The mitochondrial transporter SLC25A25 links ciliary TRPP2 signaling and cellular metabolism

Cilia are organelles specialized in movement and signal transduction. The ciliary transient receptor potential ion channel polycystin-2 (TRPP2) controls elementary cilia-mediated physiological functions ranging from male fertility and kidney development to left–right patterning. However, the molecular components translating TRPP2 channel–mediated Ca2+ signals into respective physiological functions are unknown. Here, we show that the Ca2+-regulated mitochondrial ATP-Mg/Pi solute carrier 25 A 25 (SLC25A25) acts downstream of TRPP2 in an evolutionarily conserved metabolic signaling pathway. We identify SLC25A25 as an essential component in this cilia-dependent pathway using a genome-wide forward genetic screen in Drosophila melanogaster, followed by a targeted analysis of SLC25A25 function in zebrafish left–right patterning. Our data suggest that TRPP2 ion channels regulate mitochondrial SLC25A25 transporters via Ca2+ establishing an evolutionarily conserved molecular link between ciliary signaling and mitochondrial metabolism.


Introduction
Cilia are cellular appendages that coordinate several signaling pathways during development and tissue homeostasis [1]. Various signaling proteins, such as G protein-coupled receptors and transient receptor potential (TRP) channels, are compartmentalized to cilia to sense the cellular environment [2]. Yet little is known about how ciliary signals are propagated to control cellular functions.
TRP channels present in cilia have been proposed to regulate local Ca 2+ levels [3,4]. Ca 2+ signaling regulates many cellular processes, including mitochondrial function and cellular metabolism [5][6][7]. Hundreds of proteins have evolved to undergo conformational changes upon Ca 2+ binding to control cellular signaling [7,8]. At the core of these versatile signaling pathways are Ca 2+ -permeable ion channels and Ca 2+ -regulated effector proteins. However, there is little known about the functional connection between specific channels and corresponding effector proteins.
Transient receptor potential channel polycystin-2 (TRPP2) is a ciliary Ca 2+ -permeable nonselective cation channel [9,10]. Loss of TRPP2 causes severe phenotypes throughout the animal kingdom, including polycystic kidney disease in humans, left-right patterning defects in vertebrate model organisms, and male infertility in invertebrates [9,[11][12][13][14]. At the cellular level, TRPP2-mediated Ca 2+ signals have been proposed to regulate processes such as transcription, proliferation, and differentiation [15][16][17]. However, the molecular components connecting TRPP2 to specific physiological functions have remained elusive. This raises the fundamental question of how ciliary TRPP2 ion channel activity controls processes like left-right patterning and fertility-or more specifically, what are the Ca 2+ -regulated proteins that act downstream of TRPP2 to control cellular metabolism? We have addressed this issue with a forward genetic screen in D. melanogaster followed by a targeted analysis of left-right pattering in Danio rerio (Fig 1).

Results and discussion
A genome-wide screen in D. melanogaster identifies the Ca 2+ -regulated mitochondrial carrier SCaMC in TRPP2 signaling We have previously shown that ciliary TRPP2 (Amo) is required for male fertility in flies [12,13]. Amo-deficient males produce motile sperm that are transferred to the uterus but do not reach the female sperm storage organs, which is critical for reproductive success (Fig 2A  and S1 Movie). Based on this phenotype, we performed an unbiased forward genetic screen in flies to identify novel components of the TRPP2 signaling pathway in vivo (S1A Fig). Since we were interested in ciliary signaling rather than ciliary structure, we excluded mutations with defects in structure or motility of sperm. Out of a collection of 2,216 ethyl methanesulfonateinduced male sterile lines, we found that 404 produced mature sperm [18,19]. Among the latter set, 90 lines produced normal amounts of motile sperm but failed to be stored in the female sperm storage organs, similar to the amo 1 mutant phenotype (S1B-S1E Fig) [12,13]. Using deficiency mapping and exon sequencing, we identified several of the corresponding genes. We focused our attention on the line Z3-2147, since the mutated gene CG32103 contained Ca 2+ -binding EF-hands, suggesting a putative function downstream of TRPP2.
Similar to Amo-deficient males, homozygous Z3-2147 males were infertile due to a sperm storage defect (Fig 2B and S1D-S1G Fig). We found a mutation in CG32103 causing replacement of the conserved arginine 308 by a tryptophan in CG32103 (CG32103 R308W ) (Fig 2C and  2D). The deleterious impact of CG32103 R308W was validated by complementation experiments and full rescue of the male sterile phenotype with a transgene expressing wild-type CG32103 cDNA (Fig 2B, S2A and S2B Fig). This demonstrates that the missense mutation CG32103 R308W causes the Z3-2147 phenotype.
The function of Drosophila CG32103 is unknown [20]. The protein encoded by CG32103 belongs to the family of evolutionarily conserved short Ca 2+ -binding mitochondrial carriers   [21,22]. The specific activation of these carriers by Ca 2+ is based on conformational changes mediated by Ca 2+ binding to the N-terminal EF-hands [23,24]. To test whether Ca 2+ regulation of SCaMC (CG32103) is required in vivo, we performed rescue experiments comparing wild-type SCaMC and Ca 2+ bindingdeficient transgenes in D. melanogaster. In contrast to wild type, transgenes lacking the Nterminal EF-hands of SCaMC (SCaMC Δ1−268 ) or missense mutations abolishing Ca 2+ binding (SCaMC EF ) significantly impaired the rescue of the SCaMC 1 (CG32103 R308W ) phenotype, suggesting that SCaMC activity is regulated through Ca 2+ in vivo (Fig 2E and S2A-S2D Fig). The identical phenotypes of amo 1 and SCaMC 1 mutant flies in combination with the Ca 2+ regulation of SCaMC support a model in which both proteins act in a common pathway.

