Sorting nexin (SNX) family proteins are best characterized for their abilities to regulate protein trafficking during processes such as endocytosis of membrane receptors, endosomal sorting, and protein degradation, but their in vivo functions remain largely unknown. We started to investigate the biological functions of SNXs using the zebrafish model. In this study, we demonstrated that SNX7 was essential for embryonic liver development. Hepatoblasts were specified normally, and the proliferation of these cells was not affected when SNX7 was knocked down by gene-specific morpholinos; however, they underwent massive apoptosis during the early budding stage. SNX7 mainly regulated the survival of cells in the embryonic liver and did not affect the viability of cells in other endoderm-derived organs. We further demonstrated that down-regulation of SNX7 by short interfering RNAs induced apoptosis in cell culture. At the molecular level, the cellular FLICE-like inhibitory protein (c-FLIP)/caspase 8 pathway was activated when SNX7 was down-regulated. Furthermore, overexpression of c-FLIPS was able to rescue the SNX7 knockdown-induced liver defect.
SNX7 is a liver-enriched antiapoptotic protein that is indispensable for the survival of hepatoblasts during zebrafish early embryogenesis. (HEPATOLOGY 2012;55:1985–1993)
Liver development begins with the specification of hepatoblasts within the anterior foregut endoderm during early embryogenesis. Previous studies in mice and chickens have demonstrated that fibroblast growth factors (FGFs) from the adjacent cardiac mesoderm and bone morphogenetic protein (BMP) signals from the septum transversum mesenchyme are both required for the onset of hepatogenesis.1, 2 Other signaling pathways, including transforming growth factor beta (TGF-β) and WNT, have also been implicated in early liver development.3, 4 After their specification, the newly formed hepatoblasts form the liver primordium, which proliferate rapidly and further differentiate into mature hepatocytes or cholangiocytes during later stages of liver development. Transcriptional factors, including hematopoietically expressed homeobox (Hhex), prospero homeobox protein 1 (Prox1), and hepatocyte nuclear factor (HNF) family members, play important roles during these processes.5, 6
Recently, teleosts, including zebrafish and medaka, have become valuable model animals for studying the mechanisms of liver development.7-9 Liver development in zebrafish can be described in three steps comparable to those in other vertebrates. First, the endodermal cells migrate to the midline and form a rod-like structure at 24 hours postfertilization (hpf).10 Two earliest hepatoblast markers, hhex and prox1, are induced by signals from the nearby mesoderm. Several such signaling molecules have been identified so far. For example, Wnt2bb, from the lateral plate mesoderm, is required for liver specification.11 Inhibition of BMP or FGF signaling blocks liver specification,12 and BMP2b regulates the hepatic versus pancreatic fate of hepatopancreatic progenitors.13 Variant hepatocyte nuclear factor 1 and retinoic acid (RA) are reported to regulate liver specification as well.14, 15 RA regulation of wnt2bb is reported to be essential for liver specification in medaka as well.16 Shortly after the specification of hepatoblasts, hepatogenesis enters the “budding stage”: Hepatoblasts aggregate and form a thickened structure, termed liver bud. The intestinal primordium undergoes a leftward bend (i.e., gut-looping) at approximately 30 hpf, which places the liver bud to the left side of the midline.17 The liver primordium continues to develop and enters the “expansion growth” stage at approximately 50 hpf: Hepatoblasts proliferate rapidly and undergo further morphogenesis to reach the shape and place of the mature liver. It is in this period that hepatoblasts differentiate into mature hepatocytes as well as bile duct cells. Several recent studies have identified genes specifically required for the budding and growth of liver in the zebrafish. For example, mutation in def18 or myosin phosphatase target subunit 1 (mypt1)19 does not affect the specification of hepatoblasts, but inhibits the proliferation of these cells. The expansion growth of the liver requires genes, including liver-enriched gene 1 (leg1),20 npo,21 ubiquitin-like protein containing PHD and ring finger domains-1 (uhrf1),22 or DNA methyltransferase (dnmt)2.23 Embryos with mutation in translocase of outer mitochondrial membrane 22 (tomm22)24 or dnmt125 have normal early hepatogenesis, but show liver degeneration at later stages. Epigenetic-related genes, such as histone deacetylase (hdac)1/3, are involved in the regulation of liver development as well.26, 27 Although many critical regulators of hepatogenesis have been identified, detailed understandings of liver development at the molecular and cellular levels remain to be established.
