https://orcid.org/0000-0002-9331-8143
Abstract
Endocytosis is a fundamentally important process through which material is internalized into cells from the extracellular environment. In the renal proximal tubule, endocytosis of the abundant scavenger receptor megalin and its co-receptor cubilin play a vital role in retrieving low molecular weight proteins from the renal filtrate. Although we know much about megalin and its ligands, the machinery and mechanisms by which the receptor is trafficked through the endosomal system remain poorly defined. In this study, we show that inositol phosphatase interacting protein of 27 kDa (Ipip27A), an interacting partner of the Lowe syndrome protein oculocerebrorenal syndrome of Lowe (OCRL), is required for endocytic traffic of megalin within the proximal renal tubule of zebrafish larvae. Knockout of Ipip27A phenocopies the endocytic phenotype seen upon loss of OCRL, with a deficit in uptake of both fluid-phase and protein cargo, which is accompanied by a reduction in megalin abundance and altered endosome morphology. Rescue and co-depletion experiments indicate that Ipip27A functions together with OCRL to support proximal tubule endocytosis. The results therefore identify Ipip27A as a new player in endocytic traffic in the proximal tubule in vivo and support the view that defective endocytosis underlies the renal tubulopathy in Lowe syndrome and Dent-2 disease.
Introduction
The renal proximal tubule is the site of retrieval of numerous solutes from the renal filtrate, which includes glucose, amino acids and low molecular weight proteins. The latter are retrieved by endocytic uptake into renal proximal tubular cells, mediated by the abundant endocytic receptor megalin [also known as LDL receptor-related protein 2 (LRP2)] and its co-receptor cubilin, which undergo continuous rounds of internalization and recycling to retrieve a multitude of low molecular weight proteins from the renal filtrate (1,2). Loss of megalin in various animal models results in a block in endocytosis from the apical pole of proximal tubular cells and in a striking reduction in the abundance of endocytic organelles found within these cells (3–5). Mutation of megalin in humans is responsible for Donnai-Barrow syndrome, with the hallmark sign of low molecular weight proteinuria, in addition to other symptoms that arise as a consequence of megalin loss from other tissues (5,6). Endocytosis of megalin is mediated by clathrin-coated vesicles and requires interaction between NPXY motifs in the cytoplasmic tail of megalin with the clathrin adaptor protein DAB adaptor protein 2 (Dab2) (7). The high endocytic activity of the proximal tubule is underscored by the fact that megalin and Dab2 are among the most highly expressed transcripts within proximal tubule cells (8). Despite the importance of megalin endocytosis for renal tubular physiology, we still have a poor understanding of how it is trafficked within the endosomal system of proximal tubule cells.
Lowe syndrome and Dent-2 disease are rare X-linked disorders that result in a selective renal tubulopathy with the hallmark trait of low molecular weight proteinuria as well as ocular and neurological symptoms that are more severe in Lowe syndrome (9–11). Both disorders arise from loss of function mutations in oculocerebrorenal syndrome of Lowe (OCRL), an inositol 5-phosphatase that preferentially hydrolyzes PtdIns (4,5) P2 (12,13). Loss of OCRL results in a number of phenotypes in cultured cells, affecting endocytic traffic (14–20), lysosome function and autophagy (20–22), cytokinesis (23,24), cell migration and polarity (25,26) and ciliogenesis (27–29), but the varying extent to which these processes contribute to disease pathology is poorly defined (30,31). We previously found that loss of OCRL in vivo causes defective endocytosis in the zebrafish pronephros, the larval renal tubule (32), which shares a high degree of functional conservation with the mammalian nephron (33,34). Defective endocytosis can explain the low molecular weight proteinuria seen in Lowe and Dent-2 patients (9,10) and in mouse models where OCRL has been knocked-out in combination with knockout of the homologous 5-phosphatase INPP5B (35), or in which a humanized version of INPP5B has been expressed in the OCRL knockout background (20,36). Interestingly, loss of OCRL in zebrafish also affects lysosomal homeostasis in the renal tubule, which mimics the situation in Lowe patients, further confirming the validity of the zebrafish model for the study of Lowe syndrome and Dent-2 disease (21).
Although OCRL is required for endocytosis in the proximal tubule, the underlying mechanisms remain to be identified. We therefore turned our attention to inositol phosphatase interacting protein of 27 kDa (IPIP27A) and IPIP27B, also known as Ses1 and Ses2 or PH domain-containing endocytic traffcking adapto PHETA1 and PHETA2, which are endosome-associated OCRL binding partners (37,38). Previous work has revealed a role for these proteins in endocytic trafficking, where they contribute to the process of receptor recycling from endosomes (38). IPIP27A can link OCRL to actin-associated Src homology 3 (SH3) domain proteins, including the -Fes/Cip4 homology Bin/amphiphysin/Rvs (F-BAR) protein Pacsin 2 (also known as Syndapin 2), suggesting a role in coupling OCRL to actin machinery during the generation of trafficking intermediates (39). To determine the functional importance of IPIP27A in vivo and to better understand how OCRL functions in endocytic traffic in the renal tubule, we generated a zebrafish knockout of IPIP27A. Strikingly, this mutant phenocopies the renal tubular endocytic defect seen in OCRL-deficient zebrafish, which is consistent with the two proteins acting together in trafficking, which is confirmed in rescue and co-depletion experiments. The results indicate that IPIP27A is a new player in renal tubular endocytosis and support the view that defective endocytic trafficking is responsible for the proteinuria seen in Lowe syndrome and Dent-2 disease.