Solute carrier 25 A 25 (SLC25A25), the vertebrate ortholog of D. melanogaster SCaMC, is a Ca 2+ -regulated ATP transporter
To investigate the evolutionary conservation of SCaMC function, we tested whether its human homolog SLC25A25 could substitute for Drosophila SCaMC. Human SLC25A25 fully rescued the male infertility in SCaMC 1 flies (Fig 2F).
Drosophila SCaMC and SLC25A25 share similar transport and Ca 2+ -binding domains, but like SCaMC, the molecular function of human SLC25A25 has not been established. SLC25A25 is closely related to the Ca 2+ -regulated exchanger of cytosolic adenine nucleotides for phosphate (APC) from the mitochondrial matrix [21,22]. To test this hypothesis and determine the function and regulation of SLC25A25, we purified the transporter protein and characterized it in vitro ( Fig 3A). Thermostability assays were used to assess the effect of Ca 2+ on the apparent melting temperature (T m ) of SLC25A25 (Fig 3B). A biphasic unfolding profile was observed, indicating that SLC25A25 was in a mixed population consisting of a Ca 2+ -bound state (apparent T m = 49.5˚C) and a Ca 2+ -free state (apparent T m = 60.6˚C), with similar values to those observed for the paralog SLC25A24 [24]. After addition of Ca 2+ , the apparent T m shifted to a single peak (apparent T m = 50.4˚C), indicating that the mixed population had shifted to a homogeneous population of Ca 2+ -bound protein. These results indicate that SLC25A25 undergoes a Ca 2+ -dependent conformational change (Fig 3B). To test whether Ca 2+ regulates the transport activity and whether SLC25A25 transports Mg-ATP, purified transporter protein was reconstituted into proteoliposomes, and the uptake of radio-labeled [ 14 C]-Mg-ATP was monitored with and without added Ca 2+ . Upon addition of Ca 2+ , a significant increase in the uptake rate of [ 14 C]-Mg-ATP was observed, showing that the SLC25A25 transport activity was up-regulated by Ca 2+ (Fig 3C). Taken together, these experiments demonstrate that SLC25A25 is a Ca 2+ -regulated Mg-ATP transporter.

SLC25A25 is required for left-right organ patterning in zebrafish
We addressed the question whether SLC25A25 function is required in vertebrate TRPP2 signaling, using zebrafish as a model organism. In zebrafish, Trpp2 is required for left-right patterning (S4A-S4D Fig) [25][26][27][28][29]. Rotating motile cilia in the left-right organizer generate a fluid flow-mediated directional signal that is sensed by primary cilia [30,31] required to translate this signal into correct left-right patterning [9,11,32]. Here, we show that loss of Slc25a25b (slc25a25b) phenocopied loss of Trpp2 (pkd2), resulting in randomization of left-right asymmetry (

SLC25A25 regulates left-right patterning upstream of Nodal
Asymmetric expression of members of the Nodal cascade is required for left-right patterning [34,35]. TRPP2-mediated Ca 2+ signaling has been reported to control this cascade through an unknown mechanism [11,27,32,36]. We, therefore, tested whether SLC25A25 represents a missing link connecting asymmetric TRPP2 activity to Nodal. In support of this hypothesis, knockdown of slc25a25b resulted in randomization of southpaw (Nodal) expression in the lateral plate mesoderm (Fig 4D-4G   Nodal cascade, providing a molecular link between TRPP2-mediated Ca 2+ signals and asymmetric gene expression.

