Cystin Localizes to Primary Cilia via Membrane Microdomains and a Targeting Motif

Abstract

Primary cilia are dynamic, complex structures that contain >500 proteins, including several related to polycystic kidney disease. How these proteins target to cilia and assemble is unknown. We previously identified Cys1 as the gene responsible for disease in Cys1^cpk mice, a mouse model of autosomal recessive polycystic kidney disease; this gene encodes cystin, a 145–amino acid cilium-associated protein. Here, we characterized the localization of cystin in the embryonic kidney and liver, in isolated renal collecting ducts, and in an inner medullary collecting duct mouse cell line. Because endogenous levels of cystin expression are low, we generated inner medullary collecting duct cell lines that stably express enhanced green fluorescence protein–tagged constructs of wild-type cystin or various truncation mutants. We determined that cystin is myristoylated at its G2 residue and that N-myristoylated cystin fractionates with membrane microdomains. Furthermore, the N-myristoylation signal is necessary but not sufficient to target cystin to the primary cilium. Analysis of deletion and chimeric constructs identified an AxEGG motif that is necessary to target and retain cystin in the cilium. Derangement of these localization motifs may lead to cystic kidney disease.

Renal tubular cysts develop in a number of inherited, developmental, and acquired diseases.¹ Inherited single-gene disorders include (1) autosomal dominant polycystic kidney disease (PKD), (2) autosomal recessive PKD, (3) juvenile nephronophthisis–medullary cystic disease complex, (4) Bardet-Biedl syndrome, and (5) several malformation syndromes (reviewed in references^2,3). In addition, numerous murine (mouse and rat) PKD models have been described in which the mutant phenotypes closely resemble human renal cystic disease with respect to cyst morphology, cellular and extracellular matrix abnormalities, and disease progression (reviewed in reference⁴).

In some studies, proteins involved in human renal cystic disease, as well as murine models, have been localized, at least in part, to the primary cilium.^5–7 Primary cilia are hair-like structures that emerge from the plasma membrane as single projections from one of the basal bodies.^8,9 Although previously considered to be vestigial organelles, multiple lines of evidence now indicate that primary cilia function in left-to-right embryonic patterning,¹⁰ as well as mechanosensors (renal tubular epithelia),^11,12 photosensors (photoreceptor cells),¹³ and chemosensors (olfactory neurons).¹⁴ In renal tubular epithelia, a primary cilium projects from the apical membrane of almost every cell type.⁸

We have identified the disease gene Cys1¹⁵ in the Cys1^cpk mouse model of autosomal recessive PKD and demonstrated that the gene product cystin localizes to the primary cilium in association with the ciliary membrane.^15,16 Cystin is a novel, hydrophilic protein that contains a predicted N-terminal myristoylation site coupled to a polybasic domain. This motif is used by proteins such as c-Src, K-Ras, and myristoylated alanine-rich C-kinase substrate proteins for membrane anchoring.^17,18

In this study, we show that cystin is myristoylated at its G2 residue and N-myristoylated cystin fractionates with membrane microdomains. We used a lentiviral–green fluorescence protein (GFP) vector system, quantitative Western blotting, and FACS methods to generate a suite of stably transfected mouse inner medullary collecting duct (IMCD) cell lines that express wild-type cystin, the Cys1^cpk allele, and a series of point and truncation mutants as enhanced GFP (eGFP)-tagged fusion proteins at levels comparable to the endogenous protein. We have determined that G2 myristoylation is necessary for cystin association with the plasma membrane, but a novel second domain encoded within amino acid (AA) residues 28 to 35 is required for the stable targeting and retention of cystin in the cilium.

Results

Cystin Is a Developmentally Regulated Protein that Localizes to the Primary Cilium

Cystin is a novel, hydrophilic protein of 145 AAs with a calculated molecular mass of 15.6 kD. Similar to other cystoproteins (e.g., polycystin-1 [PC1], fibrocystin/polyductin [FPC]), cystin expression seems to be developmentally regulated. Whole-mount in situ hybridization in mouse embryos ranging from 7.5 to 11.5 d post conception (dpc) revealed no Cys1 expression (data not shown). In sections of whole embryo and adult tissues, specific Cys1 expression was observed in embryos at 14.5 and 16.5 dpc (Figure 1, A and C). Expression was confined to the ureteric bud–derived tubules in the developing kidney and the biliary ductal epithelia in the developing liver (Figure 1B). No expression was observed in pancreas, brain, heart, or lung. Cys1 was not detected in neonatal or adult kidney, suggesting that the highest expression occurs in developing renal collecting and biliary ducts with rapid downregulation in postnatal life.