Results
Conservation of IPIP27 in zebrafish
IPIP27 exists as two paralogues in mammalian species, termed IPIP27A and IPIP27B (also known as Ses1 and Ses2 or PHETA1 and PHETA2, respectively) (37,38). Analysis of the zebrafish genome indicated the presence of two genes encoding IPIP27 homologues (Fig. 1A and Supplementary Material, Fig. S1), both of which contain an F&H motif that was previously shown in other species to bind OCRL and INPP5B (37,38). We refer to the first of these as Ipip27A based upon overall sequence conservation (Fig. 1B and Supplementary Material, Fig. S2A), the presence of a conserved PPPxPPRR motif that is found in all mammalian IPIP27A proteins (Supplementary Material, Fig. S1A) and synteny around the IPIP27A locus between zebrafish and mammals (Supplementary Material, Fig. S2B). A similar observation was made by Ates et al., who describe the zebrafish orthologue as Pheta1 (40). In mammalian IPIP27A, the PPPxPPRR motif is required for binding to SH3 domain-containing actin-associated endocytic proteins (38). Notably, this motif is conserved in the single IPIP27 orthologue found in flies (37,38,41), indicating that binding to SH3 domain proteins is a conserved feature of IPIP27 across all species. The second zebrafish IPIP27 protein shares greater similarity at the amino acid level with human IPIP27A than human IPIP27B (Fig. 1B), and it also comprises a proline-rich C-terminus that is present in human IPIP27A but not IPIP27B (Fig. 1A and Supplementary Material, Fig. S1B). Moreover, the second zebrafish protein lacks a six amino acid insertion found in all mammalian IPIP27B orthologues (Supplementary Material, Fig. S1B), and the gene locus lacks synteny with IPIP27B from other species (Supplementary Material, Fig. S2B). Thus, although we refer to this variant as Ipip27B, and Ates et al. refer to it as Pheta2 (40), it is unclear whether it represents a true functional orthologue of IPIP27B or a variant of IPIP27A that has arisen within teleosts (Supplementary Material, Fig. S2A) as a likely consequence of the additional genome duplication event that occurred in this lineage (42). Interestingly, zebrafish Ipip27B contains a putative clathrin-binding site at its extreme C-terminus that is absent in mammalian IPIP27 proteins (Fig. 1A and Supplementary Material, Fig. S1A) but present in the single IPIP27 found in flies (37,38,41).
Zebrafish Ipip27 expression, localization and interaction with OCRL. (A) Schematic representation of human and zebrafish Ipip27A and B proteins indicating functional domains and total length. Different colours show predicted PH domain (red), coiled-coil region (grey), proline-rich region (blue), PPPxPPRR motif (orange), F&H motif (yellow) and clathrin-binding motif (fuchsia). (B) The table shows homologies between human and zebrafish IPIP27 proteins as percentage identity at the amino acid level. (C and D) Expression analysis of zebrafish ipip27a and ipip27b. RT-PCR of mRNA prepared from adult zebrafish tissues (C) or nine embryonic developmental stages (1–96 h postfertilization, hpf; D) using primers against ipip27a and ipip27b. Results are representative of three repeats. (E and F) Localization of GFP-Ipip27A (E, green) or GFP-Ipip27B (F, green) when transiently expressed in PAC-2 zebrafish fibroblasts. Cells were labelled with antibodies to EEA1 or Golgin 84, Alexa594-phalloidin to detect actin or were co-transfected with mCherry-clathrin light chain (mCh-clathrin) (red), as indicated. (G) Co-expression of GFP-Ipip27A or Ipip27B (green) with mCherry-OCRL (red) in PAC-2 cells. Scale bars in (E–G) represent 10 μm. (H) The indicated recombinant MBP-tagged Ipip27A or B constructs were incubated with PAC-2 cell extract and binding of OCRL was monitored by western blotting.
Expression analysis of zebrafish Ipip27
The tissue expression of IPIP27 paralogues has yet to be described for any species. We therefore performed RT-PCR analysis of adult zebrafish tissues to determine the expression profile of ipip27a and ipip27b. As shown in Figure 1C, ipip27a is expressed in all tissues at a comparable level. In contrast, although ipip27b is also expressed in all tissues, its level of expression is more heterogeneous, with highest expression in the ovaries and lowest expression in the liver and muscle (Fig. 1C). Both ipip27a and ipip27b are expressed throughout early development (Fig. 1D), which was reported previously (40), consistent with a function during embryonic development.
