Lack of de novo phosphatidylinositol synthesis leads to endoplasmic reticulum stress and hepatic steatosis in cdipt-deficient zebrafish


  • Prakash C. Thakur,

    1. Division of Hematology/Oncology, Department of Medicine, University of Pittsburgh, Pittsburgh, PA
    Search for more papers by this author
    • Supported by a Department of Veterans Affairs Senior Research Career Scientist Award (to J. K. Y.), Cancer Center Support Grant P30CA047904, and National Institutes of Health Grant R21DK073177 (to N.B.).

  • Carsten Stuckenholz,

    1. Division of Hematology/Oncology, Department of Medicine, University of Pittsburgh, Pittsburgh, PA
    Search for more papers by this author
  • Marcus R. Rivera,

    1. Division of Pediatric Gastroenterology, Childrens Hospital, University of Pittsburgh, Pittsburgh, PA
    Search for more papers by this author
  • Jon M. Davison,

    1. Department of Pathology, University of Pittsburgh, Pittsburgh, PA
    Search for more papers by this author
  • Jeffrey K. Yao,

    1. Medical Research Service, VA Pittsburgh Healthcare System, and Departments of Psychiatry and Pharmaceutical Sciences, Pittsburgh, PA
    Search for more papers by this author
  • Adam Amsterdam,

    1. David H. Koch Institute for Integrative Cancer Research, Massachusetts Institute of Technology, Cambridge, MA
    Search for more papers by this author
  • Kirsten C. Sadler,

    1. Department of Medicine-Liver Diseases, Mount Sinai School of Medicine, New York, NY
    Search for more papers by this author
  • Nathan Bahary

    Corresponding author
    1. Division of Hematology/Oncology, Department of Medicine, University of Pittsburgh, Pittsburgh, PA
    2. Department of Microbiology and Molecular Genetics, University of Pittsburgh School of Medicine, Pittsburgh, PA
    • Division of Hematology/Oncology, Department of Medicine, Biomedical Science Tower 3, Room 5058, 3501 Fifth Avenue, Pittsburgh, PA 15260
    Search for more papers by this author
    • fax: 412-648-9852


Hepatic steatosis is the initial stage of nonalcoholic fatty liver disease (NAFLD) and may predispose to more severe hepatic disease, including hepatocellular carcinoma. Endoplasmic reticulum (ER) stress has been recently implicated as a novel mechanism that may lead to NAFLD, although the genetic factors invoking ER stress are largely unknown. During a screen for liver defects from a zebrafish insertional mutant library, we isolated the mutant cdipthi559Tg/+ (hi559). CDIPT is known to play an indispensable role in phosphatidylinositol (PtdIns) synthesis. Here we show that cdipt is expressed in the developing liver, and its disruption in hi559 mutants abrogates de novo PtdIns synthesis, resulting in hepatomegaly at 5 days postfertilization. The hi559 hepatocytes display features of NAFLD, including macrovesicular steatosis, ballooning, and necroapoptosis. Gene set enrichment of microarray profiling revealed significant enrichment of endoplasmic reticulum stress response (ERSR) genes in hi559 mutants. ER stress markers, including atf6, hspa5, calr, and xbp1, are selectively up-regulated in the mutant liver. The hi559 expression profile showed significant overlap with that of mammalian hepatic ER stress and NAFLD. Ultrastructurally, the hi559 hepatocytes display marked disruption of ER architecture with hallmarks of chronic unresolved ER stress. Induction of ER stress by tunicamycin in wild-type larvae results in a fatty liver similar to hi559, suggesting that ER stress could be a fundamental mechanism contributing to hepatic steatosis. Conclusion: cdipt-deficient zebrafish exhibit hepatic ER stress and NAFLD pathologies, implicating a novel link between PtdIns, ER stress, and steatosis. The tractability of hi559 mutant provides a valuable tool to dissect ERSR components, their contribution to molecular pathogenesis, and evaluation of novel therapeutics of NAFLD. (HEPATOLOGY 2011;)

Nonalcoholic fatty liver disease (NAFLD), one of the most common causes of chronic liver disease, represents a spectrum of liver disorders extending from simple hepatic steatosis to steatohepatitis, cirrhosis, and fibrosis in the absence of significant alcohol abuse.1, 2 Although this disease is highly prevalent, its molecular pathogenesis is poorly understood, hindering the development of effective therapeutics. Hepatic steatosis is believed to be the initial stage that progresses to a more severe form of NAFLD. Currently, there is a lack of genetic models to investigate molecular mechanisms of hepatic steatosis. In this study, we present a zebrafish model to identify the potential mechanisms of hepatic steatosis.