TRPP2 and SLC25A25 are compartmentalized in microdomains
Mechanistically, our data support a model in which TRPP2-mediated Ca 2+ signals regulate SLC25A25 activity via its four EF-hand domains (Figs 2E and 3, S3A Fig). Ca 2+ is a In contrast to control fish, in which southpaw expression was largely restricted to the left side, slc25a25b-morphant fish showed randomized southpaw expression ( Ã P = 0.00008). Numbers of embryos are indicated above bars. L = left; S = symmetric; R = right. This denomination for heart looping is equivalent to wt d-loop, symmetric no-loop, and reversed, sinistral s-loop [33]. For numerical values, see S1 Data. cmlc2, cardiac myosin light chain 2; hpf, hours post fertilization; wt, wild type. TRPP2 and SLC25A25-Cilia and metabolism promiscuous second messenger regulating many different cellular processes. Specificity of Ca 2+ signals is achieved through tight spatial coupling of the local Ca 2+ source and Ca 2+dependent effector proteins in microdomains [8]. Ca 2+ -permeable TRPP2 channels are found in primary cilia and in the endoplasmic reticulum (ER), whereas SLC25A25 is expressed in the inner mitochondrial membrane ( Fig 5A) [37,38]. Close proximity of these cellular compartments is required for functional coupling of TRPP2 and SLC25A25 in a signaling microdomain. We found that mitochondria cluster at the ciliary base in KV ( Fig 5B and S9A Fig) [39]. Hence, we suggest that TRPP2-mediated Ca 2+ influx at the ciliary base may control SLC25A25 activity. Furthermore, we observed that ER-resident TRPP2 and mitochondrial SLC25A25 showed tight spatial coupling in epithelial cells (Fig 5C and S9B Fig). This is in line with the close contacts between the ER and mitochondria at mitochondria-associated ER membranes [40]. In the ER, which is the main intracellular Ca 2+ store, TRPP2 contributes to Ca 2+ release and may amplify ciliary Ca 2+ signals [3,41]. Ca 2+ release from the ER results in mitochondrial Ca 2+ transients that regulate cellular metabolism [40]. Taken together, the proximity of TRPP2 and SLC25A25 supports the functional interaction of both proteins. In D. melanogaster sperm, TRPP2 is exclusively expressed in the ciliary membrane and SCaMC in mitochondria. Therefore, the critical cellular location for the TRPP2/SCaMC interaction is the cilia/mitochondria interphase (S1F-S1G, S3B and S3C Figs). The additional ER expression of TRPP2 in vertebrate cells, on the other hand, precludes an unambiguous localization of the functional coupling of SLC25A25 and TRPP2 to an ER/mitochondria or cilia/mitochondria microdomain.

