CLIC5A, a component of the ezrin-podocalyxin complex in glomeruli, is a determinant of podocyte integrity

Abstract

The chloride intracellular channel 5A (CLIC5A) protein, one of two isoforms produced by the CLIC5 gene, was isolated originally as part of a cytoskeletal protein complex containing ezrin from placental microvilli. Whether CLIC5A functions as a bona fide ion channel is controversial. We reported previously that a CLIC5 transcript is enriched ∼800-fold in human renal glomeruli relative to most other tissues. Therefore, this study sought to explore CLIC5 expression and function in glomeruli. RT-PCR and Western blots show that CLIC5A is the predominant CLIC5 isoform expressed in glomeruli. Confocal immunofluorescence and immunogold electron microscopy reveal high levels of CLIC5A protein in glomerular endothelial cells and podocytes. In podocytes, CLIC5A localizes to the apical plasma membrane of foot processes, similar to the known distribution of podocalyxin and ezrin. Ezrin and podocalyxin colocalize with CLIC5A in glomeruli, and podocalyxin coimmunoprecipitates with CLIC5A from glomerular lysates. In glomeruli of jitterbug (jbg/jbg) mice, which lack the CLIC5A protein, ezrin and phospho-ERM levels in podocytes are markedly lower than in wild-type mice. Transmission electron microscopy reveals patchy broadening and effacement of podocyte foot processes as well as vacuolization of glomerular endothelial cells. These ultrastructural changes are associated with microalbuminuria at baseline and increased susceptibility to adriamycin-induced glomerular injury compared with wild-type mice. Together, the data suggest that CLIC5A is required for the development and/or maintenance of the proper glomerular endothelial cell and podocyte architecture. We postulate that the interaction between podocalyxin and subjacent filamentous actin, which requires ezrin, is compromised in podocytes of CLIC5A-deficient mice, leading to dysfunction under unfavorable genetic or environmental conditions.

glomerular filtration depends upon the highly specialized structure and function of glomerular podocytes, mesangial cells, and endothelial cells (ECs). The complex ultrastructure of these intrinsic glomerular cells is regulated by specific interactions between plasma membrane proteins with subjacent cortical actin filaments. Podocytes extend actin-rich foot processes that form an interdigitating scaffold around the exterior of the glomerular capillary loops. Nephrin-based filtration slit diaphragms that span the space between adjacent podocytes are anchored to cortical actin in podocytes (47). The pericyte-like mesangial cells form a glomerular interstitium and are essential for the formation of glomerular capillary loops (28, 37, 56). The function of mesangial cells in maintaining the capillary loop structure is mediated, at least in part, by their attachment to the glomerular basement membrane (GBM) through contacts that couple to intracellular actin-based microfibrils (33). Glomerular ECs are perforated by transcelluar fenestrae, each ringed by cortical actin (3, 42), and their fenestrae are required for the high hydraulic conductivity of the glomerular capillary wall (17). Mechanisms that form and maintain the specialized actin-dependent architecture of glomerular cells are as yet incompletely understood.

We have reported that the transcript for chloride intracellular channel 5 (CLIC5) is enriched ∼800-fold in human glomeruli compared with pooled nonglomerular tissues and cells by serial analysis of gene expression (SAGE) (44). A similar enrichment of the CLIC5 transcript is also evident in a human glomerular SAGE library that was compared with libraries from microdissected nephron segments (13), and the CLIC5 transcript is enriched in a mouse glomerular expressed sequence tag library compared with whole kidney (27). In cultured glomerular ECs, a SAGE tag that corresponds to the DKFZp564B076 cDNA (accession no. AL049313), representing the 3′ end of the CLIC5 gene, was also enriched compared with nonglomerular ECs (52). Enrichment of the CLIC5 transcript in glomeruli, which is similar to that observed for transcripts of podocyte-specific proteins (44), suggests that CLIC5 might play an important functional role in the glomerulus.

CLIC5 belongs to a family of six mammalian CLIC proteins (2), all sharing a highly conserved ∼220-amino acid COOH terminus with variability at the NH₂ terminus. Human CLIC5 is transcribed from two alternative exons 1 (1A and B) producing ∼32 kDa CLIC5A (251 amino acids) and ∼49 kDa CLIC5B (410 amino acids), the latter having a much longer NH₂ terminus with a unique amino acid sequence (7, 53). The first reported mammalian CLIC protein, p64, was purified from bovine kidney (35, 36) and most closely resembles human CLIC5B.

Mammalian CLIC proteins are synthesized without leader sequence and exist in a soluble form that shares structural homology with Omega class glutathione S-transferases (16, 18, 25). All CLICs have a conserved and highly reactive cysteine residue in the NH₂-terminal region that may confer redox sensitivity (40, 41, 55, 66). CLICs exist in invertebrates and plants (5, 19, 39), and certain protein domains are functionally interchangeable among invertebrate and vertebrate CLICs as well as some glutathione S-transferases (6). Despite several lines of evidence showing that CLICs can function as ion channels in vitro (25, 38, 39, 61, 65), there is little evidence to establish their function as bona fide Cl⁻ channels in vivo (2, 8, 23, 24, 30, 43, 48). Collectively, the CLIC proteins have been found in diverse subcellular locations, including actin-based structures of the plasma membrane such as microvilli and stereocilia (7, 23), various intracellular organelles, including the Golgi apparatus, mitochondria, and secretory vesicles (5, 15, 22, 50, 53), and nonmembranous organelles such as the centrosome and the nucleus (10, 58, 64).

Evidence is emerging to indicate that CLIC5A is involved in the control of the three-dimensional organization of actin-based cell surface projections. CLIC5A was first purified in a complex with actin, ezrin, and several other actin-associated proteins using a preparation of microvilli isolated from human placenta (7). Ezrin is a member of the ezrin-radixin-moesin (ERM) protein family involved in coupling plasma membrane proteins to the subjacent cortical actin cytoskeleton (12). CLIC5A colocalizes with ezrin at the apical surface of placental epithelium in tissue sections (7) and also in transfected placental epithelial cells expressing a CLIC5A fusion protein (8). More recently, it has been shown that CLIC5A is essential for hearing and balance in mice and that it is concentrated within stereocilia of mechanosensory hair cells of the cochlear and vestibular organs. In these highly specialized epithelial cells, CLIC5A concentrates at the base of the stereocilia bundle, a distinct morphological domain of the stereocilia where the ERM protein radixin is also concentrated (23, 46). In CLIC5-deficient jitterbug (jbg/jbg) mice, CLIC5A protein is absent from the cochlear and vestibular hair cells, radixin protein levels are reduced, and stereocilia begin to degenerate soon after birth, leading to vestibular dysfunction and complete deafness by 7 mo of age (23). Interestingly, radixin deficiency also results in deafness associated with progressive degeneration of cochlear hair cell stereocilia in mice (34), and mutations in the human radixin gene are associated with nonsyndromic hearing loss (32).

