Overexpression of autotaxin, a lysophosphatidic acid-producing enzyme, enhances cardia bifida induced by hypo-sphingosine-1-phosphate signaling in zebrafish embryo

Lysophosphatidic acid (LPA) and sphingosine-1-phosphate (S1P) are lysophospholipid mediators with a number of similar characters. First, the structures of LPA and S1P are quite similar. Second, they are both present in various biological fluids such as plasma, serum and cerebrospinal fluids (1–4). Third, they share similar receptors: three LPA receptors (LPA_1–3) and all five S1P receptors (S1P_1–5) belong to the endothelial differentiation gene (Edg) family and share ∼35% sequence similarity with each other, although LPA has an additional three receptors belonging to the P2Y family (5). Fourth, they have several similar functions that have been demonstrated both in vivo and in vitro. They both stimulate cell proliferation and motility of various cell types (6). They also have critical roles in the vasculature (7–9) and are vasoactive (10, 11). Indeed, LPA and S1P regulate blood pressure, both positively and negatively (12, 13). Thus, LPA and S1P appear to share common features and have similar biological roles. However, it is unclear whether there is an interaction between LPA and S1P signaling. Genes involved in LPA and S1P signaling including receptors, producing enzymes, degrading-enzymes and transporters are highly conserved in vertebrates. For example, zebrafish and mammalian autotaxin (Atx) have ∼65% amino-acid identity and have similar biochemical and biological roles (14). Recent studies of zebrafish and mouse mutants revealed the essential cardiovascular functions of S1P signaling through the S1P transporter spns2 are conserved from fishes to mammals (15–19). These studies have indicated that zebrafish is a useful model organism for elucidating LPA and S1P functions. It may be possible to examine the interaction between LPA signaling and S1P signaling in zebrafish embryos by manipulating the expressions of several genes simultaneously. In this study, we investigated the functional interaction between LPA signaling and S1P signaling by manipulating LPA- and S1P-related genes in zebrafish. Here, we describe the first in vivo evidence showing that LPA signaling affects S1P signaling.

Materials and Methods

Maintenance of zebrafish and drug treatment

Wild-type strain (AB, TU) and transgenic (cmlc2:mRFP) were obtained from the Zebrafish International Resource Center (University of Oregon, Eugene, OR) and National BioResource Project, Zebrafish (Riken Brain Research Institute, Wako, Japan). Fish were maintained at 27–28°C under a controlled 13.5-h light/10.5-h dark cycle. Embryos were obtained from natural spawnings and staged according to morphology as described (20). Ki16425 was diluted in embryo media supplemented with 1% DMSO and added to zebrafish embryos between 12-h post-fertilization (hpf) and 24 hpf.

mRNA and morpholino injection

The sequences of antisense morpholino oligonucleotides (MO) (Gene Tools, LLC, Corvallis, OR) were designed as previously described (14, 16, 21). The mRNAs for zebrafish wild-type atx, catalytically inactive atx (T205A) and mouse wild-type atx were synthesized using mMESSAGE MACHINE kit (Ambion, Austin, TX). Morpholinos and synthetic mRNAs were dissolved in water with 0.2% phenol red. Synthetic mRNAs were injected into the blastomere of one cell stage embryos. Morpholinos were injected into the yolk of 1–8 cell stage embryos.

Whole-mount in situ hybridization

An antisense RNA probe labeled with digoxigenin for cmlc2 was prepared with an RNA labeling kit (Roche Applied Science, Penzberg, Germany). Whole-mount in situ hybridization was performed as previously described (22).

Evaluation of LPA receptor activation

Activation of LPA receptor was evaluated by a transforming growth factor-α (TGFα) shedding assay as described previously (23, 24). Briefly, each zebrafish LPA receptor gene was introduced into HEK 293T cells together with DNA encoding TGFα fused to alkaline phosphatase. Upon the addition of a ligand for LPA receptors, the TGFα proform expressed on the plasma membrane was cleaved by tumour necrosis factor α converting enzyme (TACE) proteases endogenously expressed in HEK293T cell and thereby released into the culture medium. Then, the activation of LPA receptor was evaluated by measuring the alkaline phosphatase activity in the conditioned medium of the cells.