Sorting nexin (SNX) family proteins are phox homology domain-containing proteins involved in diverse intracellular processes, such as endocytosis, protein sorting, and endosomal signaling.28, 29 The first SNX family member, SNX1, was discovered as an epidermal growth factor receptor (EGFR)-binding partner required for the lysosomal degradation of EGFR.30 Further studies demonstrated that SNX1, 2, 5, and 6 are components of the retromer that mediates the retrograde transport of transmembrane cargo from the endosome to the trans-Golgi network.31 SNX4 regulates the endosomal sorting of the transferrin receptor32 and SNX27 regulates the endosomal trafficking of G-protein–gated potassium channels, such as inwardly rectifying K, in neuronal cells.33 SNX17 enhances the endocytosis of the low-density lipoprotein (LDL) receptor as well as LDL-receptor–related protein.34, 35 SNX9 interacts with Wiskott-Aldrich syndrome protein and regulates actin-dependent fluid-phase endocytosis and T-cell signaling.36 The in vivo functions of a few SNX genes have been investigated. For example, SNX1 and 2 have been knocked out in the mouse, and mice lacking either one of them are viable and fertile. However, the double-knockout mice die at midgestation, which complicates the detailed analysis of the in vivo functions of SNX1 and 2.37 SNX13 knockout mice are also embryonic lethal,38 whereas SNX27 plays essential roles during postnatal growth and survival.39
We started to investigate the in vivo functions of SNXs in the zebrafish model. We identified six SNX genes expressed in the embryonic liver and found that one of them (SNX7) was indispensable for hepatogenesis. The specification and proliferation of hepatoblasts were normal when SNX7 was blocked. However, these cells underwent extensive apoptosis during the budding stage of hepatogenesis. We concluded that an antiapoptotic activity of SNX7 was crucial for the survival of hepatoblasts during liver budding.
Materials and Methods
Detailed protocols, including zebrafish manipulation, cell culture and short interfering RNA (siRNA) treatment, immunostaining, fluorescence-activated cell sorting (FACS) analysis, real-time reverse-transcription polymerase chain reaction (RT-PCR), and western blotting, can be found in the Supporting Materials and Methods.
Identification of SNX Genes Expressed in the Embryonic Liver.
We performed a BLAST search against the zebrafish genome and EST databases, using human SNX sequences as references, and identified 38 zebrafish SNX family genes. Phylogenetic analysis revealed that SNX1, 8, 9, 10, 18, 19, 27, and 28 were duplicated in the zebrafish genome, presumably because of the partial genome duplication in teleost (Supporting Fig. 1). To evaluate their potential roles during early embryogenesis, we examined the expression patterns of all SNXs by in situ hybridization. We focused on SNXs expressed in the embryonic liver in this report. Six SNX genes were expressed in the livers of 3-day-old embryos. SNX1a, 3, and 7 were highly expressed in the liver and gut. SNX17 was present in the liver, eye, and brain, but not the gut (Fig. 1). SNX25 was more abundant in the eye and brain than in the liver and gut. SNX29 was also detectable in the liver, gut, and brain regions.
SNX7-Regulated Liver Development.
