Requirement of Npc1 and availability of cholesterol for early embryonic cell movements in zebrafish.

Niemann-Pick disease, type C (NP-C), often associated with Niemann-Pick disease, type C1 (NPC1) mutations, is a cholesterol-storage disorder characterized by cellular lipid accumulation, neurodegeneration, and reduced steroid production. To study NPC1 function in vivo, we cloned zebrafish npc1 and analyzed its gene expression and activity by reducing Npc1 protein with morpholino (MO)-oligonucleotides. Filipin staining in npc1-morphant cells was punctate, suggesting abnormal accumulation of cholesterol. Developmentally, reducing Npc1 did not disrupt early cell fate or survival; however, early morphogenetic movements were delayed, and the actin cytoskeleton network was abnormal. MO-induced defects were rescued with ectopic expression of mouse NPC1, demonstrating functional gene conservation, and by treatments with steroids pregnenolone or dexamethasone, suggesting that reduced steroidogenesis contributed to abnormal cell movements. Cell death was found in anterior tissues of npc1 morphants at later stages, consistent with findings in mammals. Collectively, these studies show that npc1 is required early for proper cell movement and cholesterol localization and later for cell survival.

during epiboly, one of the earliest morphogenetic movements of gastrulation that ultimately generate the embryo's complex body plan ( 36,37 ). The morphogenetic process of epiboly involves coordinated movements of each of the embryonic cell layers that are present during late blastula ( 1 ): the deep cell layer, which gives rise to the embryo proper ( 2 ); the enveloping layer (EVL), an extraembryonic superfi cial epithelial layer covering the deep cells; and ( 3 ) the yolk syncytial layer (YSL), an extra-embryonic cytoplasmic cell layer within the yolk cell ( 38,39 ). Epiboly commences when the yolk cell bulges toward the animal pole and the deep cell blastomeres radially intercalate. This process continues with the thinning and vegetal migration of the blastoderm over the yolk cell until 50% of the yolk surface is covered (50% epiboly) (40)(41)(42), at which time the deep cells begin the second phase of gastrulation involving dorsal convergence and involution movements which form the germ cell layers. Concomitant with this, each of the three cell layers continues to spread over the yolk in the epiboly process until the yolk cell is completely covered and internalized (41)(42)(43)(44). Reducing the levels of cholesterol metabolites in zebrafi sh embryos results in an epiboly-delay phenotype ( 36 ).
In this study we have identifi ed and cloned the zebrafi sh npc1 gene and found that it is widely present during early embryonic development. Using targeted morpholino (MO) antisense oligonucleotides, we have demonstrated that loss of npc1 leads to sterol localization defects in early embryos, similar to defects observed in fl y and mammalian cells lacking Npc1. Our gene knockdown studies further revealed that npc1 is required for normal epiboly movement. Epiboly defects may be in part due to abnormal cytoskeletal structures, as we observed disruptions in the actin cytoskeleton in npc1 morphants. Unlike some zebrafi sh mutants and morphants that have epiboly delay, Npc1morphant zebrafi sh did not show a reduction in EVL integrity ( 45,46 ) or any decline to the ability of the leading cells to undergo necessary shape changes during epiboly ( 47 ). The phenotype cannot be attributed to cell death during early developmental stages as we did not detect any increases in apoptosis in these morphant embryos. The epiboly defect in morphants was rescued by the coinjection of mouse Npc1 mRNA at the 1-cell stage or into the yolk cell of a 1,000-cell stage embryo, showing conservation of function between the fi sh and mammalian orthologs. Moreover, two downstream components of steroid synthesis, including the cholesterol derivative pregnenolone (P5) and glucocorticoid dexamethasone (Dex), could partially rescue the epiboly defects, demonstrating that such defi cits are likely due, at least in part, to a steroid defi ciency. Embryos that make it through epiboly continue to show defi ciencies in gastrulation movements and also demonstrate cell death in the developing central nervous system. In full, this study indicates that zebrafi sh possess a functional ortholog of the mammalian Npc1 gene, which functions in sterol traffi cking and is required for early epiboly movements in developing embryos. Furthermore, the morphant studies reported herein demonstrate that the zebrafi sh is an excellent vertebrate model system to further Recent studies have established that NP-C pathology is enhanced by a physiological defi ciency state (21)(22)(23), as cholesterol sequestration in late endosomes and lysosomes consequently limits substrates for steroid biosynthesis ( 23 ). Cholesterol-derived steroid hormones include mineralocorticoids, glucocorticoids, sex hormones, and neurosteroids that collectively control important physiological functions, such as electrolyte balance, glucose homeostasis, and sexual development ( 24,25 ). Normal steroid hormone metabolism is disrupted in fl ies and mice where Npc1 and Npc2 have been genetically inactivated (21)(22)(23)26 ). Steroid hormone replacement therapy may be an option for alleviating NP-C patient symptoms as this method of treatment has proven to be successful in studies involving Npc1-defi cient mice. Injection of allopregnanolone perenterally to Npc1-defi cient mouse embryos results in delayed onset of symptoms, prolonged life span, and improved neurological function ( 21,27 ), possibly by quenching elevated levels of reactive oxygen species in the brain ( 28 ). Moreover, chronic treatment with the sex steroid estradiol in Npc1-defi cient mice improves defective pituitary development and is capable of reversing ovarian defects and infertility in Npc1-defi cient females ( 29,30 ). These studies suggest that further insight into the pathophysiology of NP-C may be gained by having a better understanding of cholesterol metabolism and steroid hormone action during development and adulthood.
