Distinct Functions for Different scl Isoforms in Zebrafish Primitive and Definitive Hematopoiesis

The stem-cell leukemia (SCL, also known as TAL1) gene encodes a basic helix-loop-helix transcription factor that is essential for the initiation of primitive and definitive hematopoiesis, erythrocyte and megakarocyte differentiation, angiogenesis, and astrocyte development. Here we report that the zebrafish produces, through an alternative promoter site, a novel truncated scl (tal1) isoform, scl-β, which manifests a temporal and spatial expression distinct from the previously described full-length scl-α. Functional analysis reveals that while scl-α and -β are redundant for the initiation of primitive hematopoiesis, these two isoforms exert distinct functions in the regulation of primitive erythroid differentiation and definitive hematopoietic stem cell specification. We further demonstrate that differences in the protein expression levels of scl-α and -β, by regulating their protein stability, are likely to give rise to their distinct functions. Our findings suggest that hematopoietic cells at different levels of hierarchy are likely governed by a gradient of the Scl protein established through temporal and spatial patterns of expression of the different isoforms.


Introduction
On the basis of anatomic locations of development, time of initiation, and cell type produced, vertebrate hematopoiesis can be divided into primitive and definitive programs [1][2][3]. In mouse, the primitive, or first, wave of hematopoiesis initiates in the yolk sac at about embryonic day 7.5 and produces primarily nucleated embryonic erythrocytes and macrophages [4,5]. The definitive, or second, wave of hematopoiesis is believed to originate from the intraembryonic aorta-gonads-mesonephros at approximately embryonic day 8.5 and give rise to all the mature blood cell types [6][7][8].
Similar to that of mammals, zebrafish hematopoiesis also consists of primitive and definitive programs, and produces differentiated cells analogous to most of the mature blood lineages found in mammals [9][10][11]. Zebrafish primitive erythropoiesis originates from the posterior lateral mesoderm (PLM) as a pair of bilateral stripes at approximately the five-somite stage [9,10,12]. These bilateral stripes extend anteriorly and posteriorly, and converge in the midline at the 20-somite stage to form the main structure of the intermediate cell mass (ICM), where the erythroid progenitors further develop. On the other hand, primitive myelopoiesis is believed to arise from the rostral blood island of the anterior lateral mesoderm (ALM) region at around the ten-somite stage, and produces mainly macrophages [10,13]. Compared to the onset of primitive hematopoiesis, the onset of zebrafish definitive hematopoiesis is less well defined. Preliminary studies indicate that the earliest definitive hematopoietic stem and progenitor cells arise from the ventral wall of dorsal aorta (DA) at around 26 to 30 h postfertilization (hpf) and subsequently migrate to the kidney, the adult hematopoietic organ in zebrafish, by 5 d postfertilization (dpf) [10,14,15].
Stem-cell leukemia (SCL, also known as TAL1) was originally identified as a proto-oncogene through the study of T cell acute lymphoblastic leukemia patients with a chromosomal translocation at the breakpoint of t(1;14) (p32;q11) [16][17][18]. The importance of SCL in normal hematopoiesis and angiogenesis was revealed by gene targeting analysis in mouse embryonic stem cells. Mice lacking SCL function failed to form vitelline vessels in the yolk sac and died at embryonic day 8.5 of development because of the complete absence of primitive hematopoiesis [19][20][21]. SCL-null embryonic stem cells, when injected into blastocysts, failed to contribute to any hematopoietic lineage in mouse chimeras [22,23]. These results demonstrate that SCL is essential for the generation of primitive and definitive hematopoietic cells as well as for the formation of yolk sac vessels. In addition to its pivotal role in early hematopoiesis, SCL also exerts important biological functions in subsequent hematopoietic lineage specification. Enforced SCL expression in hematopoietic cell lines favors erythroid differentiation [24,25], and ablation of SCL in adult mice impairs erythropoiesis and megakarypoiesis [26,27]. Despite its important functions, the molecular mechanisms of how SCL mediates these multiple functions remain obscure. Previous in vitro studies in human and mouse malignant hematopoietic cell lines have described several SCL isoforms involved in T cell leukemia development and differentiation of erythrocytes and megakaryocytes [28][29][30][31][32][33]. However, the existence and biological functions of these SCL isoforms in vivo have not been demonstrated.