Subcellular localization of zebrafish Ipip27 and binding to OCRL
Mammalian IPIP27A and IPIP27B are diffusely localized in the cytoplasm but also associate with early and recycling endosomes and the trans-Golgi network (37,38). Attempts to raise antibodies for immunofluorescence detection of endogenous Ipip27A and Ipip27B were unsuccessful. To determine the subcellular localization of the zebrafish proteins, we therefore expressed GFP-tagged versions of Ipip27A and Ipip27B at low levels (as determined by fluorescence intensity) in PAC-2 zebrafish fibroblasts. As seen with the mammalian protein, Ipip27A was mainly diffused in the cytoplasm but was also localized to cytoplasmic structures corresponding to early endosomes and the Golgi apparatus (Fig. 1E). We also observed a pool of the protein associated with filamentous and punctate actin structures, the latter of which may correspond to endocytic compartments (Fig. 1E). A similar localization was seen with FLAG-tagged Ipip27A, indicating that GFP-tagging did not affect localization (Supplementary Material, Fig. S3A). Ipip27B localization was noticeably more punctate than Ipip27A (Fig. 1F and Supplementary Material, Fig. S3B). Some of the larger puncta coincided with early endosomes, but the majority of puncta did not. Instead, there was a striking co-localization of Ipip27B with clathrin, suggesting that the majority of puncta correspond to clathrin-coated pits or buds, or transport vesicles (Fig. 1F). This observation is consistent with the presence of a putative clathrin-binding site in Ipip27B (LIDL), the deletion of which partially reduced co-localization with the clathrin puncta (Supplementary Material, Fig. S3C). In contrast, there was little apparent co-localization of Ipip27B with the Golgi apparatus.
Upon co-expression of zebrafish OCRL, both Ipip27A and Ipip27B gave near total overlap with the expressed OCRL (Fig. 1G), consistent with an interaction between the proteins. To confirm interaction with OCRL, pull-down experiments were performed. Maltose binding protein (MBP)--tagged recombinant C-terminal regions of Ipip27A and Ipip27B were coupled to beads and incubated with PAC-2 cell extract. As shown in Figure 1H, both Ipip27A and Ipip27B pulled-down endogenous OCRL. Binding via the F&H motif was confirmed for Ipip27A since mutation of this motif completely abolished interaction with OCRL (Fig. 1H).
Knockout of zebrafish Ipip27A does not affect viability or growth
In this study, we aimed to determine the functional importance of Ipip27A in vivo. We focussed on this protein as it has the best evolutionary conservation (Fig. 1B and Supplementary Material, Figs. S1 and S2) and is the best understood in terms of cellular function, linking OCRL to actin-associated SH3 domain endocytic proteins (39). To assess Ipip27A function in vivo, we used transcription activator-like effector nuclease (TALEN)-mediated genome editing to generate a knockout strain. TALEN target sequences were selected at the 5′ end of the ipip27a gene (0–63 bp), which exists as a single exon, to maximize the likelihood of generating a null allele (Fig. 2A). The target sequence contained a MwoI restriction enzyme site such that successful editing would result in a loss of cleavage. Germline transmission and generation of the mutant strain was confirmed using MwoI digestion of PCR fragments generated from the genomic locus of F2 animals (Fig. 2B). DNA sequencing confirmed the presence of a 10 bp deletion at the target site (Fig. 2C), which would result in a frameshift mutation and generation of a severely truncated protein, encoding only the 10 most N-terminal amino acids of Ipip27A, followed by 27 nonsense amino acids (Fig. 2D). In-crossing of F2 ipip27a+/− animals generated offspring at the expected Mendelian ratios, with no increased mortality at the embryonic or larval stages of development of the ipip27a−/− fish (Fig. 1E). The ipip27a−/− fish did not display any gross morphological abnormalities at the larval or juvenile stages of development and were viable and fertile at adult stages (Fig. 2F and Supplementary Material, Fig. S4). Similar results were obtained by Ates et al., who generated ipip27a/pheta1 as well as ipip27b/pheta2 and double knockout zebrafish using CRISPR-Cas9 (40). Thus, Ipip27A is not essential for zebrafish development, nor is it required for viability or fertility of adult zebrafish.
Knockout of Ipip27A in zebrafish. (A) The two TALEN arrays designed to target zebrafish ipip27a are shown bound to the genomic DNA strands. (B) ipip27a genotyping of F2 embryos using MwoI restriction digestion showing, in order, a wild-type, ipip27a−/− (100% uncut), +/− (50% uncut) and +/+ (100% cut) embryo. (C) ipip27a genomic sequence showing the TALEN target sequences (blue boxes) and a 10 bp deletion (red dashes) present in the F2 embryos that results in a frame-shift mutation in the coding region. (D) Alignment of wild-type Ipip27A protein sequence with the mutated version produced by TALEN mutagenesis. Green indicates the alternative protein sequence generated by frame-shift mutation, with a premature stop codon highlighted in red. (E) Viability of ipip27a−/− F2 embryos assessed at 48 hpf by genotyping offspring from an in-cross of ipip27a+/− F1 parents. (F) Morphology of ipip27a−/− F2 embryos at 48 hpf.