The Npc1 gene has been found to be strongly conserved in many experimental model organisms ( 31 ), playing a critical role in some aspect of sterol homeostasis in each species examined to date. Structurally, Npc1 is a 13 transmembrane-spanning protein ( 32 ), containing a sterolsensing domain (SSD) composed of fi ve transmembrane helices. The SSD is found in a number of different proteins, some of which are also involved in binding cholesterol and in sterol metabolism [e.g., sterol regulatory elementbinding protein cleavage-activating protein (SCAP)], and others of which are involved in moving and binding to the cholesterol-modifi ed protein Sonic hedgehog (e.g., Dispatched and Patched) ( 33 ). Despite this, the mechanistic role of this family of proteins in vertebrate development has not been widely examined. Our interest in the contribution of Npc1 to vertebrate development stems from our overall interest in the role of SSD-containing proteins during development ( 34 ). To this point, most studies concerning Npc1 function have focused on either the cell biology of sterol movements within a cell or the neuropathological outcomes that result from disruption of this process. Less is known about the contribution of this protein to vertebrate development.
Because it is a vertebrate animal and genetic morphants can be produced in large numbers, the zebrafi sh embryo is an ideal organism to examine the role of Npc1 in development. Additionally, zebrafi sh closely resemble mammals in their development and in cholesterol metabolism (reviewed in Ref. 35 ), making them highly amenable for studying proteins involved in lipid and sterol metabolism. In zebrafi sh development, early studies have revealed a requirement for cholesterol in promoting cell migration suppress apoptotic effects induced by MO, was also occasionally used ( 50 ). Co-injection of an Npc1 MO and p53 MO did not alter any of the phenotypes examined, suggesting that they were not due to off-target activation of p53 activity. MOs were diluted to working concentrations of 0.5-3 mg/ml in Danieau solution (58 mM NaCl, 0.7 mM KCl, 0.4 mM MgSO 4 , 0.6 mM Ca (NO 3 )2, 5 mM HEPES, pH 7.6).
For detecting spliced products in MO2-injected embryos, total RNA extraction, cDNA synthesis, and RT-PCR were performed as described above. Oligonucleotides were designed to fl ank exon 13, which is removed due to a MO2-induced RNA splicing deficit. The forward primer is in exon 12, 5 ′ GGCATGGGAGA-AAGAGTTTATTAGG 3 ′ , and the reverse primer is in exon 14, 5 ′ CATGTTCCTCTCTTCCTTCTCTGAG 3 ′ . The wild type product, which includes exon 13 (178 nucleotides), is 480 nucleotides in length. Mis-splicing produced by MO2 will result in a fragment that is 302 nucleotides in length, as well as the wild-type fragments. To determine the identity of npc1 fragments in MO2injected fi sh, PCR was performed with the primers above, and products were separated on an agarose gel. The DNA bands were extracted and purifi ed using the Qiaquick Gel Extraction kit (Qiagen, Valencia, CA), and DNA sequencing was performed (Genewiz, Inc., South Plainfi eld, NJ).
Capped mRNAs, of either mouse Npc1 or synthetic transpose (for control injections), were synthesized using the mMESSAGE mMACHINE kit (Ambion, Inc., Austin, TX) according to manufacturer's instructions. Nanoliter quantities of 100 pg/nl mRNA solutions, suspended in 0.2 M KCl, were pressure-injected at the 1-cell stage, alone or in conjunction with MO1, or at the 1,000cell stage, depending on the experimental design. Diluted MOs were injected into the yolk of 1-cell stage embryos at a concentration of 2.5-7.5 ng per embryo or into the yolk syncytial layer of a 1,000-cell stage embryo at a concentration of 5.0 ng per embryo, along with 0.1% rhodamine dextran solution (Sigma-Aldrich, St. Louis, MO).
Actin staining was performed essentially as previously described ( 63 ). Actin was visualized by incubating fi xed embryos with Alexa Fluor 568 phalloidin (Invitrogen, Carlsbad, CA). Alexa Fluor 568 phalloidin was diluted to 1:1000 in 1% goat serum. Embryos were blocked in blocking buffer [1% goat serum, 10% BSA in PBT (phosphate buffered saline + 0.1% Tween)], then incubated in phalloidin at room temperature for 1-2 h, followed by multiple washes in PBT. Images were acquired using the Zeiss 510 META confocal laser scanning microscope.
Filipin complex from Streptomyces fi lipinensis (F9765; Sigma-Aldrich) was diluted to a stock concentration of 25 mg/ml in DMSO and stored at 4.0°C. The stock concentration was further diluted to 0.05 mg/ml in 1% goat serum (fi lipin staining solution). Embryos study the in vivo mechanisms related to altered cholesterol traffi cking and steroid biosynthesis, as well as a potential model for future therapeutic and genetic interaction studies for Npc1.