In this study, we report that the zebrafish produces, through an alternative promoter site within exon 2, a novel scl (tal1) isoform, scl-b, that encodes a truncated protein lacking the N-terminal 118 amino acids (aa). Whole-mount in situ hybridization (WISH) demonstrated that scl-b exhibits temporal and spatial expression that overlaps but is clearly distinctive from that of the full-length scl-a: scl-b emerges first and expresses in the entire ALM and PLM and the ventral wall of DA, where the first definitive hematopoietic progenitors arise; scl-a expresses later in the posterior of the ALM and the PLM, possibly in a manner overlapping with scl-b, and is subsequently restricted to the ICM region. Loss-of-function analysis using an antisense morpholino oligonucleotide (MO) knockdown approach revealed that while scl-a and -b are redundant in the initiation of primitive hematopoiesis, these two isoforms appear to function at different stages of primitive erythroid cell differentiation. Analysis of definitive hematopoiesis of scl-a and -b MO-injected embryos (morphants) showed that the knockdown of Scl-b, but not Scl-a, protein expression resulted in the loss of c-myb (cmyb) and runx1 expression in the ventral wall of DA as well as rag1 expression in the thymus, demonstrating that scl-b but not scla is essential for definitive hematopoietic stem cell development. Interestingly, we found that scl-b and -a exhibited a remarkable difference in their protein expression levels both in vitro and in vivo, indicating that differences in the protein expression level of scl-a and -b isoforms likely confer their distinct functions in the regulation of hematopoietic cell development.

Results
Identification of a Novel Zebrafish scl Isoform, scl-b To investigate whether different scl isoforms exist in zebrafish, RNA samples were prepared from 18-somite-stage embryos and kidney, the adult hematopoietic organ in zebrafish [10], and subjected to Northern blot analysis. The result showed that two transcripts, one 2.6 kilobases (kb) and the other 2.2 kb, were specifically hybridized to the probes corresponding to the coding sequence and the 39 untranslated region (UTR) of the zebrafish scl cDNA (data not shown), suggesting that the 2.6-kb and 2.2-kb transcripts may represent two different scl isoforms. To characterize the nature of these two transcripts, we carried out a rapid amplification of cDNA ends (RACE) experiment and obtained one 39 RACE and two 59 RACE products (data not shown). DNA sequencing revealed that the larger 59 RACE product was identical to the published full-length scl sequence [34,35], whereas the smaller fragment was also identical except that it lacked the first 438 base pairs at the 59 end of the full-length scl, indicating that the 2.6-kb transcript is the full-length scl and the 2.2-kb transcript represents a novel scl isoform. This was confirmed by Northern blot analysis, which showed that while both transcripts were hybridized to the scl 39 UTR probe (39-probe), only the larger 2.6-kb species was recognized by the scl 59 414-bp probe (59-probe) ( Figure 1A). We hereafter designate the 2.6-kb full-length and the 2.2-kb truncated forms as scl-a and scl-b, respectively. Time course

Author Summary
Hematopoiesis is the process that generates all the body's blood cells. In vertebrates, hematopoietic development occurs in two phases: a transitory embryonic, or primitive, wave produces only erythrocytes (red blood cells) and myeloid cells; an adult, or definitive, wave gives rise to at least three blood cell lineages, including erythrocytes and two types of immune cells-myeloid cells and lymphocytes. Previous studies have shown that the stemcell leukemia (SCL) gene is essential for hematopoietic stem cell specification and erythrocyte maturation. Yet how SCL regulates these distinct processes is not fully understood. This study demonstrates that zebrafish produce a smaller isoform of scl, scl-b, that plays an overlapping role with the full-length scl-a in the initiation of primitive hematopoiesis, but possesses distinct functions in regulating primitive erythrocyte maturation and definitive hematopoietic stem cell development. We further show that the distinct functions of scl-a and -b are likely due to differences in their protein expression level through the regulation of protein stability. We postulate that hematopoietic cell development at different levels of hierarchy is governed by a gradient of the Scl protein created by temporal and spatial expression of different scl isoforms.