Ipip27A knockout causes impaired renal tubular endocytosis
We and others have previously shown that loss of OCRL in zebrafish, achieved either through generation of a genetic mutant or via injection of morpholino anti-sense oligonucleotides, results in impaired endocytic uptake into the proximal tubule of the larval zebrafish kidney (pronephros) (32,40). We therefore investigated whether Ipip27A may function in endocytic traffic within the proximal tubule. Tubular uptake was assessed using fluorescent Alexa 488-conjugated 10 kDa dextran, which is efficiently filtered by the glomerulus and internalized by endocytosis into epithelial cells lining the proximal tubule. Confocal analysis of cross-sections through the proximal tubule of wild-type larvae indicated clear uptake of dextran into cytoplasmic puncta corresponding to endosomal and lysosomal compartments (Fig. 3A). The accumulation of dextran within these compartments was greatly reduced in ipip27a−/− larvae, indicating a deficit in renal uptake (Fig. 3A). The reduction in renal uptake of dextran could also be visualized using stereofluorescence microscopy (Fig. 3B). Categorical scoring of the phenotype indicated a strong loss of renal uptake in the mutant (Fig. 3C). The uptake defect was stronger than that described by Ates et al. in their ipip27A/pheta1 knockout zebrafish (40). Importantly, re-expression of Ipip27A in the pronephros, achieved using the enpep promoter (43), restored dextran uptake, confirming specificity of the phenotype (Fig. 3B and C). Reduced dextran uptake was also observed in larvae injected with a morpholino to suppress Ipip27A expression, further indicating that Ipip27A is required for efficient proximal tubular endocytosis (see Fig. 5E). To further assess the renal phenotype, we injected fluorescently conjugated receptor-associated protein (RAP), which is a ligand of the major endocytic receptor megalin. Although not as efficiently internalized into wild-type larvae as dextran, possibly owing to reduced glomerular filtration (RAP is 39 kDa compared with 10 kDa for dextran), we again observed a strong loss of uptake into the ipip27a−/− larvae (Fig. 3D). Loss of Ipip27A therefore leads to an impairment of both fluid phase and receptor-mediated endocytosis in the proximal tubule.
Defective endocytosis and reduced megalin abundance in the pronephros of Ipip27A knockout zebrafish. (A) Confocal transverse sections of the proximal pronephric tubule of 3 dpf wild-type and ipip27a−/− larvae showing accumulation of injected Alexa 488-conjugated 10 kDa dextran (green) in endocytic and lysosomal compartments at 2.5 h postinjection. Scale bar represents 10 μm. (B) Fluorescence stereomicroscope images of the pronephric tubule (white dashed lines) of 3 dpf wild-type, ipip27a−/−, or ipip27a−/− larvae rescued with enpep-driven wild-type Ipip27A, which were injected with Alexa 488–10 kDa dextran (green) and imaged after 2.5 h. (C) Quantification of pronephric uptake of Alexa 488–10 kDa Dextran. Data are presented as the mean of ± SD from three experiments. (D) Top, fluorescence stereomicroscope images of the pronephric tubule of 3 dpf wild-type or ipip27a−/− larvae injected with Cy3-conjugated RAP (green) and imaged after 1 h. Bottom, quantification of RAP uptake. Data are presented as the mean of ± SD from three experiments. ***P < 0.0001, **P < 0.001. (E) Quantification of the relative megalin fluorescence levels in confocal sections of WT and ipip27a−/− larvae. Data are the mean ± SD from 15 sections (three repeats/5 sections each). **P < 0.001. (F) Transverse confocal sections of the renal proximal tubule of WT and ipip27a−/− 3dpf embryos stained with anti-megalin antibody. The images have been inverted, and black dashed lines indicate the edges of the pronephric tubule, black squares indicate the magnified regions on the right and pink arrowheads indicate megalin-positive vacuolar endosomal structures. Scale bar represents 10 μm.
Normal glomerular function and renal tubular polarity in Ipip27A knockout zebrafish
The consequence of loss of Ipip27A upon another major aspect of renal function, namely glomerular filtration, was assessed next. The integrity of the glomerular filtration barrier can be analyzed in zebrafish by injecting larvae with fluorescently conjugated 500 kDa dextran (Supplementary Material, Fig. S5A), which is unable to pass through a normally functioning intact glomerular barrier, resulting in retention within the bloodstream over prolonged periods. As expected, 500 kDa dextran was retained in the bloodstream of wild-type larvae 24 h following injection (Supplementary Material, Fig. S5B). The same result was obtained with ipip27a−/− larvae and was reported previously (40), indicating that loss of Ipip27A does not disrupt the integrity of the glomerular filtration barrier (Supplementary Material, Fig. S5C).
The gross morphology and apicobasal polarity of the renal tubule were analyzed next. As shown in Supplementary Material, Figure S6A, the overall renal tubular morphology was comparable between wild-type and ipip27a−/− larvae, although there was some dilation of the more proximal segment. Similarly, there was no effect upon renal tubular cell polarity in ipip27a−/− larvae, assessed using immunolabelling of cryosections (Supplementary Material, Fig. S6B) and electron microscopy, which indicated the presence of an apical brush border and tight junctions (Supplementary Material, Fig. S6C; see also Fig. 4). Loss of Ipip27A therefore does not affect renal tubule formation or polarity, although some dilation of the proximal segment is discernible.
Block face electron microscopy of the proximal tubule of Ipip27A knockout zebrafish. (A-C) Block face scanning electron microscopy images of transverse sections of the proximal renal tubule of 3 dpf wild-type, ipip27a−/− or ipip27a−/− larvae rescued with enpep-driven wild-type Ipip27A. (A) Regions containing vacuolar endosomes and lysosomes are highlighted by black and pink squares, respectively. and shown at higher magnification in (B) and (C). Scale bar represents 10 or 2 μm for the low and high magnification images, respectively. (D and E) Quantification of the average area, and number and total area per section of vacuolar endosomes (D) and lysosomes (E). Data are presented as the mean of ± SD (10 sections analyzed for three kidneys for a total number of 30 sections per category). **P < 0.001, *P < 0.01.
