Loss of the RNA-binding protein Rbm15 disrupts liver maturation in zebrafish

Liver organogenesis begins with hepatic precursors in the foregut endoderm, followed by hepatoblast specification, differentiation, outgrowth, and maturation for the formation of functional hepatocytes. Although several signaling pathways and critical factors that regulate liver specification, differentiation, and proliferation have been identified, little is known about how liver maturation is regulated. Here, we used a screen for mutations affecting liver development in zebrafish and identified a cq96 mutant that exhibits a specific defect in liver maturation. Results from positional cloning revealed that cq96 encodes an RNA-binding protein, Rbm15, which is an evolutionarily conserved Spen family protein and known to play a crucial role in RNA m6A modification, nuclear export, and alternative splicing. However, a function of Rbm15 in embryonic liver development has not been reported. We found that Rbm15 is specifically expressed in the liver after its differentiation. CRISPR/Cas9-mediated loss of rbm15 repressed hepatic maturation, but did not affect hepatoblast specification, differentiation, and hepatocyte proliferation and apoptosis. Additional experiments disclosed that the mTOR complex 1 (mTORC1) pathway is highly activated in rbm15-deficient hepatocytes. Moreover, rapamycin treatment partially restored normal hepatic gene expression as well as the nuclear location of the transcription factor Hnf4a. Taken together, these results reveal an unexpected role of Rbm15 in liver maturation.

