Fundc1 is necessary for proper body axis formation during embryogenesis in zebrafish

FUN14 domain-containing protein 1 (FUNDC1) is a mitochondrial outer membrane protein which is responsible for hypoxia-induced mitophagy in mammalian cells. Knockdown of fundc1 is known to cause severe defects in the body axis of a rare minnow. To understand the role of Fundc1 in embryogenesis, we used zebrafish in this study. We used bioimaging to locate zebrafish Fundc1 (DrFundc1) with MitoTracker, a marker of mitochondria, and/or CellLight Lysosomes-GFP, a label of lysosomes, in the transfected ovary cells of grass carp. The use of Western blotting detected DrFundc1 as a component of mitochondrial proteins with endogenous COX IV, LC3B, and FUNDC1 in transgenic human embryonic kidney 293 T cells. DrFundc1 induced LC3B activation. The ectopic expression of Drfundc1 caused cell death and apoptosis as well as impairing cell proliferation in the 293 T cell line, as detected by Trypan blue, terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) and incorporation of BrdU. DrFundc1 up-regulated expression of both autophagy- and apoptosis-related genes, including ATG5, ATG7, LC3B, BECLIN1, and BAX in transgenic 293 T cells. A knockdown of Drfundc1 using short hairpin RNA (shRNA) led to midline bifurcation with two notochords and two spinal cords in zebrafish embryos. Co-injection of Drfundc1 mRNA repaired defects resulting from shRNA. Knockdown of Drfundc1 resulted in up- or down-regulation of genes related to autophagy and apoptosis, as well as decreased expression of neural genes such as cyclinD1, pax2a, opl, and neuroD1. In summary, DrFundc1 is a mitochondrial protein which is involved in mitophagy and is critical for typical body axis development in zebrafish.


Results
Location and function of DrFundc1 in vitro. Use of a multiple-sequence alignment of selected proteins as well as a molecular phylogenetic tree show that FUNDC1 is highly conserved in vertebrates (Fig. S1A,B). The open reading frame (ORF) of zebrafish fundc1 (Drfundc1) is 459 base pairs (bp) in length. This encodes DrFundc1, a protein of 152 amino acids (AAs). DrFundc1 contains a putative LIR motif (YEVV) and a FUN domain (AA 51-132; Fig. S1A). DrFundc1 is a putative transmembrane protein with three α-helical stretches and the LIR motif in the N-terminal out of the mitochondrial outer membrane (Fig. S1C).
DrFundc1's location and function in vitro were studied in two available cell lines: the grass carp (a relative of zebrafish and rare minnow in Cyprinidae family) ovary (GCO) cell line and the human embryonic kidney (HEK) 293 T cell line. This was due to a lack of both a proper zebrafish cell line for gene transfer and antibodies for detection of zebrafish proteins. We ask whether DrFundc1 is located in mitochondria and if it can work as its homolog FUNDC1 in mammalian cells to induce mitophagy.
Through the use of bioimaging, we found that the red fluorescence of DrFundc1-Cherry overlapped with the green fluorescence from MitoTracker Green (Thermo Fisher Scientific, Carlsbad, CA, USA; M7514), a reagent labeling mitochondrion, in transgenic GCO cells (Fig. S2A). Use of Western blotting showed a clear band (~44 kD) of DrFundc1-Cherry-His in the mitochondrial extract of transgenic GCO cells (Fig. S2B). It was also observed that DrFundc1-Cherry co-located with CellLight Lysosomes-GFP (Thermo Fisher Scientific; C10596), a reagent labeling lysosome, in transgenic GCO cells (Fig. S2C).
GCO cells grew poorly following Drfundc1 transfection. The more Drfundc1 was transfected, the poorer the cell growth (Fig. S2D). Cell numbers decreased significantly in the dosage of 400-500 ng of pCS2 + -Drfundc1-Cherry-His compared to control cells transfected with pCS2 + -Cherry plasmid. Cells transfected with Drfundc1 displayed low density, while some cells were round in shape, floating, and aggregating into clusters.
Use of Western blotting detected DrFundc1-Cherry fusion protein, mainly in the mitochondrial extract of transgenic 293 T cells which had been transfected with pCS2 + -Drfundc1-Cherry-His plasmid (Fig. 1A). DrFundc1-Cherry levels were significantly higher in mitochondria than in the cytoplasm of transgenic cells (P < 0.001), while endogenous cytochrome C oxidase subunit 4 (COX IV), LC3B, and endogenous FUNDC1 were primarily detected in mitochondria. Due to expression of DrFundc1 in the cells, more LC3B-I was converted to LC3B-II than in control cells (P < 0.001) (Fig. 1A). This suggests that DrFundc1 is a component of mitochondria and can lead to LC3 activation, which in turn causes mitophagy.