Reduced megalin abundance and altered endocytic compartments upon loss of Ipip27A
We previously showed that loss of OCRL results in reduced megalin abundance and its accumulation in endocytic compartments within the zebrafish proximal tubule (32). A similar phenotype was seen in OCRL-deficient mice, consistent with defects in endocytic recycling of megalin (20). We therefore assessed megalin abundance and distribution in the pronephros of ipip27a−/− larvae. As shown in Figure 3E and F, megalin abundance was decreased upon loss of Ipip27A, and there was shift in its distribution from the apical membrane to sub-apical vacuolar compartments, most likely corresponding to endosomes and lysosomes.
We next wanted to assess whether Ipip27A knockout impacted upon the morphology of endocytic and lysosomal compartments of proximal tubular cells, as seen previously upon loss of OCRL (20,32). The endocytic compartments of the proximal tubule are well defined morphologically, comprising sub-apical endocytic vesicles of <0.25 μm diameter, larger electron-lucent vacuolar endosomes (≥0.5 μm diameter) and numerous electron-dense recycling tubules. Lysosomes are also discernible by morphology, comprising electron-dense oval or spherical organelles of at least 0.5 μm diameter. The endocytic and lysosomal compartments were present in both wild-type and ipip27a−/− larvae. However, there was an increase in the size and total area of the sub-apical vacuolar endosomes (black squares) in the ipip27a−/− larvae (Fig. 4A, B and D). This effect was specific as it was rescued by re-expression of Ipip27A in the renal tubule (Fig. 4A, B and D). The other striking observation was that the abundance, size and total area of lysosomes (pink squares) were increased in the ipip27a−/− larvae compared with wild-type (Fig. 4A, C and E). Again, this phenotype could be rescued by renal tubule-specific expression of Ipip27A (Fig. 4A, C and E). The expansion of both apical vacuolar endosomal and lysosomal compartments is similar to that observed in zebrafish lacking functional OCRL (21,32). Hence, loss of Ipip27A phenocopies the loss of OCRL in terms of effects upon endosomal and lysosomal morphology. The results are consistent with defects in endocytic trafficking and lysosomal homeostasis (31). Interestingly, in the ipip27a−/− larvae, many of the lysosomes were found in close contact with the enlarged electron-lucent vacuolar endosomes, suggestive of a defect in endosome-lysosome fusion, as has been shown previously for cultured cells lacking OCRL (21).
Ipip27A function in endocytic traffic is dependent upon interaction with OCRL and SH3 domain proteins
To obtain a better mechanistic understanding of Ipip27A function, in vivo rescue experiments were performed with mutant versions of the protein deficient in binding to either OCRL- (F234AH238A, termed F&H) or SH3 domain-containing partners (R215AR216A, termed RRAA) (Fig. 5A). These mutants were expressed in the renal tubule of ipip27a−/− larvae using enpep-driven expression, with wild-type Ipip27A serving as a positive control and dextran endocytosis being monitored by stereofluorescence microscopy. RT-PCR confirmed that the ectopically expressed wild-type Ipip27A and mutant variants were expressed at comparable levels (Fig. 5B, lower bands marked with an asterisk). As expected, re-expression of wild-type Ipip27A efficiently restored dextran uptake in the proximal tubule (Fig. 5C and D). In contrast, both F&H and RRAA mutants were deficient in rescue (Fig. 5C and D). This result indicates that Ipip27A must interact with OCRL- and an SH3 domain-containing binding protein (s) to function in endocytic traffic within the proximal tubule. The former is consistent with the previous observation that OCRL deficient in Ipip27A and APPL1 binding (both proteins share the same binding site in OCRL) is unable to support endocytic traffic in the proximal tubule (32). The results support the view that OCRL and IPIP27A act together in endocytic traffic in the renal tubule in vivo. To further test this idea, the proteins were depleted from zebrafish larvae using morpholino injection into wild-type or mutant zebrafish strains. Embryos were injected with a sub-maximal dose of morpholino to give a partial phenotype and to assess any additive effects that may occur in the mutants. Morpholino-induced depletion of OCRL or Ipip27A in a wild-type background resulted in impaired dextran uptake into the proximal tubule, and dextran uptake was also impaired in ocrl−/− or ipip27a−/− mutants, as expected (Fig. 5E). Loss of both proteins, either through morpholino-induced depletion of Ipip27A in the ocrl−/− mutant, or depletion of OCRL in the ipip27a−/− mutant, resulted in a more severe impairment of dextran uptake (Fig. 5E). Together the results indicate that OCRL and Ipip27A act together in supporting renal tubular endocytosis. The results also indicate that Ipip27A must also engage with an SH3 domain protein (s) to fulfil its endocytic function.
Interaction between Ipip27A and OCRL is required for renal tubular endocytosis. (A) Schematic diagram of zebrafish Ipip27A showing domains and the point mutations used in functional rescue experiments. (B) RT-PCR of mRNA or PCR of genomic DNA prepared from wild-type, ipip27a−/− or ipip27a−/− larvae rescued with enpep-driven wild-type Ipip27A or the indicated point mutants. PCR products were digested with MwoI prior to electrophoresis to distinguish between the endogenous WT and ipip27a−/− alleles. Rescue constructs ectopically expressed from the enpep promoter are marked with an asterisk. (C) Uptake of Alexa 488–10 kDa dextran (green) into the pronephric tubule (dashed lines) of the indicated 3 dpf larvae imaged at 2.5 h postinjection. (D) Quantitation of dextran uptake. Data are presented as the mean ± SD from three experiments. ***P < 0.001. (E) Quantitation of dextran uptake in 3 dpf wild-type, ocrl−/− or ipip27a−/− larvae injected with 2 ng of either control (Ctrl) morpholino (MO) or MO targeting ocrl or ipip27a. Data are mean ± SD, n = 23–29 from two experiments. **P < 0.01, *P < 0.05.
