n=37) and 92.7% (n=41), respectively. However, no GFP signals were observed at primary cilia of cells transfected with the Inv (842-1002) (n=31), Inv (742-1002) (n=34) or Inv (842-1031) constructs (n=26) (Fig. 4A). These results suggest that the C-terminal amino acids 1031-1062 are required for the localization of the C-terminal Inv protein to cilia. Thus, there are two regions in the C-terminus of the Inv protein: the IQ2 domain (amino acids 915-935) and amino acids 1031-1062, which are required for localization to the primary cilia. The ciliary GFP signal in cells transfected with the Inv (915-1062) construct was weaker than the signal observed in the Inv (842-1062)-transfected cells. Therefore, the region between amino acids 842-914 might also have some function in the ciliary localization of the Inv protein. The region between amino acids 902-916 is rich in basic amino acids (lysine and arginine), is highly conserved among species and was named the basic region (BR) (Yasuhiko et al., 2006).
|Name .||Vector .||MCS .|
|Inv (742-1062: ΔIQ2)||pEGFP-C2||ApaI-BamHI|
|Inv (742-1062: I921E)||pEGFP-C2||ApaI-BamHI|
|Inv (1-1062: ΔIQ2)||pEGFP-N3||EcoRI-BamHI|
|Inv (1-1062: ΔBR)||pEGFP-N3||EcoRI-BamHI|
|Inv (742-1062: ΔIQ2)||pGEX-5-1||EcoRI-XhoI|
|Inv (742-1062: I921E)||pGEX-5-1||EcoRI-XhoI|
|Name .||Vector .||MCS .|
|Inv (742-1062: ΔIQ2)||pEGFP-C2||ApaI-BamHI|
|Inv (742-1062: I921E)||pEGFP-C2||ApaI-BamHI|
|Inv (1-1062: ΔIQ2)||pEGFP-N3||EcoRI-BamHI|
|Inv (1-1062: ΔBR)||pEGFP-N3||EcoRI-BamHI|
|Inv (742-1062: ΔIQ2)||pGEX-5-1||EcoRI-XhoI|
|Inv (742-1062: I921E)||pGEX-5-1||EcoRI-XhoI|
The numbers in parentheses denote the position of amino acids. MCS, multiple cloning site
Immunocytochemistry confirmed that the GFP signal in Inv (842-1062)-transfected cells was detected at the base of primary cilia (Fig. 4B). Expression levels of GFP-fused proteins were always high enough to be observed in live cells without increased exposure during image acquisition. Exogenously expressed GFP-fused proteins were also present in significant quantities in the cytoplasm or nucleus of all cells.
Calmodulin-Inv protein binding and calmodulin localization
The IQ domain is known to bind calmodulin (Cheney and Mooseker, 1992). Since the IQ2 domain is important for ciliary localization of C-terminal Inv, we examined calmodulin localization in primary cilia of the renal epithelial cell line. Calmodulin was detected with anti-calmodulin antibody. Calmodulin immunostaining was detected in the membrane, cytoplasm, basal body and primary cilium in both wild-type and inv cells. The calmodulin signal around the peri-basal body area was the highest in both cell types. The signal in primary cilia was weak and was observed over the entire length of the primary cilium (Fig. 5A). The IQ2 domain of C-terminal Inv is conserved among mammals (Fig. 5B). Substitution of a Glu for an Ile amino acid (Ile to Glu) in the IQ2 domain resulted in loss of cilium targeting of GFP-fused Inv (amino acids 742-1062) (n=33). The expression level of GFP-fused proteins was high enough to be observed in live cells without increased exposure during image acquisition. Exogenously expressed GFP-fused proteins were also present in significant quantities in the cytoplasm and nucleus of all cells (Fig. 5C). Calmodulin binding to the IQ2 domain was analyzed by gel overlay assay. GST-fused C-terminal Inv proteins were expressed in Escherichia coli under isopropyl β-D-thiogalactoside (IPTG) induction. The GST-Inv (742-1062) fusion protein did bind to calmodulin, but deletion or mutation of the IQ2 domain resulted in loss of calmodulin binding (Fig. 5D).