To date, a functional role for CLIC5 in renal glomeruli has not been documented. However, it is known that ezrin, via the adaptor protein EBP50 (NHERF2), connects the cytoplasmic tail of podocalyxin to the actin cytoskelton in glomerular podocytes (45, 59). In this study we observe that CLIC5A protein is highly expressed in both glomerular podocytes and ECs. In podocytes, CLIC5A localizes to the apical surface of foot processes, with a similar distribution as that previously shown for ezrin and podocalyxin (45, 59). We confirm that CLIC5A protein colocalizes with ezrin and podocalyxin and show that podocalyxin coimmunoprecipitates with CLIC5A. In glomeruli of jitterbug mice lacking CLIC5, ezrin abundance is markedly reduced, podocyte foot processes are broadened, and glomerular ECs contain large vacuoles. These ultrastructural changes are associated with greater vulnerability to adriamycin-induced glomerular injury. These data are consistent with a functional role for CLIC5A in maintaining the normal glomerular podocyte and EC ultrastructure.

MATERIALS AND METHODS

Cells, tissues, and animals.

Human kidney tissue was obtained from the uninvolved portion of tumor nephrectomy specimens (Human Subjects Protocol No. 6196, University of Alberta). Bovine glomerular ECs were isolated and cultured as described previously (4). All procedures in mice were approved by the University of Alberta Animal Care and Use Committee (protocol no. 545). Control C3H/HeJ mice and jitterbug (jbg/jbg) mice (23) on the C3H/HeJ background were purchased from The Jackson Laboratory (Bar Harbor, ME), and offspring were derived by breeding wild-type or +/jbg heterozygous females with jbg/jbg homozygous male mice. The jbg/jbg mutant mice exhibit a 97-bp deletion in the CLIC5 gene (87 bp at the 3′ end of exon 5 plus 10 bp in the adjacent intron) that leads to skipping of exon 5 and a translational frame shift with premature stop codon. Western blot and immunohistochemistry show that CLIC5 protein is absent in jbg/jbg mutant mice. Male 12- to 13-wk-old +/+ or jbg/jbg mutant mice were anesthetized with isoflurane inhalation. Adriamycin (10 mg/kg, doxorubicin hydrochloride; Alexis Biochemicals, San Diego, CA) prepared in 200 μl of 0.9% NaCl was then injected via the jugular vein. Mice were allowed to recover, and then water and chow were provided ad libitum. Urine was collected from conscious mice 5 times/wk. Spontaneous voiding was stimulated by stroking the suprapubic area. The creatinine concentration in urine and serum was measured with the Creatinine Enzymatic Assay Kit (DZ072B; Diazyme, San Diego, CA). Urine creatinine was adjusted to 20 mg/dl with distilled water. Urine was diluted 3:1 (vol/vol) in 4× Laemmli buffer and boiled. For mice not treated with adriamycin, the equivalent of 10 μl of urine was subjected to SDS-PAGE followed by Western blotting with goat anti-mouse albumin antibodies (Bethyl Laboratories, Montgomery, TX). For mice treated with adriamycin, the equivalent of 0.25 μl of urine (1:40 dilution) was subjected to SDS-PAGE. Urinary albumin was quantified by densitometry of Western blots against a standard curve of mouse albumin on the same blots. Statistical comparison of albuminuria after adriamycin treatment was by Student's t-test.

Northern blot analysis.

cDNA probes homologous to human CLIC5 exon 1A through the beginning of exon 6 or to the untranslated region (UTR) of exon 6 were generated by RT-PCR with primer pairs Hs cDNA probes 1 and 2 (Table 1), respectively, from human kidney RNA. Likewise, cDNA probes homologous to bovine CLIC5 exons 3–5 and the bovine sequence homologous to the 3′-UTR of CLIC5 were generated with primer pairs Bt cDNA probes 1 and 2 (Table 1), respectively, from bovine glomerular EC RNA. Labeling of probes with [α-³²P]dATP was performed by linear PCR. Probes were hybridized under high-stringency conditions to blots containing human kidney or bovine glomerular EC RNA.

Table 1.

Primer pairs used for probe generation and RT-PCR

	Forward Primer	Reverse Primer
Hs. cDNA probe 1	5′-CCAACGGCATGACAGACTC-3′	5′-AACTCGATCTCACTGTCAGC-3′
Hs. cDNA probe 2	5′-AAATCATGATGGATCCAAAG-3′	5′-TCTGGGACTCTCATATGGTC-3′
Hs.1	5′-CCAAGTTTTCTGCCTACATC-3′	5′-AACTCGATCTCACTGTCAGC-3′
Hs.2	5′-AAATCATGATGGATCCAAAG-3′	5′-CCAAATAAGCTGATGGACTC-3′
Hs.3	5′-CCAAGTTTTCTGCCTACATC-3′	5′-CCAAATAAGCTGATGGACTC-3′
Hs.4	5′-AGTGTTGATGCCAAAATACC-3′	5′-CCAAATAAGCTGATGGACTC-3′
Hs.5	5′-CCAACGGCATGACAGACTC-3′	5′-CCAAATAAGCTGATGGACTC-3′
RNAse protection probe	5′-AAGCCACATTTAGGAGGTG-3′	5′-GGGAGAAGGAGTTGATGAG-3′
Bt. cDNA probe 1	5′-ACTCATCCACCCTTCCTG-3′	5′-GACCCCTTTTCATCATCC-3′
Bt. cDNA probe 2	5′-CCAAAATGTGCATAGTCTTAC-3′	5′-CAAGTTCAAACATCAGTCAG-3′
Bt.3	5′-GCATCGACATCTTCGTTAAG-3′	5′-CAAGTTCAAACATCAGTCAG-3′

Hs., Homo sapiens; Bt., Bos taurus.