Animals
Zebrafi sh ( Danio rerio ) embryos were obtained from natural crosses and staged as previously described ( 48 ). The Tübingen strain of fi sh was originally obtained from the International Zebrafi sh Resource Center (Eugene, OR), and Ekkwill wild-type fi sh were obtained from EkkWill Waterlife Resources (Gibsonton, FL). Zebrafi sh were maintained according to IACUC regulations and standard practices ( 49 ).

Cloning and RT-PCR
To obtain the npc1 gene, data mining revealed a partial clone (NCBI Accession number BG799923). An oligonucleotide sequence recognizing this partial clone was designed and used for 5 ′ RACE and 3 ′ RACE (SMART Race kit, Clontech, Mountain View, CA). cDNA was synthesized (SuperScript III RT First-Strand Synthesis System, Invitrogen, Carlsbad, CA) using RNA isolated and purifi ed (PerfectPure RNA Tissue Kit, 5 Prime Inc., Gaithersburg, MD) from 24 h post fertilization (hpf) Ekkwill zebrafi sh.

Morpholino design and testing
Anti-sense MOs were developed (GeneTools LLC, Philomath, OR) to target the translational start site of Npc1 (MO1, TGTG-GTTTCTCCCCAGCAGAAGCAT) or the splice site acceptor for Npc1 exon 13 (MO2, GAGTCCACCTGTACAACATTTACAG). MO1 can disrupt translation from both zygotic and maternal mature mRNAs; however, some maternal mRNAs may become translated prior to injection of MO1. MO2 disrupts an mRNA processing event, and thus, it can only inhibit the proper translation of zygotic transcripts. Two 5-basepair-mismatch morpholinos were also generated to serve as controls for MO1 and MO2. A p53 MO (GCGCCATTGCTTTGCAAGAATTG), reported to Npc-like region (NpcL) and the 5-transmembrane-spanning SSD motif ( Fig. 1C ). These protein characteristics have been described in all NPC1 orthologs studied to-date ( 64 ). Previous genetic analysis has shown that the NpcL region and the SSD motif are both necessary for protein function ( 65,66 ). Protein sequence alignments of these two domains between zebrafi sh and other species revealed a high level of conservation ( Fig. 1D, E ), strongly implying that the ascribed role for human NPC1 in lipid homeostasis and traffi cking may be conserved in the zebrafi sh Npc1 protein. In further support of this prediction, the presence of a dileucine target motif (LLSY) in the carboxy terminus of the zebrafi sh Npc1 protein (data not shown) is highly reminiscent of similar motifs identifi ed in mammalian Npc1 that have been suggested to promote protein localization to the endosomal and lysosomal pathways ( 64,67,68 ).
To determine if zebrafi sh npc1 is required for sterol traffi cking, we performed sterol localization studies. To do this, we employed a whole-embryo, fi lipin-staining protocol following the reduction of Npc1 protein levels by MO injection. Filipin stains free 3-␤ -hydroxysterols, which include unesterifi ed cholesterol ( 69 ). Npc1 protein levels were reduced by injecting a translation-blocking morpholino (MO1) into one-cell-stage embryos. MO1-injected embryos exhibited early embryonic defects (discussed in detail below); however, these defects did not prohibit us from determining the distribution of endogenous cholesterol in npc1 morphants. We found that cells within the dorsal trunk of 12 hpf noninjected wild-type or control embryos injected with 2.5-5.0 ng/embryo of a 5-basepairmismatch MO1 (Con MO1) at the one-cell stage displayed a diffuse staining pattern for fi lipin, revealing that sterols were evenly distributed throughout the cell ( Fig. 1F, H ). In contrast, one-cell-stage MO1 injections produced embryos that displayed uneven, patchy sterol distribution in cells ( Fig. 1G, I ). Fluorescent fi lipin staining in npc1 morphants was punctate within individual cells, suggesting that sterols were accumulating in subcellular locations. Moreover, while dark areas (free of fl uorescent fi lipin signal) were not apparent in cells of control embryos ( Fig.  1F, H ), these areas were commonly seen in npc1-morphant cells, further refl ecting the uneven distribution of sterols ( Fig. 1G, I ; arrows). The sterol localization defi cit in npc1morphant cells is similar to what has been shown in NPC1 mutant fl y and mammalian cells stained with fi lipin ( 23,70 ), suggesting that the zebrafi sh npc1 gene shares a conserved function in cholesterol traffi cking.

Expression of zebrafi sh npc1
npc1 could be detected throughout embryogenesis and early larval development, from 1 hpf up to 7 dpf, by RT-PCR ( Fig. 2A ). No signal was seen in minus RT control samples (data not shown). npc1 antisense riboprobes were generated to examine the spatiotemporal expression of npc1 by in situ hybridization during embryogenesis. Consistent with RT-PCR data, we found that npc1 was present in the blastodiscs at the two-cell stage, ‫ف‬ 1 hpf ( Fig. 2B ), revealing that the earliest npc1 transcripts are maternally were fi xed at 12 hpf in 4% paraformaldehyde for 1 h at room temperature, followed by multiple washes in PBT. After the last wash, PBT was removed and replaced with fi lipin staining solution. Fixed embryos were incubated in the dark, at room temperature, in the fi lipin staining solution for 2 h. After incubation, embryos were rinsed in PBT and gradually move into 80% glycerol/20% PBT. Embryos were visualized in PBT:glycerol by confocal microscopy with a Zeiss 510 META confocal laser scanning microscope.
DAPI staining to visualize cell nuclei was performed essentially as previously described ( 63 ). Fixed embryos were incubated in a DAPI solution (Sigma-Aldrich), which was diluted 1:50,000 in equal parts PBT and glycerol. Embryos were blocked in blocking buffer for 2 h at room temperature prior to incubation in the DAPI solution for 30 min, followed by multiple washes in PBT. Images were acquired using a Zeiss 700 META confocal laser scanning microscope.
Cell death was visualized in live embryos by incubation in 5 mg/ml Acridine Orange (Sigma-Aldrich) for 1 h at room temperature, followed by three rinses in PBS. Images were acquired using the Zeiss 700 META confocal laser scanning microscope.

Steroid treatments
Pregnenolone (P5) and Dexamethasone (Dex) (Sigma-Aldrich) were solubilized in DMSO to obtain stock solutions of 20 mM that were stored at Ϫ 20°C. Stock solutions were further diluted at 1:500-1,000 in zebrafi sh embryo water to obtain a fi nal working concentration of 20-40 M. Manually dechorionated embryos were bathed in the P5 or Dex solutions from the 16-cell stage until the time they were visually analyzed or fi xed (generally between 50-95% epiboly) in 4% paraformaldehyde, and then analyzed by in situ hybridization. Vehicle-only controls were included for each steroid treatment experiment by diluting DMSO alone at 1:500-1,000 in zebrafi sh embryo water.

Measurements of epiboly
Epiboly progression was measured as previously described ( 37 ). The percentage of epiboly was calculated by determining the distance between the animal pole to the blastoderm margin (which was marked using in situ hybridization staining with an ntl riboprobe) divided by the distance between the animal pole and the vegetal pole. Measurements were acquired using Adobe Photoshop software. Statistical signifi cance between variant groups was determined using either a two-tailed Student's t -test or multivariate ANOVA followed by Tukey's posthoc analysis (Excel; PASW 18/SPSS).