The Inv C-terminal region contains a ninein homologous region
Using the ciliary proteome website (http://www.ciliaproteome.org/) (Gherman et al., 2006), we found a sequence of 18 amino acids in the Inv C-terminus (amino acids 1028-1045 SVN-SLQSIH-LDNSG-RSKK; bold residues indicate common amino acids between Inv and ninein; `-' indicates space for alignment) that is highly homologous to the ninein C-terminus (amino acids 1921-1939 SLERGLETIH-LENEGLKKK) (Fig. 6A), and this domain is conserved among mammals (Fig. 6A). Inv (1001-1061) encodes for amino acids 1001-1061 of Inv and GFP-ninein (1875-2035) encodes for amino acids 1875-2035 of mouse ninein. The GFP signals of Inv (1001-1061) and GFP-ninein (1875-2035) were detected at two centrosomes and this was confirmed by anti-γ-tubulin antibody staining (Fig. 6B,C). Centrosome localization of the Inv (1001-1061) fusion protein was independent of formation of primary cilia. During mitosis, localization of the Inv (1001-1061) fusion protein to the spindle pole was evident from prophase to metaphase. During anaphase, a weak signal at the spindle pole was detected but, during telophase, the signal disappeared completely (supplementary material Fig. S2).
The Inv C-terminal region is required for localization of Inv to the Inv compartment
Truncation experiments with C-terminal Inv identified three regions that affected localization of Inv to the cilium. However, it is unknown whether these regions are also required for the targeting of the full-length Inv to the cilium, and in particular for the targeting to the Inv compartment. To identify sequences for Inv-compartment localization, three GFP constructs, Inv (1-1062: IQ2), Inv (1-1062: BR) and Inv (1-1002), were generated. Inv (1-1062: IQ2) and Inv (1-1062: BR) contained the full-length Inv, but lacked the IQ2 domain (amino acids 915-935) or the BR region (amino acids 844-914), respectively. Inv (1-1002) lacked the C-terminal amino acids 1003-1062. Unexpectedly, Inv (1-1062: IQ2) and Inv (1-1062: BR) showed the same intraciliary localization as full-length Inv-GFP [Inv (1-1062)] (Fig. 7). By contrast, Inv (1-1002) localized to the entire cilium with punctate staining. These results indicate that the 60 C-terminal amino acids, but not the BR or the IQ2 domain, are required for the regulation of Inv localization in the Inv compartment.
Inv is a dynamic molecule in the Inv compartment of primary cilia
FRAP was carried out to examine the intraciliary dynamics of the Inv protein using Inv-GFP cells. Primary cilia were viewed from the side and FRAP was applied (Fig. 8A). Inv-GFP in the Inv compartment was partially photobleached at time 0. Inv-GFP fluorescence recovered after photobleaching and the bleaching gap was almost completely filled after 60 seconds (Fig. 8B).
Morphologically, cilia have been divided into several segments, including the tip, ciliary necklace, transitional zone and basal body (Gilula and Satir, 1972; Wheatley, 1967; Wheatley, 1995). A segment between the tip and ciliary necklace, which is called the body (shaft), has not been defined well or analyzed extensively. Inv localization indicated that the body of the primary cilium is not uniform along its long axis. Analysis of Inv-GFP cells as well as immunocytochemical examination of endogenous Inv protein in wild-type cells revealed that the Inv protein is localized at the proximal segment of the cilia body (the Inv compartment), indicating that the body is divided into at least two segments based on Inv protein localization. Our study identified intraciliary sub-compartments based on molecular distribution that have not previously been identified by morphological analysis.
Examination of Inv further supported the idea of a distinct molecular compartment in the primary cilium. We identified that 60 amino acids in the C-terminus are required for localization of the full-length Inv to the Inv compartment. Since targeting sequences are essential for localization to the Inv compartment, this suggests a presence of mechanisms that make Inv localize into the compartment.
Furthermore, dissection of Inv showed a complex mechanism for its ciliary and intraciliary localization. Signals from constructs containing N-terminal Inv fused to GFP [Inv (1-1002) and Inv (1-743)] were detected throughout the entire cilia, with a punctate appearance. Ciliary proteins are known to be transported from the cytoplasm by intraflagellar transporters (IFTs). IFTs in Chlamydomonas and mammals show a punctate staining pattern along the entire cilium (Cole et al., 1998; Deane et al., 2001; Follit et al., 2006; Taulman et al., 2001). The staining patterns of IFTs and GFP-fused N-terminal Inv constructs [Inv (1-743) and Inv (1-1002)] are quite similar, suggesting that IFTs are a good candidate for carrying N-terminal Inv. However, IFT alone cannot explain the intraciliary localization of the full-length Inv protein. IFTs are distributed throughout the entire cilium, whereas the Inv protein is restricted to the Inv compartment.