RT-PCR and PCR analysis.

cDNA was generated from cultured glomerular EC RNA with SuperScript III RT (Invitrogen, Carlsbad, CA). Reactions without RT for each primer set served as controls. PCR was performed using “FirstChoice PCR-Ready Human Kidney cDNA” (Ambion) or cDNA from bovine glomerular ECs using SuperTaq Plus Polymerase (Ambion) with a 100-ng template [94°C, 3 min; 35 cycles: 94°C, 30 s; 55°C, 30 s; 68°C, 30 s + 1 min for each kilobase pair (kbp) of PCR product to be amplified; 72°C, 7 min]. Purified PCR fragments were cloned into TOPO T/A cloning vector PCR II (Invitrogen), transformed into One Shot TOP 10 Competent cells (Invitrogen), and sequenced.

RNase protection assay.

A 296-bp RT-PCR fragment (RNase protection probe, Table 1) corresponding to nucleotides 2,231–2,527 of the CLIC5A reference sequence NM_016929 was amplified and cloned into the pCR II TOPO vector (Invitrogen). The cDNA was sequenced to verify fidelity and orientation. The plasmid was linearized with EcoRV and SpeI (Invitrogen) for generation of antisense and sense RNA probes, respectively. Digoxigenin-UTP-labeled antisense RNA probe (420 bp; Roche Applied Science, Laval, QC, Canada) was generated with the Maxiscript Kit (Ambion) using SP6 polymerase. Digoxigenin-UTP-labeled sense probe was generated using T7 polymerase to yield a 610-bp riboprobe. Transcribed probes contained vector sequences at their 3′ and 5′ ends, allowing their distinction from each other and from the protected mRNA by size. RNase protection assay was performed with the RPA III Ribonuclease Protection Assay Kit (Ambion) with ∼3 ng of sense or antisense probes and 10 μg of human kidney mRNA. Hybridization occurred at 42°C for 18 h. When appropriate, samples were treated with RNAse T1 (1:50 dilution of 5 U/μl stock) for 30 min at 37°C. Protected fragments were separated on a 5% denaturing polyacrylamide gel in 1× TBE, 8.0 M urea buffer at 140 volts. They were transferred to a Hybond-N membrane (Ambion) and detected by chemiluminescence with alkaline phosphatase-conjugated anti-digoxigenin antibodies (Roche Applied Science).

Western blot analysis.

Full-length untagged CLIC4 and CLIC5 were produced in E. coli, as described previously (7). Bovine kidney cortex and glomerular lysates were boiled in 2× Laemmli buffer. SDS gels were stained with Coomassie Blue (Sigma, St. Louis, MO), or proteins were transferred to membranes (Immobilon; Millipore, Bedford, MA). Western blots were probed with unfractionated B132 rabbit antiserum (1:2,000) that recognizes both CLIC4 and CLIC5 (7), with mouse monoclonal anti-β-actin antibody (1:6,000; Sigma) followed by appropriate peroxidase-conjugated antibodies. Secondary antibodies were detected with enhanced chemiluminescence (Amersham Biosciences).

Immunoprecipitation followed by Western blot analysis.

To determine whether CLIC5A is found in the same complex as podocalyxin, glomeruli were isolated from mouse kidney cortex, as described by Takemoto et al. (60). Glomeruli were used only if the purity was >90%. Coimmunoprecipitation was performed as described by Takeda et al. (59). Briefly, glomeruli were homogenized by a Dounce homogenizer in RIPA buffer [0.1% SDS, 1% sodium deoxycholate, 1% Triton X-100, 150 mM NaCl, 2 mM EDTA, 25 mM Tris, pH 7.4, complete protease inhibitors (Roche, Lavalle, QC, Canada)], passed through a 26-G needle, and then sonicated. After sedimentation of insoluble material, podocalyxin or CLIC5 was immunoprecipitated in immunoprecipitation (IP) buffer (50 mM Tris, pH 7.4, 150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate, and complete protease inhibitors). One hundred micrograms of lysate in 1,000 μl of IP buffer was precleared for 30 min with 10-μl agarose beads [either protein A-Sepharose (Sigma-Aldrich, St. Louis, MO) for CLIC5 IP or protein G plus/protein A-Agarose (Calbiochem, San Diego, CA) for podocalyxin IP]. After preclearing, lysates were incubated overnight at 4°C with 5 μl of rabbit B132 serum, 5 μg of rat anti-podocalyxin antibody, (R & D Systems, Minneapolis, MN), or corresponding rabbit preimmune serum or normal rat IgG (Jackson ImmunoResearch Laboratories, West Grove, PA). Lysates were then incubated with 50 μl of appropriate beads for 6 h at room temperature. Bead complexes were washed six times for 2 min each at room temperature. Protein complexes were eluted by boiling for 10 min at 95°C in 2× Laemmli buffer. Precipitated proteins were separated by SDS-PAGE, transferred to Immobilon membranes, and blotted with goat anti-podocalyxin (R & D Systems) and mouse anti-CLIC5 (Abnova, Taipei, Taiwan) antibodies. Blots of whole lysate served as “input controls.” All IPs were repeated three or more times using tissue from distinct mice.

Immunofluorescence microscopy.

Kidney cortex (∼0.5-cm³ cubes) was incubated in 30% sucrose at 4°C overnight, frozen in optimal cutting temperature (Sakura Finetek, Torrence, CA), and stored at −80°C. Frozen sections (∼6 μm thick) were thaw-mounted on SuperFrost Microscope slides (Microm International, Kalamazoo, MI), air-dried, and fixed for 10 min in acetone at 4°C. Slides were washed in PBS (3 × 5 min) at room temperature and blocked for 30 min with 3% BSA in PBS, followed by incubation with the primary antibodies in 3% BSA-PBS overnight at 4°C. Slides were washed (3 × 5 min) in PBS and incubated for 1 h in the dark with 1:500 dilutions of the appropriate secondary antibodies coupled with Alexa fluor 594 or 488 (Molecular Probes, Eugene, OR) in 3% BSA-PBS. Slides were washed three times in PBS, mounted with ProLong Gold Antifade (Molecular Probes), and viewed on a Zeiss Axioplan 2 microscope and a Zeiss LSM 510 laser-scanning confocal microscope (Zeiss, Toronto, ON, Canada) at ×20, ×40, or ×100 magnification. Antibodies against Tie-2 (mouse, 1:100 dilution; Upstate Biotechnology, Lake Placid, NY) and eNOS (mouse, 1:1,000 dilution; Biomol International, Plymouth Meeting, PA) and anti-platelet endothelial cell adhesion molecule-1 (PECAM-1; goat, 1:200 dilution, no. S1506; Santa Cruz Biotechnology, Santa Cruz, CA) were used as EC markers. Antibodies directed against synaptopodin (mouse, undiluted; Biodesign International, Saco, ME) and podocin (goat, 1:200 dilution, no. A11058; Santa Cruz Biotechnology) were used as podocyte markers. Goat anti-podocalyxin (1:10,000 dilution, no. AF1556; R & D Systems), rabbit anti-ezrin (1:2,000 dilution; Cell Signaling, Danvers, MA), and mouse monoclonal anti-CLIC5 (no. H00053405-M03, 1:2,000 dilution; Abnova, Taipei City, Taiwan) were used to define colocalization of CLIC5 with ezrin and podocalyxin. All immunofluorescence studies were repeated three or more times using tissue from distinct samples or mice.