Liver organogenesis begins with hepatic precursors in the foregut endoderm, followed by hepatoblast specification, differentiation, outgrowth, and maturation for the formation of functional hepatocytes. Although several signaling pathways and critical factors that regulate liver specification, differentiation, and proliferation have been identified, little is known about how liver maturation is regulated. Here, we used a screen for mutations affecting liver development in zebrafish and identified a cq96 mutant that exhibits a specific defect in liver maturation. Results from positional cloning revealed that cq96 encodes an RNA-binding protein, Rbm15, which is an evolutionarily conserved Spen family protein and known to play a crucial role in RNA m6A modification, nuclear export, and alternative splicing. However, a function of Rbm15 in embryonic liver development has not been reported. We found that Rbm15 is specifically expressed in the liver after its differentiation. CRISPR/Cas9mediated loss of rbm15 repressed hepatic maturation, but did not affect hepatoblast specification, differentiation, and hepatocyte proliferation and apoptosis. Additional experiments disclosed that the mTOR complex 1 (mTORC1) pathway is highly activated in rbm15-deficient hepatocytes. Moreover, rapamycin treatment partially restored normal hepatic gene expression as well as the nuclear location of the transcription factor Hnf4a. Taken together, these results reveal an unexpected role of Rbm15 in liver maturation.
The liver is the biggest digestive organ of the body and plays a central role in metabolism and systemic homeostasis. It mainly consists of hepatocytes, stellate cells, macrophages, sinusoidal endothelial cells, and cholangiocytes (1). The hepatocyte is the most populous cell of the liver; nearly 80% of hepatic cells are hepatocytes. It undertakes multiple physiological functions, such as bile secretion, glycol metabolism, lipid storage, protein secretion, and detoxification (2). Therefore, it is crucial to understand the mechanisms underlying liver development using animal models. Although the molecular regulation on liver organogenesis is starting to be unraveled, the critical factors to regulate liver development specifically have not been completely known.
In zebrafish, hepatogenesis can be divided into two steps. In the first stage, mesoderm around foregut endoderm induces hepatic precursor's proliferation and migration out from the gut endoderm and then specification to hepatoblast. Multiple layers of hepatoblast and the surrounding septum transversum form the liver bud. This progress lasts from 24 h post-fertilization (hpf) to 34 hpf (3). Mesoderm-derived FGF, WNT, and BMP signaling orchestrate this progress (4)(5)(6). Many transcription factors, such as GATA6, Hhex, and Foxas, are implicated in liver bud formation and hepatoblast specification (7)(8)(9). In the second stage, hepatoblast proliferates, differentiates, and maturates into hepatocytes and cholangiocytes, eventually forming a functional liver (2). In this progress, Wnt signaling is the key regulator during liver formation via increasing cell proliferation and repressing apoptosis (10,11). The Hippo/ YAP pathway is crucial for liver size control; hyper-YAP1 activity in the liver stimulates hepatocyte and cholangiocyte specification, whereas loss of YAP1 decreases hepatocyte survival and impairs biliary duct morphogenesis (12). Notch signaling and esr2b are also involved in hepatocyte differentiation and bile duct formation (13,14). Numerous transcription factors govern cell fate choices. Hnf4a, Prox1, Hnf1a, and c/EBPa are important for hepatocyte differentiation and hepatic gene expression (15)(16)(17)(18). Hnf1b and Hnf6 are responsible for bile duct formation (18). The mTOR pathway is also involved in liver development (19). Although many mechanisms underlying the hepatic fate decisions in vivo and in vitro have been reported, the key transcriptional factors that regulate hepatocyte maturation have been addressed less.
Rbm15 is an evolutionarily conserved Spen family protein involved in cell fate decisions. It can bind to RNA and recruit other proteins to specific binding sites to regulate post-transcriptional modification (20). In mice, RBM15 regulates megakaryocyte terminal differentiation by affecting alternative RNA splicing of a group of genes such as Gata1, C-mpl, Runx1, and Tal1, which are important for megakaryopoiesis (21). RBM15 can bind to specific sites in the intron of pre-mRNA, recruit SF3B1 to the splicing region, and facilitate alternative RNA splicing in different developmental stages (21). RBM15 can affect promoter activity of Notch target genes, such as Hes1, in a cell-specific manner. It can repress the expression of Notch-induced gene Hes1 in nonhematopoietic cells but enhance the Hes1 mRNA level in hematopoietic cell lines (22). RBM15 can also affect hematopoietic stem cell and megakaryocyte development via regulating c-myc expression (23). m6A is the most widespread RNA modification; the establishment of m6A relies on RBM15 (24,25). RBM15 is also involved in RNA nuclear export (26). However, the role of Rbm15 in liver development is unknown.
Zebrafish has been developed into a popular model organism in recent years because of advantages of their transparent embryos developing outside of the mother, allowing constant visualization, and 69% of all zebrafish genes have a clear human orthologue (27). High genetic conservation among vertebrates makes zebrafish an excellent model system to study liver development and disease (27,28). Here we used chemical mutagenesis to screen genes involved in liver development and got a specific liver maturation mutant, cq96. The mutant cq96 revealed normal liver specification, differentiation, and outgrowth, but the liver maturation was repressed. The development of the pancreas and intestine showed normal in cq96 mutant. Positional cloning and knockout experiments confirmed this mutation site located in gene rbm15. Deficiency of Rbm15 leads to up-regulation of the activity of mTORC1 signaling, and inhibition of the mTORC1 pathway by rapamycin can rescue liver maturation defect. Furthermore, the treatment of rapamycin enhanced the nuclear import of Hnf4a and restored hepatic gene expression. Our data revealed an unexpected function of rbm15 in liver development in zebrafish.

The cq96 mutant confers a liver maturation defect
We have screened mutagenized zebrafish larvae using transgenic line Tg(lfabp:Dendra2-NTR) (29) and identified the cq96 mutant. At 5 dpf, this mutant showed normal body morphology but weaker Dendra2 expression (Fig. 1A). The confocal image exhibited relatively normal liver size but heterogeneous Dendra2 expression in the mutant (Fig. 1B). The expressions of digestive organ-specific markers trypsin (exocrine pancreas) and ifabp (intestine) were normal, but the hepatocyte-specific gene cp was repressed (Fig. 1C). The expressions of functional markers of hepatocytes such as uox, gc, and gys2 were repressed in the mutant ( Fig. 1D and Fig. S1  (A and B)). These data indicate that the cq96 mutant specifically affects liver development and especially regulates liver maturation.