Discussion
The molecular machinery and mechanisms responsible for endocytic trafficking of megalin in the proximal tubule are poorly understood. Our results identify Ipip27A as a new player in this process. Loss of Ipip27A causes a deficit in endocytic uptake, accompanied by a reduction in megalin abundance and alteration of endosome morphology. It remains to be determined precisely how loss of Ipip27A leads to these effects, but our favoured hypothesis is that defective recycling is responsible, leading to a reduction in megalin abundance at the apical membrane and to a consequent impairment of endocytosis. This would be consistent with the observed redistribution of megalin to endosomal compartments, as well as reduced megalin abundance owing to mis-sorting to the lysosome, and enlargement of vacuolar endosomes as a likely consequence of continued membrane input but reduced formation of recycling carriers. Moreover, cell culture studies support a role for mammalian IPIP27A in receptor recycling (38). However, it is also possible that Ipip27A could function directly in endocytosis at the plasma membrane. Mammalian IPIP27A is enriched in clathrin-coated vesicles (38), and its known binding partners OCRL and Pacsin 2 have also been localized to clathrin-coated pits (44–46). Further experiments will be required to pinpoint precisely where Ipip27A functions within the pathway.
How might Ipip27A function mechanistically within the proximal tubule? Rescue experiments indicate that binding to an SH3 domain protein is functionally important. Previous work has shown that mammalian IPIP27A can functionally link OCRL to the actin-associated SH3 domain protein Pacsin 2 and that this complex is important for endocytic traffic, most likely by coordinating actin dynamics and membrane morphogenesis during transport carrier formation (39). Pacsin 2 is therefore a candidate for the SH3 domain-containing Ipip27A-binding partner within the proximal tubule. It is expressed in proximal tubular cells (47), although at a lower level than that seen during nephrogenesis (48). It is therefore conceivable that another SH3 domain-containing endocytic protein is involved, which could include Myosin 1E, CD2AP or Abi1, which have been identified as possible IPIP27A-binding partners (39). It will be important to identify the relevant partner in order to better understand how Ipip27A functions in proximal tubule endocytic traffic.
In a recent study, Ates et al. generated single and double knockout ipip27a/pheta1 and ipip27b/pheta2 zebrafish (40). They also observed reduced renal tubular uptake, although the phenotype in the single ipip27a knockout was milder than the one we report here. There may be several possible explanations for this. It is known that stable knockout can trigger adaptation or compensatory responses in vivo (49), and it is conceivable that the degree of adaptation, which would be expected to ameliorate the phenotype, is different in the two knockout strains. Indeed, we have observed a progressive decrease in phenotypic severity through maintenance of the Ipip27A line (data not shown). The ability of an Ipip27A morpholino, which acutely depletes the protein, to phenocopy our Ipip27A knockout, together with rescue experiments, supports the veracity of our conclusions using the knockout line. Regardless of the different phenotypic severities observed, the two independent studies support a role for Ipip27A in renal tubular endocytosis.
Interestingly, mutation of IPIP27A in humans is associated with a human disorder affecting various tissues including the kidney, where oligosacchariduria was described (40). Changes in oligosaccharides in the urine are associated with lysosomal storage diseases (50), consistent with the view that IPIP27A deficiency may affect lysosomal homeostasis. We may have expected low molecular weight proteinuria, a hallmark of defective tubular endocytosis, in the patients, but this was not reported (40). Thus, we currently lack evidence that IPIP27A deficiency results in a tubular endocytosis defect in humans. The patient mutation (R6C) does not disrupt OCRL binding (40), so it is possible that some IPIP27A functions are maintained while others are lost in the patient cells, which would be different to a complete loss of the protein. This could account for the craniofacial abnormalities seen in patients, which were phenocopied in ipip27a/pheta1 and ipip27b/pheta2 knockout zebrafish and were attributed to changes in cartilage development, although the mechanistic details underlying this phenotype remain to be determined (40). Further work will be required to ascertain the effect of the patient mutation on IPIP27A function.
The mechanisms underlying the Lowe syndrome and Dent-2 disease phenotype remain poorly defined and may be multifactorial (30,31). Our data support the view that defective endocytosis is a major pathological mechanism in the proximal tubule. Defective trafficking of megalin can account for the low molecular weight proteinuria seen in patients, but defective endocytosis of other cell surface proteins such as ion transporters may also occur, which could account for the other more variable aspects of the tubulopathy (10,11). Other mechanisms may also contribute to the tubular phenotype, including dysregulation of lysosome homeostasis and autophagy (20–22), ciliogenesis (27–29) and cell division (23,24), which can cause a shortening of proximal tubule length (51). IPIP27A may also play a role in these processes alongside OCRL. Indeed, we observed a dramatic accumulation of lysosomes in the Ipip27A knockout zebrafish, which is similar to that seen upon loss of OCRL, where its role in lysosomal homeostasis is separate to that in endocytosis (20,21). Ates et al. reported defective ciliogenesis in their ipip27a/pheta1 and ipip27b/pheta2 knockout zebrafish, while other work has shown that IPIP27A is required for efficient cytokinesis in human and Drosophila cells (41).
