The Sphingolipid Transporter Spns2 Functions in Migration of Zebrafish Myocardial Precursors

See allHide authors and affiliations

Science  23 Jan 2009:
Vol. 323, Issue 5913, pp. 524-527
DOI: 10.1126/science.1167449


Sphingosine-1-phosphate (S1P) is a secreted lipid mediator that functions in vascular development; however, it remains unclear how S1P secretion is regulated during embryogenesis. We identified a zebrafish mutant, ko157, that displays cardia bifida (two hearts) resembling that in the S1P receptor-2 mutant. A migration defect of myocardial precursors in the ko157 mutant is due to a mutation in a multipass transmembrane protein, Spns2, and can be rescued by S1P injection. We show that the export of S1P from cells requires Spns2. spns2 is expressed in the extraembryonic tissue yolk syncytial layer (YSL), and the introduction of spns2 mRNA in the YSL restored the cardiac defect in the ko157 mutant. Thus, Spns2 in the YSL functions as a S1P transporter in S1P secretion, thereby regulating myocardial precursor migration.

During the late stages of zebrafish segmentation characterized by the formation of the somites, the myocardial precursors from both sides of the anterior lateral plate mesoderm migrate toward the midline to form the heart tube (1, 2). Forward genetic analysis in zebrafish has helped to uncover genes involved in vertebrate heart formation (3). To identify additional regulators of heart development, we performed N-ethyl-N-nitrosourea (ENU) mutagenesis screening for mutations specifically affecting cardiac morphogenesis. We isolated a recessive ko157 mutant that displayed two hearts, a condition known as cardia bifida with swollen pericardial sacs (Fig. 1, A, B, E, and F). The expression of myocardial markers [nkx2.5 and cardiac myosin light chain 2 (cmlc2)] and chamber-specific markers [atrial myosin heavy chain (amhc) and ventricular myosin heavy chain (vmhc)] was detected in two separated domains (Fig. 1, C and G, and fig. S2); this finding suggests that the myocardial precursors failed to migrate but differentiated into two chambers at the bilateral positions.

Fig. 1.

Morphological phenotypes of ko157 mutants. (A, B, D, E, F, and H) Stereomicroscopic views of wild-type (Wt) embryo [(A), (B), and (D)] and ko157 mutant [(E), (F), and (H)]. Two swollen pericardial sacs (arrowheads) at 54 hours post-fertilization (hpf) were observed in ko157 mutant [(E) and (F)] but not in Wt embryos [(A) and (B)]. (B) and (F) are ventral views. (C and G) Two hearts (arrowheads) in ko157 mutants at 24 hpf were visualized (dorsal view) by whole-mount in situ hybridization with antisense cmlc2 probe. ko157 mutant (H), but not Wt embryos (D), exhibited tail blisters (arrow).

The migration of several mesodermal derivatives examined by the expression pattern of a vascular marker (fli1), an erythroid marker (gata1), a pronephric marker (pax2), and a lateral plate mesoderm marker (hand2) was not impaired in ko157 mutants (figs. S2 and S3), which suggests that the migration of myocardial precursors is dominantly affected. Besides cardia bifida, there were abnormal blisters at the tip of the tail in the mutant (Fig. 1, D and H). These two characteristic phenotypes (cardia bifida and tail blisters) in the ko157 mutant were similar to those in the miles apart (mil)/S1P receptor-2 (S1P2) mutant (4). Sphingosine-1-phosphate (S1P) is a lipid mediator involved in cell growth, death, migration, and differentiation (58). Both cardia bifida and tail blisters were observed in embryos injected with an antisense morpholino for mil/S1P2 (mil MO; 15 ng) (9) (fig. S4 and table S1), suggesting a genetic interaction between ko157 and mil/S1P2.

Genetic mapping of the ko157 mutation by means of simple sequence length polymorphism (SSLP) markers revealed that the locus of ko157 was very close to z9419 and z63525 on linkage group 5 (Fig. 2A). We found that the ko157 mutant allele contained a point mutation in the spns2 gene with a substitution of arginine to serine at amino acid position 153 (R153S). This arginine is conserved between zebrafish Spns2 and mammalian homologs of Spns2 (fig. S5). Spns1 is a member of the Spns protein family (10, 11), but injection of spns1 MO (8 ng) into wild-type embryos did not induce cardia bifida, and injection into ko157 embryos did not worsen the phenotype (fig. S6). Hence, Spns2, but not Spns1, is involved in cardiac morphogenesis.

