The Phosphoinositide Kinase PIKfyve Is Vital in Early Embryonic Development

Abstract

Gene mutations in the phosphoinositide-metabolizing enzymes are linked to various human diseases. In mammals, PIKfyve synthesizes PtdIns(3,5)P₂ and PtdIns5P lipids that regulate endosomal trafficking and responses to extracellular stimuli. The consequence of pikfyve gene ablation in mammals is unknown. To clarify the importance of PIKfyve and PIKfyve lipid products, in this study, we have characterized the first mouse model with global deletion of the pikfyve gene using the Cre-loxP approach. We report that nearly all PIKfyve^KO/KO mutant embryos died before the 32–64-cell stage. Cultured fibroblasts derived from PIKfyve^flox/flox embryos and rendered pikfyve-null by Cre recombinase expression displayed severely reduced DNA synthesis, consistent with impaired cell division causing early embryo lethality. The heterozygous PIKfyve^WT/KO mice were born at the expected Mendelian ratio and developed into adulthood. PIKfyve^WT/KO mice were ostensibly normal by several other in vivo, ex vivo, and in vitro criteria despite the fact that their levels of the PIKfyve protein and in vitro enzymatic activity in cells and tissues were 50–55% lower than those of wild-type mice. Consistently, steady-state levels of the PIKfyve products PtdIns(3,5)P₂ and PtdIns5P selectively decreased, but this reduction (35–40%) was 10–15% less than that expected based on PIKfyve protein reduction. The nonlinear decrease of the PIKfyve protein versus PIKfyve lipid products, the potential mechanism(s) discussed herein, may explain how one functional allele in PIKfyve^WT/KO mice is able to support the demands for PtdIns(3,5)P₂/PtdIns5P synthesis during life. Our data also shed light on the known human disorder linked to PIKFYVE mutations.

Introduction

Reversible phosphorylation by kinases and phosphatases at positions 3, 4, and/or 5 of the inositol head group in phosphatidylinositol (PtdIns)² generates a family of seven phosphoinositide species (1–6). They all are now found to function as versatile membrane-anchored signals that control diverse and essential cellular processes. Consequently, mutations in the genes encoding the phosphoinositide-metabolizing enzymes are associated with an increasing number of human diseases (1–6). The mammalian enzyme that makes PtdIns(3,5)P₂ from PtdIns3P and PtdIns5P from PtdIns is PIKfyve (7, 8). It is an evolutionarily conserved large protein of ∼230 kDa, a product of a single gene in the animal kingdom, whose function is required for proper performance of the endosomal system and certain signaling pathways (9–11). PIKfyve harbors a PtdIns3P-binding module, i.e. the FYVE finger domain that associates with the PtdIns3P-enriched endosomal membranes, assuring a rapid PIKfyve recruitment to this low abundance substrate of its catalytic activity (12). PIKfyve interacts physically or functionally with multiple partner proteins; the ones involved in PtdIns(3,5)P₂ homeostasis are the best studied (9). Thus, PIKfyve physically associates with the antagonistic enzyme, i.e. the PtdIns(3,5)P₂-specific phosphatase Sac3, which turns over PtdIns(3,5)P₂ to PtdIns3P (13, 14). This interaction is indirect and is promoted by ArPIKfyve, an adapter protein that associates with and stabilizes Sac3 in a heterooligomeric complex (15–17). PIKfyve binds efficiently only to the ArPIKfyve-Sac3 complex to form a triple heterooligomer in which Sac3 fulfills a dual function (17, 18). Not only does it turn over PtdIns(3,5)P₂ at the site of production, but together with ArPIKfyve, it secures robust PtdIns(3,5)P₂ synthesis through activating PIKfyve (17, 18). Clearly, the PIKfyve molecule displays a unique combination of features and binding partners, reflecting the critical requirement for its rapid recruitment at the proper endosomes and a dynamic balance between the substrate and the product of its enzymatic activity.

PIKfyve controls pleiotropic cell functions, the whole complexity of which is under intensive investigation (9, 10). In mammalian cells, the most prominent cellular phenotype under loss of the PtdIns(3,5)P₂-synthesizing activity of PIKfyve, achieved through PIKfyve dominant-negative mutation (9), pharmacological inhibition (19), or RNA interference (20, 21), is aberrant endomembrane swelling and vacuolation, which are progressively exacerbated with the dose/time of treatments. Data from cell and cell-free systems suggest that the vacuolation phenotype is a result of aberrant membrane trafficking pathways emanating from or traversing endosomes, which in turn are a consequence of compromised balance between endosome membrane fission and fusion (13, 20–25). Cellular studies in genetically modified lower eukaryotes, i.e. Saccharomyces cerevisiae (26–28), Caenorhabditis elegans (29), and Drosophila (30), have also demonstrated aberrant vacuolation of membrane compartments along the endolysosomal system upon inactivation of the respective PIKfyve orthologous gene. At the level of the whole organism, the complete loss of PIKfyve function is associated with developmental defects and embryonic lethality in C. elegans (29). Likewise, Drosophila mutants null for pikfyve die as late pupae or pharate adults in the process of hatching (30). Unlike with metazoans, deletion of PIKfyve (Fab1) in budding yeast is not lethal but causes aberrant nuclear division and growth defects (26). Genetically modified vertebrate models of PIKfyve are thus far unavailable.

Whereas PIKfyve is a single gene in mammals, evidence points to PtdIns(3,5)P₂ and PtdIns5P production, at least in vitro, by alternative lipid kinases capable of phosphorylating PtdIns5P and PtdIns at positions 3 and 5, respectively, or lipid phosphatases capable of hydrolyzing PtdIns(3,5)P₂ and PtdIns(4,5)P₂ to PtdIns5P (31–34). To address the requirement of PIKfyve for life in mammals and plausible compensatory production of PtdIns(3,5)P₂ and/or PtdIns5P steady-state levels, we generated a murine model with inactivation of the pikfyve gene using the Cre-loxP approach. We demonstrate here for the first time that whereas the PIKfyve-null embryos die during preimplantation development, the heterozygous mice appear without visible defects from birth to late adulthood despite the lack of translational compensation for the haploinsufficiency of the PIKfyve protein. MEFs derived from PIKfyve^flox/flox embryos and rendered null for pikfyve by Cre-induced excision of the loxP-flanked region, display the progressive vacuolation phenotype and arrested mitogenesis. Our study is the first genetic evidence for the essential and nonredundant function of PIKfyve in cell viability and early embryonic development of mammals.