The hepatic expression of these SNXs suggested that they could play roles during liver development. We performed loss-of-function studies on these genes using morpholino (MO) technology.41 One SNX family member, SNX7, was found to be essential for hepatogenesis. We designed MOs targeting the exon 1/intron 1 junction (MO1) or the intron 2/exon 3 junction (MO2) of the SNX7 gene. Both of them efficiently induced alternative splicing of SNX7 messenger RNA (mRNA), as determined by RT-PCR (Fig. 2A). The development of MO1-injected embryos was slightly delayed, but the general morphology of them appeared normal (Fig. 2B). However, liver development was severely disrupted in these morphants at day 3; the expression of cp (ceruloplasmin; a marker expressed in the liver after 32 hpf42) was severely reduced or not detectable in 86% of the injected embryos (Fig. 2C; N = 43). Overexpression of human SNX7 did not affect liver formation in zebrafish (data not shown); however, it was able to rescue the MO1-induced liver defect. When hSNX7 mRNA (100 pg/embryo) was coinjected with MO1, the expression of cp was restored in 79% of the treated embryos (Fig. 2C; N = 29). Similar results were observed for MO2 (data not shown). These results demonstrated that the liver defect in SNX7 morphants was not the result of off-target effects of MOs and suggested that SNX7 was essential for liver development in zebrafish. We also investigated the potential roles of SNX7 in the development of other endoderm-derived organs. The endocrine pancreas (ins; insulin), the exocrine pancreas (try; trypsin), or the intestine (intestinal fatty-acid–binding protein; ifabp) appeared normal in SNX7 morphants (Fig. 2D-F). Taken together, these results demonstrated that SNX7 was required for the liver, but not pancreas or gut, development in zebrafish.
SNX7 Was Dispensable for the Specification of Hepatoblasts.
The liver defect in SNX7 morphants could be the result of the failure to specify hepatoblasts from endodermal progenitor cells. We tested this possibility by examining the expression patterns of early endoderm- and liver-specific markers. Forkhead box protein A3 (foxA3) and GATA-binding factor 6 (gata6) are pan-endodermal markers. SNX7 morphants showed mildly an underdeveloped brain and trunk at 30 hpf; however, the expression levels and spatial patterns of foxA3 and gata6 in these morphants were comparable to those in the wild-type (WT) embryos (Fig. 3A,B). prox1 and hhex are the earliest liver-specific markers, and we found that their expressions were not affected by SNX7 knockdown as well (Fig. 3C,D). This result suggested that hepatoblasts were properly specified, and that the liver defect in SNX7 morphants might be the result of compromised development of liver during the budding or expansion growth stage. We found that prox1 staining in the liver of WT embryos increased significantly from 30 to 72 hpf, but prox1 in SNX7 morphants was not increased proportionally during the same period (Fig. 3E,F). These data suggested that the specification of hepatoblasts was SNX7 independent, but that further growth or maturation of the liver was SNX7 dependent.
SNX7 Regulated the Survival, but Not Proliferation, of Hepatoblasts.
The small liver in SNX7 morphants could be the result of either reduced proliferation or enhanced apoptosis of hepatoblasts. We investigated these two possibilities in the MP760 transgenic zebrafish line.23 In this line, the liver was distinguishable from other endoderm tissues after the formation of liver bud at approximately 30 hpf (Fig. 4A). We performed 4′,6-diamidino-2-phenylindole (DAPI) staining in MP760 to count the numbers of liver cells at 32, 48, and 72 hpf (Fig. 4B). The average number of liver cells increased from 78 to 197 (a 1.5-fold increase) in control embryos. However, that number in SNX7 morphants only increased from 52 to 86 in the same period (a 0.65-fold increase). To investigate the cellular mechanism of the liver defect in SNX7 morphants, we first analyzed the proliferation rate of liver cells by phosphorylated histone 3 (P-H3) staining. Percentages of P-H3-positive hepatoblasts were comparable between control embryos and SNX7 morphants at all stages examined (Fig. 4C) (P > 0.25 in all cases). This result suggested that down-regulation of SNX7 did not affect the growth of hepatoblasts. We next measured the ratio of apoptotic hepatoblasts by performing the terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) assay. No apoptotic cell was detected in the endoderm of WT embryos or SNX7 morphants before the formation of liver bud at approximately 30 hpf (Fig. 4D). However, extensive apoptotic signals were detected in the liver of SNX7 morphants (23.6%; N = 590), but not in control embryos (1.2%; N = 653), at 32 hpf. These results demonstrated that SNX7 was essential for the survival, but not proliferation, of hepatoblasts during the liver bud stage.