Fig. 2.

The gene responsible for the ko157 mutant encodes spns2 and is associated with S1P signaling. (A) Genetic map of the ko157 on linkage group 5 (LG5). The numbers of recombination events between SSLP markers and the ko157 locus in 400 ko157 mutant embryos (400 diploid) are shown. (B to D) Injection of spns2 MO (10 ng) but not 5-mis MO (10 ng) led to cardia bifida, as determined by mRFP expression (cmlc2 expression domain) (arrowheads). (E to G) Cardia bifida in ko157 was recovered by injection of spns2 mRNA (250 pg) but not spns2 (R153S) mRNA (250 pg) at the one-cell stage. (H and I) Cardia bifida was observed when mil MO (2 ng) was injected into heterozygous (Wt/ko157) but not Wt (Wt/Wt) embryos. (J) Cardia bifida in ko157 was restored by S1P (1 ng) injection in deep area of the blastomeres at blastula stage. Genotypes were confirmed by direct sequencing of the ko157 locus [(E) to (J)]. All views are ventral at 28 hpf.

To examine whether the mutation in Spns2 caused the functional impairment in ko157 mutants, we performed knockdown analysis with antisense morpholino (spns2 MO). The spns2 MO injection, but not control morpholino, (5-base mismatched control morpholino for spns2 MO; 5-mis MO), suppressed the production of the mature form of spns2 mRNA (fig. S7). Injection of spns2 MO resulted in cardia bifida (table S1; 86%, n = 69) with bilateral expression of cmlc2 (Fig. 2, B to D), with no cardia bifida in control 5-mis MO–injected embryos (table S1; 0%, n = 68). To evaluate whether spns2 could rescue the ko157 mutant phenotype, we injected spns2 or spns2(R153S) (spns2 mutant) mRNA (250 pg) into the embryos derived from spns2ko157 heterozygous carriers. Using more than 250 pg caused severe defects in the trunk and the tail as well as observation of one beating heart (fig. S8). Injection of spns2 mRNA, but not mutant spns2(R153S) mRNA, effectively restored both the migration of myocardial precursors and the tail blisters (Fig. 2, E to G, fig. S9, and table S2). Injection of human Spns2 (hSpns2) mRNA also rescued the cardia bifida phenotype in the spns2ko157 mutant, whereas injection of the corresponding hSpns2 mutant, hSpns2(R199S), did not (fig. S10 and table S3). The spns2ko157 mutant phenotype was not restored by injection of a construct in which human Spns1 is fused to enhanced green fluorescent protein (EGFP), hSpns1-EGFP (fig. S10 and table S3). These results show that Spns2 function is conserved from fish to mammals and that Spns1 cannot compensate for the loss of Spns2.

Because the cardia bifida phenotype in the spns2ko157 mutant was similar to that in the mil/S1P2 mutant, we investigated a possible genetic interaction between spns2 and mil/S1P2. Injection of mil MO induced cardia bifida in embryos derived from a wild-type–Tg(cmlc2:mRFP) cross in a dose-dependent manner (table S1, mil MO 15-ng injection; 90%, n = 49; table S4, mil MO 2-ng injection; 7%, n = 55). Low-dose mil MO (2 ng) injections in spns2ko157 heterozygous embryos resulted in a higher frequency of cardia bifida relative to wild-type embryos (table S4; 44%, n = 18). Genotyping revealed that most cardia bifida embryos were wild-type/ko157 (Fig. 2, H and I, and table S4). The severity of the cardiac defect was comparable between mil MO (15 ng)–injected wild-type and mil MO–injected ko157 embryos (fig. S11). These data suggest that spns2 genetically interacts with mil/S1P2.

To further examine the functional interaction of Spns2 and S1P signaling, we performed a rescue experiment of the spns2ko157 mutant by S1P injection. When S1P (1 ng) was injected deep into an area of the blastomeres of blastula-stage embryos derived from spns2ko157 heterozygous carriers, cardia bifida was effectively restored (Fig. 2J and table S2). In addition, the cardia bifida phenotype induced by mil MO (15 ng) injection into the yolk was not rescued by subsequent injection of spns2 mRNA (250 pg) into the blastomere of one- to two-cell–stage embryos (fig. S12), which suggests that Spns2 functions upstream of Mil/S1P2 (fig. S1).