EXPERIMENTAL PROCEDURES

Targeted Disruption of Mouse Pikfyve Gene

We used the Cre-loxP system to generate the targeting vector and the PIKfyve^WT/flox mice (performed by Ozgene Inc.). The essential features of the targeting vector are two loxP sequences flanking pikfyve exon 6 (encoding the FYVE finger domain peptide of PIKfyve) and two frt sites flanking the neomycin resistance selection cassette as detailed in the supplemental Experimental Procedures and supplemental Fig. 1). The excision of exon 6 by Cre recombinase leads to a frameshift and an early stop codon at residue 170, with the first 159 amino acids from the mouse PIKfyve sequence. After confirmation by restriction mapping and sequencing, the targeting vector was electroporated into C57BL/6 embryonic stem cells. Genomic DNA from neomycin-resistant cell clones was digested with suitable restriction endonucleases and then analyzed by Southern blotting with several DNA probes derived from different regions of the targeting construct. One of three positively targeted clones (I_3D8) was used for injection into the blastocysts of BALB/c (phenotypically “white”) mice. A male chimera mouse (with 90% black coat) was used for breeding with C57BL/6 females. Germ line transmission of the pikfyve flox allele was obtained and confirmed by Southern blot analysis of genomic DNA.

Crosses of PIKfyve^WT/fl to Cre-deletor Mice and Genotyping

All experiments were conducted according to institutional ethics guidelines for animal work. To eliminate the neomycin resistance selection cassette (Δneo), the PIKfyve^WT/fl mice were bred with homozygous FlpE-deletor C57BL/6 mice (Ozgene Inc.). To generate homozygous mice with global disruption in the pikfyve gene, the PIKfyve^WT/flΔneo mice were bred with global homozygous Cre-deletor C57BL/6 mice (Ozgene Inc.). Offspring were genotyped before weaning by tail DNA PCR using several primer pairs that are listed in the supplemental Experimental Procedures and in Fig. 1: (i) S3-A1, generating a 208-bp, a 378-bp, and no product for the wild-type, floxΔneo, and knock-out alleles, respectively; (ii) S2-A1, generating a 663-bp, an 852-bp and no product for the wild-type, floxΔneo, and knock-out alleles, respectively; (iii) S1-A1, generating products of 1063, 1308, and 405 bp for the wild-type, floxΔneo, and knock-out alleles, respectively; and (iv) S0-A1, generating products of 1232, 1476, and 574 bp for the wild-type, floxΔneo, and knock-out alleles, respectively. The presence of the Cre or FlpE recombinase alleles were detected with primer pairs specified in the supplemental Experimental Procedures to generate 373- and 403-bp products from the Cre and FlpE genes, respectively. The PCR conditions are detailed in the supplemental Experimental Procedures.

FIGURE 1.

Targeted disruption of mouse pikfyve gene. A, schematic representation of the strategy used for the conditional knock-out of the pikfyve gene by excision of exon 6. Shown is the locus between exons 5 and 7. Indicated are the sites of the PCR-specific primers S0–S3 and A1 used for genotyping. B, representative Southern blots of genomic DNA derived from mice with the indicated genotypes and digested with EcoRV. Membranes were probed with a genomic probe called “en probe” (indicated). C, representative genotyping of mice and embryos by PCR using the specific primer pairs indicated in A. The expected fragment sizes for the wild-type, KO, and floxΔneo alleles are indicated in parentheses. templ, template.

Embryo or Blastocyst Isolation and Genotyping

Noon of the day on which a vaginal plug was detected was considered day E0.5. Embryos E7.5–E17.5 resulted from natural mating. Following dissection and removal of the extraembryonic tissues, embryos were genotyped with one-round of PCR with primer pairs and conditions indicated in the supplemental Experimental Procedures. Embryos at morula-early blastocyst stage (E3.5) were generated by hormonal stimulation as described previously (35). Isolated E3.5 embryos were individually micrographed to analyze morphologic phenotype and then genotyped by nested PCR (supplemental Experimental Procedures).

Antibodies, Inhibitors, and MEFs

Polyclonal N-terminal (R7069) or C-terminal anti-PIKfyve (R7096), anti-Sac3, and anti-ArPIKfyve (WS047) antibodies were characterized previously (7, 13, 16, 18). Monoclonal anti-α-tubulin and polyclonal anti-actin antibodies were from Sigma. Polyclonal anti-GFP (Ab290) and anti-EEA1 (N-19) antibodies were from AbCam and Santa Cruz Biotechnology, respectively. Monoclonal anti-BrdU was from BD Pharmingen. The YM201636 inhibitor was from Symansis. For the isolation of MEFs, the head, limbs, and viscera were removed from mouse E12.5 embryo. The remaining tissue was minced and digested for 30 min at 37 °C with 0.25% trypsin, EDTA as described (36). Cells (passage 0) were collected and grown in DMEM supplemented with 10% FBS and 1% penicillin-streptomycin. Experiments performed herein were with MEFs up to passage 3.