Transmission electron microscopy and morphometry.

For immunogold transmission electron microscopy (TEM), kidney cortex from wild-type adult mice was fixed with 0.1% glutaraldehyde and 4% formaldehyde, treated with 50 mM ammonium chloride, postfixed with 2% uranyl acetate, and dehydrated with acetone before embedding in LR Gold resin, as described previously (7). Sections were incubated with affinity-purified rabbit antibody specific for CLIC5 (7) followed by gold-conjugated goat anti-rabbit IgG (10 nm, Ted Pella) and then counterstained with 2% osmium tetroxide and lead citrate.

For conventional TEM, kidneys were removed without perfusion, cortex was cut to produce ∼5 × 5 mm cubes, and specimens were prefixed in 2.5% glutaraldehyde in Millonig's buffer solution (pH 7.2) for 1.5 h at room temperature. They were then washed and postfixed with 1% osmium tetroxide in Millonig's buffer solution for 1.5 h. After a brief wash in distilled water, samples were dehydrated in a graded ethanol series (50, 70, 90, 100, and 100, 10 min each). Samples were then embedded in Araldite resin and cured at 60°C for 36 h. Thick sections (1–2 μm) were stained with methylene blue to locate glomeruli by light microscopy. Ultrathin sections were stained with 4% uranyl acetate for 30 min and lead citrate for 5 min and viewed with a Hitachi transmission electron microscope H-7000 (Tokyo, Japan). All glomeruli in a given section were photographed in a blinded fashion. Morphometry was performed by point counting, using a grid with points at 1-μm intervals on one complete glomerular cross-section from each of the four wild-type and four jbg/jbg CLIC5-deficient mice, 8 mo of age. For each image, the complete length of GBM was also quantified by tracing and digitization, and glomerular EC vacuoles were counted. Data were similarly obtained for one wild-type and one jbg/jbg CLIC5-deficient mouse aged 3 mo. For immunogold TEM, the density of gold particles associated with specific cells was quantified for glomeruli, peritubular capillaries, and renal arterioles at ×24,000 magnification. The area represented by each cell type was determined by point counting, with grid points spaced at 200-nm intervals. Data were obtained for 10–12 images from each of three separate sections of renal cortex. The total area counted was 1,933, 1,012, and 935 μm² for glomeruli, peritubular capillaries, and renal arterioles, respectively. Statistical analysis was by one-way analysis of variance followed by multiple comparison between groups using the Bonferroni method.

RESULTS

CLIC5A mRNA expression.

A SAGE tag (CATGAATCTGAACCAATTACC) corresponding to the 3′-UTR of the CLIC5 transcript was previously found to be predominant in renal glomeruli (44). Here, we first determined whether this SAGE tag indeed corresponds to CLIC5 and which of the two CLIC5 transcript variants predominate in kidney. The CLIC5 gene is composed of 7 exons (1A/1B-6) located on chromosome 6 (6p12.1 to 6p21.1). The glomerulus-enriched SAGE tag represents a short sequence in exon 6 at nt 5,617–5,637 of CLIC5A (GenBank No. NM_016929) or nt 5,885–5,905 of CLIC5B (GenBank No. NM001114086), respectively. On Northern blots, nonoverlapping cDNA probes represented the complete open-reading frame of CLIC5A (Hs. or Bt. cDNA probe 1; Table 1) or, alternatively, only the common CLIC5 3′-UTR (Hs. or Bt. cDNA probe 2; Table 1) hybridized to bands of the same size (∼5.7 kbp) on human kidney and bovine glomerular EC RNA blots (Fig. 1A). RNase protection assay confirmed the existence of a CLIC5 transcript in kidney (Fig. 1B). Forward PCR primers homologous to human CLIC5 exons 1A, 4, and 6 paired with reverse primers homologous to the CLIC5 3′-UTR (Fig. 1, C and D) generated RT-PCR products of the expected size when NM_016929 was the reference cDNA (Table 1). However, no product was observed with two distinct forward primers corresponding to the CLIC5 exon 1B (data not shown). The two primer pairs spanning the entire exon 6 (pair Hs.3, Bt.3, and Hs.5) consistently amplified two products (Fig. 1, C and D), suggesting the existence of two CLIC5A transcript variants in kidney. Sequencing of the shorter PCR products from three different reactions amplified from human kidney showed that they result from skipping of a 1,203-nt segment within the UTR of exon 6 (nt 1,238–2,439, no. NM_016929). Nonetheless, Northern blots of kidney and glomerular EC mRNA (Fig. 1A) revealed only a single 5.7-kbp CLIC5A transcript, and RNAse protection assay did not generate a shorter 100-bp fragment that would have been expected from a splice variant lacking nt 1,238–2,439 of CLIC5A exon 6. Hence, the abundance of the shorter transcript is too low to be detected by assays with lower sensitivity than RT-PCR.

Fig. 1.