Zebrafish cq96 mutation site locates in gene rbm15
To determine the target gene of the cq96 mutant, we performed genome mapping and placed the cq96 mutation site locus to gene rbm15. The genomic sequencing result showed that there are 347 bp deleted in rbm15 exon 1, leading to translation into a truncated peptide (Fig. 2, A and B). To further confirm that rbm15 affects liver maturation, we generated a new rbm15 mutant by CRISPR/Cas9 that resembled the cq96 phenotype (Fig. 2, A-F). These results revealed that the mutation gene of cq96 is rbm15. Furthermore, rbm15 expressed in the liver region from 4 dpf and much stronger at 5 dpf (Fig. 2D). The weak expressions of Dendra2 and hepatic maturation markers such as uox and cp in cq96 mutant can be rescued by using Rbm15 overexpression line Tg(hsp70l:rbm15-Flag cq97 ) (Fig. 2, E-G). These results suggested that the mutation gene in cq96 is rbm15.
Zebrafish cq96 mutant affects liver maturation but not hepatoblast specification To evaluate hepatoblast specification in the rbm15 mutant, we assayed the expression of transcriptional factors important for hepatoblast formation and specification at 60 hpf. Interestingly, the expressions of prox1, gata6, hhex, and foxa3 were not greatly different between mutants and siblings (Fig. 3A). This means that liver specification was normal in the cq96 mutant. Then we performed antibody staining for hepatic factors Hnf4a and Prox1 at 5 dpf. The nuclear signal for Hnf4a and Prox1 was much weaker compared with siblings (Fig. 3, B-E). These results suggest that the cq96 mutant has liver maturation rather than liver specification defects.

Loss of Rbm15 confers normal hepatic proliferation and apoptosis
To investigate the underlying mechanism of liver development defect in cq96 mutant, we performed antibody staining for PCNA, which labeled cells outside of the G 0 phase. We detected a similar ratio of PCNA and Dendra2 double-positive cells in siblings and mutants (Fig. 4, A and B). To assess whether cell death contributes to liver development defect in cq96 mutant, we performed a transferase-mediated dUTP nick-end labeling (TUNEL) assay on WT and cq96 at 5 dpf. Larval hepatocytes presented a low apoptotic index at 5 dpf, which was unchanged in cq96 mutant (Fig. 4C). Therefore, liver development defect in the cq96 mutant does not appear to be due to defective cell proliferation and apoptosis.

Inhibiting mTORC1 signaling pathway partially rescues the phenotype of cq96 mutant
Hyper-and hypoactivated mTORC1 pathway will impair normal liver development (30,31). To explore whether the loss of rbm15 will affect the mTORC1 signaling or not, we performed antibody staining to check the expression level of p-4Ebp1, which indicates the activity of mTORC1 signaling, and found that the rbm15 2/2 mutant larvae showed high p-4Ebp1 level in the liver (Fig. 5A). This indicates that the mTORC1 pathway was hyperactivated in the rbm15 2/2 liver. To further confirm that the liver maturation defect in the rbm15 mutant was caused by mTORC1 activation, we used 10 mM rapamycin to inhibit mTORC1 from 5 to 7 dpf. mTORC1 inhibition can partially rescue the developmental liver defect (Fig. 5 (B and C) and Fig. S2A). After rapamycin treatment, the protein level of Hnf4a in mutant hepatocyte was rescued (Fig. 5, D and E), but the mRNA level of hnf4a showed no big difference between WT, rbm15 2/2 , and sample groups (Fig. S2B). Furthermore, the expressions of hepatocyte-specific genes such as cp and gc were also restored after rapamycin treatment (Fig. 5F). These results indicate that inhibition of mTORC1 can partially rescue liver maturation defects in rbm15 mutant.