The putative 12-transmembrane domains, together with the predicted structural similarity between zebrafish Spns2 (zSpns2) and the bacterial glycerol-3-phosphate transporter (12) and the genetic interaction of Spns2 and the S1P-mediated signaling, suggested that Spns2 might function as a S1P transporter. To test this, we used Chinese hamster ovary (CHO) cells expressing a sphingosine kinase, mSphK1, essential for S1P synthesis (CHO-SphK cells) (fig. S1). Although [3H]sphingosine was taken up by the cells and effectively converted to [3H]S1P, it was not secreted because of the absence of an S1P export activity. We examined whether the expression of zSpns2-EGFP or zSpns2(R153S)-EGFP was able to induce S1P export. Both zSpns2-EGFP and zSpns2(R153S)-EGFP were predominantly localized within the plasma membrane and in the endosomes of transfected CHO-SphK cells (Fig. 3, A to C), consistent with a role in membrane trafficking. Expression of zSpns2-EGFP resulted in a time-dependent export of [3H]S1P that was not seen in either the EGFP- or zSpns2(R153S)-EGFP–transfected cells (Fig. 3, D and E, and fig. S13). Moreover, endogenous S1P release was also detected only in the medium from the zSpns2-EGFP–expressing cells (fig. S14) without altering the content of cellular sphingolipids (fig. S15). Overexpression of hSpns2-EGFP enhanced S1P export to a similar extent as zSpns2-EGFP, whereas that of hSpns2(R199S)-EGFP and hSpns1-EGFP did not (Fig. 3, D and E). S1P release was not due to cell death induced by Spns2-EGFP expression (fig. S16), and the activity of sphingosine kinase in the medium was not affected by Spns2-EGFP expression (fig. S17).

Fig. 3.

Plasma membrane–localized Spns2 exports S1P from the cells. (A to C) Confocal fluorescence microscopy images of CHO cells expressing mouse SphK1 transfected with the plasmids indicated. (D) [3H]S1P converted from [3H]sphingosine in the lipids extracted from medium (S, supernatant) and cells (P, pellet) was separated on a thin-layer chromatography plate. Cer, Sph, and S1P indicate the positions of [3H]ceramide, [3H]sphingosine, and [3H]S1P, respectively. (E) Relative amount of secreted S1P indicates the percentage of total [3H]S1P (P + S). Data are expressed as means ± SD of more than three independent experiments of (D).

Recently it was proposed that ABC transporters including ABCC1 and ABCA1 are required for S1P transport (1315). The cellular distribution of Spns2-EGFP is similar to that of ABCA1 (16). Therefore, the net S1P release to the outside of the cells would depend on the amount of Spns2 and other S1P transporters expressed on the plasma membrane. The contribution of ABC transporters in the S1P transport is still controversial in vivo because the quantity of plasma S1P is not altered in mice deficient for ABCC1 or ABCA1 (17). We propose that Spns2 is a S1P transporter essential for the S1P-mediated signaling pathway in vivo (fig. S1).

To further understand where and how Spns2 functions in vivo, we examined spns2 expression during early embryogenesis by whole-mount in situ hybridization. The expression of spns2 was induced at the marginal cells of the blastoderm at dome stage (fig. S18). During gastrulation stages, spns2 was predominantly expressed in the extraembryonic yolk syncytial layer (YSL) with a dorsal-to-ventral gradient (Fig. 4A and fig. S18). spns2 expression in the YSL was strongly detected just below the developing myocardial precursors and was maintained throughout the segmentation period (Fig. 4, B to D). Recent evidence demonstrated that both endoderm and YSL regulate cardiac morphogenesis (1820). Expression of the endoderm markers sox17 and foxa2 was not affected in spns2ko157 mutant embryos (fig. S3). In addition, spns2 expression was not affected in endoderm-defective casanova/sox32 morphants (fig. S18), which suggests that spns2 expression in the YSL is regulated independently of the endoderm. spns2 expression was detected in somites and in the tip of the tail at the 15-somite stage (fig. S18). We observed spns2 expression in the myocardial precursors and in the intermediate cell mass (fig. S18). Thus, the expression of spns2 is complex and dynamic. Although the overall morphology of the head and trunk appeared to be normal in the spns2ko157 mutant, we observed substantially increased apoptotic cells in the tail but not in the heart, head, and anterior trunk of the spns2ko157 mutant (fig. S19). These results suggest the involvement of Spns2 in the regulation of multiple organogenesis processes.