Modulation of mitochondrial metabolism by SLC25A25 and TRPP2
The evolutionary conservation of SLC25A25 in TRPP2-dependent signaling raises the question how a mitochondrial carrier controls distinct cellular outcomes. An important prerequisite to understanding the physiological role of a mitochondrial carrier protein is the clarification of its impact on cellular metabolism [42]. To elucidate the role of SLC25A25mediated transport in cellular metabolism, we performed metabolomic analyses and measured mitochondrial oxidative metabolism. We generated SLC25A25-and TRPP2-deficient epithelial cells using genome editing (Slc25a25 −/− and Pkd2 −/− ) (S10A-S10E Fig) [43]. We analyzed Ca 2+ signaling and mitochondrial oxidative metabolism in wild-type and knockout cells (S11A and S11B Fig). SLC25A25-deficient cells showed a decrease in overall cellular respiration that did not affect growth rate or survival (S11C Fig) [44]. To identify SLC25A25-dependent metabolites, we performed broad-coverage discovery metabolomics encompassing the entire metabolome (S2 Data) [45]. In SLC25A25-deficient cells, 42 metabolites were significantly altered compared to wild-type cells ( Fig 6A).
The model that TRPP2 and SLC25A25 operate in a common pathway predicts that loss of TRPP2 channel activity or loss of SLC25A25 carrier function results in concordant changes in specific metabolites. We therefore analyzed the metabolite profiles of TRPP2-deficient cells ( Fig 6B and S2 Data). We found 4 metabolites that were significantly increased and 7 metabolites that were significantly decreased in both TRPP2-and SLC25A25-deficient cells ( Fig 6C  and 6D and S1 Table). Furthermore, ATP concentrations were reduced in SLC25A25-and TRPP2-deficient cells ( Fig 6E). This observation is in line with the reported role of the SCaMC/APC carrier family in modulating the adenine nucleotide pool in the mitochondrial matrix in response to changes in energy demands [22,46,47].
Emerging evidence suggests that metabolite fluctuations regulate cellular signal transduction [48]. Several solute carriers have been implicated in metabolic signaling [42,49]. Consequently, we propose that activation of the TRPP2-SLC25A25 pathway induces downstream metabolic signals. In our experiments, we found that significantly changed metabolites are associated with amino acid metabolism, including branched-chain amino acids (BCAAs) in mitochondria (S1 Table). Several of the metabolites we discovered are involved in cellular signal transduction, but their role in Drosophila sperm movement and zebrafish left-right patterning has not been explored yet. Notably, ATP has been implicated in both processes. Sperm motility is an ATP-dependent process. Directional sperm movement in Drosophila depends on TRPP2 [12,13]. Our data suggest that TRPP2-mediated Ca 2+ signals activate SLC25A25-dependent adenine nucleotide import, which leads to an increase in ATP output from the mitochondrion to increase cytosolic ATP, which drives sperm motility via dynein motor proteins. In left-right patterning, it has been proposed that ATP is released from the left-right organizer to promote the spreading of left-sided Nodal signaling to the lateral plate mesoderm via purinergic signaling [50]. A similar purinergic signaling mechanism has been shown for TRPP2-mediated ciliary signaling in epithelial cells [51]. Ultimately, TRPP2 activity in the left-right organizer causes asymmetric gene expression of the Nodal cascade [11,32]. This transcriptional regulation might be explained by the observed metabolic changes, since almost all chromatin-modifying enzymes utilize metabolites as cofactors to control gene expression [52]. However, the precise molecular mechanism linking SLC25A25-mediated metabolic fluctuations and asymmetric gene expression in the left-right organizer remains to be determined.

SLC25A25 is an evolutionarily conserved downstream effector of TRPP2 signaling linking ciliary signaling and cellular metabolism
Finally, to test whether SLC25A25 acts downstream of TRPP2 in vertebrate left-right patterning, we evaluated TRPP2-dependent SLC25A25 function in zebrafish (Fig 7A). Parallel knockdown of slc25a25b and pkd2 did not aggravate the left-right patterning defect compared to individual suppression of either transcript, further supporting a role of both proteins in a common pathway (S12 Fig). Consequently, we performed rescue experiments overexpressing Slc25a25b in the Trpp2-deficient background. Slc25a25b expression significantly alleviated the loss of TRPP2 phenotype, suggesting a role of SLC25A25 downstream of TRPP2 ( Fig 7B). TRPP2 is thought to be activated asymmetrically in the vertebrate left-right organizer, raising the question how symmetric expression of slc25a25b can rescue laterality in pkd2-morphant zebrafish (pkd2 MO ) [27,32]. In pkd2 morphants, expression of Slc25a25b may rescue laterality by amplifying residual asymmetric Trpp2 activity ( Fig 7B). If SLC25A25 and TRPP2 act in a common pathway, Slc25a25b should not rescue Trpp2-dependent left-right determination in a pkd2-null background (pkd2 −/− ) [25]. Indeed, expression of slc25a25b in pkd2-null zebrafish did not rescue left-right patterning, corroborating a role for SLC25A25 downstream of TRPP2 ( Fig 7C).
In summary, our results establish an evolutionarily conserved link between the ciliary ion channel TRPP2 and the mitochondrial transporter SLC25A25. The identification and molecular characterization of the Ca 2+ -regulated mitochondrial ATP carrier SLC25A25 downstream of TRPP2 provides insights into the molecular mechanisms of signal transduction in ciliadependent biological processes ranging from male fertility to vertebrate morphogenesis.