Chloride intracellular channel 5 (CLIC5A) mRNA expression. A: Northern blot of human kidney mRNA (lanes 1 and 2) and bovine glomerular endothelial cell (EC) RNA (lanes 3 and 4) hybridized with nonoverlapping probes designed to detect the CLIC5A open-reading frame (lanes 1 and 3) and the CLIC5 3′-untranslated region (lanes 2 and 4). B: RNase protection assay. The sense (lane 1) and antisense (lane 2) probes span 296 nucleotides of CLIC5 (2,234–2,530 of Genbank Accession No. NM_016929). The antisense and sense probes contain 148 and 338 nt of additional vector sequence. Both probes were incubated with CLIC5A cDNA template (lanes 3 and 4) and treated with (lane 4) or without (lane 3) RNase. Kidney RNA was incubated with antisense (lane 5) or sense (lane 6) probe and treated with RNase. A 296-bp fragment was protected by antisense but not with sense probe hybridized to kidney RNA. C and D: CLIC5A PCR using human kidney or cultured bovine glomerular EC-derived cDNA as template. In C and D, primer pairs 1–5 are the same as primer pairs Hs.1–Hs.5 (Table 1). PCR products in lane 3* (D) were amplified with primer pair Bt.3 (Table 1). C: schematic representation of the CLIC5 gene. Exons: E1A/E1B-E6 (black), introns (white), location of Hs. RT-PCR primer pairs (1–5). D: RT-PCR products from human kidney mRNA (lanes 1–4 and 5) and bovine glomerular EC RNA (lane 3*). E: immunofluorescence labeling of mouse kidney with unfractionated anti-CLIC4/5 antiserum. F: Western blot analysis of recombinant CLIC4 (lane 1), recombinant CLIC5A (lane 2), kidney cortex (lane 3), and isolated glomeruli (lane 4). CLIC4 is the predominant band in kidney cortex, and CLIC5A is highly enriched in the glomerular fraction. β-actin labeling was similar for cortex and glomerular lysates.

Since PCR products were observed with primers homologous to exon 1A, but not 1B, we conclude that CLIC5A, but not CLIC5B, is expressed in kidney as a predominant 5.7-kbp transcript. Since the CLIC5A open-reading frame spans nt 319 to 1,074, the minor CLIC5A transcript variant identified by RT-PCR is predicted to shorten the 3′-UTR without alteration of the CLIC5A coding sequence.

Localization of CLIC5A protein in glomerular podocytes and ECs.

Expression of the CLIC5A protein in glomeruli was evaluated first with unfractionated rabbit antiserum (B132) and then with affinity-purified B132 antibody specific for human CLIC5A (7). Intense CLIC5A immunoreactivity was observed in mouse glomeruli of formaldehyde-fixed renal cortex after antigen unmasking (Fig. 1E). The unfractionated B132 antiserum recognizes CLIC5A (∼32 kDa) and CLIC4 (∼29 kDa) on Western blots of the recombinant proteins and kidney cortex (Fig. 1F). By contrast, only the CLIC5A isoform was observed in lysates of purified glomeruli (Fig. 1F). Staining of tubule epithelium in mouse kidney cortex (Fig. 1E, arrow) could be due to CLIC4 or CLIC5.

In frozen sections of bovine and human glomeruli, CLIC5A colocalized with the EC markers Tie-2 (Fig. 2A) and eNOS (not shown) and partially with the podocyte marker synaptopodin (Fig. 2B). To define the glomerular CLIC5A distribution in greater detail, immunogold TEM was performed with a monospecific rabbit anti-CLIC5A antibody that was cross-absorbed against CLIC4 and then affinity purified on a CLIC5A column. In podocyte foot processes, CLIC5A-reactive immunogold particles were found in close proximity to the apical plasma membrane and not at the filtration slit diaphragm or at the basal surface facing the GBM (Fig. 2C). In glomerular ECs, CLIC5A was observed in both the fenestrated (Fig. 2D) and nonfenestrated (data not shown) regions. The density of immunogold labeling was quantified by morphometric analysis (Fig. 2E). CLIC5A labeling was greatest in podocytes and glomerular ECs. Substantial immunogold labeling was also observed in fenestrated peritubular capillary ECs, but this was less than that observed in glomerular ECs (P < 0.01) or podocytes. Immunogold labeling in arteriolar ECs, mesangial cells, vascular smooth muscle cells, tubule epithelial cells, and red blood cells was less than that in peritubular ECs (P < 0.01), glomerular ECs (P < 0.001), or podocytes (P < 0.001).

Fig. 2.

CLIC5A protein expression. A: confocal microscopy with anti-CLIC5 antiserum (red) and anti-Tie-2 antibodies (green) in human glomeruli. B: confocal microscopy with anti-CLIC5 antiserum (red) and anti-synaptopodin antibodies (green) in human glomeruli. White arrows: apparent CLIC5 immunoreactivity in podocytes. C and D: CLIC5A immunogold transmission electron microscopy (TEM) with affinity-purified CLIC5A antibody in mouse kidney. C: podocytes. D: glomerular ECs. E: cell-associated gold particle density/background particle density in tissue-free areas (means ± SD; n = 3 sections, 10–12 images each; *P < 0.01, **P < 0.001 vs. glomerular ECs).

Glomerular CLIC5A is a component of the ezrin-podocalyxin complex.

Previous studies have established that ezrin connects podocalyxin to the apical cytoskeleton of podocyte foot processes (45, 51, 59). In addition, ezrin affinity chromatography was used in the original isolation and identification of CLIC5A as a cytoskeletal-associated protein (7). Given that the immunogold labeling of CLIC5A in podocytes observed here (Fig. 2C) is very similar to that described for ezrin and podocalyxin (45), we next examined the possibility that CLIC5A may be part of the ezrin-podocalyxin complex. Using immunofluorescence microscopy, we observed colocalization of CLIC5A with podocalyxin (Fig. 3A) and extensive colocalization of CLIC5A with ezrin in glomeruli (Fig. 3B). Furthermore, podocalyxin was coimmunoprecipitated with CLIC5A from lysates of purified mouse glomeruli as two molecular weight species similar in size to podocalyxin immunoprecipitated with rat anti-podocalyxin antibody (Fig. 3C). Unfortunately, in eluates from agarose and sepharose beads the CLIC5 antibodies produced marked nonspecific labeling in the 20–45 kDa range. Thus, CLIC5A/podocalyxin coimmunoprecipitation could not be confirmed by reciprocal IP. Ezrin was observed on Western blots of lysates from isolated glomeruli (Fig. 4) but was not found in immunoprecipitates from glomerular lysates prepared with CLIC5A or podocalyxin antibodies (not shown). The results may indicate that in glomeruli CLIC5A is complexed with podocalyxin but not ezrin. However, whereas the nonphosphorylated form of ezrin was abundant in glomerular lysates (Fig. 4B), the phosphorylated form was difficult to detect (data not shown) by Western blot. Since ezrin phosphorylation is required for its association with the actin cytoskeleton and for coupling of podocalyxin to actin (45, 59), it is possible that ezrin dephosphorylation during glomerular isolation or loss of ezrin complexes in the insoluble cytoskeletal residue that was removed by centrifugation precluded coimmunoprecipitation of ezrin from glomerular lysates with podocalyxin or CLIC5A.