Figure 1: In situ analyses of endogenous Cys1 expression in embryo kidney and liver are shown. (A and B) In the developing kidney of the whole embryo at 14.5 dpc, indicated by the arrowhead and at higher magnification in panels to the right, Cys1 is expressed in the ureteric bud derivatives. In the developing liver (green box), Cys1 is expressed in the ductal epithelia. (C) In the developing kidney of the whole embryo at 16.5 dpc, indicated by the arrowhead and at higher magnification in panels to the right, Cys1 is expressed in the ductal epithelia.

Cystin was detected by Western blotting in IMCD cells with an immunoreactive band of approximately 25 kD but not in whole-kidney lysates; however, after immunoprecipitation with the affinity-purified anti-cystin 70053 polyclonal antibodies (pAb), an approximately 25-kD band was detected by immunoblotting with the affinity-purified anti-cystin 71115 pAb (Figure 2A). Similar results were obtained in reciprocal experiments (data not shown). Neither antiserum recognized a specific protein in lysates from mutant kidney (Figure 2A), and the specificity of the antisera was further demonstrated by immunodepletion with affinity-purified recombinant cystin (data not shown). It seems unlikely that posttranslational modifications account for the larger than predicted size of the immunoreactive band, because both endogenous cystin and the bacterially derived recombinant protein (data not shown) migrate at the same molecular weight. Furthermore, the approximately 25-kD immunoreactive band detected in mouse tissues was not affected by reducing agents such as dithiothreitol, β-mercaptoethanol, or tributylphosphine (data not shown).

Figure 2: Cystin expression in mouse renal epithelia and IMCD cells is shown. (A) Cystin is detected in IMCD cell lysates and by immunoprecipitation of lysates from IMCD cells and wild-type kidney, but not Cys1^cpk/Cys1^cpk kidney. (B and C) In embryonic kidney (16.5 dpc) and day 0 kidney, cystin (red) co-localizes with acetylated α-tubulin (green) in the primary cilium (arrowheads). (D) In IMCD cells, cystin (green) also co-localizes with acetylated α-tubulin (red) in the primary cilium.

We examined the subcellular localization of endogenous cystin in renal epithelia by immunofluorescence. When detected by either affinity-purified antiserum, cystin co-localized with acetylated α-tubulin in the primary cilia of renal tubular epithelia sections and cultured IMCD cells, where it seemed to be most concentrated at the distal cilium tip (Figure 2, B through D). Cystin did not co-localize with γ tubulin (data not shown), suggesting that this staining did not represent centrosomal localization. In addition, a diffuse cytoplasmic signal was detected. Immunostaining was not observed in mutant kidney or with either preimmune sera or sera precompeted with recombinant cystin (data not shown).

Cystin Is an N-Myristoylated Protein that Fractionates with Lipid Raft–Enriched Fractions

BLASTP analysis revealed that cystin has no significant homologies to other eukaryotic proteins. Motif searches in PROSITE (available through the ExPASy server http://www.expasy.org/prosite/) predicted that cystin has two myristoylation sites. Further analysis through the MYR prediction server (http://mendel.imp.ac.at/myristate/) predicted an N-terminal myristoylation signal at AA positions 2 to 16 (GSGSSRSGRIPRRRRSP; overall score 1.632; probability of false positive 7.09 e-04).

To characterize potential functional domains within the protein, we generated constructs that express wild-type cystin, the predicted protein encoded by Cys1^cpk allele, and a series of mutants as eGFP-tagged fusion proteins (Figure 3). We noted that the cystin N-terminal myristoylation signal is coupled to a polybasic domain. Given that clusters of basic residues can synergize with fatty acyl groups to promote membrane binding and targeting,¹⁹ we generated two N-terminal mutant constructs, one in which the glycine at position 2 (the putative target for N-myristoyl transferase) was substituted for an alanine (cystin_G2A::GFP) and a second in which the first 20 AAs were deleted (cystin_{Δ1 to 20}::GFP).

Figure 3: Schematic representation of the expression constructs used in this study is shown. Nine different constructs are shown. The predicted N-myristoylation motif (MGSGSSR) and polybasic domain (RRRRS) are located in the first 27 AAs of cystin.

To examine whether cystin is myristoylated in vivo at the predicted N-terminal site, we performed 9,10-[³H]myristate-labeling studies. For these experiments, we transiently transfected HEK293T cells with either the cystin_wt::GFP or cystin_{Δ1 to 20}::GFP construct, allowing us distinguish the wild-type protein from the mutant construct by size in Southern blot analyses. After immunoprecipitation, the eGFP-tagged fusion proteins were resolved by SDS-PAGE and parallel blots were prepared for immunoblotting and autoradiography. The wild-type protein but not the corresponding N-terminal deletion construct was myristoylated (Figure 4A).