Mammals contain two IPIP27 paralogues, IPIP27A and IPIP27B, which are conserved between species. IPIP27A is conserved in zebrafish, whereas IPIP27B conservation is less clear. Although we have designated the zebrafish version as Ipip27B, one could argue that, because of a lack of synteny and conservation of sequence features with mammalian IPIP27B, it is not a true orthologue. It is widely expressed, and interestingly, contains a putative clathrin-binding site at its C-terminus, which is absent in mammalian IPIP27B. Based on its possible direct interaction with clathrin, and its localization to clathrin-coated pits and vesicles in cells, it is likely that zebrafish Ipip27B functions in clathrin-dependent traffic. The phenotype we see in Ipip27A knockout zebrafish indicates that Ipip27B cannot compensate for loss of Ipip27A, indicating a lack of redundancy between the proteins, which agrees with their distinct functional motifs. Nevertheless, it remains possible that some overlap in function exists in vivo. This would be consistent with the stronger phenotypes seen in double knockout zebrafish (40) and the ability of the mammalian proteins to heterodimerize (38). Further studies into the cellular roles of mammalian IPIP27A and IPIP27B and their zebrafish orthologues will be very interesting.
In conclusion, we show here that Ipip27A is an important player in endocytic traffic within the proximal renal tubule, where it functions together with OCRL to support endocytosis. The findings are relevant not only for understanding of endocytic traffic in the proximal tubule but also for a better appreciation of the mechanisms underlying the genetic disorders Lowe syndrome and Dent-2 disease.
Materials and Methods
Antibodies
The antibodies used in this study were sheep anti-zebrafish OCRL (52), goat anti-EEA1 (Santa Cruz Biotechnology, sc-6415), sheep anti-Golgin 84 (53), rabbit anti-megalin (kindly provided by Dr Michele Marino, University of Pisa, Italy), α6F mouse anti-NaK ATPase (Developmental Studies Hybridoma Bank) and rabbit anti-ZO1 (from Professor Karl Matter, UCL, London). Fluorophore-and HRP-conjugated secondary antibodies were purchased from Thermo Fisher.
Molecular biology
All constructs were made using standard molecular biology techniques. Zebrafish ipip27a and ipip27b gene sequences are designated on Ensembl as ENSDARG00000067546 and ENSDARG00000053257, respectively. Full-length zebrafish ipip27a cDNA was cloned into pT2KXIGDin-enpep vector (Dr Michael Pack, University of Pennsylvania, USA) for expression in zebrafish pronephric tubules. Point mutations were introduced by PCR using the site-directed mutagenesis Quikchange method (Stratagene). Plasmids for transfection of PAC-2 cells were generated by cloning the cDNAs for full-length zebrafish ipip27a or ipip27b into pcGlobin-GFP or pcDNA3.1-FLAG, or by cloning full-length zebrafish ocrl into pcGlobin-mCherry. MBP-tagged ipip27a constructs for bacterial expression were made by cloning the C-terminal regions of zebrafish ipip27a (encoding amino acids 127–256) or ipip27b (encoding amino acids 135–282) into pMAL-C2. Primer sequences for all manipulations are available upon request. All constructs were verified by DNA sequencing. Plasmid encoding GFP under control of the cardiomyosin light chain 2 promoter (cmlc2:GFP) was obtained from Dr Adam Hurlstone (University of Manchester, UK).
Cell culture and transfection
PAC-2 cells were maintained at 28°C with atmospheric CO2 in Lebovitz L15 medium (Gibco) containing 15% v/v Hyclone fetal bovine serum and 1% penicillin–streptomycin solution. Transfection of DNA constructs was performed with X-tremeGENE HP DNA (Roche) according to the manufacturer’s instructions by using 1 μg DNA diluted into 100 μl Opti-MEM (Gibco) and 2.5 μl X-tremeGENE HP. Cells were incubated with transfection reagent for 6 h after which the medium was replaced and the cells were incubated overnight prior to analysis.
Recombinant protein expression and pull-down experiments
MBP or MBP-tagged zebrafish Ipip27A or Ipip27B were expressed in Escherichia coli and were purified using amylose resin (New England Biolabs) according to the manufacturer’s instructions. PAC-2 cell extracts were prepared by lysis in HMNT buffer [20 mM HEPES, pH 7.4, 0.15 M NaCl, 0.5% (wt/vol) Triton X-100, 2 mM EDTA, protease inhibitors]. Purified MBP or MBP-tagged Ipip27 proteins were coupled to amylose resin and were incubated with PAC-2 cell extract for 3 h at 4°C. Beads were washed in HMNT and bound proteins were eluted with SDS sample buffer prior to analysis by SDS-PAGE and western blotting.