Zebrafish are an elegant genetic model for identifying genes and elucidating molecular pathways critical to development and disease of the digestive system. Zebrafish gastrointestinal (GI) tissues share striking similarities in anatomy, cellular composition, and function with their mammalian counterparts.3, 4 Gene expression profiles and active pathways during zebrafish GI development are also analogous to those observed in mammalian GI development and cancer.5 Furthermore, prominent histopathological similarities are seen between zebrafish and mammalian GI diseases such as fatty liver, cholestasis, and neoplasia.6, 7

An insertional zebrafish mutant library has been established,8 allowing identification of genes with a role in liver development and establishment of novel models of liver diseases.7 Here, we provide molecular characterization of the insertional mutant cdipthi559Tg/+ (hi559), which displays striking liver defects at 5 days postfertilization (dpf) and subsequent death beginning at 6.5 dpf. The mutated gene responsible for the hepatic phenotype is CDIPT (CDP-diacylglycerol- inositol 3-phosphatidyltransferase), also known as phosphatidylinositol synthase (PIS). CDIPT is a highly conserved integral membrane protein found on the cytoplasmic side of the endoplasmic reticulum (ER) and has an indispensable role in the synthesis of a critical phospholipid, phosphatidylinositol (PtdIns).9 Phosphorylated derivatives of PtdIns, known as phosphoinositides (PIs), are crucial regulators of calcium homeostasis, membrane trafficking, secretory pathways, and signal transduction. Formation and turnover of PIs are catalyzed by evolutionarily conserved families of PI kinases and phosphatases.10, 11 Improper function of several of these metabolic enzymes is associated with both benign and malignant human diseases.12, 13 We recently reported that inositol metabolism and PI3-kinase signaling pathways were enriched in the developing liver, and inhibition of PI3-kinase pathway resulted in hepatic abnormalities.5

As an integral component of the ER, PtdIns and the PI signaling components are crucial for ER and its secretory functions.14 Transmembrane, organellar, and secreted proteins are folded and modified in the ER and exit by vesicular transport. Perturbations of ER homeostasis such as elevated secretory protein synthesis and accumulation, glucose deprivation, and ER calcium depletion can cause ER stress, triggering an evolutionarily conserved response, termed the endoplasmic reticulum stress response (ERSR) or unfolded protein response (UPR).15 ER stress has been associated with a wide range of diseases, including neurodegeneration, cardiac diseases, cancer, and diabetes.16, 17 Secretory cells such as hepatocytes process large amounts of protein in their ER and hence are vulnerable to ER stress–associated pathology. Hepatocellular ER stress is believed to contribute to insulin resistance in diabetes and obesity, liver disorders such as α1-antitrypsin deficiency, and NAFLD.18 Additionally, increased expression of ER stress–related genes was recently reported in hepatocellular carcinoma.19 Although the precise molecular pathways leading to ER stress in these diseases are largely unknown, components of PI signaling play pivotal roles in vesicular trafficking at ER exit sites, suggesting that abnormal PI signaling may cause disruption of ER and subsequent pathologies.20

Analyses of hi559 larvae reveal that a lack of de novo PtdIns synthesis causes severe disruption of the ER architecture in hepatocytes with ultrastructural pathology indicating excessive ER stress and hepatic steatosis. The ER stress and cytopathologies seen in the hi559 liver resemble those seen in human NAFLD. Furthermore, Gene Set Enrichment Analysis (GSEA) of microarray data identified selective enrichment of genes involved in ERSR pathway in hi559 larvae; several of these genes are selectively overexpressed in the mutant liver. Together, these data support a model in which disrupted PtdIns synthesis leads to ER stress–mediated intracellular damage resulting in hepatic pathology similar to that seen in NAFLD.