Figure 4: Cystin is an N-myristoylated protein that partitions with lipid raft–enriched fractions. (A) HEK293T cells were transfected with cystin_wt::GFP (wt) or cystin_{Δ1 to 20}::GFP (Δ1–20), labeled with [9,10-³H]myristate, and lysed. After immunoprecipitation with the pAb 70053, proteins were resolved by SDS-PAGE on duplicated gels and transferred to nitrocellulose, and the membranes were either probed with anti-GFP mAb (left) or exposed to Hyperfilm-MPX for 3 wk. Cystin_wt::GFP but not cystin_{Δ1 to 20}::GFP is N-myristoylated. (B) With detergent-free protocol, we found that cystin_wt::GFP and cystin_cpk::GFP partition with the lipid raft marker flotillin after ultracentrifugation in an iodixanol (Optiprep) gradient, whereas cystin_{Δ1 to 20}::GFP and cystin_G2A::GFP partitioned solely with the cytosolic fractions (8, 9, and 10). (C) Semiquantitative immunoblotting indicates that cystin_cpk::GFP has approximately 5.5-fold higher affinity for the raft-enriched fractions than cystin_wt::GFP (average of n = 3 experiments).

N-myristoylated proteins, such as c-src, are thought to accumulate in plasma membrane microdomains that are enriched in low-density lipids (e.g., glycosphingolipids, cholesterol) and, thus, resistant to solubilization in nonionic detergent.^17,20 Several observations have raised concerns that extraction of cells with detergent may be generating clusters or rafts of lipids and proteins that did not exist in the intact cell²¹; therefore, we used the simplified detergent-free method described by Macdonald and Pike.²² Western blot analysis of 10 equal-volume fractions collected from each gradient detected cystin_wt::GFP and cystin_cpk::GFP raft fractions^3,4 corresponding to where the membrane microdomain marker flotillin (a raft membrane marker) also partitioned (Figure 4B). As expected, given the N-myristoylation signal defects in the cystin_{Δ1 to 20}::GFP and cystin_G2A::GFP constructs, these fusion proteins did not float with raft fractions but were found in the bottom fractions, which contain the majority of cytosolic proteins, including β-COP, a Golgi marker (Figure 4B), and the transferrin receptor, a nonraft transmembrane protein (data not shown).

To quantify the relative proportion of cystin_wt::GFP and cystin_cpk::GFP that fractionated with raft-enriched fractions, we performed parallel fractionation experiments, collected fractions from each, and resolved the proteins in each fraction by SDS-PAGE. Membranes were serially blotted first with anti-GFP, then stripped and probed with anti–flotillin-1. Specific GFP and flotillin-1 band intensities were quantified, and the ratios were calculated (Figure 4C). The affinity of cystin_cpk::GFP for raft-enriched fractions was approximately 5.5-fold that of cystin_wt::GFP, suggesting that the membrane distribution of wild-type cystin (Figure 5A) may be modulated by protein interactions directed by C-terminal domains that are not contained in the truncated protein.

Figure 5: Subcellular localization of cystin constructs is shown. (A) Cystin_wt::GFP co-localizes with acetylated α-tubulin (red) in the primary cilium and is distributed diffusely in the cytoplasm. (B) Cystin_cpk::GFP distributes along the cell membrane but is excluded from the primary cilium. (C) Cystin_G2A::GFP is primarily distributed in the cytosol and largely excluded from cilia, although localization is observed in approximately 10% of cilia examined (n >100). (D) Similar to cystin_wt::GFP, cystin_C-95::GFP co-localizes with acetylated α-tubulin in the primary cilium. Images were taken at the apical focal plane (XY plane) to capture the cilium, and the insets represent images taken at the nuclear focal plane.

N-Terminal Myristoylation Is Necessary but not Sufficient for Cystin Targeting to the Primary Cilium

Given our previous data indicating that cystin co-localized with markers of the primary cilium,¹⁵ we sought to extend our analyses using a suite of stably transformed cell lines and examine the impact of specific domains on cilium targeting. Because endogenous cystin is expressed at low levels, we used a FACS-based method to isolate single-cell transformants that express cystin_wt::GFP in a more tightly restricted range. Quantitative Western blotting indicated that these cell lines express eGFP-tagged cystin at levels approximately five-fold higher than the endogenous protein in IMCD cells (data not shown). We subsequently used the FACS method to standardize the eGFP fluorescence in all other stably transfected cell lines.

Using GFP fluorescence and staining with anti-acetylated α-tubulin antibody, we then examined the subcellular localization of the cystin constructs in each cell line. As expected, cystin_wt::GFP was detected in the primary cilium, where it seemed to be concentrated at the distal tip (Figure 5A, XY plane). A diffuse cytoplasmic distribution was also observed. In contrast, cystin_cpk::GFP was distributed diffusely along the plasma membrane but not detected in the primary cilium (Figure 5B), whereas cystin_G2A::GFP (without N-terminal myristoylation) was excluded from the plasma membrane, expressed primarily in the cytoplasm (Figure 5C), and detected in <10% of the cilia examined. Because cystin_cpk::GFP shares AAs 1 through 27 with the wild-type protein, we conclude that this region, containing the N-myristoyl–polybasic domain, is necessary for membrane association but not sufficient for cystin targeting to the primary cilium.