analysis by Northern blot revealed that the expression of scl-b began at the one-to two-somite stage, peaking at around the 11-to 18-somite stage, and maintained a lower level in the adult kidney ( Figure 1A). On the other hand, the scl-a transcript was not detected until the four-somite stage, after which it rapidly increased its level to that of scl-b, subsequently becoming the dominant form in the adult kidney ( Figure 1A). Considering the fact that the scl-b transcript initiates within the exon 2, where a potential TATA-box sequence can be found at position À31 of the predicted transcription initiation site ( Figure 1B), and expresses earlier than scl-a ( Figure 1A), we conclude that sclb is generated from an alternative promoter site within exon 2 encoding a truncated Scl protein lacking the N-terminal 118 aa ( Figure S1).

Temporal and Spatial Expression Patterns of scl-a and -b
WISH was performed to examine the temporal and spatial expression of scl-a and -b. As shown in Figure 2, the 39-probe, which recognized both scl-a and -b, exhibited a pattern identical to that of scl expression described previously [34,35]. It first emerged at around the two-somite stage as one pair of stripes in the PLM followed by the appearance of a second pair of stripes in the ALM at the four-somite stage (Figure 2A and 2B). These two pairs of stripes, which represented the combination of both scl-a and -b transcripts (referred to as scla/b stripes hereafter), extended anteriorly and posteriorly from the four-somite stage onwards ( Figure 2B and 2C). And by the 18-somite stage, the scl-a/b stripes were mainly localized in three regions: the ALM, the anterior of the PLM (APLM), and ICM ( Figure 2D). In contrast to the 39probe-positive signals, the 59-probe-positive signals, which represented only the full-length scl-a, appeared at the fourand six-somite stage as two pairs of stripes (scl-a stripes) in the PLM and ALM, respectively ( Figure 2H, and data not shown). By the 18-somite stage, the scl-a expression was restricted to the ICM ( Figure 2J), where primitive erythropoiesis actively occurs at this stage [10], suggesting that scl-a is predominantly expressed in the erythroid lineage from the 18-somite stage onwards. This possibility was further supported by the lack of scl-a expression in the erythrocyte-deficient mon (trim33) mutant embryos [36] ( Figure 2F and 2L). Notably by 26 hpf, only a weak signal of the 39-but not the 59-probe was detected in the ventral wall of DA ( Figure 2E and 2K), where the first definitive hematopoietic stem cells presumably emerge [10,14,15], suggesting that scl-b is the main isoform expressed in the definitive hematopoietic stem cells. To test this possiblity, we carried out double staining analysis in which 26 hpf embryos were stained with anti-Scl-a (Ab-Scl-N) or anti-Scl-a/b (Ab-Scl-C) antibodies together with c-myb WISH. The transverse section through the trunk region of these embryos showed that Scl-b but not Scl-a protein was present in the ventral wall of DA in a manner partially overlapping with c-myb-positive cells ( Figure 2M-2R), indicating that the definitive hematopoietic stem cells predominantly express sclb. Based on these observations, we conclude (as illustrated in Figure 2S) that scl-b appears first and expresses in the entire ALM and PLM regions; scl-a emerges later in the ALM and ICM, possibly in a manner overlapping with scl-b, and is subsequently restricted to the ICM by the 18-somite stage. Notably, only scl-b expresses in the ventral wall of DA where the first definitive hematopoietic stem cells arise.