MEF Transfection and Microscopy

MEF transfection with cDNA constructs encoding a pEGFP-2xFYVE domain derived from the PIKfyve sequence or pCAG-GFP-Cre-recombinase (a gift by Dr. Connie Cepko; Ref. 37) was performed by electroporation as described elsewhere (14, 22). Cells were then reseeded on coverslips placed in 35-mm dishes. Twenty-four hours post-transfection, MEFs were fixed and processed for immunofluorescence microscopy with anti-EEA1 and Cy3-conjugated rabbit anti-goat IgG as specified elsewhere (18). Texas Red-dextran endocytosis (10 min) was followed as detailed previously (23). Coverslips were mounted on slides and observed on a motorized inverted confocal microscope (Olympus IX81) using a 40× water immersion objective. GFP and red signals were captured by standard green and red fluorescence filters. Images were taken by a Hamamatsu Orca 12-bit digital charge-coupled device camera. Cre-expressing MEFs were observed for induction and progression of the vacuolation phenotype with a Nikon Eclipse TE 200 inverted fluorescence microscope using a Hoffman Modulation Contrast System with 40× objective and a standard green fluorescence filter as described previously (22). Images were captured with a SPOT RT Slider charge-coupled device camera (Diagnostic Instruments).

DNA Synthesis

Thirty-six hours post-transfection, Cre-transfected MEFs were serum-deprived for 36 h. Mitogenesis was then stimulated by the addition FBS (20%) for 16 h. BrdU (Sigma) was added at 20 μm during the last 6 h as described elsewhere (22). MEFs were then washed and fixed in formaldehyde. DNA was denatured with 0.07 n NaOH for 8 min. Cells were then washed in PBS, permeabilized, and incubated with a monoclonal anti-BrdU antibody followed by Alexa Fluor 568-conjugated anti-mouse secondary antibody. Cells were observed on a Nikon Eclipse TE 200 fluorescence microscope with the 40× Hoffman objective and the standard green and red channels for GFP and Alexa Fluor 568, respectively. To quantify the results, >50 Cre-GFP-expressing MEFs per condition were counted, and the percentage of nuclei incorporating BrdU was calculated.

myo-[2-³H]Inositol Labeling, Lipid Extraction, and HPLC

MEFs were labeled with myo-[2-³H]inositol (PerkinElmer Life Sciences) following our protocols for adipocytes (14) with a slight modification (38). Briefly, cells (35-mm dishes) maintained for 22 h in glucose- and inositol-free DMEM containing 10% dialyzed FBS, 5 mg/ml insulin, 5 mg/ml transferrin, 2 mm pyruvate, 25 mm HEPES, pH 7.4, 100 units/ml penicillin, and 100 μg/ml streptomycin were labeled for 26 h with 25 μCi/ml myo-[2-³H]inositol in glucose- and inositol-free DMEM containing 5 mg/ml transferrin, 2 mm pyruvate, and 25 mm HEPES, pH 7.4. Lipids were extracted in the presence of EDTA and TBAS (5 mm each), deacylated, and analyzed by HPLC (Waters 5215) on a 5-μm Partisphere SAX column (Whatman; 4.6 mm × 250 mm) as detailed previously (7, 13, 14). [³²P]GroPIns5P, [³²P]GroPIns3P, [³²P]GroPIns(4,5)P₂, and [³²P]GroPIns(3,5)P₂ prepared by enzymatic synthesis with [γ-³²P]ATP as described elsewhere (7, 8, 13, 14) were co-injected as internal HPLC standards. Fractions were collected every 0.25 min and analyzed for ³H and ³²P radioactivity after addition of the scintillation mixture. Data evaluation and documentation were performed by Microsoft Excel. Individual peak radioactivity was quantified by area integration and presented as a percentage of the summed radioactivity from the [³H]GroPIns3P, -4P, -5P, -(3,5)P₂, and -(4,5)P₂ peaks (“total radioactivity”).

Immunoprecipitation, Immunoblotting, and in Vitro PIKfyve Activity Assay

MEF lysates were collected in RIPA containing 1× protease inhibitor mixture (RIPA⁺). The assays were then conducted as described previously (7, 8, 13, 14) using N- or C-terminal anti-PIKfyve, anti-Sac3, and anti-ArPIKfyve antibodies.

Other Assays, Quantitation, and Statistics

Protein concentration was determined by bicinchoninic acid protein assay. Protein levels were quantified from the intensity of the immunoblot bands by a laser scanner (Epson V700) and UN-SCAN-IT software (Silk Scientific). Lipid levels were quantified by radioactive counting of the scraped silica spots. Data are presented as mean ± S.E. Statistical analysis of variance was performed by t test for independent samples and a one-tail t test for paired samples. Ninety-five percent confidence intervals were calculated as mean ± 1.96 × S.E. p < 0.05 was considered significant.

RESULTS

Disruption of pikfyve Gene and Generation of WT/flox Mice

To begin elucidating the role of PIKfyve at the level of the whole organism, we generated a mouse model with targeted disruption of the pikfyve gene using the Cre-loxP approach. The targeting vector was constructed based on sequence information (ENSMUSG00000025949) from mouse genome databases. The pikfyve gene, located on chromosome 1, harbors 42 exons spread over ∼92 kbp with a translation start at exon 2. We designed the targeting vector with two loxP sites flanking exon 6 (residues 159–215) that encodes the evolutionarily conserved FYVE finger domain (residues 164–230) of PIKfyve (Fig. 1A and supplemental Fig. 1). Southern blot analysis confirmed the germ line transmission of the pikfyve flox allele in the WT/flox mice (Fig. 1B). The WT/flox mice were then crossed with C57BL/6J mice expressing FlpE recombinase to remove the neomycin resistance selection cassette. The offspring containing the pikfyve floxΔneo allele, confirmed by Southern blot analysis, was bred with C57BL/6J mice expressing Cre recombinase to establish a heterozygous line of mice with a single knock-out allele (PIKfyve^WT/KO; Fig. 1, A and B).