A Novel AxEGG Domain Encoded in AA Residues 28 through 35 Is Required to Direct Cystin Trafficking to the Primary Cilium

To define the minimal protein backbone required for the cilium targeting and retention, we generated a series of cystin C-terminal truncation constructs—cystin_C-30::GFP, cystin_C-60::GFP, and cystin_C-95::GFP—and used the FACS method to isolate stably transformed cell lines expressing the fusion constructs at levels approximating the endogenous protein. When examined by immunofluorescence, the distribution of each truncated fusion protein in these cell lines strongly resembled that of cystin_wt::GFP (data shown for cystin_C-95::GFP; Figure 5D). Taken together, the data from cystin_cpk::GFP and cystin_C-95::GFP suggest that the minimal cilium-targeting domain required is encoded in cystin AAs 27 through 50.

Resh and co-workers²³ showed previously the first 30 residues of the HIV type 1 Gag protein (HIV Gag) can function independently as a membrane-targeting domain when fused to heterologous proteins. Like cystin, this sequence contains a membrane-targeting motif consisting of an N-terminal myristoylation site coupled to a highly basic region that binds acidic phospholipids on the cytoplasmic leaflet of the plasma membrane; therefore, to determine whether the first 50 residues of cystin contain a unique cilium-targeting signal, we generated a chimeric construct containing the HIV Gag N-myristoyl–polybasic domain coupled to cystin AAs 22 through 50 (HIV Gag_{AA1 to 30}-cystin_{AA22 to 50}::GFP). As a corresponding control, we used a construct encoding the first 50 AAs of HIV Gag (HIV Gag_{AA1 to 50}::GFP). Using sequential staining in stable cell lines, we observed that the HIV Gag_{AA1 to 30}-cystin_{AA22 to 50}::GFP chimeric protein trafficked to the primary cilium and co-localized with acetylated α-tubulin (Figure 6A), whereas the HIV Gag_{AA1 to 50}::GFP was excluded from the primary cilium (Figure 6B). Extensive bioinformatic analysis failed to identify homologies between cystin AAs 22 through 50 and cilium-targeting domains described in other proteins.^24–26 Further analysis with the C-terminal truncation construct cystin_C-110::GFP (Figure 6Ci) and a derivative chimeric construct containing the HIV Gag N-myristoyl–polybasic domain coupled to cystin AAs 28 through 35 (HIV Gag_{AA1 to 30}-cystin_{AA28 to 35}::GFP; Figure 6Cii) demonstrated that the cystin AAs 28 through 35 contains the cilium-targeting domain.

Figure 6: The AxEGG motif is required for cystin cilium-targeting. (A) The chimeric construct HIV Gag_{AA1 to 30} + cystin_{AA22 to 50}::GFP traffics to the primary cilium and co-localizes with acetylated α-tubulin (red). (B) In contrast, the control construct HIV Gag_{AA1 to 50} is excluded from the primary cilium. (C) Similar to cystin_wt::GFP and cystin_C-95::GFP, cystin_C-110::GFP (Ci) and HIV Gag_{AA1 to 30} + cystin_{AA28 to 35}::GFP (Cii) co-localize with acetylated α-tubulin in the primary cilium. When combined with the data from cystin_cpk::GFP (which shares AAs 1 through 27 with the wild-type protein), these data indicate that both the N-myristoyl–polybasic domain and a novel domain encoded in AAs 28 through 35 are required for cystin targeting to the primary cilium. (D) Clustral X alignment identifies a highly conserved motif, A-x-E-G-G, within this AA sequence. When this motif is substituted by a series of alanines (AAAAA), the chimeric protein no longer targets to the primary cilium (Ciii).

On the basis of data sets from stable cell lines expressing cystin_wt::GFP, cystin_cpk::GFP, cystin_C-110::GFP, and the chimeric construct HIV Gag_{AA1 to 30}-cystin_{AA28 to 35}::GFP, we concluded that cystin contains a novel motif encoded in AAs 28 through 35 that is required for cystin trafficking to the primary cilium. Comparison of this region in mouse cystin with cystin orthologues in five other species identified a highly conserved motif, A-x-E-G-G (Figure 6D), suggesting that these residues are required for the stable targeting and retention of cystin in the primary cilium. In support of this conclusion, when we mutate the AxEGG motif in the HIV Gag_{AA1 to 30}-cystin_{AA28 to 35}::GFP chimera to a series of alanines (HIV Gag_{AA1 to 30}-cystin_{AA28 to 35(AAAAA)}::GFP), the chimeric protein no longer targets to the primary cilium (Figure 6Ciii).