CDIPT, CDP-diacylglycerol-inositol 3-phosphatidyltransferase; dpf, days postfertilization; ER, endoplasmic reticulum; ERSR, endoplasmic reticulum stress response; GI, gastrointestinal; GSEA, Gene Set Enrichment Analysis; ISH, in situ hybridization; mRNA, messenger RNA; NAFLD, nonalcoholic fatty liver disease; ORO, Oil Red O; PCR, polymerase chain reaction; PI, phosphoinositides; PIS, phosphatidylinositol synthase; PtdIns, phosphatidylinositol; RT-PCR, reverse-transcription PCR; UPR, unfolded protein response.

Materials and Methods

Zebrafish Mutant.

The zebrafish line hi559 was obtained from a large-scale insertional mutagenesis screen.8 All fish husbandry was performed in accordance with local institutional animal care and use committee protocols. Heterozygous and homozygous fish were confirmed by way of genotyping using multiplex polymerase chain reaction (PCR).

Whole-Mount Staining.

CY3-streptavadin (CY3-SA) labeling was performed as illustrated previously.7 For whole-mount in situ hybridization (ISH), embryos were processed as described.5 Probes and their corresponding accession numbers are provided in the Supporting Information. Alkaline phosphatase staining for vasculature and whole-mount Oil Red O (ORO) staining were performed as described.21, 22

Reverse-Transcription PCR, Morpholino Knockdown, and Messenger RNA Rescue.

Total RNA was extracted from 5-dpf wild-type and hi559 larvae using RNAeasy (Qiagen). Oligo dT–primed complementary DNA was then synthesized using SuperScript II RT (Invitrogen) and probed by way of reverse-transcription PCR (RT-PCR). cdipt mRNA was synthesized from a full-length linear DNA template using mMessage (Ambion) and purified by RNA clean (Zymo Research). cdipt and gfp mRNAs were injected into 1-cell stage embryos. For knockdown analyses of Cdipt, two zebrafish cdipt splice-blocking morpholinos were coinjected with tp53 morpholino into wild-type embryos. tp53 morpholino alone was injected as a control morpholino. See Supporting Information for primer and morpholino sequences.

Radioactive PIS Assay.

The PIS assay was performed essentially as described.23, 24 The assay was conducted in 100 μL total volume containing 0.2 mM CDP-DAG, 0.5 mM myo-[3H]inositol (5,000 cpm/nmol), 2 mM MnCl2, 50 mM Tris-Hcl (pH 8.0), 0.15% Triton X-100, and 50 μg of total protein isolated either from wild-type or hi559 larvae. After 1 hour of incubation at 37°C, the reaction was terminated by adding 0.35 mL chloroform and 0.5 mL 1 M MgCl2. The organic phase was separated for lipid extraction. PIS activity was measured as amount of myo-[3H]inositol incorporation into PtdIns per milligram of protein as determined by scintillation counting.

Histology and Transmission Electron Microscopy.

Wild-type and hi559 larvae were fixed in 4% paraformaldehyde/phosphate-buffered saline at 4°C overnight, dehydrated with ethanol, and embedded in JB-4 (Polysciences). Serial sagittal and transverse sections (4 μm) were stained with hematoxylin and eosin. For semithin sections, epoxy resin–embedded embryos were sectioned (20 nm) and stained with Toluedene blue. For lipid staining, freshly collected embryos were embedded in OCT (Tissue-Tek), frozen in liquid nitrogen, sectioned (5 μm) using a cryostat at −20°C, and stained with ORO. Sectioning and transmission electron microscopy imaging was performed by the Renal Pathology Service at the University of Pittsburgh Medical Center (Pittsburgh, PA). See Supporting Information for further details.

Microarray Analyses.

Total RNA was extracted from three samples each of 5-dpf wild-type and mutant larvae (n = 25) using RNAeasy (Qiagen). Hybridization of Affymetrix GeneChips, microarray data collection, and analyses were performed as described using Ingenuity's pathway analysis (http://www. and GSEA (, 25 Microarray data have been deposited with Gene Expression Omnibus (GSE17711).


hi559 Mutants Exhibit Defects in Liver Development.

Heterozygous hi559 carriers were phenotypically indistinguishable from their wild-type siblings; the hi559 phenotype was completely penetrant in homozygotes. The hi559 embryos hatched and were phenotypically normal until 5 dpf, when homozygous hi559 larvae became easily distinguishable from wild-type siblings by a globular (abnormally shaped), darkish liver, as seen on bright-field microscopy and CY3-SA labeling (Fig. 1A-C). hi559 larvae also displayed a smaller intestine and slightly smaller eyes. The pancreas did not exhibit any noticeable defects (Supporting Fig. 1). hi559 larvae began to die around 6.5 dpf.