In the presence of the Cre recombinase, exon 6 of the pikfyve flox allele is deleted, causing a frameshift mutation at the beginning of exon 7 and a translation stop at residue 171. Only the first 159 amino acids are from the PIKfyve sequence (supplemental Fig. 1). To test whether this N-terminal PIKfyve fragment is expressed, we performed immunoprecipitation and Western blotting with anti-PIKfyve N-terminal antibodies, detecting epitopes in the first 100 amino acids. No specific immunoreaction with a band within the 15–25-kDa range was observed (not shown), indicating that the partial PIKfyve protein was either not expressed at all, expressed but rapidly degraded, or expressed at a level below the sensitivity of the detection method. It should be emphasized, however, that, even if expressed at undetectable levels, a putative dominant-negative effect of such a fragment is highly unlikely because it lacks all evolutionarily conserved functional domains in PIKfyve, i.e. the FYVE finger, DEP, Cpn60_TCP1, CHK homology, and phosphoinositide kinase domains (supplemental Fig. 1).

Homozygous pikfyve Gene Deletion Is Lethal at Preimplantation Embryo Stage

To generate homozygous knock-out mice, we intercrossed the PIKfyve^WT/KO mice and genotyped the progeny before weaning using several sets of PCR primers (Fig. 1C). The litter size born from this intercross was 6.2 on average in 24 litters, which was lower than the average for control mice in a C57BL/6J background (8.6 in seven litters). Intriguingly, genotyping of more than 200 progeny before weaning did not detect any PIKfyve^KO/KO mice (Fig. 1C and Table 1). Likewise, of the 11 postnatal day 1–2 neonatal progeny assayed, none were the KO/KO genotype (Table 1), demonstrating embryonic lethality of systemic ablation of the pikfyve gene.

TABLE 1.

Genotype of progeny from PIKfyve^WT/KO intercross

Developmental stagesa	No. and percentage (in parentheses) of the given genotype
	Wild type	Heterozygote	Null
Postnatal
P17–20 (n = 210)	71 (34)	139 (66)	0
P1–2 (n = 11)	4 (35)	7 (65)	0
Embryonic
Postimplantation
E17.5 (n = 7)	5 (71)	2 (29)	0
E12.5 (n = 26)	13 (50)	13 (50)	0
E11.5 (n = 22)	6 (27)	16 (73)	0
E9.5 (n = 10)	3 (30)	7 (70)	0
E7.5 (n = 35)	8 (23)	27 (77)	0
Preimplantation
E3.5 (n = 58b)	17 (29)	37 (64)	2 (3.5)

^a Mice and embryos were genotyped by one-round PCR as indicated in Fig. 1C. Blastocysts were genotyped by nested PCR with primer pairs depicted in Fig. 1A.

^b Genotype of two embryos was not confirmed because of low DNA recovery.

To determine the time of embryonic death, we time-mated the heterozygous mice and removed embryos from maternal horns at different times of gestation. Genotyping of 90 embryos from late embryonic day E17.5 to as early as E7.5 revealed no PIKfyve^KO/KO at these postimplantation stages (Table 1). Increased embryo resorption in mothers' uterine horns was also not apparent, suggesting that the mutant embryos died during early embryogenesis prior to implantation. This hypothesis was confirmed by genotype/phenotype analyses of embryos at E3.5 (32–64 cells). From the total of 58 E3.5 embryos dissected from five hormone-induced mothers, we genotyped only two homozygous PIKfyve^KO/KO, both from a single mother (Table 1). An additional embryo from another mother, whose genotype was not confirmed because of insufficient DNA recovery, appeared irregular in size and morphology and was deemed to be PIKfyve^KO/KO. Morphologic analyses by phase-contrast microscopy indicated that the PIKfyve^KO/KO preimplantation embryos were clearly distinguishable from the wild-type and heterozygous embryos (Fig. 2, A and B). They were morbid with increased cellular fragmentation and reduced cell number in both inner cell mass and trophectoderm. In addition, the embryos had either failed to cavitate or not maintained the blastocoel if they had cavitated. These are hallmarks of early preimplantation lethality in other null mutant lethals (35, 39). Intriguingly, close inspection of these embryos revealed an aberrant vacuolation phenotype within the cell mass in both the genotyped and the predicted PIKfyve^KO/KO embryo (Fig. 2A), a defect routinely used as a sensitive morphological indicator of an ablated PtdIns(3,5)P₂ intracellular pool (9, 19, 20). Together, these data indicate a significantly lower proportion of the PIKfyve^KO/KO genotype than that predicted by Mendelian inheritance, consistent with the notion that the vast majority of the PIKfyve^KO/KO mutants died at the preimplantation period prior to the 32–64-cell stage.

FIGURE 2.

Preimplantation PIKfyve^KO/KO embryos are abnormal. A and B, embryos from the PIKfyve^WT/KO intercrosses were collected at the E3.5 stage, observed by phase-contrast, and then genotyped by nested PCR. Arrows in A denote the presence of aberrant vacuoles in the PIKfyve^KO/KO embryos. Also apparent is the decreased cell number in both the inner cell mass and trophectoderm.

Arrested DNA Synthesis in MEFs Rendered pikfyve-null by Cre Expression

The observation of only two or three PIKfyve^KO/KO embryos, instead of the predicted 14, at the 32–64-cell stage (Table 1) suggests that PIKfyve function is likely required as early as the first rounds of divisions in the fertilized egg. To address this prediction, we sought to generate pikfyve-null MEFs by Cre-induced excision of exon 6 from the pikfyve flox alleles in PIKfyve^flox/flox cells (see Fig. 1). For this purpose, MEFs isolated from E12.5 flox/flox or wild-type embryos were transfected with pCAG-Cre-GFP, a construct enabling expression of active Cre recombinase in the nucleus (37). Successful excision of the pikfyve flox alleles was morphologically monitored by the formation of aberrant cell vacuoles (9, 19, 20) as a sensitive functional indicator of localized PtdIns(3,5)P₂ reduction, PIKfyve protein depletion, and hence pikfyve gene deletion. Concordantly, under these conditions, we observed progressively exacerbating cytoplasmic vacuolation 72–120 h post-transfection only in the Cre-positive MEFs but not in the Cre-negative flox/flox MEFs or in the Cre-positive wild-type MEFs (Fig. 3A). The progression of the aberrant phenotype correlated well with the expected PIKfyve protein depletion, which, based on the half-life of the endogenous PIKfyve protein (t½ = 32 h; Ref. 15), was calculated to decrease by 75% (after 72 h) and 94% (after 120 h). Under these conditions, we measured basal and serum-induced DNA synthesis by indirect immunofluorescence microscopy of BrdU nuclear incorporation. As seen from the quantified data, presented in Fig. 3B, the Cre-positive flox/flox MEFs displayed a profound arrest in BrdU incorporation. In contrast, the Cre-negative flox/flox MEFs incorporated BrdU at the level of the wild-type MEFs with or without Cre expression (data not shown). Together, these data indicate markedly arrested DNA synthesis in MEFs null or nearly null for pikfyve. However, whether this defect is secondary to the aberrant vacuolar morphology induced by PtdIns(3,5)P₂ depletion under pikfyve gene deletion could not be ruled out.