Redundant Roles for scl-a and -b in the Initiation of Primitive Hematopoiesis
To determine the biological functions of scl-a and -b in hematopoiesis, two MOs, scl-a MO and scl-b MO ( Figure 1B), that specifically inhibited the protein syntheses of scl-a and -b, respectively, were injected into wild-type zebrafish embryos. Immunohistochemistry staining showed that scl-a MO and -b MO specifically abolished the Scl-a and -b protein expression as indicated by lack of anti-Scl-N staining in the ICM of scl-a morphants ( Figure 3N We reasoned that the lack of phenotypes in both morphants was likely due to the functional redundancy of scl-a and -b. To test this possibility, we co-injected scl-a MO and scl-b MO to block both protein syntheses ( Figure 3P, n ¼ 45/45; and 3T, n ¼ 50/50), and found that the expression of gata1, be1-globin, and pu.1 were either absent or drastically reduced in the co-injected morphants ( Figure 3D, n ¼ 42/42; 3H, 46/46; and 3L, n ¼ 47/ 47). This phenotype is very similar to that found in the scl-sp morphants, in which both Scl-a and -b protein expression were eliminated by scl-sp MO ( Figure 1B) that interfered with the splicing between exon 2 and 3 of the scl gene [37,38]. Furthermore, injection of in vitro synthesized scl-a or -b mRNA was sufficient to rescue the expression of gata1, be1-globin, and pu.1 in the scl-sp morphants (data not shown). Taken together, we conclude from these results that scl-a and -b are functionally redundant in the initiation of primitive hematopoiesis.

scl-a and -b Play Distinct Roles in Primitive Red Blood Cell Differentiation
To address their functions in the late developmental stages of primitive erythropoiesis, we examined the scl-a and -b morphants beyond 30 hpf. Although primitive erythropoiesis initiated normally, o-dianisidine staining revealed that the red blood cells (RBCs) in the scl-b morphants were significantly reduced by 2 dpf (data not shown) and finally not detectable by 3 dpf (Figure 4C). On the other hand, RBCs in the scl-a morphants were normal before 3 dpf ( Figure 4B) but began to decrease by 4 dpf (data not shown) and were severely reduced by 5 dpf ( Figure 4E). These data indicate that the loss of either Scl-a or -b protein renders abnormal RBC differentiation at different developmental stages, eventually resulting in anemia. To provide an additional test of this possibility, circulating RBCs were collected from the scl-a and -b morphants, stained with May-Grunwald Giemsa, and compared to those from wild-type embryos. Based on the size of cell, shape of nucleus, and staining of cytoplasm, normal primitive RBCs from 30 hpf to 5 dpf can be classified into four main stages: stage I, basophilic erythroblast; stage II, polychromatophilic erythroblast; stage III, orthochromatophilic erythroblast; and stage IV, mature erythrocyte ( Figure  4G). In the 2-dpf wild-type embryos and 2-dpf scl-a morphants, more than 98% of the circulating RBCs were at stage II ( Figure 4H and 4I). However, 90% of the circulating RBCs arrested at stage I in the 2-dpf scl-b morphants ( Figure  4J). As expected, although the circulating RBCs developed normally before 3 dpf, 80% of them were blocked at stage II in the 4-dpf scl-a morphants ( Figure 4L), compared to the number of RBCs at stage III in the 4-dpf wild-type embryos ( Figure 4K). These phenotypes concurred with the virtual Northern blot result showing that circulating RBCs from 30hpf wild-type embryos contain both scl-a and -b, whereas those from 2-dpf wild-type embryos express predominantly scl-a ( Figure 5). Collectively, we conclude that scl-b plays a critical role in the differentiation of basophilic erythroblasts to polychromatophilic erythroblasts, whereas scl-a is pivotal for the transition from polychromatophilic erythroblasts to orthochromatophilic erythroblasts.   Figure 6B, n ¼ 45/46; and 6E, n ¼ 43/45). As artery endothelial cells appeared to be retained in both scl-a and -b morphants, as indicated by two artery-specific markers deltaC (dlc) and grl (hey2) ( Figure S2), these data indicate that scl-b, but not -a, is essential for definitive hematopoietic stem cell development. This result is consistent with the finding that only scl-b is expressed in the ventral wall of DA (Figure 2). To further test whether scl-b is indeed required for the development of definitive hematopoietic stem cells, we investigated T cell development in both morphants by examining rag1 expression at 5 dpf. As expected, rag1 was detected in the thymus of control embryos and scl-a morphants ( Figure 6G, n ¼ 30/30; and 6H, n ¼ 42/45) but not scl-b morphants ( Figure 6I, n ¼ 46/50). Taken together, these data demonstrate that scl-b is essential for the development of definitive hematopoietic stem cells while scl-a is dispensable.