Fig. 3.

Relationship of glomerular CLIC5A to ezrin and podocalyxin. A and B: confocal immunofluorescence microscopy of human renal glomeruli. A: rabbit anti-ezrin (green) and mouse monoclonal anti-CLIC5 (red). B: goat anti-podocalyxin (green) and mouse monoclonal anti-CLIC5 (red). C: immunoprecipitation of CLIC5A followed by Western blot (WB) analysis with goat anti-podocalyxin (PC) or mouse anti-CLIC5 antibodies. All data are representative of 3 or more separate experiments.

Fig. 4.

Reduced ezrin abundance in glomeruli and phospho-ezrin-radixin-moesin (ERM) abundance in podocytes of jbg/jbg mutant mice. A: confocal immunofluorescence of wild-type (+/+) and CLIC5-deficient (jbg/jbg) glomeruli with affinity-purified rabbit anti-CLIC5 antibody (representative of 3 experiments, ×40 magnification). B: Western blot analysis of lysates of exactly 20 glomeruli/lane prepared from 3 separate sets of age-matched wild-type (+/+), heterozygous (+/jbg), and CLIC5-deficient (jbg/jbg) mice (9 mice total). Individual glomeruli were isolated by micromanipulator to remove contaminating tubules. Blots were probed with rabbit anti-podocalyxin, monospecific rabbit anti-ezrin, and affinity-purified rabbit anti-CLIC5 antibodies. C and D: dual immunofluorescence confocal microscopy of TCA-fixed renal cortex from age-matched wild-type (+/+) and CLIC5-deficient (jbg/jbg) mutant mice. Representative of 3 separate experiments. C: goat anti-podocin and rabbit anti-phospho-ERM antibody (×40 magnification). The white arrow shows preserved phospho-ERM in a proximal tubule. D: goat anti-platelet endothelial cell adhesion molecule-1 (PECAM-1) and rabbit anti-phospho-ERM antibody (×100 magnification). The white arrow shows a dual band of phospho-ERM labeling along a glomerular capillary loop.

Reduced ezrin abundance in glomeruli of CLIC5A-deficient mice.

We next determined the level of expression of podocalyxin and ezrin in glomeruli from wild-type, (+/+), heterozygous (+/jbg), and CLIC5A-deficient (jbg/jbg) mutant mice. Glomerular CLIC5A immunoreactivity was absent in jbg/jbg mutant mice (Fig. 4A). Glomeruli were collected individually by micromanipulator to remove any contamination by proximal tubules, which contain abundant ezrin in their brush border (9, 14). By Western blot analysis of lysates prepared from pools of 20 glomeruli/mouse, podocalyxin protein abundance was similar, but the ezrin protein abundance was reduced in glomeruli of three separate jgb/jgb mutant mice compared with that in age-matched wild-type controls (Fig. 4B). In +/jbg heterozygous mice, glomerular ezrin abundance was intermediate between that of wild-type controls and jbg/jbg mutant mice. Immunofluorescence studies also showed a qualitative reduction in glomerular ezrin immunoreactivity in jbg/jbg mutant mice compared with age-matched wild-type mice (data not shown). As expected, no CLIC5A protein was detected on Western blots of glomerular lysates from jbg/jbg mutant mice, and in +/jbg heterozygous mice, CLIC5A protein abundance was less than that observed in the wild-type controls. The amount of podocin, another podocyte-specific protein, did not differ between +/+, +/jbg, and jbg/jbg mice on these blots (not shown).

The reduced ezrin levels in glomeruli from CLIC5-deficient mice observed by Western blotting, coupled with the previous finding that the ERM protein radixin is reduced in stereocilia bundles of CLIC5-deficient mice (23), led us to compare the localization of phospho-ERM immunoreactivity in glomeruli. To preserve ERM phosphorylation, tissue was rapidly fixed in trichloroacetic acid (26). The COOH-terminal phosphorylation site (Thr⁵⁶⁷ in ezrin) is highly conserved among the ERM proteins; consequently, phosphospecific antibodies do not distinguish between them. As shown in Fig. 4C, phospho-ERM labeling was observed in glomeruli of wild-type mice and overlapped partially with the podocyte marker podocin. In glomeruli of jbg/jbg mutant mice, phospho-ERM staining was much less intense, and colocalization with podocin was reduced compared with the appearance in wild-type mice. Since CLIC5A is also expressed in glomerular ECs, and moesin is the predominant ERM in ECs (9), phospho-ERM colocalization with the EC marker PECAM-1 was also evaluated. Immunofluorescence labeling for phospho-ERM was observed as dual bands along glomerular capillary loops in wild-type mice. One of the phospho-ERM bands strongly colocalized with PECAM-1. Interestingly, in glomeruli from age-matched jbg/jbg mutant mice, the phospho-ERM band colocalizing with PECAM-1 was preserved, whereas labeling for phospho-ERM presumably associated with podocytes was markedly reduced. Taken together with the observation that ezrin abundance in glomeruli is reduced (Fig. 4B), the data suggest strongly that reduced phospho-ERM immunoreactivity in glomeruli is due to reduced podocyte ezrin levels. By contrast, EC phospho-ERM is preserved in CLIC5A-deficient jbg/jbg mutant mice.

Ultrastructural abnormalities in glomeruli of CLIC5-deficient mice.

In homozygous jitterbug (jbg/jbg) mice, serum creatinine and blood urea nitrogen concentrations revealed no discernable differences between jbg/jbg mutant and wild-type mice ≤1 yr of age (data not shown), and renal histology appeared normal by light microscopy ≤18 mo of age. However, by TEM there was a conspicuous, patchy broadening and effacement of podocyte foot processes (Fig. 5A). The fenestrated portion of the glomerular endothelium appeared normal (Fig. 5A), but large vacuoles were observed in many glomerular EC bodies (Fig. 5B).

Fig. 5.