Deliberate expression of DrFundc1 was harmful to 293 T cells, which decreased in density, changed in morphology, aggregated, and even died following Drfundc1 transfection (Fig. 1B,C, S3A,B). Use of a 3-(4,5-dime thylthiazol-2-yl)−2,5-diphenyltetrazolium bromide (MTT) assay found a significant decrease in proliferation of transgenic 293 T cells following Drfundc1 transfection (P < 0.001) (Fig. 1B). Use of trypan blue staining and cell counting indicated a significant increase in mortality (P < 0.01) in cells by Drfundc1 expression (Fig. 1C,D). We used a 5-bromo-2′-deoxyuridine (BrdU) incorporation and found a decrease of cell proliferation in the transgenic cells with Drfundc1 expression (P < 0.001), although no apparent changes to cell proliferation in total cells were observed (Fig. 1E, S3C). This suggests that DrFundc1 caused cell death and impaired cell proliferation.
Transfection of Drfundc1 led to apoptosis of transgenic 293 T cells, as revealed by terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) assay. TUNEL-positive cells were observed in cells transfected with Drfundc1-Cherry (Fig. 1F), while use of a quantitative reverse transcription polymerase chain reaction (qRT-PCR) demonstrated expressional change of autophagy-and apoptosis-related genes in cells. Autophagy-related genes ATG5, ATG7, ATG12, and LC3B, as well as apoptosis-related genes BCL2-associated X (BAX), B-cell CLL/lymphoma 2 (BCL2), and BECLIN1 were significantly up-regulated by use of DrFundc1 (Fig. 1G). Therefore, decreased cell viability was due to DrFundc1-induced autophagy and apoptosis. However, CASPASE3 expression did not change, suggesting that apoptosis may not depend on CASPASE3.

Expression of Drfudnc1 in adult tissues and embryos of zebrafish. Use of a qRT-PCR detected
Drfundc1 in selected tissues (brain, eye, heart, intestine, liver, muscle, kidney, testis, and ovary; see Fig. S4A). Expression of Drfundc1 was highest in the brain, followed by a moderate expression in the liver, ovary, testis, and kidney, while the lowest expression was in the heart and muscle. Use of a qRT-PCR also detected Drfundc1 in zygotes throughout zebrafish embryogenesis (Fig. S4B). Expression of Drfundc1 increased from the 1-cell stage, peaked at the gastrula stage (6 h post fertilization [hpf]), decreased at 12 hpf, and was then maintained at a low level from 24 hpf until hatching. We further studied the expression pattern of Drfundc1 using a whole-mount in situ hybridization (WISH; see Fig. S4C). Use of WISH detected Drfundc1 in embryos from zygote until hatching. Drfundc1 was found in all blastomeres at early stages, from the 1-cell stage to the gastrula stage. Expression of Drfundc1 was enriched in embryos' heads -including brains and eyes -from 24 hpf onwards.

Knockdown of Drfundc1 formed two notochords and two spinal cords as well as affecting expression of autophagy-, apoptosis-, and neural development-related genes.
To understand midline bifurcation, we subjectively studied collagen 2a1a (col2a1a) and col8a1a as notochord markers and sonic hedgehog protein a (shha) as a spinal cord marker [26][27][28][29] . Use of WISH clearly demonstrated that interference with Drfundc1 induced formation of two notochords as well as two spinal cords in the embryo, while the control embryo was typical, with one notochord and one spinal cord (Fig. 3A), suggesting that midline bifurcation is a malformation of the body axis.
Use of a qRT-PCR found that atg5, atg7, ambra1a, and ambra1b were down-regulated, while p62 and atg12 were significantly up-regulated by interference of Drfundc1 (P < 0.01 or 0.001; Fig. 3B). In addition, beclin1, bcl2a, bcl2b, and caspase3a were significantly down-regulated (P < 0.001) by interference of fundc1. Interference of fundc1 down-regulated baxb but up-regulated baxa (P < 0.001) and did not affect expression of caspase9, p53, and fox3a. This suggests that knockdown of Drfundc1 interfered with autophagy and apoptosis, leading to body axis malformation in zebrafish embryos, and that this process depends on Caspase3 rather than Caspase9.