Ethics statement
All animal experiments were conducted according to international guidelines and the German law for the welfare of animals. Approval for animal studies was obtained from the regional authorities (Regierungspräsidium Freiburg, reference number G-16/89) [53].

Cell culture and transfection
HeLa and mIMCD3 cells were obtained from the American Type Culture Collection. Cells were cultivated as adherent monolayers in DMEM (Lonza) and DMEM F-12 (Lonza) media supplemented with 10% heat-inactivated fetal bovine serum (FBS; Biochrom). Cell lines were maintained in a humidified 10% CO 2 incubator at 36.5˚C. Cells were passaged every 3-4 d, using 0.05% and 0.25% Trypsin-EDTA (Gibco), respectively. Cells were transfected using Lipofectamine 2000 (Invitrogen) for HeLa and Cell Line Nucleofector Kit R (Lonza) for mIMCD3 cells.

Construction of Saccharomyces cerevisiae expression strains and sitedirected mutagenesis
The slc25a25 isoform b gene was codon-optimized for expression in S. cerevisiae (Genscript). The gene was truncated Δ1-15, and an 8-histidine tag followed by a Factor Xa cleavage site (IEGR) was introduced to the N-terminus of the gene. Vectors were isolated by miniprep (Real Genomics), and the gene was confirmed by sequencing (Source Bioscience). The gene was cloned into a modified S. cerevisiae expression vector pYES3/ CT, with the inducible galactose promoter replaced by a constitutively active promoter for the S. cerevisiae phosphate carrier PIC2, as described previously [59]. Expression vectors were transformed into the S. cerevisiae W303-1B strain based on the LiAc/SS carrier DNA/PEG method [60]. Transformed cells were selected on synthetic complete tryptophan dropout (SC-Trp) plates supplemented with 2% glucose.

Growth of S. cerevisiae and mitochondrial isolation
A 5-liter preculture of SC-Trp medium (Formedium) supplemented with 2% glucose media was used to inoculate 50 L of YPD in an Applikon bioreactor. For large-scale cultures, cells were harvested by continuous centrifugation (CEPA Z41). Mitochondria were prepared using established methods, snap frozen in liquid nitrogen, and stored at −80˚C [61].

Solubilization and protein purification of SLC25A25
The mitochondria were solubilized and purified by nickel affinity chromatography as described previously [23], with the following amendments: The nickel sepharose resin (GE Healthcare) was washed with 50 column volumes of buffer A containing 50 mM HEPES, pH 7.4, 200 mM sodium chloride, 50 mM imidazole, 0.1% (w/v) lauryl maltose neopentyl glycol, 0.1 mg/mL tetraoleoyl cardiolipin, and 30 column volumes of buffer B (Buffer A without imidazole). Cleavage of the protein from the histidine tag was carried out using Factor Xa protease (New England Biolabs) for 2.5 h at 10˚C. The purified protein was incubated with 10 mM EGTA overnight at 4˚C, and EGTA was removed using a PD10 desalting column (GE Healthcare).

SDS-PAGE analysis of SLC25A25
Proteins were separated by using 4%-12% TruPAGE precast gels SDS-PAGE gels (Merck KGaA) loaded with 5 μg of protein at a 3:1 mix of sample to loading buffer and stained using InstantBlue Coomassie according to manufacturer's instructions (Merck KGaA). Protein concentration was determined using a Spectrophotometer (NanoDrop ND-1000, NanoDrop-Technologies) at absorbance 280 nm.

Thermostability assays
A microscale method, based on the thiol reactive fluorophore N-[4-(7-diethylamino-4methyl-3-coumarinyl)-phenyl]-maleimide (CPM), was used to measure the thermostability of SLC25A25 [62,63]. CPM dissolved at 5 mg/ml in DMSO was diluted into assay buffer (purification buffer B) and incubated at room temperature in the dark for 10 min. The protein stock was diluted to 40 μg/ml in the assay buffer (±1 mM CaCl 2 ) and incubated on ice for 10 min. CPM working solution (5 μL) was added, and the sample was vortexed briefly and incubated at 4˚C in the dark for a further 10 min. Data were collected using a high-resolution melt (HRM) channel on a RotorGene Q 2plex HRM qPCR cycler with a 36-sample rotor (Qiagen). Measurements were made every 15 s at 1˚C intervals from 25-90˚C. Protein denaturation profiles were analyzed using the Rotor-gene Q software, and the peak in the derivative of the fluorescence was plotted from which the apparent melting temperature (T m ), a relative measure of protein stability, was determined.