FIGURE 3.

Arrested DNA synthesis in flox/flox MEFs rendered pikfyve-null by expression of Cre recombinase. A and B, cultured MEFs (100-mm dish) isolated from E12.5 PIKfyve^WT/WT and PIKfyve^flox/flox embryos were transfected by electroporation with the Cre recombinase-GFP cDNA and then seeded on coverslips. A, 72 h post-transfection, cells were observed by a 40× Hoffman objective. Shown are typical phase-contrast images of the defective vacuolar phenotype due to Cre recombinase-induced pikfyve gene deletion (panels a–c) seen in 96 ± 4% (mean ± S.E.) of Cre-transfected flox/flox MEFs. Panels d–f illustrate the normal morphology that was observed in nearly all of the Cre-transfected wild-type cells. Predominant Cre expression is seen in the nucleus (panels c and d). B, 36 h post-transfection, cells were serum-deprived for 36 h and then stimulated for 18 h in the presence or absence of FBS (20%). BrdU was included during the last 6 h of incubation, and its cell incorporation was detected by indirect immunofluorescence with an anti-BrdU and Alexa Fluor 568-conjugated anti-mouse IgG as primary and secondary antibodies, respectively. GFP/BrdU fluorescence-positive MEFs from three experiments were counted, and the mitogen response is presented as a percentage of the total GFP-Cre-transfected cells. In Cre-expressing flox/flox MEFs, serum-induced BrdU incorporation was practically not apparent in contrast to the Cre-expressing wild-type cells, which readily incorporated BrdU under these conditions. Error bars, ± S.E.

No Ostensible Defects in PIKfyve^WT/KO Mice, Embryos, and Endosomal Trafficking Despite 2-Fold Decreased PIKfyve Protein or Activity

To characterize the heterozygous genotype in more detail, we examined the progeny of the PIKfyve^WT/KO intercrosses. Consistent with PIKfyve^KO/KO embryonic lethality, at weaning, the heterozygous PIKfyve^WT/KO versus wild-type mice were at a 2:1 ratio as expected by Mendelian inheritance (Table 1). The PIKfyve^WT/KO embryos from either preimplantation (E3.5) or postimplantation development periods (E7.5 through E17.5) were indistinguishable from the wild-type littermates with respect to macro- or microscopical appearance, weight, size, and embryo-to-placenta ratios (Figs. 2B and 4A and data not shown). The mature PIKfyve^WT/KO mice were fertile and showed no significant difference versus wild-type littermates with regard to body weight, organ morphology, and organ size when fed with mouse chow containing ≥11% fat (Fig. 4B and data not shown). Together, these data indicate that PIKfyve expression from a single wild-type allele is sufficient to support embryo development, survival, and maturation into adulthood.

FIGURE 4.

PIKfyve^WT/KO embryos, mice, and endosomal system are similar to wild type. A, embryos from the PIKfyve^WT/KO intercrosses were collected at E11.5, photographed, and genotyped. B, body weights of male PIKfyve^WT/KO and wild-type littermates born from PIKfyve^WT/KO intercrosses fed on mouse chow (11% fat); error bars, ± S.E. C, E12.5 embryos from the PIKfyve^WT/KO intercrosses were dissected. MEFs were then isolated and cultured. MEFs collected from a confluent 100-mm dish were transfected by electroporation with the pEGFP-2xFYVE cDNA for staining the PtdIns3P-enriched endosomes. Cells were then reseeded and processed 24 h post-transfection by confocal microscopy with anti-EEA1 antibodies. The morphologic appearance of the endosomal system and the extent of colocalization (insets in the merged images c and f) between EEA1- (panels b and e) and 2xFYVE-associated fluorescence (panels a and d) in PIKfyve^WT/WT and PIKfyve^WT/KO MEFs are similar. Bar, 10 μm.

Lack of significant defects associated with the PIKfyve^WT/KO genotype was paralleled at the intracellular level as revealed by confocal microscopy inspection of the integrity and performance of the endosomal system in cultured MEFs. Thus, eGFP-2xFYVE, a probe comprising two tandem FYVE domains of PIKfyve that binds to PtdIns3P-enriched endosomal membranes with high affinity (12, 24), was found to similarly colocalize with internalized dextran (10 min) or with the early endosomal marker EEA1 in both PIKfyve^WT/KO and PIKfyve^WT/WT MEFs (Fig. 4C and data not shown).