Zebrafi sh Npc1 has a conserved protein identity and function in sterol traffi cking
Zebrafi sh npc1 was identifi ed by homology searches using human NPC1 to screen genomic and cDNA zebrafi sh databases, followed by 3 ′ and 5 ′ RACE PCRs to obtain the putative mRNA sequence [deposited in the National Center for Biotechnology Information (NCBI) as JF951427]. Zebrafi sh npc1 is predicted to encode a 1,278 amino acid protein with 60% identity and 66% similarity to human NPC1, and 59% identity and 65% similarity to mouse Npc1 ( Fig. 1A , B ). Peptide analysis software predicted that zebrafi sh Npc1 is a 13-transmembrane-spanning domain protein, harboring a putative leucine zipper motif with surrounding sequences that collectively form the so-called higher doses (7.5 ng/embryo), well over half of the embryos (25/44, 57%) receiving MO1 exhibited a class 3 phenotype ( Fig. 3O ), meaning that embryos were still in the earliest stages of epiboly when control embryos had reached 100% epiboly. The majority of embryos receiving a moderate dose (5.0 ng/embryo) of MO1 exhibited class 2 phenotypes (109/180, 60%) ( Fig. 3N ). Embryos were assigned a class 2 phenotype if their epiboly progression lagged behind at 70-89% epiboly when control embryos had reached 100% epiboly. Finally, we found that low doses (2.5 ng/embryo) of MO1 was a suffi cient amount of morpholino to delay epiboly in 79% of embryos (120/153), with the majority of embryos exhibiting a class 1 phenotype (78/153, 51%) ( Fig. 3M ), characterized by an embryo that had only progressed to 90-95% epiboly when wild-type embryos had reached 100% epiboly. In contrast to this, the vast majority of embryos injected with a moderate dose of Con MO1 displayed normal epiboly phenotypes, as did noninjected controls (77/95, 82% and 130/136, 96%, respectively), suggesting that MO injection itself does not lead to adverse epiboly effects ( Fig. 3L ). Those rare embryos in these two groups that showed an epiboly-delay phenotype could invariably be characterized as a class 1 phenotype ( Fig. 3M ). Overall, the severity of the epibolydelay phenotype was proportional to the amount of MO1 injected ( Fig. 3Q ).
To confi rm that MO1 was targeting Npc1 and to examine the conservation of action of Npc1 across species, RNA rescue experiments were performed using mouse Npc1 (m Npc1 ) RNA. m Npc1 mRNA cannot be recognized by MO1 due to sequence variation between the fi rst exon of zebrafi sh and mouse Npc1 gene transcripts. Embryos injected with 200-400 pg m Npc1 mRNA alone were indistinguishable from controls throughout all stages of epiboly, indicating that overexpression of Npc1 had no apparent impact on epiboly progression ( Fig. 4H ). However, when 400 pg of m Npc1 mRNA was co-injected with 5 ng MO1, the negative effects of MO1 on epiboly progression were greatly reduced ( Fig. 4I, J ). To quantify the extent of epiboly rescue in npc1 morphants following m Npc1 mRNA injection, we stained embryos with no-tail ( ntl ), which is localized within the blastoderm margin. By marking the blastoderm margin we could measure the extent of blastoderm cell migration over the ventral yolk (see Experimental Procedures about how epiboly was measured in this study). To initiate the study, noninjected controls or sibling embryos that were injected with either MO1, m Npc1 mRNA, or both were fi xed uniformly (time-matched) once the control siblings had progressed to ‫ف‬ 65-85% epiboly and the percentage of epiboly progression for each group was measured. We found that epiboly progression was not signifi cantly altered between controls (epiboly progression of 69.1% ± 4.8%, n = 22) and their siblings injected with 200 pg of m Npc1 mRNA (epiboly progression of 67.1% ± 8.2%, n = 12) or 400 pg of m Npc1 mRNA (epiboly progression of 69.4% ± 4.5%, n = 10) ( Fig. 4A-C ). In contrast, siblings receiving a moderate dose of MO1 had signifi cantly delayed epiboly progression (epiboly progression of 47.5% ± 7.6, n = 19) compared with controls ( Fig. 4D ), deposited. During epiboly stages, npc1 was ubiquitously expressed throughout the blastomeres extending from the animal to the vegetal pole ( Fig. 2C-F ). Sectioning stained embryos revealed that npc1 was strongly expressed in the extra-embryonic YSL ( Fig. 2E, E ′ ; arrowheads). As embryos ended epiboly and began somitogenesis, npc1 expression was expressed ubiquitously at a low level, yet the expression levels were more intense in tissues bordering the yolk ( Fig. 2G ). npc1 expression became intense in anterior regions of the embryo proper by 24 hpf ( Fig. 2H ), wherein the staining was strongest in neural tissues ( Fig. 2H ′ ). Importantly, background signal with npc1 sense RNA probe was minimal at all stages tested, which included 24 hpf ( Fig. 2I ). This expression data revealed that npc1 is expressed dynamically throughout early development in embryonic and extra-embryonic tissues.

Morpholino knockdown of npc1 causes epiboly delay
To examine the function of npc1 during embryonic development, we continued our gene knockdown studies using morpholino oligonucleotides. We injected two separate function-blocking MOs, one designed to interrupt protein translation (MO1) and the other to reduce Npc1 protein levels by altering RNA splicing, specifi cally to remove the SSD (MO2) (see Experimental Procedures for further details). Injection of either MO at the one-or two-cell stage of embryonic development led to a reduced progression of epiboly.
To describe the dose-dependent epiboly-delay effects in npc1 morphants, MO1-and Con MO1-injected embryos were collected, and their epiboly delay was scored at the point at which noninjected controls had completed epiboly (approximately 10.25 hpf). If injected embryos had completed epiboly at the same time as noninjected controls, they were scored as normal; however, if the embryo had delayed epiboly progression, it was characterized as a class 1, 2, or 3 depending on the severity ( Fig. 3P ). At  ( Fig. 5A ), chosen because it contains SSD amino acids, a motif required for Npc1 function in higher vertebrates. While the exact function of the SSD motif in Npc1 is not fully understood, biochemical evidence has revealed that the domain is necessary for sterol traffi cking, possibly by directly binding cholesterol ( 65,66 ) Thus, this MO also tests the necessity of the SSD domain in zygotic transcripts of the Npc1 protein.
To determine the extent of MO2 activity, we developed primers complimentary to sequences contained in exon 12 and exon 14 that amplify 480 bp of the npc1 gene. PCR with these primers revealed the presence of smaller ( ‫ف‬ 300 bp) fragments in MO2-injected embryos ( Fig. 5B ) resulting from abnormal slicing. DNA sequencing of the gel bands confi rmed that the loss of exon 13 nucleotides (supplementary Fig. I), ‫ف‬ 180 bp in length, was the specifi c consequence of MO2. Perturbing normal RNA splicing could be achieved with a concentration of 2.5 ng MO2; increasing the concentration seemingly did not lead to an increased amount of mis-spliced product ( Fig. 5B ). Importantly, a which could be rescued upon co-injection with 400 pg m Npc1 (68.4% ± 7.4%, n = 11) ( Fig. 4F ), but not by 200 pg of m Npc1 (52.1% ± 7.2%, n = 15) ( Fig. 4E ). These results are summarized in Fig. 4G and represent the results of a single experiment, which was replicated twice ( Table 1 ). Taken together, these studies strongly suggest that MO1 produces defects, specifi cally by reducing Npc1, and that zebrafi sh Npc1 and mouse Npc1 share functionality. Furthermore, they show that Npc1 functions to infl uence epiboly progression.