Discussion

A central feature in renal cystic disease pathogenesis is the inability of epithelia to achieve or maintain a state of terminal differentiation.²⁷ Studies indicated that the primary cilium is a major effector of differentiation signals, including those directing planar cell polarity (reviewed in references^28,29). Cystoproteins localize at least in part to the primary cilium/centrosome complex, thus implicating this complex in the pathogenesis of renal cystic disease.³

How cystoproteins target to the primary cilium is largely uncharacterized, and few cilium-specific targeting motifs or specific trafficking mechanisms have been identified. Geng et al.²⁴ described an amino-terminal RVxP motif in PC2 that is required for ciliary localization. Interestingly, the same motif positioned in the carboxy-terminus is required for ciliary trafficking of the cyclic nucleotide–gated (CNG) channel CNGB1b subunit in olfactory neurons,²⁵ and the closely related VxPx motif directs the ciliary targeting of rhodopsin²⁶; however, bioinformatic analyses have failed to identify this same motif in other cystoproteins (L.M.G.-W., unpublished observations). Other cilium-specific targeting motifs reported to date seem to be cystoprotein specific.^30–32

In addition to specific cilium-targeting motifs, studies in experimental models indicate that the cilium targeting of some proteins involves fatty acylation. For example, SMP-1 is a member of a family of small myristoylated proteins (SMPs) that require N-terminal myristoylation and palmitoylation for flagellar localization in Leishmania.³³ Analysis of eGFP-tagged rhodopsin expression in transgenic frogs has revealed that a 44-AA C-terminal sequence, which includes two palmitoylated cysteines, is required for proper targeting of the chimeric protein to the photoreceptor rod outer segment (ROS).³⁴ In rat olfactory sensory neurons, key components of the olfactory G protein–coupled signal transduction cascade are enriched with caveolin-1 and caveolin-2 in low-density fractions derived from cilia. Thus, caveolins seem to play an important role in olfactory signaling by assembling elements of the transduction machinery in cilium-associated lipid microdomains.³⁵ Taken together, these data suggest that the lipid modification may play a key role as a cilium localization signal.

This study demonstrates that, similar to other cystoproteins (e.g., PC1,^36,37 fibrocystin/polyductin³⁸), cystin is developmentally regulated and expressed at low levels in fully differentiated epithelia. Cystin is myristoylated at its G2 residue and partitions in fractions with caveolin and flotillin, markers of lipid microdomains. Immunofluorescence of cells stably expressing our eGFP-tagged constructs indicates that this myristoylation motif is necessary for cystin membrane association. Because of the limited hydrophobicity of the myristolyation anchor, N-myristoylated proteins require an additional membrane attachment factor for membrane targeting.³⁹ In cystin, the membrane attachment factor likely involves the cluster of positively charged AAs directly adjacent to the N-myristoylated glycine, because this hydrophilic protein has neither an adjacent cysteine for putative palmitoylation nor a transmembrane domain.

Cystin is not distributed diffusely along the plasma membrane but seems to be selectively targeted to the membrane of the primary cilium. This observation is notable because, on the basis of its biochemical characterization, cystin is predicted to associate with membrane rafts or microdomains. Membrane rafts are small (10 to 200 nm), heterogeneous, and dynamic, with sterol- and sphingolipid-enriched domains.²¹ Small rafts can be stabilized to form larger platforms through protein–protein and protein–lipid interactions. These highly ordered structures are proposed to function in dynamic cellular processes such as membrane trafficking, signal transduction, and regulation of membrane protein activity.⁴⁰ Although the existence of membrane microdomains in live cells has been controversial, some studies^41–43 provided experimental evidence for these microdomains in living cell membranes.

Interestingly, both endogenous cystin and the eGFP-tagged protein are often (although not always) observed at the distal tip of the cilium. Haycraft et al.⁴⁴ demonstrated that proteins in the mammalian hedgehog signaling pathway (Gli proteins and Sufu) co-localize to the distal tips of cilia in primary limb bud cells. On the basis of these data, they speculated that the tip of the cilium may be a specialized domain for regulating signal transduction. Of note, Yoder et al.¹⁶ showed that PC1 traffics to the primary cilium in renal epithelia and often localizes to the distal tip. Although there are limited data implicating the localization of cystoproteins such as PC1 to membrane microdomains, Wandinger-Ness and co-workers⁴⁵ showed using sucrose gradient analyses that the polycystins and their associated proteins E-cadherin and β-catenin distribute in a complex with the raft marker flotillin-2. Bhunia et al.⁴⁶ also reported that PC1 fractionates with raft-enriched fractions and is likely to be palmitoylated.