Differences in the Protein Expression Level of Scl-a and -b Give Rise to Their Distinct Functions
To gain insight into the molecular basis underlying the distinct functions of scl-a and -b in hematopoietic cell development, we performed rescue experiments-by coinjecting in vitro synthesized scl-a or scl-b mRNA with scl-sp MO into wild-type embryos-to test whether the lack of the Nterminal 118 aa could lead to the differences in their biological functions. Examination of the expression of be1globin and c-myb by WISH showed that either isoform was sufficient to rescue both primitive and definitive hematopoietic defects in the scl-sp morphants ( Figure 7A The data indicate that the N-terminal 118-aa segment is not essential for determination of the functions of these two Scl protein isoforms, consistent with previous findings showing that the basic helixloop-helix domain of the murine SCL protein is sufficient for hematopoiesis [39]. As recent studies have suggested that different SCL protein expression levels are required at different levels of hematopoietic hierarchy [40,41], these findings raise the possibility that the distinct functions of scla and -b are due to differences in their protein expression levels, resulting in establishment of a gradient of Scl protein at different stages of hematopoietic hierarchy. To test this hypothesis, we performed immunoblotting of protein extracts from 18-somite-stage wild-type embryos, in which the transcription levels of the scl-a and -b isoforms are similar (the scl-b RNA is slightly higher) ( Figure 1A). We found that the protein expression level of Scl-b was much lower than that of Scl-a in these embryos (the Scl-b protein was hardly detectable at all by Western blot; Figure 7I). When COS7 cells were transiently transfected with construct expressing either the full-length scla or scl-b, the protein expression level of Scl-b was also much lower than that of Scl-a, while their RNA levels were comparable ( Figure 7J). These data strongly indicate that a post-transcriptional mechanism is involved in the regulation of Scl-b protein expression level. To test this, equal amounts of in vitro synthesized scl-a and -b mRNA were injected into one-cell-stage wild-type embryos, and protein levels were examined at different time points post-injection ( Figure 7K). As anticipated, real-time reverse transcriptase PCR analysis revealed that mRNAs of both isoforms behaved similarly in these injected embryos (data not shown). Immunoblotting of whole embryo protein extracts showed that protein expres- sion levels were comparable at 3 h post-injection, indicating that both isoforms are effectively translated. However, by 4 h post-injection, Scl-b protein level was greatly reduced, while Scl-a level increased ( Figure 7K), showing that the low protein expression level of Scl-b was likely due to the rapid degradation of its protein. Taken together, these data strongly indicate that differences in the protein expression levels of Scl-a and -b isoforms, via regulation of their protein stabilities, likely confer their distinct functions in the regulation of hematopoietic cell development.