The ciliary-membrane-targeting signal RVxP has been recently identified in polycystin-2 and cyclic nucleotide-gated channels (Geng et al., 2006; Jenkins et al., 2006). The N-terminal 15 amino acids of polycystin-2, which contain this sequence, are sufficient to localize heterologous proteins in cilia (Geng et al., 2006). Different from this ciliary-membrane-targeting mechanism, the mechanism involved in the localization to the Inv compartment is complex. One simple hypothesis for the Inv-compartment localization mechanism is that one carrier, maybe IFT, transports Inv to the cilia by binding to the N-terminal Inv region. Then, the other carrier, which binds to the C-terminal 60 amino acids, confines the Inv protein to the Inv compartment. The second carrier might also have the ability to carry the target protein to the centriole, since the C-terminal 60 amino acids fused to GFP, alone, is localized to centrioles. A computational search of the C-terminal sequences revealed a ninein homologous region that is conserved among mammals. Alternatively, the C-terminal 60 amino acids allow for a negative regulatory mechanism that excludes Inv from other regions within the primary cilium. Our next step in understanding the mechanisms for ciliary and intraciliary localization is to identify factors that could confine Inv to the primary cilium and the Inv compartment.
Two previous studies suggested a functional significance for the Inv compartment. First, C-terminal-deleted Inv failed to rescue renal-cyst formation in inv mice (Watanabe et al., 2003). Second, mutations in most human NPHP2 families had a C-terminal truncation containing the ninein homologous region (Otto et al., 2003). Therefore, a loss of the compartment-targeting sequences of Inv in renal epithelial cells might have some role in renal-cyst formation. These reports suggest that the Inv compartment has a functional role in maintaining normal renal architecture.
Deletion and mutation analysis of C-terminal Inv (742-1062) showed that the IQ2 domain is essential for cilia targeting of Inv (742-1062). A mutation in the IQ2 domain abolished calmodulin binding, suggesting a role for calmodulin in cilia targeting of Inv (742-1062). However, the full-length Inv construct lacking the IQ2 domain localized in the Inv compartment, as was the case for full-length Inv. Therefore, the IQ2 domain is unlikely to regulate Inv-compartment localization.
Immunoelectron microscopy showed that positive labeling for Inv was seen predominantly between the ciliary membrane and the axoneme. No morphological structure was detected between the ciliary membrane and the axoneme. Inv-GFP signals disappeared after detergent treatment (supplementary material Fig. S3), which suggests that Inv does not have strong connection with the axoneme. FRAP analysis revealed that Inv was dynamic within the Inv compartment. These results demonstrate that Inv is a dynamic molecule rather than a stable structural component of primary cilia.
In summary, we found that Inv localization is restricted to the proximal segment of cilia body (designated as the Inv compartment). Inv ciliary sub-localization is dependent on 60 amino acids in the C-terminus of Inv. Our results suggest that the body of the primary cilium can be divided into sub-compartments along its long axis based on molecular (Inv) protein distribution. The C-terminal 60 amino acids of Inv contain a ninein homologous region that is required for full-length Inv localization to the Inv compartment. Human mutations as well as a mouse-model rescue study suggest that the Inv compartment has a functional significance.
Materials and Methods
Anti-acetylated α-tubulin antibody (clone 6-11B-1) and anti-γ-tubulin (clone GTU-88) from Sigma (St Louis, MO), anti-polycystin-2 antibody (D-3) and anti-calmodulin antibody from Santa Cruz Biotechnology (Santa Cruz, CA), fluorescein-conjugated LTA (LTA-FITC) and Rhodamine-conjugated DBA (DBA-Rhodamine) from Vector Laboratories (Burlingame, CA), and cell culture supplements from Invitrogen (Carlsbad, CA) were used. Restriction enzymes were purchased from Takara (Sigma, Japan), and the pEGFP vectors and pGEX expression vectors were purchased from Clontech (Mountain View, CA). Unless otherwise stated, all other chemicals were purchased from Sigma (St Louis, MO) or Wako Pure Chemical (Osaka, Japan).
Identification of proximal convoluted tubules and collecting ducts
Freshly isolated kidneys from wild-type or inv mice (postnatal day 5) were frozen on carbon ice in OCT compound (Tissue-Tek) and then cryosectioned at 10 μm. Sections were stained with LTA-FITC (diluted 1:1000) for proximal tubules, or with DBA-Rhodamine (diluted 1:100) for collecting ducts (supplementary material Fig. S1A) (Laitinen et al., 1987).