For cystin, like rhodopsin,³⁴ acylation alone is not sufficient for cilium targeting, and a second targeting signal is required. Analysis of stably transfected cell lines expressing eGFP-tagged deletion constructs indicate that the cilium targeting signal is encoded in the first 50 AAs of cystin. The selective cilium trafficking of the HIV Gag_{AA1 to 30}-cystin_{AA22 to 50}::GFP chimeric protein suggests that the cilium-targeting domain is encoded in cystin AAs 22 through 50; however, eGFP-tagged cystin_cpk, which contains the first 27 AAs of the wild-type protein, is excluded from cilia, whereas cystin_C-110::GFP, containing only the first 35 AAs, localizes to the primary cilium. Furthermore, an eGFP tagged fusion protein encoded by chimeric construct containing the HIV Gag N-myristoyl–polybasic domain coupled to cystin AAs 28 through 35 (HIV Gag_{AA1 to 30}-cystin_{AA22 to 50}::GFP) also localizes to the primary cilium.

Taken together, these data indicate that a cilium-targeting domain is encoded in cystin AAs 28 through 35. Within this primary sequence, we identified a novel AxEGG motif that is highly conserved among all vertebrate cystins, and when this motif is mutated to AAAAA, protein targeting to the primary cilium is disrupted. This motif seems to be unique to cystin, because it is not associated with other flagellar-associated proteins predicted in the Chlamydomonas genome⁴⁷ or in the ciliary proteome database, an integrated and curated compendium of all existing ciliary and basal body proteomics data from Trypanosoma, Chlamydomonas, Caenorhabditis, Drosophila, and humans.⁴⁸

In summary, our study demonstrates that cystin is directed to the primary cilium through the combined action of its acylation motif and a unique cilium-targeting signal. We propose that cystin is organized in membrane microdomains through protein–lipid interactions and functions in signal transduction pathways that direct and maintain epithelium differentiation. We note that renal epithelia derived from Cys1^cpk kidneys have structurally intact cilia,⁴⁹ suggesting that the renal cystic disease in mutant mice results from a defect in cilium function(s). Given that cilia are structurally intact in many other renal cystic diseases, we speculate that this model of cilium targeting and function is applicable to other membrane-associated cystoproteins.

Concise Methods

Mice

Mice used in these experiments were bred from a stock colony of C57BL/6J-+/Cys1^cpk (B6-+/Cys1^cpk) maintained at the University of Alabama at Birmingham in accordance with Institutional Animal Care and Use Committee regulations and National Institutes of Health Guidelines for the Care and Use of Laboratory Animals. The University of Alabama at Birmingham is fully accredited by the American Association of the Accreditation of Laboratory Animal Care.

In Situ Hybridization Assay

Whole-mount and tissue section in situ hybridization was performed on the basis of the method described by Wilkinson and Nieto.⁵⁰ Digoxygenin-UTP labeled antisense riboprobes were generated from the full Cys1 open reading frame (ORF). For these hybridization experiments, whole mouse embryos (7.5 to 11.5 dpc), as well as tissue sections from embryonic (14.5 and 16.5 dpc), neonatal, and adult mice, were used. After color development, embryos and tissue sections were washed in PBT (PBS + 0.1% Tween-20), fixed in 4% paraformaldehyde, dehydrated in methanol, and photographed in benzyl alcohol:benzyl benzoate (1:1). Images were captured with a Nikon DXM1200 digital camera on a Nikon SMZ1500 stereoscope and assembled using Photoshop 7.0.

Antibodies

Anti-cystin pAb 70053 and 71115 were raised in rabbits using a fusion protein containing the Cys1 ORF. Briefly, a cDNA encoding the ORF was subcloned into the pET21b vector (Invitrogen, San Diego, CA) to generate His-tagged cystin. The sequence-verified construct was used to transform the Escherichia coli host, BL21, and the recombinant protein was isolated via its poly-His tag using a Ni²⁺-charged matrix (Ni-NTA Agarose; Qiagen) and serially injected (n = 5) into two rabbits (70053 and 71115; Invitrogen). Affinity purification was performed by immunoaffinity chromatography using the recombinant protein coupled to an AminoLink Plus Column (Pierce). Monoclonal mouse anti-HA and rabbit anti-His were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Monoclonal mouse anti–caveolin-1, anti–flotillin-1, and anti-GFP (JL-8) were obtained from BD Transduction Laboratories (San Diego, CA). Monoclonal mouse anti-transferrin receptor was purchased from Zymed (San Francisco, CA). Monoclonal mouse anti-acetylated tubulin and anti–β-COP were obtained from Sigma (St. Louis, MO).

Cell Lines, DNA Constructs, Transfection, and Infection

HEK293T cells were maintained in DMEM containing 4.5% glucose and L-glutamine supplemented with 10% heat-inactivated FBS (Hyclone, Logan, UT) and penicillin/streptomycin. Mouse IMCD-K2 cells⁵¹ were grown in DMEM/F-12 containing Earle's balanced salt solution (Cellgro, Mediatech, Washington, DC). Transfected cell lines were maintained in puromycin (lentiviral vector) or G418 (pEGFP-N-1 vector) containing medium.