The SSD domain of Npc1 is required for protein function during epiboly
Since npc1 is expressed as both maternal and zygotic transcripts (see Fig. 2 ), both may be involved in epiboly. Zygotic RNA is subject to processing events, including RNA splicing, while maternally deposited RNA transcripts are already spliced, allowing exclusive reduction of zygotic npc1 expression by utilizing a morpholino designed to perturb npc1 RNA splicing (MO2). MO2 was designed to target  compared with sibling embryos injected with Con MO2 (28/98, 29% with epiboly-delay phenotypes) or noninjected controls (17/169, 10% with epiboly-delay phenotypes) ( Fig. 5C ). Epiboly delay in MO2-injected embryos was most commonly characterized as a class 1 phenotype (132/200, 66%), with a smaller percentage showing a class 2 phenotype (10/200, 5%) (refer to Fig. 3M-O for representative images of class 1-3 phenotypes). Moreover, epiboly-delay phenotypes in control embryos were invariably class 1 ( Figs. 3M and 5C ). These data suggest that zygotic npc1 is essential for normal epiboly movements and points to an important role for the SSD, which is specifically disrupted by MO2, in protein function during epiboly.

The role of npc1 in the YSL
The YSL is one critical location of early npc1 expression ( Fig. 2 ). It has long been recognized that the YSL is required for teleost epiboly movements ( 38 ), although the molecular mechanisms behind this requirement is poorly understood. To examine the functional contributions of YSL-localized npc1 to epiboly progression, we specifi cally reduced YSL-localized Npc1 by co-injecting 5.0 ng/embryo MO1 and 0.1% rhodamine dextran directly into the YSL at the 1,000-cell stage. When viewing these embryos at 30% epiboly, the distribution of the dextran solution was limited to the yolk and YSL, and we only selected those embryos with YSL-limited injections for our analysis ( Fig.  6C , D ). This is in contrast to all other injections in this 5-basepair-mismatch control morpholino for MO2 (Con MO2) had no effect on splicing when it was injected into fi sh at low or moderate concentrations ( Fig. 5B ).
In a manner similar to MO1, animals injected with 5.0 ng/embryo MO2 displayed a higher incidence of epiboly delay (142/200, 71% with epiboly-delay phenotypes),  to develop normally until 3 hpf/1,000-cell stage, at which point m Npc1 mRNA or control mRNA encoding the Tol2 transposase was injected into the YSL ( Fig. 6F ). Control mRNA injections had little impact on epiboly progression; however, YSL-targeted m Npc1 mRNA microinjections were capable of rescuing a majority of the epiboly-delay phenotypes evident in embryos injected at the one-cell stage with MO1 ( Fig. 6F ). In contrast, MO1-injected embryos receiving control mRNA into the YSL showed high incidence of epiboly delay, consistent with the epiboly delay seen in nonrescued MO1-injected embryos ( Fig. 6F , compare with Fig. 3Q ). In full, these data strongly imply that the YSL is a critical site of action for Npc1 to promote normal epiboly progression during early embryonic development.
Since directed injection of MO into the yolk cell resulted in epiboly delay, we next sought to determine if epiboly delay in npc1 morphants injected with MO1 at the one-cell stage could be rescued by microinjection of m Npc1 mRNA into the YSL. Sibling embryos were injected with MO1 or Con MO1 at the one-cell stage and then allowed  Fig. 4G . C, clutch; N, number of embryos analyzed per group.
a Posthoc analysis using Tukey's HSD indicated that these groups had signifi cantly reduced ( P < 0.001 in all cases) epiboly progression compared with control embryos. b Posthoc analysis using Tukey's HSD indicated that these groups had no signifi cant difference when compared with controls (C1: P = 1.00; C2: P = 0.993; C3: P = 0.309).
d Posthoc analysis using Tukey's HSD indicated that these groups were not different from control embryos ( P values between 0.116 and 1.0), indicating that injection alone did not result in a change in epiboly progression. by dapi staining were associated appropriately with the epiboly leading edge (data not shown). These results suggest that, in the case of npc1 morphants, the EVL integrity is maintained and all cell layers show roughly equivalent epiboly delay.