Figure 1.

hi559 larvae exhibit defects in liver development. (A) Larval morphology at 5 dpf (top, wild-type; bottom, mutant). (B) hi559 larvae show globular liver (yellow outline) and smaller intestine (red dotted line). (C) Globular liver is apparent on CY3-SA labeling (arrows). (D-F) Lateral view of ISH at 5 dpf showing expression of the liver-specific markers sepp1b (D), cp (E), and fabp10a (F). gb, gas-bladder; ib; intestinal bulb; L, liver; y, yolk.

To analyze developmental abnormalities in the liver, we characterized the 5-dpf hi559 larvae by way of ISH using RNA probes against three liver-specific transcripts: sepp1b, cp, and fabp10a (Fig. 1D-F). Although their expression appeared similarly intense in wild-type and hi559 larvae, the abnormal shape of the liver was apparent. We did not notice any difference in expression of the liver markers in clutches of embryos between 2 and 4 dpf (data not shown), indicating that there were no overt defects in liver formation at early stages. We observed no noticeable differences in the expression of markers specific to exocrine (try) and endocrine (ins) pancreas (Supporting Fig. 1A). The defects in hi559 liver at 5 dpf suggest an important role of the wild-type gene product in hepatic development and function.

Retroviral Insertion Disrupts Cdipt Expression.

Using inverse PCR, the retroviral insert was mapped to the first intron (35 nucleotides past the first exon) of cdipt (Fig. 2A).8 RT-PCR results revealed lack of cdipt mRNA in hi559, but products of expected sizes in wild-type siblings (Fig. 2B). Bioinformatic analyses of the zebrafish genome (Build 7, Ensembl) did not reveal a second copy of cdipt. To investigate whether knockdown of Cdipt can replicate the hi559 phenotype, we injected two different splice-blocking morpholinos against cdipt into wild-type embryos. To reduce nonspecific effects due to activation of tp53, we coinjected tp53 morpholino.26 Approximately 60% (53/90) of the injected embryos displayed hepatic abnormalities similar to hi559 (Supporting Fig. 2A-D). Injection of 100 pg of cdipt mRNA resulted in significant reduction of larvae with hepatic phenotype from a clutch, rescuing presumably mutant embryos (Supporting Fig. 2E). In a clutch (n = 147) of cdipt mRNA–injected embryos, only 5 (3.4%, expected 25%) displayed the typical hi559 phenotype at 5 dpf; all remaining larvae appeared normal. In the control group (injected with gfp mRNA), ≈25% (19/70) showed the typical hi559 phenotype. These results strongly suggest that the hi559 phenotype is due to loss of Cdipt function.

Figure 2.

Disruption of cdipt expression and PtdIns synthesis in hi559 larvae. (A) The retroviral insertion (black triangle) was mapped to the first intron of the cdipt gene.8 White boxes represent exons; gray boxes represent introns. (B) cdipt expression is disrupted in hi559 larvae (RT-PCR amplification of three different regions, indicated in A). (C-E) Developmental expression pattern of cdipt in wild-type embryos. Note the absence of cdipt expression in hi559 larvae (E, left: wild-type; E, right: mutant). Arrows indicate intestine; arrowheads indicate liver. (F) PtdIns synthesis is disrupted in hi559 larvae. The bar chart represents values from three biological replicates. Abbreviation: wt, wild-type.

Cdipt Is Expressed in the Liver During Development.

At 24 hours postfertilization, cdipt mRNA expression was ubiquitous, but became restricted to the developing liver and intestine by 48 hours postfertilization and remained high in these tissues through 5 dpf (Fig. 2C-E). Cdipt was also expressed in the brain, retina, and branchial arches throughout development. As expected, hi559 embryos lacked cdipt expression (Fig. 2E [right] and Supporting Fig. 3).

hi559 Larvae Are Deficient in De Novo PtdIns Synthesis.