To generate the cystin::GFP construct, we amplified the complete cystin ORF from wild-type kidney cDNA and cloned the product into the pEGFP-N1 vector (Clontech) in-frame with carboxy-terminal eGFP. From the cystin::GFP backbone, we generated the following mutant constructs: (1) Cystin_cpk::GFP, containing the truncated Cys1^cpk protein that shares only the first 27 AAs with wild-type cystin; (2) cystin_{Δ1 to 20}::GFP, deleting the N-terminal 20 AAs (containing the predicted N-myristoylation signal and the polybasic domain); and (3) cystin_G2A::GFP, with a glycine-to-alanine substitution at the second position. Using the same protocol, we generated a series of C-terminal deletion constructs: cystin_C-30::GFP, cystin_C-60::GFP, cystin_C-95::GFP, and cystin_C-110::GFP.

To generate the chimeric construct containing the HIV Gag N-myristoyl-polybasic domain (the first 30 AAs) and cystin AAs 22 through 50 or AAs 28 through 35 and the corresponding control, we used a PCR-based strategy with long 3′ end primers that contained the HIV Gag sequence or both HIV Gag and cystin sequence. HIV Gag_{AA1 to 30} + cystin_{AA22 to 50}::GFP contains the N-myristoyl–polybasic domain of HIV Gag and cystin AAs 22 through 50; HIV Gag_{AA1 to 30} + cystin_{AA28 to 35}::GFP contains the N-myristoyl–polybasic domain of HIV Gag and cystin AAs 28 through 35; HIV Gag_{AA1 to 50}::GFP contains the first 50 AAs of HIV gag. In HIV Gag_{AA1 to 30} + cystin_{AA28 to 35(AAAAA)}::GFP, the AxEGG motif (AAs 29 through 33) was substituted by a series of alanines (AAAAA). All constructs were verified by sequence analysis and are listed in Figure 3.

Transfections were performed in either IMCD-K2 cells or HEK293T cells. Briefly, cells were grown to 90% confluence and transfected using LipofectAMINE 2000 (Invitrogen) as described previously.¹⁵ Although the pEGFP-N1 vector has been used extensively, we observed a diminution of GFP expression after 10 cell passages, perhaps because of progressive silencing of the orthotopic cytomegalovirus promoter; therefore, we regenerated our GFP constructs in lentiviral vector LVV No. 1880 (gift of J. Kappes). Cystin::GFP and corresponding mutant fragments were amplified from pEGFP-N1 vector and subcloned into BamHI and XhoI sites of the lentiviral vector and sequence verified. To generate virus-like particles, we co-transfected constructs into HEK293T cells by the calcium phosphate DNA precipitation method with two helper vectors, VSV/G and Lentiviral Backbone, according to the manufacturer's recommendations (Stratagene). VSV/G vector encodes the envelope glycoprotein, and the Lentiviral Backbone vector offers the essential viral genes for virus-like particle packaging. Forty-eight hours after co-transfection, supernatants were harvested, clarified by low-speed centrifugation (1000 × g for 10 min), filtered through 0.45-μm pore-size filters, aliquotted, and frozen at −80°C. IMCD cells were infected with the virus-like particle stocks in DMEM containing 1% FBS and 10 μg/ml of DEAE-dextran for 4 h at 37°C. The medium was then replaced with fresh DMEM containing 10% FBS. Infected cells were selected with 1 μg/ml puromycin for stable cell line generation.

N-Terminal Myristoylation Assay

HEK293T cells were transfected with either the cystin::GFP or cystin_{Δ1 to 20}::GFP construct. After 24 h, the medium was changed to DMEM containing 10% delipidated FBS, 10 mM sodium pyruvate, and 0.1 mCi/ml 9,10-[³H]myristate (30 to 40 Ci/mmol; NEN, Dupont, Boston, MA). Cultures were incubated at 37°C for 5 h (myristate labeling), then washed several times with PBS to remove excess label and scraped into PBS. Cells were lysed in 1 ml of RIPA buffer and centrifuged at 14,000 rpm for 10 min. After homogenization, the lysate aliquots were immunoprecipitated with anti-GFP mAb. The proteins were resolved on 12% SDS-PAGE in duplicate gels and transferred to nitrocellulose. One membrane was subjected to immunoblotting. The duplicate membrane was treated for fluorography (Amplify; Amersham, Arlington Heights, IL) and exposed to Hyperfilm-MP (Amersham) film for 14 d.