Disruption of Npc1 function results in actin cytoskeleton disorganization
Epiboly defects are often associated with disruption to the actin cytoskeleton ( 47,72 ), and examination of the actin cytoskeleton can reveal alterations in cell shape that contribute to epiboly delay ( 47 ). To further investigate the loss of npc1 and its possible effect on epiboly, we visualized the actin cytoskeleton with Alexa Fluor-labeled phalloidin in control embryos for comparison with embryos injected with a low dose (3 ng/embryo) of MO1. In initial experiments, we collected both control and morphant embryos when they had reached 90-100% epiboly. To do this, we allowed our morphant embryos to continue to develop after collection of the control embryos, making them stagematched. Upon examination of these embryos, we found that the actin cytoskeleton was disorganized in the npc1morphant embryos, with both actin clumping as well as localized loss of the actin microfi laments in other cells ( Fig. 7C , C ′ ; n = 32). In contrast, this phenotype was not seen with control MO injections ( Fig. 7B, B ′ ). Moreover, the leading edge of the actin, which forms a contractile embryos. Since any changes to cell and germ layer induction would be apparent by 70-80% epiboly, we allowed npc1 -morphant embryos to reach this developmental stage prior to fi xation and analysis. At 75% epiboly, ntl normally marks the mesoderm and dorsal forerunner cells ( Fig.  4A ). We found that expression of ntl was identical between embryos injected with MO1 at the one-cell stage and their stage-matched, noninjected siblings ( Fig. 4D, D ′ ). Next, we found the expression of the mesendodermal marker gsc to be indistinguishable between MO1-injected embryos and their stage matched controls (supplementary Fig. IIA-C). Moreover, we found that expression of the early endodermal marker sox32/cas was indistinguishable between npc1 morphants and stage-matched controls (supplementary Fig. IID-F). These results strongly suggest that mesendoderm initiation and deep cell fates become specifi ed normally in npc1 morphants. During epiboly, the cell layers have characteristic movement behaviors, and these movements can be affected differently by different mutations or disruptions. In some cases, the EVL integrity is altered, which contributes to abnormal epiboly phenotypes ( 45,46,60 ). To examine the EVL in npc1 morphants and controls, a number of EVL differentiation markers were used. We found that the expression of EVL markers krt4 and krt18 were well maintained in MO1-injected embryos (supplementary Fig. IIG-N), as were the markers cldnE and krt8 (data not shown). Moreover, YSL nuclei marked  from cholesterol precursors by the mitochondrial biosynthesis enzyme cyp11a1 ( 25,75 ). Knockdown of cyp11a by MO in the zebrafi sh results in an epiboly-delay defect that can be partially rescued by incubating cyp11a morphants in P5 at early developmental stages prior to the doming of the yolk ( 37 ). We reasoned that if epiboly delay in npc1 morphants is due to reduced steroid hormone production, then treating npc1 morphants with P5 would promote faster rates of epiboly. Indeed, we found that incubating npc1 morphants in 20 or 30 M P5 solution, beginning at early cleavage stages (16-cell stage) and continuing throughout epiboly stages, partially rescued the epibolydelay defect. Fig. 8A shows results from single experiment, and Table 2 summarizes data from three replicate experiments. Concurrently, incubations of control embryos with P5 alone under identical conditions had no impact on epiboly progression ( Fig. 8A ).
Following the production of P5 from cholesterol, P5 becomes the steroidogenic precursor for glucocorticoids ( 76 ), which play a central role in vertebrate development (77)(78)(79). Zebrafi sh embryos express a glucocorticoid receptor at early stages and readily process a synthetic glucocorticoid, Dex, when delivered in the embryo water ( 79,80 ). This makes Dex an ideal candidate to determine if ectopic glucocorticoid treatment can reduce the epiboly delay seen in npc1 morphants. The npc1 morphants and control embryos were incubated in a 20 or 30 µM Dex solution in an identical fashion to the P5 treatments described above. Upon assessing epiboly progression among treated embryos, we found that 30 µM Dex solution led to a partial rescue of epiboly delay in npc1 morphants. Fig. 8B represents results from single experiment, and Table 3 summarizes three replicate experiments. Collectively, these fi ndings suggest that a reduction in steroidogenesis is at least partly responsible for the epiboly-delay phenotype in npc1 morphants.
In further experiments, we examined the potential of ectopic steroid hormone treatments to correct the actin cytoskeleton defects visualized in npc1 morphants. In these experiments, embryos were analyzed from a single clutch that was collected for analysis when the noninjected siblings reached 60-70% epiboly (supplementary Fig. IVA-C). Thus, unlike the previous actin fi gure ( Fig. 7 ), these embryos were time-matched. The actin cytoskeleton defects associated with MO1 injection could be partially rescued by incubating npc1 morphants in 40 µM Dex (supplementary Fig. IVB and C, n = 38 and 20, respectively) or 20 µM P5 (data not shown, n = 13). Collectively, these data provide evidence that the epiboly delay in npc1 morphants is at least partly due to changes in the actin cytoskeleton network. Our data further suggest that steroid hormones function to stabilize the actin cytoskeleton during epiboly in addition to stabilizing yolk cell microtubules during epiboly stages ( 76 ).

npc1 morphants that complete epiboly display a shorter body axis and neural cell death phenotypes
Most embryos receiving 2.5-5.0 ng/embryo of MO1 eventually completed epiboly. When we allowed npc1 mor-actin ring, was occasionally absent in the morphant embryos, a phenotype that was never evident in control embryos (data not shown). During normal epiboly, the leading cells undergo a distinctive cell shape change that can be seen when looking at the actin cytoskeleton. It has been demonstrated that in some cases where epiboly is disrupted, this cell shape change fails to occur ( 47 ). When we examined MO-injected embryos at different stages of epiboly, there was no difference in the cell shape of the cells on the leading edge of epiboly (data not shown). Actin clumping and disorganization is often a byproduct of cell death ( 73,74 ). Since increased cell death has been reported in many mammalian tissues in response to loss of npc1 function, we examined epiboly-staged embryos for evidence of cell death. When examining the deep cells (data not shown) and leading edge of epiboly-staged embryos, the number of cells with pyknotic nuclei was not signifi cantly increased between npc1 morphants and controls ( P < 0.05; supplementary Fig. III). By these data, we could conclude that reducing Npc1 in zebrafi sh embryos did not impact cell survival during epiboly, but it did lead to perturbations to the actin cytoskeleton network.