Because CDIPT plays a critical role in PtdIns synthesis, we wanted to confirm alteration of PtdIns levels in the absence of cdipt expression. Surprisingly, comparative phopholipid profile by way of thin layer chromatography revealed that levels of PtdIns and other phospholipids of deyolked wild-type and hi559 larvae were similar at 5 dpf (Supporting Fig. 4). However, we noticed that the embryonic yolk at 1-cell stage contained abundant PtdIns, suggesting that it was maternally deposited. We reasoned that de novo PtdIns synthesis might be disrupted in hi559 embryos and tested PIS activity in 5-dpf wild-type and mutant larvae. PIS activity was negligible in hi559, but robust in wild-type siblings (Fig. 2F). To further confirm that chemical inhibition of PIS replicates the hi559 phenotype, we treated wild-type larvae with δ-hexachlorocyclohexane, a drug with the same configuration as myo-inositol, known to inhibit myo-inositol incorporation into PtdIns.27 Treatment with δ-hexachlorocyclohexane resulted in hepatomegaly and a darkish liver similar to hi559 (Supporting Fig. 5). These results suggest that maternally deposited and de novo–synthesized PtdIns are not functionally equivalent, and that de novo synthesis of PtdIns is required for normal hepatic development.

hi559 Livers Display Features of NAFLD.

In sagittal sections, the hi559 liver appeared swollen and vacuolar, with enlarged hepatocytes and increased internuclear distance between adjacent hepatocytes compared with wild-type liver (Fig. 3A). Toluedene blue staining of semithin transverse sections revealed marginalization of nuclei with rarefied cytoplasm and vesicles often filled with dense material, suggestive of fat accumulation (Fig. 3B). Whole-mount ORO staining revealed that hi559 larvae have fatty livers (Fig. 3C). ORO staining of frozen histological sections revealed substantial fat accumulation in the form of small and large lipid droplets in hi559 hepatocytes (Fig. 3D). At the onset of NAFLD at 5 dpf, the hi559 liver displayed admixture of normal hepatocytes and foci of microvesicular and macrovesicular steatosis, without apparent necrosis (Fig. 4B,E). However, with progression of NAFLD, most hepatocytes exhibited severe macrovesicular steatosis, and some displayed fragmented nuclei (Fig. 4C,D). In some cases of severely steatotic liver, hepatic sinusoids appeared smaller (Supporting Fig. 6). The distortion of the sinusoid architecture may be attributed to the grossly enlarged hepatocyte plates compressing the adjacent sinusoids. Hepatocellular injury in the form of ballooning degeneration, apoptosis, and necrotic foci were prominent in hi559 liver by 6 dpf (Fig. 4D,F). The ballooned hepatocytes often have rarefied cytoplasm containing perinuclear hyaline inclusion bodies; therefore, many of the characteristic histological features of NAFLD, such as enlarged hepatocytes, cytoplasmic clearing, accumulation of small and large membrane-bound lipid, and subsequent necrosis are observed in hi559 livers.1, 28 Despite the severe hepatic histopathology reminiscent of NAFLD, inflammation was not conspicuous in hi559 livers at the histological level, although we noticed the presence of macrophages adjacent to the necrotic hepatocytes ultrastructurally, indicating mild inflammation. The paucity of inflammation may be attributed to an incompletely matured zebrafish immune system at this stage of larval development.

Figure 3.

Histopathological abnormalities of hi559 liver. (A) Hematoxylin and eosin–stained sagittal sections of 5-dpf wild-type and hi559 liver. Mutant liver architecture is abnormal with vacuolar appearance and smaller sinuses (asterisk). Scale bars: 20 μm. (B) Toluedene blue staining of semithin transverse sections reveals large, vacuolated hepatocytes with marginalized nuclei (arrows) in hi559 liver. Scale bars: 50 μm. (C) Whole-mount ORO staining of 5-dpf larvae shows fatty liver (black dotted line) in the mutant. (D) ORO staining of frozen tissue sections of 5-dpf larvae reveals substantial steatosis in hi559 hepatocytes. Arrows indicate lipid droplets (red); arrowheads indicate nuclei. Scale bars: 20 μm. ib, intestinal bulb; in, intestine; L, liver; wt, wild-type; y, yolk.

Figure 4.