To assess whether the absence of apparent defects associated with the PIKfyve^WT/KO genotype, observed at the level of the cell, embryo, and whole organism, is a result of presumed compensation for the PIKfyve protein and/or activity, we conducted experiments in several directions. First, we performed Western blotting to examine the PIKfyve protein levels in MEFs and in various tissues obtained from E12.5 embryos and adult mice, respectively (Fig. 5A). We consistently observed that the PIKfyve protein levels in the PIKfyve^WT/KO genotype were roughly one-half of that determined in the respective wild-type littermates as quantified by densitometry (46 ± 6%). Second, compensatory up-regulation of the PIKfyve enzymatic activity under reduced protein levels was also ruled out by our measurements of the PIKfyve lipid kinase activity in vitro. Thus, corroborating the results with the protein amounts, the TLC resolution of the PIKfyve enzymatic reactions with MEF lysates revealed also a 2-fold reduction of both PtdIns(3,5)P₂ and PtdIns5P products in PIKfyve^+/− versus PIKfyve^+/+ MEFs (Fig. 5, B and C). 2-fold decreases of the in vitro generated PIKfyve lipid products in the PIKfyve^+/− mice were also observed in various tissues, such as brain (Fig. 5C), white fat, brown fat, and spleen (data not shown). Finally, consistent with the reduced levels in PIKfyve protein/activity in the PIKfyve^WT/KO versus PIKfyve^WT/WT cells and tissues, we found that PIKfyve^WT/KO MEFs readily formed aberrant cell vacuoles (a sensitive functional measure of localized PtdIns(3,5)P₂ reduction; see Refs. 9 and 19) by low doses of the PIKfyve inhibitor YM201636. Thus, acute treatment of the PIKfyve^WT/KO MEFs with the YM201636 compound at 160 nm, a concentration that is 2.5-fold below the A₅₀ value in MEFs (400 nm; Ref. 19), induced the characteristic formation of cytoplasmic vacuoles (Fig. 5D, panels b and d). In contrast, the morphology of PIKfyve^WT/WT MEFs remained intact under these conditions (Fig. 5D, panels a and c). Together, these biochemical and morphological observations indicate a lack of translational or kinetic compensation for the decreased PIKfyve protein/in vitro activity under haploid deletion of the pikfyve gene.

FIGURE 5.

PIKfyve protein and in vitro activity are reduced by 2-fold in PIKfyve^WT/KO. A, RIPA⁺ lysates derived from cultured MEFs (E12.5) or from the indicated tissues dissected from PIKfyve^WT/KO mice and PIKfyve^WT/WT littermates were clarified by centrifugation. Equal amounts of tissue protein (200 μg) and MEF lysates (150 μg) from each genotype were examined by immunoblotting with anti-PIKfyve antibodies. Blots were reprobed with anti-tubulin or anti-actin for loading. Shown are chemiluminescence detections from representative experiments with two embryos/animals for each condition of two to three independent determinations. *, an unspecific band in the muscle samples. WB, Western blot. B and C, fresh RIPA⁺ lysates derived from MEFs or brain tissues dissected from PIKfyve^WT/KO and PIKfyve^WT/WT embryos (E12.5) and mice, respectively, underwent immunoprecipitation (IP) with anti-PIKfyve antibodies. Washed immunoprecipitates were subjected to in vitro lipid kinase activity assay. Shown are a representative autoradiogram of a TLC plate with separated radiolabeled lipids (MEFs) (B) and quantitation of the lipid kinase activities in MEFs and brain by radioactive counting of the scraped silica corresponding to PtdIns5P and PtdIns(3,5)P₂ (C). Presented are the combined decreases of both lipids as the extent of their individual decrease in PIKfyve^WT/KO versus wild type was identical. D, MEFs prepared from PIKfyve^WT/KO (panels b and d) and PIKfyve^WT/WT embryos (E12.5) (panels a and c) were seeded on coverslips. Cells were then treated for 2 h with YM201636 (0–160 nm) and observed live by the 40× objective of the Hoffman Modulation Contrast System for progression of a vacuolation phenotype. The phase-contrast images illustrate that only the PIKfyve^WT/KO MEFs exhibit the typical defective vacuolar phenotype under these conditions.

Non-linear Decreases in PIKfyve Lipid Products Versus PIKfyve Protein in PIKfyve^WT/KO

Reportedly, knock-out mice for ArPIKfyve, the PIKfyve regulator that activates PIKfyve, die perinatally, displaying a 50% decrease in steady-state levels of the PIKfyve lipid products PtdIns(3,5)P₂ and PtdIns5P as determined in myo-[2-³H]inositol-labeled fibroblasts (40). Thus, our observation of a 50% decrease in the PIKfyve protein and activity yet viable PIKfyve^WT/KO mice without noticeable defects raises the question about functional compensation at the level of intracellular PtdIns(3,5)P₂ and PtdIns5P production. Therefore, we next examined steady-state levels of PtdIns(3,5)P₂ and PtdIns5P under pikfyve haploid deletion by monitoring the phosphoinositide profiles by HPLC inositol head group analyses in MEFs subsequent to myo-[2-³H]inositol labeling (Fig. 6A). Consistent with the reduced levels in PIKfyve protein/activity under pikfyve haploid deletion, selective and significant changes were observed only in steady-state levels of the PIKfyve products PtdIns(3,5)P₂ and PtdIns5P and of the PtdIns(3,5)P₂ precursor, PtdIns3P (Fig. 6, A and B). PtdIns(4,5)P₂ and PtdIns4P steady-state levels remained practically identical between the mutant and wild-type MEFs (Fig. 6, A and B). PtdIns(3,4)P₂ and PtdIns(3,4,5)P₃ were not detected (Fig. 6A) as expected under the quiescent conditions. Intriguingly, whereas steady-state levels of PtdIns(3,5)P₂ and PtdIns5P in the PIKfyve^WT/KO MEFs were significantly decreased to 61 and 65% of the control levels, respectively (Fig. 6B), this reduction was to a lesser extent than that predicted by the observed 2-fold reduction in PIKfyve protein and in vitro activity. The deviations of PtdIns(3,5)P₂ and PtdIns5P levels from the theoretically expected 50% reduction based on the ∼50% lower PIKfyve protein in PIKfyve^WT/KO MEFs versus wild-type controls were statistically analyzed using 95% confidence intervals and found to be statistically significant (p < 0.05%; intervals: PtdIns(3,5)P₂, 54.7–67.8%; PtdIns5P, 59.1–69.9%). This non-linear correlation between reduced levels of the PIKfyve enzyme/activity versus intracellular lipid products is consistent with the notion that additional mechanisms might operate to partially mitigate the loss of lipid products under monoallelic expression of the PIKfyve protein.