Abnormal glomerular ultrastructure and enhanced susceptibility to glomerular injury in jbg/jbg mutant mice. A: TEM appearance of the peripheral capillary wall of wild-type control and age-matched CLIC5A-deficient (jbg/jbg) mice. White asterisk shows a broadened foot process (scale bar, 0.5 μm). The bar graph shows the quantification of foot process density in 8-mo-old control (+/+) and CLIC5-deficient (jbg/jbg) mice (means ± SE; n = 4 mice/group, **P < 0.001). B: TEM of glomerular capillary endothelial cells from wild-type control and age-matched CLIC5-deficient (jbg/jbg) mice. Black asterisk shows vacuolization of glomerular ECs (scale bar, 10 μm). The bar graph shows the quantification of EC vacuoles/EC area (means ± SE; n = 4 mice/group, **P < 0.01). C: Western blot of mouse urine and serum with anti-albumin antibodies at baseline; 10 μl of urine (20 mg/dl creatinine) was loaded per lane. +/+: wild-type; jbg/+: heterozygous; jbg/jbg: homozygous mice. D: Western blot of mouse urine 3 wk after adriamycin administration. Each lane represents urine from a separate mouse. The equivalent of 0.25 μl of urine (20 mg/dl creatinine) was loaded per lane. Urine albumin was quantified densitometrically and expressed as the albumin/creatinine ratio. The difference between groups was significant (P < 0.02, unpaired t-test). Not shown: data for each mouse were consistent at 3 or more additional time points through 6 wk post-adriamycin administration.

Formal morphometric analysis of glomerular ultrastructure on TEM sections of glomeruli from four wild-type and four jbg/jbg mutant mice, all 8 mo of age, was performed by a blinded observer. The total glomerular area for jbg/jbg mutant and wild-type control mice was similar (Table 2), and the fraction of intravascular and extravascular compartments was the same (Table 2). However, within the intravascular compartment, there was a significant expansion of the EC compartment and an increase in the total capillary lumen area in the jbg/jbg mutant mice (Table 2). The thickness of the GBM was similar in the wild-type and jbg/jbg mutant mice (Table 2). The fraction of the glomerular cross-sections representing podocytes was similar for wild-type and jbg/jbg mutant mice (Table 2), but there was a highly significant (P < 0.001) reduction in the number of podocyte foot processes per linear unit of GBM (Table 2 and Fig. 5A). The foot process density in one 3-mo-old jbg/jbg mouse was similarly reduced (data not shown), suggesting that the changes in foot process density are not due to progressive loss with age. There was a subtle increase in urinary albumin concentration in the jbg/jbg mice by Western blot analysis with anti-albumin antibodies (Fig. 5C). Hence, at baseline, the absence of CLIC5A in renal glomeruli of mice leads to abnormal podocyte foot process morphology and glomerular EC vacuolization but only minimal albuminuria.

Table 2.

Morphometric analysis

	Points Counted, μm2		%Total
	+/+	Jbg/jbg	+/+	jbg/jbg
Area determinations: individual components
Intracapillary Space
Capillary lumen	2,910	3,043	25.40 ± 3.17	31.69 ± 2.39*
Mesangium
Mesangial cell	1,306	697	10.58 ± 3.29	7.01 ± 2.05
Matrix	1,705	754	13.54 ± 5.06	7.57 ± 1.87
Endothelium
Cell body	526	837	4.29 ± 1.74	8.72 ± 1.81*
Fenestrated portion	431	450	3.77 ± 0.31	4.58 ± 1.60
GBM	842	661	6.92 ± 1.16	6.90 ± 1.60
Extracapillary space
Podocyte
Cell body	1,769	2,063	16.17 ± 4.16	21.42 ± 1.78
Foot process	1,100	392	9.48 ± 0.68	4.06 ± 0.78†
Urinary space	1,126	796	9.85 ± 1.10	8.01 ± 1.89
Total	11,715	9,693	100	100
Area determinations: components summed
Intracapillary space	6,878	5,781	57.57 ± 5.25	59.58 ± 1.49
Extracapillary space	3,995	3,251	35.50 ± 5.33	33.51 ± 1.39
Podocyte	2,869	2,455	25.65 ± 4.43	24.48 ± 1.46
Endothelium	957	1,287	8.06 ± 1.67	13.30 ± 3.30*
Mesangium	3,011	1,451	24.11 ± 7.94	14.58 ± 3.79
Analyses based on GBM length determinations
GBM
Area, μm2	842	661
Length, μm	1,078	1,008
Thickness, nm			172 ± 46	177 ± 16
Foot processes
No. counted	9,660	3,930
Foot processes/μm GBM			2.12 ± 0.57	1.03 ± 0.03†

Values are means ± SD (n = 4 separate mice for each +/+ and jbg/jbg). GBM, glomerular basement membrane.

^*P < 0.05;

^†P < 0.001.

Finally, the susceptibility of mice lacking CLIC5A to adriamycin-induced glomerular injury was evaluated. Three weeks after adriamycin injection, albuminuria was observed in both wild-type mice and jbg/jbg littermates treated with 10 mg/kg adriamycin, but albuminuria was significantly greater (albumin/creatinine ratio 560 ± 135 vs. 1,691 ± 350 μg/mg, P < 0.02, in +/+ vs. jbg/jbg mice; n = 5/group, means ± SE) in the jbg/jbg mutant mice compared with the wild-type controls (Fig. 5D). This finding is interpreted to indicate that CLIC5A deficiency enhances the vulnerability of mice to glomerular injury with adriamycin.

DISCUSSION

The present study shows that CLIC5A, known to be required for the maintenance of actin-based stereocilia in inner ear hair cells, is also highly expressed in renal glomerular podocytes and ECs. In podocytes, CLIC5A localizes to the apical cell membrane of foot processes and podocyte cell bodies and is not found at the filtration slit or at the basal cell membrane. CLIC5A colocalizes with ezrin and podocalyxin in glomeruli, and podocalyxin coimmunoprecipites with CLIC5A from glomerular lysates. In CLIC5-deficient jbg/jbg mutant mice, the abundance of ezrin and consequently phospho-ERM is markedly diminished in podocytes. The absence of CLIC5A is also associated with conspicuous broadening and effacement of podocyte foot processes as well as vacuolization of glomerular ECs. The CLIC5A-deficient mice have microalbuminuria at baseline, and they are more susceptible to adriamycin-induced glomerular injury than their wild-type littermates. Together, these findings suggest that CLIC5A plays a functional role in defining the ultrastructure of glomerular podocytes by maintaining the ezrin-podocalxin complex.