NAFLD progression in hi559 liver. (A-D) Hematoxylin and eosin–stained sections of wild-type liver at 6 dpf (A) and stages of NAFLD progression in hi559 liver at 5 dpf (B), 5.5 dpf (C), and 6 dpf (D). At early stage of NAFLD at 5 dpf, foci of microvesicular (arrows) and macrovesicular (arrowheads) steatosis are evident with no apparent apoptosis. At 5.5 dpf, most of the hepatocytes exhibit macrovesicular steatosis with a few apoptotic hepatocytes (arrows; the area within the yellow dotted box is magnified in the inset). At 6 dpf, hepatocellular ballooning (arrows), often with perinuclear hyaline inclusions (arrowheads), and necrotic foci (black dotted box) are apparent. (E) Bar chart showing percentages of larvae with hepatic steatosis at 5 dpf. (F) Bar chart showing percentages of apoptotic hepatocytes in 6-dpf liver. Data are representative of five biological replicates. Scale bars: 20 μm. wt/WT, wild-type

Microarray Analyses Reveal Up-regulation of ERSR in hi559 Larvae.

We performed Affymetrix array analyses to decipher dysregulated pathways and gene networks associated with the hi559 phenotype. Our analyses revealed a set of 465 genes that were significantly differentially regulated (P < 0.05) in hi559 compared with wild-type siblings at 5 dpf, 186 of which are up-regulated. GSEA revealed enrichment of a set of genes involved in ERSR/UPR (Fig. 5A,B). We noticed significant up-regulation of critical ERSR indicators in the mutant. Many of these genes encode ER resident proteins that collectively take part in UPR or in Ca2+ homeostasis, including calr, hspa5/bip/grp78, hsp90b1/grp94, caln, and atf6 (Fig. 5). We subsequently compared our gene expression profile with a previously published gene set on hepatic ER stress in mice29 and found a significant overlap between the two (Fig. 5C,D). Ingenuity's pathway analysis identified acute phase response signaling as the top-most up-regulated canonical pathways, suggesting activated transcription of immune/inflammatory response factors in hi559 larvae (data not shown). Overlaying gene expression values onto the ERSR pathway generated by Ingenuity's pathway analysis showed transcriptional up-regulation of ERSR components and ER stress-induced apoptotic signals (Supporting Fig. 7). To validate our microarray data, we analyzed the expression patterns of a set of ERSR markers by ISH. Interestingly, these genes are selectively overexpressed in hi559 liver (Fig. 6A-F). Together, these data implicate lack of PtdIns synthesis in leading to hepatocellular ER stress, causing the hepatic pathology in hi559 larvae.6

Figure 5.

Dysregulation of genes involved in ERSR. GSEA enrichment plots and expression profile (shown by heat-mapping) of genes involved in the ERSR/UPR pathway (A,B) and integrated stress-response pathway29 (C,D). Running enrichment score and signal-to-noise ratio used for ranking genes (positive, up-regulated in mutant; negative, down-regulated in mutant) are shown in the GSEA plots.25 Solid black vertical bars indicate the position of genes in the ERSR/UPR within the sorted microarray data, showing enrichment among genes up-regulated in hi559.

Figure 6.

Preferential up-regulation of ERSR/UPR genes in the liver of hi559 embryos. ISH of ERSR markers in wild-type (top) and hi559 larvae (bottom) at 5 dpf. The respective gene symbols are indicated in each panel. Arrows indicate elevated expression in the hi559 liver.

ER Architecture Is Severely Disrupted in hi559 Hepatocytes.

We performed transmission electron microscopy to analyze the ultrastructural pathology of hi559 hepatocytes. Wild-type hepatocytes exhibit a homogeneous, grainy cytoplasm, generally without clearing areas (Fig. 7A). By contrast, the hi559 hepatocytes have abnormal mitochondria, large cytoplasmic clearing areas with several membrane-bound structures containing granular materials (Fig. 7B). Irregularly shaped lipolysosomes containing lipid droplets of variable electron density are frequently seen in hi559 hepatocytes (Fig. 7F). Most strikingly, hi559 hepatocytes have large, excessively dilated (luminal swelling), abnormally distributed ER (Fig. 7C,D). It appears that the prominent clearing areas in hi559 hepatocytes may be the sequelae of excessive ER luminal swelling and vacuolation. The lumens of the expanded ER in hi559 hepatocytes are often filled with aggregates of variable electron density, suggestive of accumulated proteins (Fig. 7E). In some instances, the ER membranes are selectively sequestered and tightly packaged into autophagosome-like structures (Fig. 7G). Aggregates of macrophages are noticed adjacent to the necrotic hepatocytes, indicating mild inflammation (Fig. 7H.) These ultrastructural pathologies are consistent with chronic unresolved ER stress and resemble that seen in NAFLD.