Discussion
It is believed that the complexity in morphology and behavior of higher organism is achieved not only by higher gene numbers, but also by multiple protein isoforms being encoded by a single gene locus and by the complexity of protein-protein interactions. The most well studied phenomenon that results in the generation of multiple protein isoforms from a single gene is alternative splicing of pre-mRNA [42]. However, other mechanisms such as use of an alternative promoter-a phenomenon that is as equally widespread in higher organisms as alternative pre-mRNA splicing [43,44]-are less appreciated. In this article, we described how zebrafish produce, through alternative promoter sites, two scl isoforms, the full-length scl-a and a novel truncated scl-b (Figure 1). We further showed that these two scl isoforms manifest distinct temporal and spatial expression ( Figure 2) and exert distinct functions in the regulation of primitive and definitive hematopoiesis (Figures 3-6). The scl-a morphants (lane 2), in which Scl-a protein was not detected, were used as the controls to distinguish the Scl-a protein (arrow) and nonspecific band (asterisk). (J) Immunoblotting with Ab-Scl-C antiserum shows the expression levels of Scl-a and Scl-b proteins in transfected COS7 cells. Western blotting of whole cell protein extracts (10 lg for each sample) prepared from COS7 cells transfected with full-length scl-a, scl-b, or blank vector constructs revealed that the protein level of Scl-a was much higher than that of Scl-b. RT-PCR analysis (bottom) showed similar RNA levels of scl-a and -b in these transfected cells. Tubulin and GAPDH were used as controls for Western blot and PCR, respectively. (K) Immunoblotting with Ab-Scl-C antiserum shows the expression levels of Scl-a and Scl-b proteins in the embryos injected with in vitro synthesized scl-a or -b mRNA. Western blotting of whole embryo protein extracts (50 lg for each sample) from embryos injected with 500 pg of scl-a or -b mRNA showed that, at 3 h post-injection, the Scl-b protein was detected as being at a comparable level to that of Scl-a but dramatically reduced by 4 h post-injection. doi:10.1371/journal.pbio.0050132.g007 identification of the alternative-promoter-generated scl-b isoform in zebrafish has not only revealed new insight into the roles of scl in the regulation of hematopoietic cell development, but also provided another example to highlight the importance of alternative promoter usage in generating protein and regulatory diversity.
Previous studies have revealed that mammals contain several SCL isoforms generated by either alternative splicing, alternative promoters, or alternative translation initiation sites [28][29][30][31][32][33]. These mammalian SCL isoforms arise from alternative promoters in exon 1a and exon 1b, and encode identical proteins, the full-length SCL. In contrast, scl-b in zebrafish is generated through an alternative promoter site within the scl exon 2, the equivalent of the mammalian SCL exon 4, and encodes an N-terminal truncated protein (Figures 1 and S1). Thus, scl-b is clearly distinct from the previously described mammalian SCL isoforms. One intriguing question raised is whether the mechanism of generating different scl isoforms found in zebrafish is evolutionarily conserved in higher vertebrates, especially in mammals. Considering the facts that (1) the SCL locus is highly conserved in vertebrates [45], (2) there are observations showing that the murine SCL exon 4 has promoter activity in the context of SCL 39 stem cell enhancer [46], and (3) a truncated SCL transcript initiated from exon 4 can be detected in some of the human T cell leukemia cell lines [28,29], we speculate that a scl-b equivalent may exist in higher vertebrate species.
Another interesting issue raised by this study is the regulation of scl-a and -b transcripts during hematopoiesis. Notably, scl-b first appears in hematopoietic stem and progenitor cells and soon diminishes in the differentiated primitive RBCs (Figures 2 and 5). In contrast, scl-a emerges later and is predominantly restricted to RBCs (Figures 2 and  5). Thus, it appears that an on-off switch, from scl-b to -a expression, must occur during primitive RBC development. Considering the facts that (1) the earliest definitive hematopoietic stem/progenitor cells located in the ventral wall of DA express only scl-b and (2) scl-a becomes the predominant isoform expressed in the adult kidney marrow, where definitive hematopoiesis takes place presumably from 5 dpf onwards in zebrafish development, it is conceivable to speculate that this on-off switch may also exist during definitive erythroid cell development. However, it is unclear at this moment whether this on-off switch takes place at the transcriptional level or the post-transcriptional level, or perhaps a combination of both. Nevertheless, we believe that the on-off switch of these scl isoforms must play a crucial role in normal hematopoietic cell development, at least for RBCs, and that the underlying molecular basis of this regulation warrants further studies.