Fig. 4.

Spns2 in the YSL is required for the migration of myocardial precursors. (A to D) Whole-mount in situ hybridization with antisense spns2 probe at different stages of development. (B) and (D) show transverse sections of spns2-stained embryos. spns2 was expressed in the YSL with a dorsal-to-ventral gradient at 80% epiboly stage [(A); dorsal right, lateral view]. spns2 expression was maintained in the YSL at bud and 15s stages [(B) and (D); red arrowheads] and detected under the anterior midline [(C); dorsal up, anterior view]. (E to H) Injection of spns2 MO (10 ng) with fluorescein isothiocyanate (FITC)–dextran into the YSL of shield-stage embryos led to cardia bifida, whereas injection of mil MO (15 ng) did not. (I to L) Cardia bifida in spns2ko157 mutant was recovered by mRNA injection of spns2 (250 pg) with FITC-dextran into the YSL of shield-stage embryos but not by mRNA injection of spns2(R153S) (250 pg). Genotypes were confirmed by direct sequencing of the ko157 locus. Injection into the YSL was confirmed by the distribution of FITC-dextran in the YSL at gastrulation stages [(F), (H), (J), and (L)]. White arrowheads indicate the positions of cmlc2 expression domain [(E), (G), (I), and (K)].

Because spns2 was strongly expressed in the YSL below the developing myocardial precursors, we further examined whether Spns2 in the YSL contributes to the migration of myocardial precursors. When spns2 MO (10 ng) was injected into the YSL at shield stage, cardia bifida was observed (Fig. 4, E and F, and table S5). In contrast, mil MO (15 ng) injection in the YSL at shield stage did not induce cardia bifida (Fig. 4, G and H, and table S5). Because mil/S1P2 is expressed in the mesoderm just lateral to the midline (4), mil/S1P2 is proposed to function in mesoderm over the YSL. Further, cardia bifida in the spns2ko157 mutant was restored by the injection of spns2 mRNA but not spns2(R153S) mRNA into the YSL at shield stage (Fig. 4, I to L, and table S6).

The rescue frequency by injection of spns2 mRNA into the YSL was slightly lower than for injection into the blastomere (tables S2 and S6). One explanation is that spns2 mRNA injected into the blastomere at the one-cell stage is widely distributed in the YSL because the YSL is constituted by marginal blastomeres collapsing onto the yolk around the 1000-cell stage. Another explanation is that the function of Spns2 in embryonic tissues as well as in the YSL may be partly required for the migration of myocardial precursors. Furthermore, transplantation analysis showed that Spns2 at least functions in a cell-nonautonomous manner, because ko157-derived donor cells were incorporated into single beating hearts of wild-type recipients, and wild type–derived donor cells were incorporated into one of two beating hearts of ko157 recipients (movies S1 to S3). One attractive interpretation is that Spns2 in the YSL regulates the S1P export from the yolk to the embryonic body, leading to the activation of Mil/S1P2 in mesoderm just lateral to the midline (fig. S1). Recent reports have pointed out the importance of ferroportin1 (fpn1) as a transporter of iron from the yolk to the embryonic body (21) and the clinical relevance to hypochromic anemia and hemochromatosis in humans (22, 23).

By investigating characteristic features of the zebrafish spns2ko157 mutant and analyzing the biological activity of Spns2, we have demonstrated that Spns2 functions as a S1P transporter and that Spns2 in the extraembryonic YSL is a prerequisite for the migration of myocardial precursors, presumably mediated by the S1P-Mil/S1P2 pathway. The identification of Spns2 not only contributes to our understanding of the molecular mechanism of biological S1P delivery, but may also elucidate the physiological importance of Spns2 in autoimmune disease (24), cardiovascular diseases, and cancer (25) in which S1P plays a central role.

Supporting Online Material

Materials and Methods

Figs. S1 to S19

Tables S1 to S6

Movies S1 to S3


References and Notes

View Abstract

Stay Connected to Science

Navigate This Article