Discussion
We have described a liver developmental mutant caused by rbm15 mutation, and inhibition of mTORC1 was an efficient strategy to rescue hepatic maturation defects. We showed that loss of Rbm15 specifically affected hepatic maturation, but not the developmental progress of intestine, pancreas. The mutant liver exhibited relative normal proliferation and apoptosis but weak hepatic gene expression. The two important hepatic transcriptional factors Hnf4a and Prox1 exhibited weak nuclear location, which could explain why rbm15 mutant showed hepatic maturation defects to some degree. Liver failure caused by rbm15 mutation also showed abnormal mTORC1 activation. Inhibition of mTORC1 can partially recover hepatic gene expression; this progress may rely on the enhancement of Hnf4a nuclear import.
Rbm15 is an important post-transcriptional regulator involved in RNA nuclear export, m6A modification, and alternative splicing (21,25,26). It is indispensable for megakaryocyte differentiation, and rbm15 defect can induce acute megakaryoblast leukemia (20). mTORC1 acts as a metabolic regulator important for cell growth and differentiation (19). Abnormal mTORC1 activation is associated with liver developmental defects and enhancement of liver damage (30,32). Our finding first reveals that rbm15 is essential for hepatic maturation and that loss of rbm15induced liver developmental failure partially depends on aberrant high mTORC1 activation.
We point out the importance of rbm15 in hepatic maturation, but we still do not know the target genes of rbm15. The relationship between rbm15 deficiency and high mTORC1 activity is still a mystery. RNA immunoprecipitation sequencing experiments will be indispensable to further answer these questions.

Ethics statement
All experimental protocols were approved by the Institute of Developmental Biology and Regenerative Medicine, Southwest University (Chongqing, China), and the methods were carried out in accordance with the approved guidelines. The zebrafish facility and study were approved by the Institutional Review Board of Southwest University (Chongqing, China). Zebrafish were maintained in accordance with the Guidelines of Experimental Animal Welfare from the Ministry of Science and Technology of the People's Republic of China (2006) and the Institutional Animal Care and Use Committee protocols from Southwest University (2007).

Zebrafish lines
Zebrafish (Danio rerio) AB strain-derived Tg(lfabp:Dendra2-NTR) cq1 was used as WT, and rbm15 cq96 mutant was generated by ENU treatment. The IND line was used for mapping. These zebrafish lines were raised under standard conditions, and embryos/larvae for the experiment were treated with 0.003% PTU (Sigma) from 24 hpf.

CRISPR/Cas9-targeted rbm15 knockout
The CRISPR/Cas9 was carried out essentially as reported previously (33). The sequence for CRISPR RNA is shown in Fig.  2A. We used the following primers to identify the genotype of

Antibody staining and TUNEL assay
Antibody staining and TUNEL assay were performed as described previously (29). The following antibodies were used: antibodies against Dendra2

Generation of transgenic line for rescue experiments
Full-length rbm15 cDNA was amplified by PrimeSTAR HS DNA Polymerase (Takara) and cloned into pBluescript vector. The full-length rbm15-Flag CDS was driven by hsp70l promoter, and the plasmid was injected into AB strain embryos to generate the transgenic line Tg(hsp70l:rbm15-Flag) cq97 . Positive embryos showed cerulean expression in the eyes of the offspring.

Rapamycin treatment and heat shock
Embryos were treated with 10 mM rapamycin (Sangon Biotech, Shanghai, China) in PTU egg water from 5 to 7 dpf and replaced rapamycin solution every 24 h. The control group was treated with 0.2% DMSO. To induce rbm15 overexpression from Tg(hsp70l: rbm15-Flag) cq97 , larvae were placed in egg water and then incubated in a 38.5°C water bath for 30 min once a day from 5 to 7 dpf.

Data collection and analysis
All images were taken on a SteREO DiscoveryV20 microscope (Carl Zeiss, Germany) and LSM880 confocal microscope  (Carl Zeiss). The intensities of fluorescent images were measured with ImageJ. The statistical analysis were performed with GraphPad Prism 8. Variation of individual data points was represented in S.D.

Data availability
All the data are contained within the article.