Treating npc1 morphants with steroid hormones P5 and Dex partially rescues epiboly delay
The synthesis of neurosteroids and steroid hormones is signifi cantly reduced in vertebrate and invertebrate animals lacking a functional Npc1 ( 21,23,26 ), likely due to reduced amounts of cholesterol available for steroid metabolism ( 23 ). The precursor for all steroid hormones in the zebrafi sh is pregnenolone (P5), which is metabolized somitic mesoderm in papc -stained npc1 morphants was clear by visual assessment, we quantifi ed the width of the notochord in npc1 morphants to verify that these embryos had ML expansion. To determine the extent of notochord widening in morphants, the notochord width was quantifi ed by measuring the span of the ntl -stained notochord. We found the notochord of npc1 morphants to be signifi cantly wider compared with noninjected, stage-matched siblings at six somites. Later in development, at nine somites, npc1 morphants and their control siblings that were deyolked and physically fl attened revealed the extent of the AP shortening and ML expansion in the morphant embryos ( Fig. 9I-L ). In these embryos, fkd6 staining marks anterior structures while ntl staining marks both the posterior tailbud and the somites, thus allowing for proper stage matching based on the number of somites expressing ntl ( Fig. 9I, K ). Comparison of these embryos showed that npc1 morphants were relatively well developed; however, they continued to be shorter ( Fig. 9I, K ) than control embryos. Moreover, npc1 morphants continued to display a signifi cantly widened notochord ( Fig. 9J, L ). These results suggest that npc1 may be required for proper gastrulation movements in cells undergoing convergence and extension during somitogenesis.
At later stages of somitogenesis (22 hpf), MO1-injected embryos remained shortened and exhibited dark staining phants to develop beyond epiboly, they consistently displayed a shorter and wider body axis. These morphant embryos invariably died prematurely, prior to 2 dpf. The severity of the phenotypes were morpholino dosedependent, and embryos receiving low doses of MO1 (2.5 ng/embryo) were consistently healthier and more viable than siblings receiving moderate MO1 doses (5.0 ng/embryo). For this reason, the remaining analyses were performed on morphants receiving low doses of MO1.
In npc1 morphants, cells in the anterior and posterior body axis migrated a shorter overall distance by the sixsomite stage compared with stage-matched control embryos ( Fig. 9A , E ). Moreover, npc1 morphants displayed a wider notochord and somites than their stage-matched controls ( Fig. 9B, F ). Analysis with various cell markers along the anterioposterior (AP) axis throughout somitogenesis supported earlier evidence (supplementary Fig.  II) that all germ layers became properly specifi ed in npc1 morphants. However, comparing expression patterns between npc1 morphants and wild-type, noninjected siblings suggested that gastrulation cell movements were negatively affected in morphant embryos. At the six-somite stage, npc1 morphants displayed laterally expanded midline mesoderm and presomitic mesoderm tissue, as evidenced by ntl and papc staining, respectively ( Fig. 9C, D, G, H ). While the mediolateral (ML) expansion of the pre-  a Posthoc analysis using Tukey's HSD indicated that these groups had signifi cantly reduced ( P < 0.001 in all cases) epiboly progression compared with control embryos. b Posthoc analysis using Tukey's HSD indicated that these groups were signifi cantly different to embryos injected with MO1 alone (C2: P < 0.001; C3: P = 0.006).  Fig. 8A . C, clutch; N, number of embryos analyzed per group.
a Posthoc analysis using Tukey's HSD indicated that these groups had signifi cantly reduced ( P < 0.001 in all cases) epiboly progression compared with control embryos.