Detergent-Free Lipid Raft Preparation

Detergent-free rafts were prepared using the simplified method described by Macdonald et al.²² All procedures were carried out on ice. Two 100 × 20-mm plates of cells were washed and scraped into base buffer (20 mM Tris-HCl [pH 7.8]) to which had been added 1 mM CaCl₂ and 1 mM MgCl₂. Cells were pelleted by centrifugation for 2 min at 250 × g and resuspended in 500 μl of base buffer containing 5 μl of protease inhibitor cocktail (from Sigma). Cells were lysed by 20 strokes in a Dounce homogenizer using a tight-fitting pestle, followed by 10 passages through a 23-G needle and, finally, sonication two times for 15 s with intensity level 3 of ultrasonic dismembrator model 100 (Fisher Scientific). A total of 400 μl of the homogenate was mixed with an equal volume (400 μl) of base buffer containing 90% sucrose and placed in the bottom of a test tube (13 × 51-mm polycarbonate centrifuge tube from Beckman). Two milliliters of base buffer containing 35% sucrose was layered on top, followed by 0.9 ml of base buffer containing 5% sucrose. Gradients were centrifuged for 16 h at 175,000 × g in an SW 50.1 Ti rotor. Tubes were fractionated into ten 370-μl fractions. Proteins from each fraction were resolved by SDS-PAGE, followed by immunoblotting with anti-cystin, anti-GFP, anti–flotillin-1, anti-transferrin, and anti–β-COP.

Flow Cytometry Analysis and Cell Sorting

Transfected cells were seeded into 10-cm tissue culture dishes and grown for 24 to 48 h to confluence. Three days after confluence, cells were trypsinized, resuspended in PBS containing 1% BSA to a final density of 0.35 to 1 × 10⁶ cells/ml, and filtered through a nylon membrane to remove cell aggregates. The sorting of single cells into 96-well tissue culture plates was performed by using a FACStar Plus (Becton Dickinson, San Jose, CA) at a flow rate of 1000 to 3000 cells/s. GFP signals were detected with 530/30-nm band-pass filter.

The single cells were cultured in growth medium with puromycin in 96-well tissue culture plates until confluence, then trypsinized and resuspended in fresh growth medium and seeded in appropriate six-well tissue culture plate. To isolate stably transfected cell subpopulations, we allowed the cells to reattach to tissue culture dishes. The fluorescence intensity of the clones or cell subpopulations was measured at 5 d after confluence using either a FACScan or a FACSCalibur flow cytometer (Becton Dickinson). A minimum of 20,000 events was collected for each analysis. Data acquisition and analysis were performed with CellQuest software. The geometric mean of cell fluorescence intensity was calculated and used to compare the level of GFP signal between cell lines.

Immunofluorescence

Stable transformants were cultured in DMEM/F-12 with 10% FBS, 100 U/ml penicillin, 100 mg/ml streptomycin, 1 μg/ml puromycin, and 5% CO₂/95% air at 37°C. Cells were seeded onto Transwell tissue culture polyester membrane (Costar 3470 for en face imaging, Corning 3801 for folded orientation) and cultured for 5 d after confluence.

Cells were washed with PBS and fixed with 4% paraformaldehyde in PBS at room temperature for 10 min, permeabilized with 0.2% Triton X-100, and subsequently washed with PBS. After blocking nonspecific binding with 1% BSA in PBS for 30 min, cells were incubated for 1 h with primary antibodies and subsequently incubated with Alexa Fluor 488–conjugated (1:1000) and Alexa Fluor 594–conjugated secondary antibodies (1:5000; Molecular Probes, Eugene, OR) for 1 h. Cells were counterstained with Hoechst 33258 and washed with PBS. The membrane was cut out and mounted with ProLong Gold (Molecular Probes) on slides for en face imaging of the monolayer, as well as in a folded orientation allowing side-view imaging of cells in the XZ plane at the apex of the fold. Fluorescently labeled cells were analyzed on a Leica scanning laser confocal microscope configured with both an Argon Ion (5 mW, 488 nm) and a Krypton Ion (10 mW, 568 nm) laser.

Kidneys were harvested from B6 wild-type mice, fixed in ice-cold 4% paraformaldehyde, cryoprotected in 30% sucrose, and frozen overnight in Tissue-Tek OCT Compound (Sakura Finetek, Torrance, CA). Serial 10-μm sections were prepared in a Leica CM3050 cryostat, permeabilized with 0.2% Triton X-100, incubated first with primary antibodies as described already, and then appropriate fluorophore-conjugated secondary antibodies (Molecular Probes). Sections were mounted with p-phenylene-diamine (Sigma) and examined with conventional (×100) and/or confocal microscopy. Images were acquired using either a Nikon TE200 fluorescence microscope equipped with a Hamamatsu C5810 digital camera or a Leica TCS SP confocal microscope configured with both an Argon Ion (5 mW, 488 nm) and a Krypton Ion (10 mW, 568 nm) laser. The captured images were processed using Adobe Photoshop 7.0 (Adobe Systems, San Jose, CA).

Disclosures

None.

This work was supported by National Institutes of Health grant DK55534 (L.M.G.-W.).

We thank Karen Johnson, Mary Lou Watkins, and Zhiqian Gao, PhD, for technical assistance. We acknowledge technical support from the Genetically Defined Microbe and Expression Core of the UAB Mucosal HIV and Immunobiology Center. We are grateful to Bradley K. Yoder, PhD, for critically reviewing the manuscript.