Measurement of lysophospholipase D activity and Western blotting

Twenty-eight hours after injection, embryos were deyolked as described previously (14) and homogenized in lysis buffer [10 mM Tris-HCl (pH 7.4), 10% Triton X-100, 10 µg/ml PMSF, 50 µg/ml Leupeptin, 50 µg/ml Aprotinin] using an ultrasonic homogenizer (Smurt NR-50M; Microtec Nition, Funabashi, Japan). The homogenate was sequentially centrifuged at 21,500 × g and the resulting supernatant was collected. Protein concentration was measured with BCA Protein Assay Reagent (Thermo, Waltham, MA) and 5 µg of protein was used for each of lysophospholipase D assay and Western blotting. Lysophospholipase D activity was measured as described previously (25). Briefly, the extracts were mixed with 14:0 lysophosphatidylcholine (LPC) (100 mM Tris-HCl, 5 mM MgCl₂, 500 mM NaCl, 0.05% Triton X-100, pH 9.0) and incubated for 66 h at 37°C. Liberated choline was quantified using choline oxidase (Wako, Osaka, Japan), peroxidase (TOYOBO, Osaka, Japan) and TOOS reagent (Dojindo, Kumamoto, Japan). The activity was indicated by the generation rate of choline per unit time and protein mass (pmol/µg/h). Western blotting was performed using anti-zebrafish Atx rat polyclonal antibody that was generated as previously described (14). Proteins bound to the antibody were visualized with an enhanced chemiluminescence kit (GE Healthcare, Waukesha, WI).

Microscopic analysis and live imaging

Embryos were positioned in 3% methylcellulose (Sigma) on slide glass. Images were captured with a Leica M80 stereomicroscope equipped with Leica DFC425 digital camera (Leica Microsystems, Wetzlar, Germany).

Results

Overexpression of Atx enhances the cardia bifida phenotype in zebrafish embryos

We previously showed that down-regulation of Atx in zebrafish embryos resulted in malformation of the vasculature (14). When we tried to rescue the phenotype by injecting mRNA encoding zebrafish atx in the embryos, we accidentally found that introduction of atx mRNA resulted in an abnormal heart formation (Supplementary movies S1 and Supplementary Data). Compared with control embryos, blood flow and heart beating were severely impaired at 36 hpf. This phenotype is known as a two-heart or cardia bifida phenotype. In addition, at 54 hpf the tail region of the embryos frequently had blisters (Fig. 1E and F). These are the most characterized phenotypes in previous studies when S1P signal via either S1P receptor (s1pr2/mil) or S1P transporter (spns2) is attenuated (15–17). To examine the cardia bifida phenotype in more detail, we visualized cardiomyocytes by whole-mount in situ hybridization using cardiac myosin light chain 2 (cmlc2) probe. At 24 hpf, in control embryos, cardiomyocytes were detected in one cluster in the center of the embryos, whereas two clusters of cardiomyocyte were detected in atx mRNA-treated embryos (Fig. 1A and B). Similar cardia bifida and tail blister phenotypes were also observed in embryos treated with mouse atx mRNA (Fig. 1C and G). Both lysophospholipase D activity and zebrafish Atx protein increased with the dose of injected atx mRNA (Fig. 2A and B) and the phenotype was observed when atx mRNA higher than 0.03 ng was employed (Fig. 1I). The cardia bifida phenotype required catalytic activity of Atx because either the cardia bifida phenotype or tail blister phenotype was not induced when mRNA for catalytically inactive zebrafish atx (T205A) was injected (Fig. 1D, H and I), suggesting that the cardia bifida phenotype was induced via the product of Atx, that is LPA.

Down-regulation of S1P signaling enhances the Atx-induced cardia bifida phenotype

To address the possible link between S1P and LPA signaling, we examined the Atx-induced cardia bifida phenotype when S1P signal was attenuated. We injected atx mRNA simultaneously with mil or spns2 MO. First we injected MO for either mil or spns2 and confirmed that significant cardia bifida phenotype was induced at MO concentrations 3.2 ng or more (mil), and 5.0 ng or more (spns2), but rarely observed at MO concentrations lower than these dosages (Fig. 3A–E and K). Intriguingly, injection of a low dosage of mil (1.6 ng) or spns2 (1.0 ng) MO with atx mRNA (0.1 ng) induced an even more severe bifida phenotype (Fig. 3F, G, I, L and M). Indeed, only 7.0% and 3.0% of embryos displayed the cardia bifida phenotype in mil and spns2 MO-treated embryos, respectively, whereas 90% and 60% of embryos displayed the phenotype when atx mRNA was co-injected. We observed that the tail blister phenotype was also synergistically induced when atx mRNA was co-injected (data not shown). This synergistic effect by atx mRNA was catalytically dependent since mRNA for catalytically inactive zebrafish atx (T205A) did not show such effect (Fig. 3H, J and N).

LPA₁ mediates Atx-induced cardia bifida

To confirm the involvement of LPA signaling in Atx-induced cardia bifida, we next tried to identify the LPA receptor mediating the Atx-induced cardia bifida. Addition of a LPA_1–3 antagonist Ki16425 in embryo medium, which also worked on zebrafish Lpar1–3 (14), significantly decreased the occurrence frequency of the cardia bifida phenotype induced by injecting atx mRNA and mil MO in a dose-dependent manner (Fig. 4A). We recently showed that (R)-Ki16425 was more potent in antagonizing mammalian LPA_1–3 than (S)-Ki16425 (26). We also confirmed that this is also true for zebrafish Lpar1–3 (Fig. 5). Consistent with this, the cardia bifida phenotype was effectively rescued by (R)-Ki16425, but not by (S)-Ki16425 (Fig. 4B). We further confirmed that the lpar1 MO partially rescued the cardia bifida phenotype induced by co-injection of atx mRNA and mil MO (Fig. 4C). The rescue was not observed with lpar2–6 MO (data not shown). Together, these findings indicate that the cardia bifida phenotype induced by Atx overexpression in zebrafish embryos is mediated by overproduction of LPA and consequent activation of mainly Lpar1. Thus, these results showed that atx mRNA treatment affected cardiomyocyte migration that is regulated by S1P signaling in zebrafish embryos at least through Lpar1.