Liposome preparation, reconstitution, and ATP transport assays
L-α-phosphatidylcholine (Avanti Polar Lipids) and tetraoleoyl cardiolipin (Avanti Polar Lipids) were mixed in a 20:1 (w/w) ratio and dried under a stream of nitrogen. The lipid mixture was washed with 500 μL of methanol and dried under nitrogen.
Approximately 65 μg of SLC25A25 in LMNG was reconstituted into liposomes loaded with 20 mM HEPES, pH 7.4, 1 mM DTT, 1 mM MgCl 2 , and 1 mM ATP as the internal buffer. The detergent pentaethylene glycol monodecyl ether was added to a final concentration of 1.6% (v/v), and the lipids were solubilized by vortexing and incubated on ice for 30 min. The protein was added, and the samples were incubated on ice for 5 min. The pentaethylene glycol monodecyl was removed by multiple additions of SM-2 bio-beads (Bio-Rad). Four additions of 60 mg and 1 addition of 480 mg of bio-beads were added to the sample every 20 min with inversion at 4˚C. The samples were incubated overnight at 4˚C with inversion. Bio-beads were removed by passage of the sample through empty micro-bio spin columns (Bio-Rad). The external substrate was removed using a PD10 desalting column (GE Healthcare).
Transport assays were carried out with a Hamilton MicroLab Star robot (Hamilton Robotics). One hundred μL of proteoliposomes in external buffer was loaded per well in a Multi-Screen HTS -HA 96-well filter plate (pore size = 0.45 μm; Millipore). Uptake assays of radiolabeled ATP in the presence or absence of 1 mM CaCl 2 were initiated by the addition of 100 μl of HEPES buffer with 2 μM [ 14 C]-ATP (Perkin Elmer) per well. The uptake of [ 14 C]-ATP was stopped after 0, 10, 20, 30, 45, 60, and 150 s and 5, 7.5, and 10 min incubation times with 200 μl ice-cold HEPES buffer, and the samples were filtered with a vacuum manifold, followed by 2 additional wash steps with 200 μl ice-cold HEPES buffer. Plates were dried overnight. Radioactivity was measured by adding 200 μl MicroScint-20 (Perkin Elmer) and quantification using a TopCount scintillation counter (Perkin Elmer). Uptake curves were fitted according to the one-phase association model (GraphPad, Prism).

Oxygen consumption rate (OCR)
OCR was measured using a Seahorse Biosciences XF e 96 extracellular flux analyzer. Ten thousand cells per well were seeded in XF96 cell culture plates coated with poly-D-lysine (Sigma-Aldrich). Attachment of the cells was monitored after 1 h and 3 h, and cells were incubated over night at 37˚C with 5% CO 2 . Before the assay, cells were washed twice with XF base media (unbuffered DMEM supplemented with 2 mM L-Glutamine [Sigma-Aldrich], 11 mM Glucose [Carl Roth], 1 mM Sodium Pyruvate [Sigma-Aldrich] at pH 7.4) and equilibrated for 1 h at 37˚C without CO 2 . The OCR was measured afterwards using the following inhibitors: 2 μM oligomycin (Seahorse Biosciences), 1 μM carbonyl cyanide 4-(trifluoromethoxy)phenylhydrazone (FCCP, Seahorse Biosciences), and 2 μM rotenone with 2 μM antimycin A (Seahorse Biosciences). For each condition (basal and after each inhibitor injection), cycles were performed in triplicate with 3 min mixing followed by 3 min measurement. After completion of the assay, the protein content per well was determined using the Bradford assay (Roti-Quant; Carl Roth). Absorption was measured at 595 nm using a PerkinElmer 2030 Explorer plate reader. The OCR was normalized to the protein content for each well, and the results of inner wells of the 96-well plate with the same cell density were evaluated for the analysis of the cell lines.

Nontargeted metabolomics
Nontargeted metabolomics have been described previously [70]. Briefly, mIMCD3 cells were cultivated in 6 wells (Greiner Bio-One) for 14 d (!10 d 100% confluence). Cells were homogenized using 80 mg of 0.5 mm glass beads (Precellys) in 80% v/v methanolic solution. Sample was loaded in duplicates onto 96-well 350-μL PCR plates by splitting each into 2 aliquots consisting of 105 μl. The first aliquot was used for LC-MS/MS analysis in positive electrospray ionization mode, and the second aliquot was used for that in negative mode. Metabolites were annotated by curation of the LC-MS/MS data against proprietary Metabolon's chemical database library (Metabolon) based on retention index, precursor mass, and MS/MS spectra. In this study, 199 known metabolites and 66 compounds with an unknown chemical structure, indicated by a letter X followed by a number as the compound identifier, were identified. The metabolites were assigned to cellular pathways based on PubChem, KEGG, and the Human Metabolome Database.