Ezrin is known to be a good marker of developing and mature podocytes, and its distribution changes in response to podocyte injury (29). Using the immunogold TEM approach, Orlando et al. (45) previously reported a distribution for ezrin very similar to that observed here for CLIC5A. In podocytes, ezrin, together with the scaffold protein ERM-binding phosphoprotein 50 (EBP50), is thought to connect the cytoplasmic tail of podocalyxin to the actin cytoskeleton (45, 59). Furthermore, the association between ezrin and podocalyxin is lost when podocyte foot process effacement is induced by protamine sulfate or sialidase (59). Taken together with the findings that both the ERM protein radixin and CLIC5A are essential to the development or maintenance of specialized actin-based stereocilia in the inner ear (23, 34, 54) and that CLIC5A was originally identified through interaction with ezrin-containing cytoskeletal protein complexes isolated from human placental microvilli (7), it seems likely that CLIC5A and ezrin also cooperate in regulating membrane-cytoskeletal linkages required for proper morphology of podocyte foot processes. Ezrin abundance in glomeruli was markedly reduced in mice lacking CLIC5A, similar to the finding of reduced radixin expression in inner ear stereocilia (23). Therefore, we conclude that CLIC5A is located at or near the podocalyxin-ezrin complex in podocytes and that its absence leads to a reduction in ezrin protein abundance in these cells. Whether CLIC5A serves to stabilize the ERM proteins in podocytes and inner ear hair cells or regulates ERM expression remains to be determined.

We were surprised that the significant derangement of podocyte architecture was associated with only mild albuminuria and that the overall renal cortical histology remained normal even in aged jbg/jbg mutant mice. These findings argue that the selective elimination of CLIC5A and the associated reduction in ezrin expression are not sufficient to substantially alter glomerular permselectivity for albumin or lead to discernable glomerular matrix remodeling. Although it has been proven that proteins forming the filtration slit diaphragm and those linking podocytes to the GBM are critical in maintaining glomerular protein permselectivity (31, 49), our findings suggest that the ezrin-dependent apical membrane organization is less important in this function. Nonetheless, the absence of CLIC5A and associated abnormalities in ezrin abundance and glomerular structure do confer greater susceptibility to adriamycin-induced injury.

Since deletion of CLIC5 in jbg/jbg mutant mice results in general absence of CLIC5, it might be argued that the abnormalities observed here are due to a nonspecific systemic effect. However, this seems unlikely given that CLIC5A expression in glomeruli is much higher than in all other tissues we have examined (44), that CLIC5A colocalizes with ezrin and podocalyxin in podocytes, and that its absence results in reduced ezrin and phospho-ERM abundance in podocytes. Still, absolute proof will require future podocyte-specific rescue in the jbg/jbg mutant mice or podocyte-specific CLIC5A deletion.

In addition to defective podocyte structure in the jbg/jbg mutant mice, the glomerular capillary lumen represented a larger fraction of the total glomerular cross-sectional area than in wild-type mice, the glomerular EC compartment was expanded, and many glomerular ECs contained large vacuoles. At this time, the functional significance of these changes in glomerular ECs is unclear. Excess glomerular EC vacuolization has been described in rats given VEGF antagonists (1, 57). Since podocyte-derived VEGF is critical for glomerular EC differentiation (20, 21), the possibility that podocytes of CLIC5-deficient mice produce less VEGF than controls will need to be investigated, but since glomerular endothelial fenestration was preserved, a major defect in podocyte VEGF synthesis seems unlikely.

It is of note that phospho-ERM was diminished only in podocytes and not in glomerular capillary ECs of jbg/jbg-deficient mice (Fig. 4D), although CLIC5A is expressed by glomerular ECs in wild-type mice (Fig. 2) and in culture (52). Therefore, a function of CLIC5A that does not involve ezrin in glomerular ECs must be considered. In this regard, a role for CLIC4 in EC tube formation in vitro has been described (11, 62), and an angiogenesis defect has recently been reported in CLIC4-deficient mice (63). Furthermore, acidified vacuoles that appeared to be capillary lumen precursors were observed in three-dimensional cultures of ECs from wild-type mice, and in ECs from the CLIC4-deficient mice, vacuolar acidification was defective (63). Given the role of CLIC4 in ECs, we considered whether CLIC5 immunofluorescence and immunogold labeling of glomerular ECs in this study could be due simply to cross-reactivity of the anti-CLIC5 antibody with CLIC4 in ECs. This possibility seems unlikely given that we first discovered the transcript-specific CLIC5 SAGE tag in homogeneous cultures of glomerular ECs (52) and that the CLIC5 transcript was observed in cultured glomerular ECs by Northern blot (Fig. 1A) and RT-PCR (Fig. 1D). Moreover, we did not detect CLIC4 on Western blots of lysates prepared from isolated glomeruli (Fig. 1F), and glomerular EC vacuolization and EC compartment expansion phenotypes were observed in the CLIC5-deficient mice. Disruption of the CLIC5A homolog EXCretory canal abnormal (EXC)-4, normally associated with the plasma membrane of the excretory canal of C. elegans, also results in expansion of luminal membrane and cystic enlargement of the canal (5), similar to findings in this study. It is conceivable that the vacuoles observed in glomerular ECs of jbg/jbg mutant mice are in some way related to disordered capillary lumen formation and that CLIC5A functionally substitutes for CLIC4 in glomerular ECs, since these two proteins share 76% homology at the amino acid level.

In summary, this study pursued a potential functional role for CLIC5, identified previously as a glomerulus-enriched transcript by SAGE. The CLIC5A protein is found in glomerular ECs and in podocytes, and absence of CLIC5A results in morphological abnormalities in both cell types. In podocytes, absence of CLIC5A was associated with a marked reduction of ezrin and, consequently, phospho-ezrin protein abundance. Since phospho-ezrin is known to connect integral membrane proteins, including podocalyxin, to the cortical actin cytoskeleton, and since CLIC5A is required for regulating the morphology of ERM/actin-based structures lining the apical surface of cochlear and vestibular mechanosensory cells, it is likely that it serves a similar function in podocytes. We postulate that the interaction between podocalyxin and subjacent filamentous actin, which requires ezrin, is disrupted in podocytes of CLIC5A-deficient mice, leading to dysfunction when challenged with adriamycin.

GRANTS

This work was supported by Canadian Institutes of Health Research Grant MOP 641814. B. J. Ballermann holds the Tier 1 Canada Research Chair in Endothelial Cell Biology. M. Berryman and J. Paes received support from Ohio University; J. Paes was the recipient of a Student Enhancement Award.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.