Figure 7.

Ultrastructural pathology of hepatocytes. (A-D) Electron micrographic comparison of wild-type and hi559 hepatocytes. Cytoplasmic clearing (asterisk in B), ER luminal swelling (triangle in D), and abnormal mitochondria are frequently observed in hi559 hepatocytes. (E) hi559 hepatocyte shows mitochondrial damage and excessive ER luminal vacuolization. The ER lumens are filled with granular materials. (F) hi559 hepatocytes often contain membrane-bound lipids (lipolysosomes, arrows) of variable electron density. (G) Autophagosome-like structures (arrows) containing reticular materials are noticed within hi559 hepatocytes. (H) Presence of a macrophage (arrow) adjacent to necrotic hepatocytes. er, endoplasmic reticulum; mc, mitochondria; mut, hi559 mutant; n, nucleus; wt, wild-type. Scale bars: A, B, F, 2 μm; C-E, G, H, 500 nm.

Tunicamycin-Induced ER Stress Results in Fatty Liver Similar to hi559.

While analyzing the expression of ER stress markers, we noticed elevated expression of the crucial ER stress sensor hspa5 in hi559 livers at 4 dpf prior to onset of the hepatic phenotype (Fig. 8A). This implicates that hepatocellular ER stress may be a major contributor to the hepatic steatosis seen in hi559 larvae at 5 dpf. To test whether ER stress during this developmental stage could cause hepatic steatosis, we treated wild-type larvae with tunicamycin, an inhibitor of protein N-glycosylation that induces ER stress. Chronic treatment with 1 μM tunicamycin from 3.5 dpf through 5.5 dpf induced defects similar to those seen in hi559 larvae in ≈90% of the treated larvae (Fig. 8B-E). Larvae subsequently die at 6 to 7 dpf, similar to hi559, when tunicamycin treatment was continued. Induction of ER stress upon tunicamycin treatment was confirmed by ISH with the crucial ER stress marker hspa5. The ubiquitously elevated expression of hspa5 was apparent in tunicamycin-treated larvae (Fig. 8C). Whole-mount ORO staining further confirmed the development of fatty liver in tunicamycin-treated larvae (Fig. 8D). These results indicate that ER stress triggers hepatic steatosis in zebrafish larvae and that chronic hepatocellular ER stress may be predisposing to NAFLD in hi559 larvae.

Figure 8.

Tunicamycin-induced ER stress causes fatty liver in wild-type larvae. (A) Elevated expression of hspa5 in hi559 liver at 4 dpf by ISH (arrows). (B) Chronic exposure of 1 μM tunicamycin from 3.5 dpf to 5.5 dpf causes globular darkish liver (yellow dotted lines) similar to hi559. (C) hspa5 expression is ubiquitously elevated, including liver (arrows) in tunicamycin-treated larvae. (D) ORO staining shows presence of fatty liver (yellow dotted lines) in tunicamycin-treated larvae. (E) Bar chart indicates percentage of larvae with hepatic steatosis at 5.5 dpf in dimethyl sulfoxide (DMSO) and tunicamycin (TM)-treated groups. Data are representative of three clutches (n = 55).


To our knowledge, hi559 is the first in vivo model linking PtdIns synthesis, ER stress, and NAFLD. Multiple lines of evidence support the conclusion that loss of Cdipt function eliminates PtdIns synthesis. First, Cdipt is specifically inactivated in hi559 zebrafish, evident by undetectable levels of cdipt mRNA by RT-PCR and ISH, rescue of the mutant phenotype with cdipt mRNA, and phenocopy of the hi559 phenotype by injection of cdipt morpholinos. Second, we believe that cdipt is the sole enzyme responsible for de novo PtdIns synthesis in zebrafish: in extracts from mutant larvae, no PtdIns synthesis could be detected. Third, zebrafish cdipt is highly homologous to mammalian CDIPTs (Supporting Fig. 9), and no other potential orthologs could be detected in the zebrafish genome. Finally, Drosophila embryos deficient in dPIS (the ortholog of CDIPT) were unable to synthesize PtdIns and died during embryogenesis.23 Taken together, these findings suggest that Cdipt is essential for PtdIns synthesis, and its disruption leads to the hi559 phenotype.