Zebrafish strains and husbandry
Zebrafish were raised and maintained at the University of Manchester Biological Services Unit according to the UK Animals Act 1986. The wild-type line was of the AB background. The ocrl−/− mutant line (ZDB-GENO-120531-1) was described previously (52). Generation of the ipip27a−/− mutant line, designated ipip27aumc1, was performed with TALENs by using the Golden Gate system and by following the protocol originally described by (54), and it is available at (https://www.addgene.org/static/cms/files/GoldenGateTALAssembly2011_2.pdf). Target sites and the corresponding repeat-variable diresidue (RVD) sequences were chosen using the online tool Seqbuilder TAL Effector Nucleotide Targeter 2.0 (https://tale-nt.cac.cornell.edu) and were designed to target a 63 bp sequence at the 5′ end of the gene, containing an internal MwoI restriction site. The array plasmids were fused to the Fok1 endonuclease in the GoldyTALEN backbone. After linearization, mRNAs were transcribed using the T3 mMessage mMachine Kit (Ambion) according to the manufacturer’s instructions. The two mRNAs corresponding to the left and right arms were then mixed in equal quantities and were injected into embryos at the one-cell stage. To assess mutagenesis efficiency, 10 embryos at 2 dpf were collected from each clutch, genomic DNA was extracted and the region of interest was amplified using PCR and was subjected to digestion with MwoI. If mutagenesis was successful, indicated by uncut DNA, embryos were grown to adulthood, outcrossed with wild-type fish, which were genotyped by fin clipping and PCR as described before. Mutagenesis was confirmed by DNA sequencing. In-crossing of the F1 line was used to generate homozygous mutants for experiments.
RNA isolation and PCR
Total RNA was isolated from zebrafish embryos or adult tissues using Trizol (Invitrogen) and was reverse-transcribed with Superscript First Strand (Invitrogen) to produce cDNA. For direct analysis of amplification products, cDNA was amplified using standard PCR conditions and appropriate primer pairs. For PCR of genomic DNA, genomic DNA was extracted from single or pooled embryos according to (55), or was isolated from fin clips (1 mm2) taken from juvenile fish by extraction into 50 mM NaOH, heating to 95°C, neutralization using Tris, pH 8 and by centrifugation to remove insoluble debris. PCR was performed using standard conditions and appropriate primer pairs.
RNA, DNA and morpholino injections in zebrafish
Capped mRNA encoding tol2 transposase was transcribed from the pCS2-FA vector (Dr Michael Pack) using the mMessage mMachine kit (Ambion), and approximately 1 nl of a mix of 10 ng/μl pT2KXIGDin-enpep vector containing Ipip27A, 10 ng/μl cmcl2:GFP vector and 20 ng/μl tol2 transposase mRNA was injected into one-cell stage mutant embryos. Morpholinos were obtained from GeneTools. Control and ocrl morpholinos were described previously (52). Morpholino targeting zebrafish ipip27a had the sequence CCACACTCCTCTCGTTCAGCTTCAT. Morpholinos were injected into the yolk of one cell stage embryos at 2 ng per embryo.
Injection and analysis of endocytic tracers
Lysine-fixable Alexa 488-conjugated 10 or 500 kDa dextran (Thermo Fisher) was prepared in PBS at 2 μg/μl final concentration. Recombinant Cy3-labelled His-tagged RAP (39 kDa) was prepared in PBS at 5 μg/μl final concentration. The injected volume was adjusted individually for each tracer used based on the total fluorescence in the larvae circulatory system. Zebrafish larvae at 72 hpf were anesthetized with 0.2 mg/ml MS222 (Sigma) in chorion water and were tracer-injected into the common cardinal vein using a glass micropipette PLI-90 Pico-Injector (Harvard Apparatus). Pronephric uptake was assessed at between 1 and 2.5 h after injection on whole mounts using a fluorescence dissecting stereomicroscope (Leica MZ10F). Statistical analysis was performed using the Pearson’ s chi-squared test with Prism software (Prism Software Corporation).
Fluorescence microscopy
Zebrafish larvae were fixed using 4% PFA or Dent’s Fixative, mounted in cryosectioning moulds, frozen on dry ice and sectioned using a Leica CM3050 S cryotome. Images were captured using a Leica SP5 confocal microscope and were analyzed using ImageJ software. Megalin fluorescence intensity measurements were performed with ImageJ. The region of interest was selected outlining the periphery of the kidney tubule, and background fluorescence was set at the same intensity as the internal lumen and was subtracted from the total fluorescence intensity. Statistical analysis was performed using the t-test with Prism software. Bright field images of whole embryos were taken on a Zeiss SteREO Lumar V.12 stereomicroscope.
Block face scanning electron microscopy
Serial block face scanning electron microscopy was performed according to (32). Images were analyzed using ImageJ. Endocytic and lysosomal compartments were defined by morphology. Vacuolar endosomes are oval or spherical membrane-enclosed compartments of a diameter >500 nm, with an electron-sparse lumen that contains varying degrees of granular material. Lysosomes are electron-dense oval or spherical membrane-enclosed compartments of a diameter >500 nm.
Acknowledgements
We would like to thank our colleagues for generously providing reagents as noted before. We are grateful to Dr Adam Hurlstone and Professor Rachel Lennon, both at the University of Manchester, for useful discussions and for critically reading the manuscript. We would also like to thank the Faculty Bioimaging for help with confocal microscopy and the staff in the EM Core Facility (RRID:SCR_021147) for their assistance.
Conflict of Interest statement. The authors report no competing interests relevant to this study.
Funding
This work was supported by research grants from the Lowe Syndrome Trust (ML/MU/DEC07, NoMU/ML/1010, ML/MU/2012).