Activation of Apoptosis-Related Genes in SNX7 morphants.
We investigated the molecular mechanism of SNX7 by analyzing the expression levels of cell proliferation-/apoptosis-related genes. Embryos were injected with MO1 or a standard control morpholino (4 ng), and total RNAs were prepared at 32 hpf. Relative expression levels of candidate genes were determined by real-time RT-PCR analysis, with the β-actin gene as an internal control. Expression levels of a house-keeping gene (e.g., elfa), early pan-endoderm or liver-marker genes (e.g., foxA3, gata6, hhex, and prox1), or genes crucial for the specification of hepatoblasts (e.g., wnt2bb and mypt1) were not affected by SNX7 knockdown (Fig. 4E). p27, cyclin D1, and reprimo are cell-cycle–related genes and their expressions were not significantly changed in SNX7 morphants as well. However, expression levels of proapoptotic genes, such as bax and p53, were significantly increased in SNX7 morphants (P < 0.00001 for both). Furthermore, several p53 target genes (e.g., Δ113p53, mdm2, cyclin G1, and p2118, 44) were highly up-regulated in SNX7 morphants. Interestingly, we also found that caspase 8, but not other caspases, such as caspase 3a, 3b, and 9, was induced at the transcriptional level. leg1 is a liver-enriched gene that is essential for liver development in zebrafish. The level of leg1 in SNX7 morphants was severely reduced (to 17% of control) (Fig. 4E). We tried, but failed, to rescue the liver defects in SNX7 morphants by overexpression of leg1 (data not shown).
Cellular FLICE-Like Inhibitory Protein Mediated the Antiapoptotic Activity of SNX7.
We further investigated the antiapoptotic mechanism of SNX7 in cell cultures. Two independent siRNAs to SNX7 were designed and both of them were able to induce more than 90% inhibition of SNX7 at the mRNA level in Hela cells, as measured by real-time RT-PCR analysis (Fig. 5A). Cells were transfected with these siRNAs or a universal control siRNA for 2 days, and the TUNEL FACS assay was performed to determine the level of apoptotic cells. The background level of apoptosis in a control siRNA (siCTL)-treated cells was 1.8% (Fig. 5B). Treatment of cells with siRNAs to SNX7 significantly induced apoptosis (14.4% for siSNX7-a and 11.1% for siSNX7-b). Cycloheximide (CHX) is an inhibitor of protein synthesis and regulates pathways such as tumor necrosis factor alpha (TNFα)-induced apoptosis. Treatment of Hela cells with CHX alone did not induce apoptosis, but was able to further enhance the SNX7 siRNAs-induced apoptosis (Fig. 5B,C). We performed western blotting for the apoptosis-related markers (Fig. 5D). Down-regulation of SNX7 combined with CHX treatment clearly induced the cleavage of poly(ADP-ribose) polymerase (PARP) and caspase 8, whereas caspase 9 was not activated. These results suggested that the death-receptor–mediated apoptotic pathway (the extrinsic pathway) was activated. Cellular FLICE-inhibitory protein (c-FLIP) is an inactive caspase 8 homolog that interferes with the death-ligand–induced formation of death-inducing signaling complex and subsequent activation of caspase 8.45, 46 We evaluated the c-FLIP levels after SNX7 siRNAs treatment and found that the level of c-FLIPL (the long form of c-FLIP) was not changed, whereas the level of c-FLIPS (the short form of c-FLIP) was clearly decreased when SNX7 was inhibited (Fig. 5D, bottom panel). We performed similar analysis in a human hepatocellular carcinoma–derived cell line (HepG2). Treatment of HepG2 with SNX7 siRNA plus CHX also induced the cleavage of PARP, activation of caspase 8, and down-regulation of c-FLIPS (Fig. 5E).