Neural system-related genes cyclinD1, paired box 2a (pax2a), odd-paired-like (opl), and neuroD1 were detected using WISH and qRT-PCR. CyclinD1 is associated with cell proliferation in the spinal cord 34 , pax2a is a mesencephalic marker participating in midbrain-hindbrain boundary development 35,36 , opl is a marker for forebrain cell fate which controls midline formation and forebrain patterning 37,38 , while neuroD1 is critical for neuronal cell fate 39 . Use of WISH indicated that Drfundc1 knockdown attenuated expression of cyclinD1 in the telencephalon, opl in the spinal cord, pax2a in the head and muscle, and neuroD1 in the eye (Fig. 3C). We used gray-scale analyses to find that gene expression significantly decreased in embryos which had been microinjected with Drfundc1 shRNA2 (Fig. 3D). Subsequently, qRT-PCR confirmed cyclinD1, pax2a, opl, and neuroD1 down-regulation due to Drfundc1 knockdown (Fig. 3E).

Discussion
We identified DrFundc1 as a mitochondrial protein associated with mitophagy. Additionally, DrFundc1 is critical for proper body axis formation in zebrafish. Knockdown of Drfundc1 causes down-or up-regulation of autophagy-, apoptosis-, as well as neural development-related genes.
Mitophagy is pivotal for the survival of organisms while dysfunction of mitochondria causes severe deficiencies in zebrafish 15 , mice 16 , and fruit flies 17 . Zebrafish Fundc1 is involved in mitophagy, as described above. In this study, Drfundc1 knockdown caused severe defects in the body axis, such as midline bifurcation and headlessness, which had been previously observed in rare minnows 25 . Co-injection of Drfundc1 mRNA repaired these defects, although Drfundc1's expression level did not go back to typical levels. This suggests that Fundc1 is necessary and critical for typical embryogenesis in fish.
Apoptosis is associated with mitophagy 30,31 . Knockdown of Drfundc1 decreased expression of the apoptosis-related genes baxb, bcl2a, bcl2b, and caspase3a 32,33,45,47,48 , contrasting to results of 293 T cells due to ectopic expression of Drfundc1. This means that Drfundc1 knockdown interferes with apoptosis in zebrafish embryos. Apoptosis is a major part of typical development in many organisms, including zebrafish 49 . Interference with the process of apoptosis is likely to result in defects.
Midline bifurcation had been reported in cases of forced expression of insulin-like growth factor 2a (IGF-2a) 27 as well as loss of Squint 28 in zebrafish. Overexpression of IGF-2a induced Akt phosphorylation and caused midline bifurcation 27 . Loss of Squint, a nodal-related protein, resulted in midline bifurcation due to the failing expression of wnt5b 28 . Nodal regulatory factors contribute to mitochondrial homeostasis 50 . Nodal factors can down-regulate phospho-Akt while activating Smad2 and mitophagy to induce granulosa cell apoptosis 51 . Activation of Wnt signaling and Akt maintains the mitochondrial membrane and regulates apoptosis repressor Bcl-xL 52 . Overexpression of IGF-2a and loss of Squint may impair mitophagy and apoptosis in zebrafish embryos. However, IGF-1 induces mitophagy through BNIP3 accumulation in mitochondria, stimulates mitochondrial biogenesis 53 , and protects mitochondria from apoptosis 54 . IGF-IIR, the receptor of IGFII, induces mitophagy through Rab9-dependent alternative autophagy 55 . In contrast, it has been reported that IGF-IIRα disrupts mitochondrial membrane potential, induces perturbation of mitophagy, and can lead to mitochondrial oxidative stress 56 . Further study is needed to clarify these controversial results. Drfundc1 knockdown impaired mitophagy and apoptosis and resulted in a similar phenotype and midline bifurcation in our experiment, suggesting that midline bifurcation results from impairment of mitophagy and apoptosis during embryogenesis.
In addition to midline bifurcation, Drfundc1 knockdown led to other severe defects in the anterior section of zebrafish embryos. Knockdown of Drfundc1 decreased expression of cyclinD1, pax2a, opl, and neuroD1, which are involved in development of the neural system [34][35][36][37][38][39] . However, the ways in which these genes are affected remains unclear.
Overall, we found that Fundc1 is critical for proper body axis formation in fish. Fundc1 knockdown induces insufficient mitophagy or autophagy, interferes with apoptosis, and leads to body axis malformation. Interference with Fundc1 expression impacts on genes which are important for typical embryogenesis.