Establishment of immortalized mouse renal epithelial cells
Wild-type, inv and Inv-GFP transgenic mice (Watanabe et al., 2003; Yokoyama et al., 1993) were maintained in an animal facility according to experimental procedures that were approved by the Committee for Animal Research, Kyoto Prefectural University of Medicine. Mice (postnatal day 5) were anesthetized by intraperitoneal administration of sodium pentobarbital at a dose of 50 mg/kg body weight. Kidneys were isolated and dissociated with Krebs buffer containing 10% BSA and 1 mg/ml collagenase for 30 minutes with gentle shaking at 37°C. Digested tissue fragments were passed through 125 μm, 105 μm and 45 μm sieves, and centrifuged at 1000 g for 10 minutes at room temperature. The pellet was resuspended in Dulbecco's modified Eagle's medium (DMEM) /F-12 medium containing 10% fetal bovine serum, and cells were seeded on plastic dishes or glass coverslips coated with human collagen IV (50 mg/ml). Cells were incubated at 37°C and equilibrated with 5% CO2 in humidified air. After 24 hours of incubation, cells were trypsinized and labeled with LTA-FITC [1:1000]. FITC-labeled cells were sorted with the EPICS ELITE ESP (Coulter, Hialeah, FL) flow cytometer (supplementary material Fig. S1B). Following sorting, cells were transfected by SV40 large T antigen (Yanai et al., 1991). Medium was changed daily and single colonies were picked. Established cell lines were named as wild-type cells (clone Dai1), inv cells (clone Dai2) and Inv-GFP cells (clone Dai3). Genotypes were determined by PCR analysis (Mochizuki et al., 1998) (supplementary material Fig. S1C). All established cell lines exhibited a long primary cilium after growing to confluency (supplementary material Fig. S1D). The constitutive expression level of Inv-GFP in Inv-GFP cells was not toxic to the cells.
Cells were cultured on glass coverslips coated with human collagen IV (50 mg/ml) at 33°C in DMEM/F-12 medium containing 0.5% fetal bovine serum, 100 μM MEM non-essential amino acid solution, 5 mg/l insulin, 5 μg/l sodium selenite, 5 mg/ml transferrin, 400 mg/l dexamethasone, 10 ng/ml epidermal growth factor, 5 pg/ml 3,3,5-triido-l-thyronine, 10,000 U/l penicillin, 100 mg/l streptomycin and 250 μg/l Fungizone.
Imaging of primary cilia and GFP fluorescence
Cells were cultured on collagen-Type-IV-coated glass or Formvar film coated with collagen Type IV until primary cilia were formed. To view primary cilia from the side, Formvar film was folded over and on itself, cell side out as according to the method previously developed by Roth et al. (Roth et al., 1988). Unless otherwise stated, living cells were observed by microscope (IX70, Olympus, Tokyo, Japan) with Nomarski differential interference contrast (DIC) objectives equipped with a CCD camera (UIC-QE, Molecular Devices Corporation, Sunnyvale, CA). GFP fluorescent images were also obtained with an IX70 microscope with a GFP filter and a CCD camera. Digital images were processed by MetaMorph (Molecular Devices, Downingtown, PA). For scanning z-axis images of GFP fluorescence and primary cilium, the FV1000 confocal laser microscope (Olympus, Tokyo, Japan) was used.
Preparation and characterization of antibody against the Inv protein
A PCR fragment coding for amino acids 561-724 of mouse Inv was inserted into the pGEX-2T expression vector. The encoded recombinant protein was expressed as glutathione S-transferase (GST)-Inv fusion protein in the Escherichia coli host BL21(DE3), purified by chromatography on glutathione-Sepharose 4B. The purified protein was injected into rabbits. The antiserum was collected and affinity purified. Renal inv cells were transfected with or without GGS-Inv (Mochizuki et al., 1998), washed twice with ice-cold PBS and lysed in 10 mM Tris-HCl and 1% sodium dodecyl sulfate (SDS), pH 7.4, with boiling for 5 minutes. Equal amounts of sample were separated on 7.5% SDS-polyacrylamide gel electrophoresis (PAGE) gels. After electrophoresis, proteins were transferred to polyvinylidene difluoride membrane from Amersham Pharmacia. Inv was detected by incubation with the first antibody against Inv followed by enhanced chemiluminescence detection using horseradish-peroxidase-conjugated anti-rabbit IgG second antibody. Pictures were captured by Versa Doc MODEL 5000 (BIO-RAD) (Fig. 2C).