Our study has provided evidence indicating that differences in the protein expression levels of the scl-a and -b isoforms are likely to confer their distinct functions in regulating hematopoietic cell development (Figure 7). Although we cannot rule out the possibility that translational control may contribute to the regulation of their protein expression levels, the fact that the Scl-b protein was initially expressed at a level comparable to that of Scl-a but soon reduced dramatically upon injection of equal amounts of in vitro synthesized scl-a and -b mRNA ( Figure 7K) strongly indicates that differences in the protein expression levels of the scl-a and -b isoforms are largely due to the rapid turnover of the Scl-b protein. However, the triggers causing the onset of rapid degradation of Scl-b protein in vivo are unclear. Given the differences in their N-terminal residues, one could speculate that, perhaps, the short half-life of the Scl-b protein is mediated through the N-end rule degradation, a common proteolytic pathway that is present in prokaryotes, fungi, plants, and animals [47]. Further biochemical analyses are required to clarify this issue.
Finally, our data strongly suggest that the establishment of an appropriate Scl protein gradient at different levels of hematopoietic hierarchy-a low level in hematopoietic stem and progenitor cells and a high level in differentiated RBCsis essential for hematopoietic cell development. The phenomenon of a lower Scl protein level in hematopoietic stem and progenitor cells correlating to the importance of hematopoietic cell development is intriguing. One possibility is that the specification of definitive stem and progenitor cells requires the low concentration of Scl protein, which occurs by preferential expression of scl-b during early hematopoiesis. However, the fact that injection of either scl-a or -b mRNA is sufficient to rescue the c-myb expression at 30 hpf in the ventral wall of DA in scl-sp morphants (Figure 7) suggests that this may not be the case. A high Scl protein level, which is known to be required for the maturation of RBCs [40,41], has the tendency to promote hematopoietic stem and progenitor cell differentiation into erythroid lineage, so a more likely explanation is that the low concentration of the Scl-b protein ensures the proper expansion of these cell pools by promoting their proliferation rather than their differentiation. In addition, the low concentration of Scl-b may also be crucial for maintaining an unbiased differentiation potential of hematopoietic stem and progenitor cells during ontogeny. Uncovering the molecular basis of scl-aand -b-mediated actions will provide further insight into our understanding of the specification, proliferation, and differentiation of hematopoietic lineages.

Materials and Methods
Zebrafish. Zebrafish were maintained at 27 to 28 8C as described in [48]. The clo s5 and mon tg234 mutants were kindly provided by Didier Y. R. Stainier (University of California San Francisco, United States) and Artemis Pharmaceuticals (Germany), respectively.
Sequence alignment analysis. The T-Coffee method [49] was used for protein sequence alignment.
Northern and virtual Northern blot. Total RNA from different stages of embryos, adult kidney, and RBCs were prepared using the RNeasy Kit (Qiagen, http://www.qiagen.com) according to the manufacturer's instructions. mRNA was purified using the Nucleo-Trap mRNA mini kit (Macherey-Nagel, http://www.macherey-nagel. com) according to the manufacturer's instructions. Three micrograms of each embryonic mRNA and 30 lg of total kidney RNA were used for Northern blot analysis ( Figure 1A). For virtual Northern blot, total RNA from 30-hpf and 2-dpf embryos and RBCs were reverse transcribed into cDNA and then amplified by the SMART PCR cDNA synthesis kit (Clontech). The amplified cDNAs were used as targets for hybridization with the DIG-labeled scl-59 and scl-39 probes. DIG labeling was carried out by PCR amplification with the DIG Probe Synthesis Kit (Roche Applied Science, http://www. roche-applied-science.com) according to the manufacturer's instructions. Northern blot and virtual Northern blot analyses were performed as previously described [50,51].