Materials and Methods
Ethics. This study was conducted in strict accordance with recommendations for the Regulation for the Cell culture and transfection. An ovary cell line (GCO) of grass carp (Ctenopharyngodon idellus) was obtained from the National Key Laboratory of Fresh Water Ecology and Biotechnology at the Institute of Hydrobiology, Chinese Academy of Sciences. GCO cells were cultured in M199 medium (Thermo Fisher Scientific, Carlsbad, CA, USA; C11150500BT) and supplemented with 10% inactivated fetal bovine serum (FBS) (Thermo Fisher Scientific) 58 . HEK 293 T cells were cultured in high-glucose Dulbecco's modified Eagle medium (Thermo Fisher Scientific; C11995500BT) and supplemented with 10% FBS 59 . GCO and 239 T cells were incubated at 28 °C and 37 °C respectively in a humidified atmosphere of 95% air and 5% CO 2 .
A fusion gene of Drfundc1 with Cherry and His tags (Drfundc1-Cherry-His) was subcloned into pCS2 + vector as pCS2 + -Drfundc1-Cherry-His. pCS2 + -Cherry was constructed as a control. Plasmids were transfected into GCO or 239 T cells using PolyJet (SignaGen Laboratories, Rockville, MD, USA; SL100688) at a cell density of 70-80% in plates, or at a cell density of 50% on a coverslip, following manufacturer's instructions. Transfection medium was replaced with a fresh complete medium at 12-18 h post-transfection.
Cell staining and fixation. In order to label mitochondria, GCO and 239 T cells were incubated with pre-warmed (37 °C) MitoTracker Green FM probes (Thermo Fisher Scientific; M7514) at a concentration of 200 nmol/L for 40 min under growth conditions in the dark. Nuclei were stained with Hoechst 33258 (Thermo Fisher Scientific; H3569). After staining, the solution was replaced with a fresh pre-warmed culture medium. Images were taken using an EVOS FL auto fluorescence microscope (Thermo Fisher Scientific).
In order to label lysosomes, CellLight Lysosomes-GFP reagent (Thermo Fisher Scientific; C10596) was added to cells in a complete culture medium, following the manufacturer's instructions, and was gently mixed. Cells were incubated with the reagent overnight (≥16 h) in the dark. Images were taken with a confocal microscope (Leica, Germany; SP5) following cell fixation. (2019) 9:18910 | https://doi.org/10.1038/s41598-019-55415-0 www.nature.com/scientificreports www.nature.com/scientificreports/ TUNEL assay. A one-step TUNEL apoptosis assay kit (Beyotime Biotechnology; C1088) was utilized to detect apoptosis in 293 T cells. The TUNEL reaction was carried out according to the protocol 60 and the manufacturer's instructions. After fixation with freshly prepared 4% paraformaldehyde (PFA) for 30 min, cells were washed once with phosphate-buffered saline (PBS) and then incubated at room temperature with PBS containing 0.3% Triton X-100 for 5 min. Next, cells were incubated with 100 μL TUNEL detection solution in a 24-well plate at 37 °C for 60 min in the dark. Following Triton X-100 treatment and PBS immersion, positive controls were treated with DNase I reaction solution (Beyotime Biotechnology; C1082) at room temperature for 10 min, while negative controls were incubated without the terminal transferase (TdT) enzyme reaction solution during the labeling reaction. Coverslips were washed three times with PBS and mounted onto slides using antifade mounting medium (Beyotime Biotechnology; P0126). Images were taken with an EVOS FL auto fluorescence microscope (Thermo Fisher Scientific).
Cell death and cell proliferation assay. An MTT proliferation assay kit (Beyotime Biotechnology; C0009) was used to measure cell viability. Briefly, 293 T cells were seeded into 96-well plates and cultured. Cell viability was measured according to the manual of the kit post-transfection.
Trypan Blue (Thermo Fisher Scientific; T10282) was used to detect dead cells; 10 μl 0.4% Trypan Blue solution was added into 100 μl cell suspensions. Blue staining cells and total cells were counted using a hemocytometer. Cells were also stained with Trypan Blue in plates after fixation with 4% PFA followed by a wash with PBS.