Discussion

In this study, we found that excess LPA signal in zebrafish embryos led to cardia bifida (two heart) phenotype. Because the same phenotype was induced when S1P signaling was down-regulated, we explored the functional interaction between LPA and S1P signaling in the zebrafish heart morphogenesis and found that LPA signaling down-regulated the S1P signaling that led to the cardia bifida phenotype. To our knowledge, this is the first to demonstrate the functional interaction between LPA and S1P signaling.

Among various LPA receptors Lpar1 appeared to be involved in the LPA-induced cardia bifida phenotype. First, the cardia bifida phenotype induced by the co-administration of atx mRNA and mil MO was almost completely rescued by Ki16425 (Fig. 4A and B), which were found to antagonize all Edg LPA receptors in zebrafish including Lpar1, Lpar2a, Lpar2b and Lpar3 (Fig. 5). Unlike lpar2a, lpar2b and lpar3, down-regulation of lpar1 alone rescued the cardia bifida phenotype, showing that Lpar1 is the major LPA receptor involved. Second, we speculate that the functional interaction between LPA and S1P signaling occurs downstream of each LPA (lpar1) and S1P (mil) receptor. Mil is the zebrafish ortholog of mammalian s1pr2. A body of evidence showed that LPA₁ mainly activates Gα_i-Rac1 signaling whereas S1P₂ mainly activates Gα_12/13-RhoA signaling (5, 27). Interestingly, Gα_i-Tiam1-Rac1 pathway downstream of LPA₁ was shown to inhibit Gα_12/13-mediated RhoA activation in various cell types (28–30). Moreover, RhoA and its downstream effector Rho kinase (ROCK) were shown to be essential in cardiac cell migration in both mice and zebrafish (31, 32). Thus, excess Rac1 activation downstream of LPA₁-Gα_i signaling might interfere with the RhoA-ROCK activation downstream of mil-Gα_12/13 signaling, leading to the cardiac cell migration defect and the two heart phenotype. Third, we speculate that endoderm cells are the cells in which the functional interaction occurs. It was reported that the endoderm cells that are associated with migrating cardiac cells expressed significant amount of s1pr2/mil mRNA (15, 27). In addition, the endoderm cells are in the vicinity of yolk syncytial layer that expresses spns2 and thus produces S1P (16). Lpar1 and s1pr2/mil showed a similar expression pattern in the heart field of developing zebrafish embryos (21), supporting the hypothesis.

We also examined if endogenous LPA signaling down-regulates S1P signaling. However, the mil or spns2 MO-induced cardia bifida phenotype was not rescued by lpar1 MO or Ki16425 (data not shown), suggesting that endogenous LPA₁ signaling does not suppress the S1P signaling in zebrafish cardiac cell migration. Recently, several studies have indicated that excessive Atx-LPA₁ signaling leads to the development of several chronic diseases such as lung fibrosis and arthritis (33–35). It is interesting to examine if S1P signaling is suppressed in such diseases and up-regulation of S1P signaling leads to the treatment.

In conclusion, we found two opposite effects of LPA and S1P signaling in zebrafish cardiomyocyte migration. The present results raise the possibility that LPA signaling acts as a modulator of S1P signaling in vivo. Further analyses will be necessary to elucidate the precise molecular mechanism of the interaction between LPA and S1P signaling at the cellular level.

Funding

This study was supported by Grant in aid for Scientific Research from the Ministry of Education, Culture, Sports, Science, and Technology of Japan (to K.H., A.I. and J.A.), the National Institute of Biomedical Innovation (to J.A.), Japan Science and Technology Agency Precursory Research for Embryonic Science and Technology (PRESTO) (to A.I.), Core Research for Evolutional Science and Technology (CREST) (to J.A.).

Conflict of interest

None declared.

Abbreviations

Abbreviations
 
• Atx

  autotaxin

 
• cmlc2

  cardiac myosin light chain 2

 
• Edg

  endothelial differentiation gene

 
• hpf

  hours post fertilization

 
• LPA

  lysophosphatidic acid

 
• LPC

  lysophosphatidylcholine

 
• MO

  antisense morpholino oligonucleotides

 
• ROCK

  Rho kinase

 
• S1P

  sphingosine-1-phosphate

References