O-dianisidine staining. Fish embryos were stained for 15 min in the dark in o-dianisidine staining solution as previously described [12].
Whole-mount in situ hybridization. Generation of the DIG-labeled anti-sense RNA probes and whole-mount in situ hybridization were performed as described in [48]. For transverse cryosection, embryos were embedded in OCT solution (Sakura, http://www.sakura.com) after WISH and immunochemistry staining. Embedded embryos were sectioned at a thickness of 10 lm using a Leica (http://www.leica.com) microtome. The sections were mounted on glass slides and imaged using a Zeiss (http://www.zeiss.com) AxioPhot 2 imaging system.
Immunohistochemistry analysis. Anti-zebrafish Scl-a N-terminal (aa 17 to 62, referred to as Ab-Scl-N) and Scl-a/b C-terminal (aa 255 to 325, referred to as Ab-Scl-C) antisera were generated by immunizing rabbits with the GST-Scl-N and GST-Scl-C fusion proteins using standard protocol. Antibody purification and immunohistochemistry staining were carried out as previously described [52].
MO injection. Anti-sense MOs (Gene Tools) were designed as follows: scl-a MO, 59-gctcggatttcagtttttccatcat-39; scl-b MO, 59-gcggactcaactgcaccattcgagt-39; scl-sp MO, 59-agatttaaaatgctcttaccatcgt-39. Three nanograms of scl-a MO, 8 ng of scl-b MO, and 8 ng of scl-sp MO mixed with phenol red were separately injected into wild-type embryos at the one-cell stage. Similarly, 3 ng of scl-a MO and 8 ng of scl-b MO were used for the co-injection experiment. Wild-type embryos injected with a mixture of sterile water and phenol red were used as control.
May-Grunwald Giemsa staining of embryonic RBCs. Fish embryos were anesthetized in calcium-and magnesium-free PBS (pH 7.4) containing 0.02% tricaine (Sigma-Aldrich, http://www.sigmaaldrich. com) and 1% BSA (Sigma-Aldrich). After tail clipping using surgical scissors, blood cells were collected by pipetting and cytospun onto slides by centrifugation at 450 rpm for 3 min using a Cytospin 4 (Thermo Scientific, http://www.thermo.com). The slides were then airdried and subjected to May-Grunwald Giemsa staining according to the standard protocol.
In vitro RNA synthesis and micro-injection. In vitro transcription was carried out using the mMESSAGE mMACHINE sp6 kit (Ambion, http://www.ambion.com) according to the manufacturer's instructions. For rescue experiments, 200 pg of scl-a or -b RNA was co-injected with 8 ng of scl-sp MO into wild-type embryos at the one-cell stage.
Protein stability analysis. For protein stability analysis in fish embryos, 500 pg of in vitro synthesized scl-a or -b mRNA was injected into embryos at the one-cell stage, and protein extracts were prepared as described in [48]. Total RNA from both scl-aand -binjected embryos at 6 h post-injection were extracted and reverse transcribed using random hexamer as primer. The cDNAs were examined with real-time PCR using scl-specific primers (scl-961F, 59ctagtgcgggacgacctc-39; scl-1528R, 59-ggaactaaactgtgccga-39), which can amplify both injected scl-a and -b mRNA. COS7 cells were maintained in DMEM (Gibco, http://www.invitrogen.com/content. cfm?pageid¼11040) supplemented with 10% bovine calf serum (Hyclone, http://www.hyclone.com). Transient transfection was carried out by SuperFect Reagent (Qiagen) according to the manufacturer's protocol. Real-time RT-PCR was performed to ensure that the transfection efficiency was similar using the same RT-PCR protocol for protein stability analysis in fish embryos. Cell extract preparation and Western blot were carried out as described previously [50].