We next tested whether SNX7 would regulate the c-FLIP protein level in zebrafish embryos. Embryos were injected with either a control MO or MO1, and the c-FLIP protein amount in whole embryonic lysate was determined by western blotting. MO-CTL treatment did not affect the level of c-FLIP, whereas MO1 injection down-regulated c-FLIPS to 36% of the control level (Fig. 6A). More important, we found that coinjection of mRNA encoding human c-FLIPS (75 pg/embryo) could rescue the MO1-induced liver defect in 75% of the injected embryos (Fig. 6B; N = 44). Taken together, these results suggested that knockdown of SNX7 induced the degradation of c-FLIPS, which led to the activation of the caspase 8–dependent pathway and subsequent cell death.
Many molecules involved in hepatogenesis have been identified from various model systems. The majority of them can be grouped into one of the following categories: (1) cell-signaling molecules, such as FGF, BMP, Wnt, Hedgehog, or RA pathway-related genes; (2) transcriptional factors, such as Gata and HNF family members, Hhex, Prox1, and so on; and (3) epigenetics-related molecules, such as Dnmt1/2, Hdac1/3, and Uhrf1. We report here that SNX7, a SNX family member supposed to be involved in vesicular trafficking and protein sorting, is crucial for embryonic liver development in zebrafish. SNX7 is an early endosome and multivesicular-body–distributed protein (data not shown). Interestingly, a recent study reveals that tomm22, a regulator of protein traffic from cytoplasm into the mitochondria, is required for liver development in zebrafish.24 Disruption of this gene induces extensive apoptosis of hepatocytes, which is similar to what we observed in SNX7 morphants. On the other hand, mutation in vacuolar protein sorting protein 18 (vps18), a class C vacuolar protein-sorting gene, causes hepatomegaly (i.e., large liver) in zebrafish.47 Vps18 is involved in the regulation of vesicles from late endosome to lysosome, and mutation in vps18 causes the accumulation of cytoplasmic vacuoles, which eventually leads to the hepatomegaly phenotype. These observations suggest that different subcellular protein-traffic pathways could affect different aspects of liver development. Thus, SNX7 could provide us with new opportunities to study the molecular mechanism of liver development.
The specification of hepatoblasts was normal in SNX7 morphants; however, these cells underwent apoptosis during the budding stage. Knockdown of SNX7 by siRNAs in Hela or HepG2 cells induced apoptosis as well. We revealed that SNX7 regulated the death-receptor–mediated caspase 8 pathway, but not the mitochondrion-related caspase 9 pathway. c-FLIP is a catalytically inactive homolog of caspase 8 and is able to interfere with the activation of caspase 8. We demonstrated that down-regulation of SNX7 decreased the intracellular level of c-FLIPS, and this regulation appeared to be proteasome dependent (data not shown). Proteasomes are large protein complexes involved in ubiquitin-dependent protein degradation. c-FLIP has been reported to be regulated by E3 ubiquitin ligases, including Itch48 and mindbomb 1 (Mib1)49 during TNF-induced cell death. It will be interesting to test whether SNX7 regulates Itch, Mib1, or other E3 ubiquitin ligase-dependent degradation of c-FLIP. Further studies are needed to reveal the exact molecular mechanism of SNX7 in the caspase 8–/c-FLIP-dependent apoptotic pathway.
The authors are grateful to Hanbing Zhong, Yonglong Chen, and Xingguo Liu for reagents and helpful discussions. The authors thank Yi Zheng and members of the Pei Lab for technical assistance.