DISCUSSION
Several lines of evidence suggest that the zebrafi sh npc1 gene identifi ed in this report is functionally similar to mammalian NPC1. First, the protein encoded by the zebrafi sh npc1 gene is a multi-transmembrane protein that is highly conserved when compared with NPC1 proteins from other species, including human. Analysis of the zebrafi sh Npc1 protein sequence revealed the strongest homology to other Npc1 species in known functional motifs of the protein, including the SSD motif, which encompasses transmembrane spanning domains 3-7, as well as the NpcL region that is present at the protein's N terminus and contains a leucine zipper motif ( 64 ). Second, zebrafi sh npc1 morphants exhibited phenotypes similar to those reported in other Npc1-defi cient model systems, including mislocalization of endogenous cholesterol within cells, reduced cell mobility, increased cell death, decreased in cells in anterior regions, characteristic of cell death ( Fig. 9M, P ). Cell death can be monitored in live fi sh using acridine orange (AO), a vital dye that stains dying cells, including apoptotic cells ( 81 ). AO staining indicated that knocking down npc1 led to increased cell death in embryos, becoming apparent near the end of the fi rst day of development ( Fig. 9N, O, Q, R ). In contrast to these fi ndings, control MO injections did not result in increased cell death ( Fig. 9 ), nor was increased cellular death seen in npc1 morphants at earlier stages of development during epiboly (supplementary Fig. III). Cell death in npc1 morphants was followed by a variety of abnormal phenotypes during the second day of development, with morphant embryos rarely surviving beyond 48 hpf. Because npc1 knockdown was lethal, even at low doses, in the early embryonic development of zebrafi sh, further investigations at later embryonic or larval stages was precluded. control additional cytoskeleton behaviors during epiboly. Along these lines, we found that Dex, a synthetic glucocorticoid, was capable of partially rescuing the epiboly-delay phenotype in npc1 morphants to a similar degree as was shown for phenotypic rescue with P5. Like humans, zebrafi sh express a single glucocorticoid receptor ( 83 ) likely to interact with endogenous cortisol, and in zebrafi sh this receptor is strongly expressed as a maternal and zygotic transcript during epiboly and gastrulation stages ( 84,85 ). In fact, the glucocorticoid receptor is the most highly expressed steroid receptor in zebrafi sh embryos, suggesting that glucocorticoids have an important role in embryogenesis ( 85 ); however, such a requirement has not been examined in zebrafi sh. This is the fi rst report, to our knowledge, that suggests a requirement for glucocorticoids or any P5 metabolite during epiboly. This is in contrast to progesterone, which was shown to worsen the epiboly-delay phenotype when added to zebrafi sh embryos with reduced steroidogenic enzyme activity ( 37 ). Considering that npc1 morphants would have reduced levels of both P5 and its metabolites (i.e., glucocorticoids), it is not surprising that these embryos experience epiboly-delay phenotypes as severe as cyp11a1 morphants.
Overall, npc1 -depleted embryos had a disorganized actin cytoskeleton network within the blastoderm cells. P5 or Dex treatments were suffi cient to partially rescue the changes to actin fi lament organization, leading us to suggest that npc1 likely regulates the cytoskeletal network either directly or indirectly in the early embryo. It is conceivable that P5 and/or its metabolites function to stabilize or maintain normal actin fi laments during epiboly, similar to the ascribed function for P5 on microtubule stabilization during epiboly ( 37 ); however, more work will be required to determine the exact nature of steroid regulation on the actin cytoskeleton network. Importantly, the disordered actin skeleton in npc1 morphants was not associated with any increases in cell death, at least during early epiboly stages.
npc1 -depleted embryos that received low to moderate doses (2.5 ng-5.0 ng/embryo) of MO1 often completed epiboly but were signifi cantly shorter, with wider somites and notochord tissues and a high degree of cell death, especially in anterior neural tissues. In spite of this, these morphant embryos showed essentially normal expression levels and appropriate relative localization of a number of differentiation markers that we examined, including gsc , foxd3 , sox32 , and ntl. These fi ndings suggest that Npc1 affects cell behavior but not cell specifi cation during gastrula and segmentation stages. In the future it would be interesting to study the specifi c cellular behaviors of npc1depleted cells to determine what role, if any, disrupting normal sterol-cellular biology has on in vivo cell behavior during early embryonic development. Moreover, it may be interesting to determine if earlier cell movement defects are present in other NP-C models.
This study demonstrates that zebrafi sh possess an ortholog of the mammalian Npc1 gene that encodes a transmembrane protein that displays functional conservation with higher vertebrate proteins. The npc1 gene is expressed lifespan, and epiboly delay, a phenotype indicative of reduced steroid hormone production ( 37,76 ). Thus, the severe phenotypic consequences seen in npc1 morphants likely refl ect the dual requirements for Npc1 function in cholesterol traffi cking and steroid production, reminiscent of fi ndings reported in mouse and fl y embryos that fi rst suggested a connection between NP-C disease and reduced steroid production. Finally, protein function conservation between zebrafi sh Npc1 and mammalian NPC1 was confi rmed by rescuing the primary developmental defect observed in zebrafi sh npc1 -morphant embryos with mouse Npc1 RNA.
To gain insight into the requirement for npc1 during embryonic development, we studied its gene expression pattern and examined the consequences of reducing protein function in embryos injected with morpholinos. Zebrafi sh npc1 was present from fertilization up to 7 dpf. During gastrulation, we found that zebrafi sh npc1 was expressed abundantly, specifi cally in the extra-embryonic YSL cells, but it was also present in the embryonic cells. Reducing Npc1 protein levels in the whole embryo by a protein translation-blocking MO (MO1), as well as in YSL cells only, impacted the normal epiboly movements. In npc1 morphants, we observed that epiboly progression of the YSL and deep blastodermal cells was equally delayed during epiboly stages; however, involution of marginal cells occurred normally as evidenced by gsc staining during late epiboly stages. Importantly, adding m Npc1 mRNA to the YSL cells in embryos where Npc1 protein levels were reduced throughout the embryo restored epiboly progression, suggesting that YSL-localized Npc1 protein is critical for normal epiboly movements. Finally, a second MO (MO2), which acts by altering normal RNA splicing, also led to slower rates of epiboly upon microinjection at the one-cell stage, although its effect on epiboly progression was not as strong MO1. This may be due to maternal stores of npc1 , which bypass splicing, as well as incomplete levels of splicing by MO2.
The npc1 gene expression pattern during epiboly stages is similar to the expression patterns of two crucial steroidogenic enzymes, cyp11a1 and hsd3b , which, respectively, catalyze cholesterol to P5 before converting P5 to subsequent steroid hormones ( 36,76 ). Like npc1 morphants, the primary developmental defect in cyp11a1depleted embryos is a slower rate of epiboly ( 37 ). In either morphant embryo, adding exogenous P5 was suffi cient to partially rescue the epiboly delay, indicating that the epiboly defi cit in npc1 morphants may be at least partly due to reduced steroid levels. Zebrafi sh epiboly is driven by yolk microtubules and the forces exerted by the actin cytoskeleton. Disrupting either actin or microtubule fi laments in zebrafi sh embryos signifi cantly delays or arrests epiboly progression ( 63,82 ). Since hsd3b -depleted embryos, which cannot metabolize P5, display normal epiboly movements, it has been suggested that P5 alone infl uences proper epiboly function and its metabolites do not play a suffi cient role in promoting epiboly ( 37 ). However, these fi ndings do not exclude two further possibilities: i ) that P5 metabolites share a redundant function with P5 in promoting epiboly movements and ii ) that P5 and/or its metabolites may in early embryonic development and is essential for normal morphogenetic movements at these stages. One hallmark of NP-C disease in human patients is a progressive degeneration of brain function due to cholesterol accumulation in neural tissues. Although early developmental defects made it harder to assess the role of npc1 in central nervous system (CNS) development, it was possible to see increased cell death in the CNS of embryos treated with a low dose of morpholino. While the CNS cell death phenotype in zebrafi sh is reminiscent of the human disease condition as well as mouse models for the disease, at present it is unclear how much of the early developmental defects in npc1 morphants contribute to later CNS defects. It would be interesting to study CNS development at later stages (1 dpf or older) in npc1 morphants that had received doses of fi sh or mammal Npc1 genes at early stages to bypass the initial defects associated with Npc1 functional loss during epiboly. The early defects found in the zebrafi sh happen in less than 12 h. We have established that drug treatment, in this case P5 and Dex, can reduce the severity of this early Npc1-related phenotype. Finally, since mouse Npc1 can functionally rescue the morphant embryos, taken together our fi ndings suggest that the zebrafi sh model system might be an appropriate venue for rapid testing of potential therapeutics and for distinguishing between cholesterol accumulation and metabolism defects as casual agents in the debilitating outcomes associated with NP-C.