Cells were fixed in ice-cold methanol/acetone (1:1) for 10 minutes, permeabilized in 0.1% Triton X-100 for 20 minutes, quenched in PBS (137 mM NaCl, 2.6 mM KCl, 6.5 mM Na2HPO4 and 1.5 mM KH2PO4) with 1% BSA for 1 hour at room temperature. Cells were then incubated with primary antibodies at room temperature for 2 hours [mouse anti-acetylated α-tubulin (1:2000), mouse anti-γ-tubulin (1:2000), rabbit anti-Inv (1:500), mouse anti-polycystin-2 (1:1000), rabbit anti-calmodulin (1:1000)]. Cells were washed in PBS and incubated with Hoechst 33342 and Alexa-Fluor-488 or -555-conjugated goat anti-mouse or rabbit IgG for 1 hour. Fluorescence was visualized on an IX70 microscope with U-MWU2 (BP 330-385 nm, BA 420 nm), U-MGFPHQ (BP 460-480 nm, BA 495-540 nm) or U-MWIG2 (BP 520-550 nm, BA 580 nm) filters with a xenon light source. Digital images were processed by MetaMorph.
For immunoelectron microscopy, cells were processed according to the pre-embedding silver-intensified immunogold method. Wild-type cells on culture dishes were fixed with 3% paraformaldehyde for 15 minutes at room temperature, permeabilized in 0.1% Triton X-100 for 25 minutes, washed in PBS containing 50 mM glycine and blocked for 30 minutes with 3% BSA. The cells were then immunostained with the rabbit anti-Inv antibody (1:500) at room temperature for 2 hours, washed in PBS containing 0.1% BSA, incubated with 1.4-nm Nanogold-labeled Fab′ fragment against rabbit IgG (Nanoprobes, Yaphank, NY; 1:100) at 4°C overnight, washed in PBS and fixed with 1% glutaraldehyde for 10 minutes. After silver enhancement (HQ silver; Nanoprobes), the specimens were post-fixed with 0.4% OsO4 for 1 hour, rinsed in distilled water, stained with 2% uranyl acetate, dehydrated through a graded series of ethanol and directly embedded in Quetol812. After the embedded cells were detached from culture dishes, ultrathin sections were prepared for observation under a TEM (JEM-200CXII; Jeol, Tokyo, Japan).
Preparation of GFP constructs and transfections
A full-length inv cDNA has been described previously (Mochizuki et al., 1998). Full-length Inv (1-1062) was subcloned into the pEGFP-N3 vector. Constructs encoding truncated forms of Inv and ninein were engineered using PCR-based methods to amplify Inv or ninein fragments that contained 5′ and 3′ restriction enzyme cut site. These fragments were then subcloned into the pEGFP vectors as indicated in Table 1. Transfections were performed using Gene juice (Novagen, Darmstadt, Germany) according to the manufacturer's instructions. Briefly, the day before transfection, cells were plated at sub-confluent density on glass dishes. Cells were incubated overnight in 1000 μl culture medium and 50 μl serum-free medium that contained 5 μl of Gene juice and 0.5 μg of expression construct DNA per well.
Gel overlay assay for calmodulin-Inv binding
Calmodulin-Inv binding was analyzed according to the method previously described (Yasuhiko et al., 2001). The GST-Inv fusion plasmids used in the present study are shown in Table 1. The encoded recombinant protein was expressed as a GST-Inv fusion protein in the E. coli host BL21(DE3) under isopropyl β-D-thiogalactoside (IPTG; final concentration 1 μM) induction. Collected cells were lysed in a SDS sample buffer, and the samples were electrophoresed on SDS-PAGE gels. Gels were transferred onto PVDF membranes and the membranes were incubated with 100 ng/ml biotin-conjugated calmodulin (catalogue number 208697; Calbiochem, San Diego, CA) in binding buffer (50 mM Tris buffer, pH 7.5; 200 mM NaCl; 5 mM EGTA). Membranes were then washed and incubated with avidin-conjugated peroxidase and visualized by the ECL system. After the gel overlay assay, expression of fusion proteins was confirmed by anti-GST-HRP conjugate antibody (GE Healthcare).
An FV1000 confocal microscope (Olympus) with a heated stage and a 60× oil-immersion objective (N.A.=1.42) was used for FRAP analysis of cells cultured in 35-mm dishes containing a coverslip insert. Inv-GFP cells were used for FRAP analysis. A boxed region of interest (ROI), containing the primary cilium, was bleached using 50% laser power at an excitation wavelength of 488 nm. Images were acquired during a 90-second period for side-view FRAP using an excitation wavelength of 488 nm and at 2% laser power. Images were acquired as the fluorescence recovered at the indicated times. The fluorescent intensity was analyzed in MetaMorph.
We thank Taku Tsukamoto and Sayaka Kamiko for their technical assistance. We also thank Ritsu Kamiya (Department of Biological Sciences, Graduate School of Science, University of Tokyo), Daisuke Kobayashi and Noriyuki Sugiyama for their valuable suggestions. The work reported here was supported by Grants-in-Aid for Young Scientists from the Ministry of Education, Science, Sports and Culture (19790154) to D.S.