FIGURE 6.

PIKfyve lipid products are decreased to lesser degree than PIKfyve protein in PIKfyve^WT/KO MEFs. MEFs isolated from E12.5 PIKfyve^WT/KO and PIKfyve^WT/WT embryos and seeded on 35-mm dishes were subjected to inositol and glucose starvation (22 h). Cells were then labeled with myo-[2-³H]inositol (26 h) as described under “Experimental Procedures.” Lipids were extracted, deacylated, and fractionated by HPLC. Presented are the radioactivity from the HPLC profiles of a representative experiment with both genotypes (A) and the quantitation (B) from four independent experiments calculated as a percentage of the total phosphoinositide radioactivity and presented as mean ± S.E. The total phosphoinositide (PI) radioactivity was obtained by summing the counts within the elution times corresponding to the indicated [³H]GroPIns peaks. The changes in PtdIns(3,5)P₂, PtdIns5P, and PtdIns3P in PIKfyve^WT/KO versus PIKfyve^WT/WT were statistically significant (*) with p = 0.0014, p = 0.0076, and p = 0.0488, respectively.

Reduced Formation of PIKfyve-ArPIKfyve-Sac3 (PAS) Regulatory Complexes in PIKfyve^WT/KO

We have recently revealed that PIKfyve associates with the ArPIKfyve scaffold and Sac3 phosphatase in a regulatory ternary assembly, called the PAS complex, that relays two opposing enzymatic activities, synthesis and breakdown of PtdIns(3,5)P₂ (17, 18). Considering a potential regulatory mechanism at the level of the Sac3 phosphatase as a means to counterbalance the loss of PtdIns(3,5)P₂ (and probably PtdIns5P) in the PIKfyve^WT/KO genotype, we next examined the total protein levels of Sac3 and ArPIKfyve as well as their relative amounts engaged in the association with PIKfyve. To this end, fresh lysates from cultured MEFs derived from PIKfyve^WT/WT or PIKfyve^WT/KO embryos were subjected to Western blotting directly or subsequent to immunoprecipitation with anti-PIKfyve antibodies. As illustrated in Fig. 7, in contrast to the 2-fold reduction in PIKfyve protein levels in PIKfyve^WT/KO embryos, those of Sac3 or ArPIKfyve remained unchanged. Importantly, however, despite unaltered protein levels, the amounts of both Sac3 and ArPIKfyve co-immunoprecipitated with PIKfyve were reduced by 50% in PIKfyve^WT/KO versus PIKfyve^WT/WT MEFs proportionally to the 50% decrease in immunoprecipitated PIKfyve (Fig. 7). These observations suggest that under pikfyve haploid deletion, the reduction in PIKfyve levels is associated with a commensurate decrease in the formed PAS complexes.

FIGURE 7.

Decreased PAS complexes in PIKfyve^WT/KO MEFs. Cultured MEFs isolated from E12.5 PIKfyve^WT/KO and PIKfyve^WT/WT embryos were solubilized in RIPA⁺. Clarified lysates underwent immunoprecipitation (IP) with anti-PIKfyve antibodies. After washes in the same buffer, immunoprecipitates were analyzed by SDS-PAGE and immunoblotting. Blots, cut horizontally above the 110-kDa standard, were probed with anti-PIKfyve (top part), anti-Sac3, and anti-ArPIKfyve antibodies (bottom part) with a stripping step in between. The top and bottom parts were probed with anti-IRAP (insulin-responsive aminopeptidase, ∼160 kDa) and anti-α-tubulin antibodies (∼55 kDa), respectively, for equal loading. Shown are chemiluminescence detections of a representative experiment of three independent determinations.

DISCUSSION

In the present study, we have characterized the first mammalian model with systemic deletion in the pikfyve gene by the Cre-loxP approach. Our key findings are that whereas nearly all homozygous pikfyve-null mice die as early as preimplantation embryonic development, the heterozygous mice are without ostensible defects and develop into late adulthood. Furthermore, we document an arrest of DNA synthesis in PIKfyve^flox/flox MEFs rendered null for the pikfyve gene by Cre-induced excision of the loxP-flanked region, suggesting that the early lethality of the pikfyve null embryo is a result of defective cell division. Thus, our data provide the first genetic evidence for the essential and nonredundant function of PIKfyve in preimplantation embryonic development in mice.

Our searches of the literature and publicly available databases reveal expression of PIKfyve mRNAs/expressed sequence tags in the mouse oocyte and fertilized egg (41–43). Thus, serial analysis of mouse gene expression by Affymetrix microarray detects PIKfyve transcripts as early as the eight-cell embryo. Likewise, mouse UniGene dbEST libraries from various stages of embryogenesis reveal the presence of PIKfyve cDNA sequences in preimplantation development at two-cell embryo (library 862), 16-cell embryo (library 1532), and other preimplantation stages (libraries 13911, 850, and 10026). These observations, together with the herein documented only 3.5–5.2% homozygous knock-outs at the E3.5 preimplantation embryo stage (instead of 25% by the Mendelian prediction; Table 1) indicate the requirement for PIKfyve in an early stage of preimplantation embryo development in mice. This conclusion is corroborated by recent genetic studies in Arabidopsis thaliana that carry two orthologous forms of PIKfyve. Pollen homozygous for mutations in both genes is unviable as early as the first mitotic division of development (44). Putative null mutants in C. elegans and Drosophila die during different embryonic and late pupal stages, respectively (29, 30), but a requirement for earlier stages has not been examined. It is possible that maternally provided mRNA or protein is sufficient to drive development as is common for other genes that are required throughout development.