Although Cdipt is indispensable for PtdIns synthesis, widespread developmental abnormalities are not observed in hi559 embryos during early development, possibly due to maternally deposited PtdIns in the yolk (Supporting Fig. 4). The later phenotypic abnormalities reflect a requirement of de novo PtdIns synthesis, because pools of PtdIns are locally made and used in intracellular PI signaling almost instantly after synthesis.30 Thus, despite an abundant supply of maternal PtdIns, cells may still require de novo synthesis of PtdIns for appropriate PI signaling and PtdIns function. Hence, we surmise that lack of de novo PtdIns synthesis during development causes aberrant PtdIns function and PI signaling in secretory hepatocytes of hi559 larvae.

The dynamics and function of PtdIns and their pathophysiological roles in various human diseases remain elusive. In this study, disruption of PtdIns de novo synthesis results in persistent hepatocellular ER stress, evident by robust activation of ER stress sensors and chaperones in the hi559 liver, and grossly expanded ER lumens. Aberrant PtdIns functions can affect ER homeostasis and cause subsequent ER stress–associated cytopathologies in several ways, such as calcium misregulation, alteration of secretory pathways and accumulation of proteins in the ER (Supporting Fig. 8). First, intracellular Ca2+ signaling and Ca2+ homeostasis in the ER are dependent on the PtdIns breakdown products, IP3.31, 32 Aberrant Ca2+ results in dysfunction of ER chaperones, thus affecting proper folding of proteins in ER. Second, protein kinase C signaling requires PtdIns and its breakdown products, inositol and diacylglycerol, and calcium. Altered protein kinase C signaling can cause elevated transcription of secretory proteins.33 Third, the continual exchange of proteins and lipids between the ER and Golgi apparatus through vesicular transport is also a Ca2+ and PtdIns-dependent process.34 One of the transient PIs, PtdIns-4-phosphate, is critical in vesicular trafficking and ER-associated degradation, and its deficiency may contribute to the accumulation of secretory proteins in the ER lumen, causing ER stress.20, 35 Therefore, we hypothesize that disrupted PtdIns synthesis alters one or more of these molecular processes, resulting in unresolved ER stress and consequent hepatic pathology. Consistent with this hypothesis, UPR is activated when yeast is cultured on inositol-deficient media and inactivated upon inositol supplementation as a result of modulation of PtdIns levels.36, 37

Concurrent with ER stress, the hi559 liver displays NAFLD pathologies, which we believe are a consequence of unresolved ER stress. Hepatocytes cope with ER stress through UPR, but chronic unresolved ER stress can unleash pathological consequences, including hepatic fat accumulation, cell death, and inflammation, thus contributing to NAFLD.18, 38 XBP1, a critical mediator of ERSR, is reported to be involved in increased hepatic lipogenesis, and we found selective up-regulation of xbp1 in the hi559 liver. Up-regulation of hspa5, the master ER stress sensor, is apparent in the hi559 liver at 4 dpf, before onset of the hepatic phenotype (Fig. 8A). Additionally, pharmacological induction of ER stress by tunicamycin caused hepatic steatosis similar to hi559. These results suggest that chronic unresolved ER stress may predispose the secretory hepatocytes to hepatic steatosis in hi559 larvae.

Hyperlipidemia, obesity, and diabetes may predispose to NAFLD, a disease with increasing prevalence in Western societies and currently without effective therapy.1, 28 The similarity of cytopathological features of hi559 liver to NAFLD emphasizes the potential of this mutant as an in vivo model for unraveling molecular pathogeneses of this disease. Here, we report a novel association between PtdIns, ER stress, and hepatic steatosis, suggesting that modulation of PtdIns may mitigate the contribution of ER stress to the pathology of NAFLD. With the increasing recognition of the role of ER stress in human disease, including hepatocellular carcinoma, several ER stress–modulating compounds are being explored for their therapeutic potential.16, 38 The hi559 mutant described in this study is uniquely positioned to aid in the functional characterization of these compounds in a live animal model and in the identification and analyses of potentially new treatment paradigms.


We thank Christine Sciulli, Ardith Ries, Patricia Snyder, Lisa Chedwick, and Lili Lu for excellent technical assistance and Parmjeet Randhawa, Meir Aridor, and Jeffrey Brodsky for helpful discussions. We thank Rhobert Evans and Howard Irwin for providing radioactive facilities.