Cell proliferation was also detected with BrdU incorporation 61 . The 293 T cells were seeded onto cover slips and transfected with plasmids, as described above. After transfection, cells were incubated with 10 μmol/L 5-bromo-2′-deoxyuridine (BrdU) (Thermo Fisher Scientific; B23151) in culture medium for 9 h in dark and then fixed with 4% PFA. Cells were washed with a Tris-buffered saline Tween-20 (TBST) containing 0.1% Triton X-100 (TBSTx) after fixing and were incubated in 1.5 mol/L HCl for 30 min to expose antigens. After washing, 5% BSA in TBSTx was applied onto cells for blocking. Cells were then incubated with BrdU monoclonal antibody (ABclonal; A1482) for 12 h at 4 °C. After washing three times with TBSTx, Alexa Flour 488-conjugated AffiniPure goat anti-mouse IgG (H + L) (ABclonal; AS076) was applied for 1 h in the dark. Cells were stained with Hoechst 33258 after washing and mounted onto slides with antifade mounting medium (Beyotime Biotechnology; P0126). Images were taken with an EVOS FL auto fluorescence microscope (Thermo Fisher Scientific).

Quantitative real-time PCR (qRT-PCR).
Total RNA of different cells, tissues and/or embryos at different stages was extracted with TRIzol reagent (Thermo Fisher Scientific; 15596026) and then reverse-transcribed into complementary DNA (cDNA) with a FastQuant RT Kit (with gDNase; TIANGEN Biotech, Beijing, China; KR106) after digestion of genomic DNA with RNase-free DNase I, following the manufacturer's instructions.
QRT-PCR was carried out in a reaction volume of 20 μL containing template cDNA, primers, RNase-free H 2 O, and 10 μL of 2х SuperReal premix plus (SYBR Green) (TIANGEN Biotech; FP205) using the following cycle settings: 95 °C for 3 min, followed by 40 cycles of 95 °C for 10 s and 62 °C for 30 s. Samples were analyzed in triplicate, and gene expression values calculated on technical triplicates and biological replicates. Expression levels of target genes were measured and normalized with that of β-actin according to the 2-DCt or 2-DDCt calculation method 62 . Table S1 shows primers used.
In situ hybridization. Whole mount in situ hybridization (WISH) was carried out on zebrafish embryos following the protocol reported previously 63 . Sense and antisense digoxigenin (DIG)-labeled RNA riboprobes were produced, as reported previously 25 . Embryos older than 20 hpf were digested with proteinase K and then hybridized with appropriate riboprobes at 70 °C for 12-16 h. After thoroughly washing and blocking, embryos were incubated with anti-DIG antibodies which conjugated with alkaline phosphatase. Embryos were then stained using nitroblue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate (NBT/BCIP). Images were taken under an MZ16F stereomicroscope (Leica) which was equipped with a digital camera. Grey scales of the genes in WISH were quantified depending on embryos' grey strengths, using the software ImageJ (https://imagej.nih.gov/ij/). Optical density of the areas stained (S) and neighboring background (B) were measured. Signal strengths were calculated as S-B 64 . The mean strength of each gene was obtained from 10 to 15 embryos in each group.
Knockdown of Drfundc1. In this study, shRNAs which interfered with Drfundc1 were designed online (http://rnaidesigner.thermofisher.com/rnaiexpress/design.do). Two shRNAs -shRNA1 and shRNA2 -targeted nucleotides of Drfundc1 ORF from 77 to 99 and 217 to 237 respectively. A mismatched shRNA (shRNAmis) as well as a random shRNA (shRNAran) without any target were designed as controls. Vector pSuper-puro (OligoEngine) was used to construct shRNA expression plasmids. shRNAs or pSuper-puro were microinjected into zygotes at a dosage of 200 pg. Drfundc1 mRNA was synthesized from pCS2 + -Drfundc1 in vitro using an mMESSAGE mMACHINE SP6 transcription kit (Thermo Fisher Scientific; AM1340). To repair defects caused by Drfundc1 shRNA, 400 pg of Drfundc1 mRNA was microinjected with 200 pg of the shRNA plasmid into zygotes. Embryos with/without injection of pSuper-puro, shRNAmis, and shRNAran were treated as controls.
Statistical analyses. Data analyses were carried out using SPSS 17.0 (IBM, Armonk, NY, USA). Differences between groups were analyzed using a one-way analysis of variance (ANOVA). A post-hoc Duncan's multiple range test was utilized to determine significant differences. All data were obtained from at least three independent experiments (n ≥ 3) and were described as mean ± standard error (SEM). P < 0.05 was considered statistically significant.