The herein observed mouse embryo lethality at such an early stage is likely related to defects in early cell division of the developing embryo. In the present study, we support this contention by findings for severely arrested DNA synthesis in pikfyve-null MEFs (Fig. 3). It is highly likely that the maternal PIKfyve protein, whose half-life is quite long (t½ = 32 h; Ref. 15), supports the early cell divisions prior to its total depletion, which could explain how the PIKfyve^KO/KO embryos survived through the initial divisions. On the other hand, variability in the maternal component among mothers may underlie the irregularity of our PIKfyve^KO/KO detection since the two genotyped PIKfyve^KO/KO embryos at the E3.5 stage were from a single mother. Our observation for the key role of PIKfyve in embryo cell division is in agreement with data in yeast where PIKfyve/Fab1 deletion was found to impair nuclear division resulting in aploid and binucleate cells (26). However, it should be taken into consideration that impaired cell proliferation could be secondary to defects in endomembrane morphology, which are inevitable under perturbed PtdIns(3,5)P₂ synthesis (9). Whereas we provide evidence for arrested DNA synthesis in pikfyve-null MEFs, a role for PIKfyve in other steps of the cell cycle of the developing embryo cannot be ruled out. Consistent with this notion are recent findings implicating the PIKfyve enzymatic substrate PtdIns3P and the PtdIns3P effector proteins containing a FYVE finger, including PIKfyve, in cytokinesis (45).

Of exceptional importance in the present study is our observation that under the lack of translational compensation for the haploid pikfyve gene deletion, evidenced herein by the 2-fold reduction of PIKfyve protein levels versus wild type (Fig. 5A), the heterozygous mice developed normally (Fig. 4). We relate this ostensible normality of the haploid deletion to certain functional compensation indicated herein by the nonlinear decrease in the PIKfyve protein/activity versus intracellular PtdIns(3,5)P₂/PtdIns5P production. Thus, we found that 50% of the PIKfyve enzyme in MEFs suffices to maintain as high as ∼61–65% of the wild-type steady-state levels of PtdIns(3,5)P₂/PtdIns5P (Figs. 5 and 6). Although this functional compensation could be only at a rate of 10–15%, it may have a vital consequence as inferred by genetic studies with arpikfyve (vac14)-null mice. Reportedly, these mice die perinatally, displaying 50% of the wild-type PtdIns(3,5)P₂ and PtdIns5P levels (40), whereas our PIKfyve^WT/KO heterozygous mice are viable to late adulthood at ∼60–65% of the wild-type lipid levels. The documented HPLC profiles in the steady-state levels of the phosphoinositide species in haploid versus wild-type conditions, whereby only the PIKfyve substrate PtdIns3P and products PtdIns3,5P₂/PtdIns5P are altered (Fig. 6), are inconsistent with PIKfyve-unrelated phosphatases/kinases functionally compensating for the PtdIns(3,5)P₂/PtdIns5P decreases. Rather, these data suggest that the subtle compensation of the loss of PIKfyve products likely occurs through a mechanism related to PIKfyve per se and/or Sac3. Thus, PIKfyve might normally operate below its capacity; therefore, half of the enzyme produces above 50% of PtdIns(3,5)P₂/PtdIns5P. An additional and/or alternative mechanism could be related to a reduced presence of the Sac3 phosphatase at key membranes as suggested herein based on our findings for a ∼50% reduction of the formed PAS regulatory complexes in the PIKfyve^WT/KO MEFs (Fig. 7). Thus, we propose that decreased levels in localized PtdIns(3,5)P₂ under pikfyve haploid deletion, together with the lower amounts of the Sac3 phosphatase at these locations create unfavorable kinetic conditions for the Sac3-hydrolyzing activity, slowing the rate of PtdIns(3,5)P₂ turnover. It should be pointed out, however, that whereas the PIKfyve^WT/KO mice showed no gross abnormalities with regard to organ morphology, size, or growth rates, pathological alterations under certain challenges are likely to be uncovered. This notion is supported by our findings for slightly increased basal glucose levels under a fasted state and delayed glucose clearance upon insulin administration in male PIKfyve^WT/KO mice.³ Further work will clarify whether this trend for mild insulin resistance under pikfyve haploinsufficiency is sex- or age-dependent.

The contribution of PIKFYVE mutations to human disease is largely elusive. One disorder linked to mutations in PIKFYVE is François-Neetens fleck corneal dystrophy, an autosomal dominant syndrome clinically manifested by the presence of enlarged/swollen vesicles (speckles) within keratocytes (46). CFD patients carry primarily protein-terminating nonsense or frameshift heterozygous mutations within the PIKFYVE gene, all localized within the PIKfyve protein region encompassing residues 705 and 1217 (46). Intriguingly, the pikfyve heterozygous mice do not display similar pathology (data not shown), indicating species variability in the outcome of the pikfyve heterozygosity and/or additional mechanism(s) underlying the CFD in humans. Thus, based on previous data for Sac3/ArPIKfyve binding regions in PIKfyve (18) and the current study, we speculate that the CFD pathogenic mechanism involves a compound reduction in PtdIns(3,5)P₂ levels due to haploinsufficiency of the PIKfyve catalytic activity and imbalanced elevation of the Sac3 phosphatase activity targeted to key endosomes by the PIKfyve partial proteins that retain some ability to bind Sac3/ArPIKfyve. Furthermore, a prepublication data release indicates PIKfyve somatic mutations in all of the examined five ovarian adenocarcinoma samples.⁴ This information, together with our observations for severely reduced DNA synthesis in pikfyve-null embryos (Fig. 3) or HEK293 cell lines expressing dominant-negative kinase-deficient PIKfyve (22) implicates PIKfyve, directly or indirectly, in cell proliferative capacity. Identification of PIKFYVE genetic abnormalities in other human cancers and diseases will further shed light as to which mutations could be tolerated and which are incompatible with life. In any case, the present study in the mouse demonstrates that whereas pikfyve-null embryos die preimplantationally, one functional allele in heterozygotes is able to support the demands for PtdIns(3,5)P₂ and PtdIns5P synthesis during life.