Protein isolation, SDS-PAGE, western blot, and CCD camera-based ECL detection
Biochemical methods have been described previously [58,71].

D. melanogaster fertility assays
D. melanogaster males of various genotypes were separated upon eclosion and maintained in isolation 2 d prior to mating with w 1118 virgin females. For each test, 2 pairs of adults were allowed to mate for 5 d and then removed from the vial. After 10 d, the number of progeny that eclosed from each vial was counted. Fertility tests were evaluated using the Mann-Whitney U test. An asterisk indicates P 0.05.

Dissection and immunofluorescence of D. melanogaster sperm
Dissection and preparation of testis and spermatozoa as well as the anti-Amo antiserum have been described [12,13]. Microscopy images were recorded using a Zeiss Axio Observer microscope (Zeiss).

Morpholino phosphorodiamidate antisense oligonucleotides (MOs)
Morpholino antisense oligonucleotides (Gene Tools) were designed to target the translation of the mRNA leading to a protein knockdown phenotype or to target an exon splice donor site causing splicing defects of the mRNA. MOs were diluted (in 100 mM KCl, 10 mM HEPES, 0.1% Phenol Red) and phenotypes were analyzed in a concentration-dependent fashion. A p53 MO was co-injected 1.5-fold with all MOs to attenuate possible off-target effects [76]. Injections were performed into 1-cell-stage embryos using a microinjector PLI-90 (Harvard Apparatus). Injection volume was approximately 2 nl comprising 8 ng pkd2 MO , slc25a23a MO , slc25a23b MO , slc25a24 MO , or slc25a25a MO ; 5 ng slc25a25b ATG ; 3 ng slc25a25b E4-I34 ; or Standard Control MO, respectively.

Analysis of heart looping in zebrafish
Heart looping of live zebrafish embryos was analyzed blinded for genotype and injection 48 hpf. At least 3 independent experiments were evaluated using a Stereo Discovery V8 microscope (Zeiss).

Rescue of MO-induced zebrafish phenotypes
The specificity of MOs was validated by mRNA-mediated rescue of respective phenotypes. Rescue experiments were done by co-injection of capped mRNA (mMESSAGE mMACHINE; ThermoFisher Scientific) and MOs. Transcripts used were pkd2 (ENSDART00000020412) and slc25a25b (ENSDART00000098163).

Immunofluorescence and live imaging of D. rerio KV
Six-somite stage embryos Tg(actb2:Mmu.Arl13b-GFP) were embedded in 1% low-temperature melting agarose (Biozym) in 30% Danieau's solution (17.4 mM NaCl, 0.21 mM KCl, 0.12 mM MgSO 4 × 7 H 2 O, 0.18 mM Ca[NO 3 ] 2 , 1.5 mM HEPES). Real-time imaging of GFP-labeled KV cilia was performed on a ZEISS LSM 510 Live confocal microscope equipped with a LD LCI Plan-Apochromat 25×/0.8 glycerine objective (Zeiss). Six hundred images were recorded per sample at 9-10 frames per second with a resolution of 512 × 512 pixels. Similarly, flow in the KV of embedded 8-somite stage embryos was visualized by particle tracking using differential interference contrast (DIC) microscopy (Zeiss Axio Observer microscope; Zeiss) for 2 min at 20 frames per second. [39]. For high-resolution 3D confocal imaging of zebrafish KV, Tg (actb2:Mmu.Arl13b-GFP) was fixed in methanol and stained anti-GFP to enhance contrast as described previously [83]. A Leica TCS SP8 STED 3X microscope was used for image acquisition. Vertical projections of recorded stacks were generated using ImageJ 2.

Transmission electron microscopy
Electron microscopy of zebrafish embryos has been described previously [84]. Thin sections (approximately 70-80 nm) were cut on a Reichert Ultracut E ultramicrotome and collected onto formvar-coated slot grids. Sections were stained with uranyl acetate and lead. Samples were examined in a Philips CM10 TEM at 80 kV.