Erythropoiesis is regulated by the transcription elongation factor Foggy/Spt5 through gata1 gene regulation


  • Takuya Taneda,

    1. Graduate School of Bioscience and Biotechnology, Tokyo Institute of Technology, Midori-ku, Yokohama, Kanagawa 226-8503, Japan
    Search for more papers by this author
  • Wenyan Zhu,

    1. Graduate School of Bioscience and Biotechnology, Tokyo Institute of Technology, Midori-ku, Yokohama, Kanagawa 226-8503, Japan
    Search for more papers by this author
    • Present address: BASF Polyurethane Specialties (China) Co., Ltd, Pudong, Shanghai 200137, China.

  • Qingfu Cao,

    1. Graduate School of Bioscience and Biotechnology, Tokyo Institute of Technology, Midori-ku, Yokohama, Kanagawa 226-8503, Japan
    Search for more papers by this author
  • Hajime Watanabe,

    1. Division of Advanced Science and Biotechnology, Graduate School of Engineering Osaka University, Yamadaoka, Suita, Osaka 565-0871, Japan
    Search for more papers by this author
  • Yuki Yamaguchi,

    1. Graduate School of Bioscience and Biotechnology, Tokyo Institute of Technology, Midori-ku, Yokohama, Kanagawa 226-8503, Japan
    Search for more papers by this author
  • Hiroshi Handa,

    Corresponding author
    1. Graduate School of Bioscience and Biotechnology, Tokyo Institute of Technology, Midori-ku, Yokohama, Kanagawa 226-8503, Japan
    2. Solutions Research Laboratory, Tokyo Institute of Technology, Midori-ku, Yokohama, Kanagawa 226-8503, Japan
    Search for more papers by this author
  • Tadashi Wada

    Corresponding author
    1. Solutions Research Laboratory, Tokyo Institute of Technology, Midori-ku, Yokohama, Kanagawa 226-8503, Japan
    Search for more papers by this author
    • Present address: Division of Molecular and Cellular Biology, Department of Supramolecular Biology, Graduate School of Nanobioscience, Yokohama City University, Tsurumi-ku, Yokohama, Kanagawa 230-0045, Japan.

  • Communicated by: Masayuki Yamamoto (Tohoku University) or


Recent studies have showed that transcription elongation factors regulate early development. Foggy/Spt5 is a subunit of DRB sensitivity-inducing factor, which negatively and positively regulates transcription elongation. Here, we report that the positive function of Foggy/Spt5 is required for gata1 expression during zebrafish embryonic hematopoiesis. Antisense morpholino oligonucleotide (MO)-mediated knockdown of foggy/spt5 has led to a reduction in the expression of gata1 and the gata1 target genes alas2 and hbae3 and inhibited proper hemoglobin production. By contrast, expression of hematopoietic stem cell and endothelial markers, including scl, lmo2, gata2, fli-1, and flk-1, and expression of biklf, whose product directs gata1 expression via its direct binding to the gata1 promoter, were unaltered, suggesting that gata1 is a functionally important target gene of Foggy/Spt5. The MO-mediated gata1 repression was relieved by forced expression of wild-type foggy/spt5, but not by a mutant lacking the positive function. Therefore, this study provides evidence that Foggy/Spt5 plays an important role in gata1 gene expression and erythropoiesis through its transcriptional activation domain.


Transcription is a critical rate-limiting process of gene expression that consists of at least three steps: initiation, elongation, and termination. Although the initiation step has been the most-studied aspect of transcriptional regulation, a number of recent studies have highlighted elongation as a highly regulated and rate-limiting step (Orphanides & Reinberg 2002; Sims et al. 2004; Price 2008). For example, before heat-shock induction, RNA polymerase II (Pol II) pauses at the promoter-proximal region of the Drosophila hsp70 gene through the action of negative elongation factors. After heat-shock induction, positive elongation factors suppress the pausing and stimulate the rate of elongation mediated by Pol II, resulting in the efficient synthesis of full-length mRNA (Andrulis et al. 2000; Kaplan et al. 2000; Wu et al. 2003). Genome-wide analyses of Drosophila embryos and human ES cells showed that promoter-proximal pausing of Pol II is prevalent at genes involved in development and response to stimuli. Although transcription of these genes is initiated normally, fully transcribed mRNA is not synthesized and therefore transcription elongation seems to function as a rate-limiting step (Muse et al. 2007; Zeitlinger et al. 2007; Core et al. 2008). Thus, transcription elongation plays physiological and developmental roles in gene expression, similar to transcription initiation.

DRB sensitivity-inducing factor (DSIF) is an evolutionarily conserved heterodimeric complex consisting of Spt4 and Spt5, and functions as both a negative and positive elongation factor in collaboration with other elongation factors (Hartzog et al. 1998; Wada et al. 1998a). DSIF and negative elongation factor (NELF) negatively regulate transcription elongation by pausing the Pol II elongation complex at the promoter-proximal region (Yamaguchi et al. 1999a). When the positive elongation factor P-TEFb is recruited by external stimuli and phosphorylates the repetitive C-terminal domains of Pol II and DSIF, the Pol II elongation complex escapes from the stalled state and enters the productive elongation mode (Wada et al. 1998b; Cheng & Price 2007). Furthermore, in downstream regions of transcribed genes, phosphorylated DSIF stimulates Pol II elongation rates in cooperation with the Paf1 complex and Tat-SF1 (Yamada et al. 2006; Chen et al. 2009). Phosphorylation of DSIF occurs at threonine residues in conserved G-S-Q/R-T-P repeats in the C-terminal region (CTR) of Spt5 and is required for its positive effect (Ivanov et al. 2000; Yamada et al. 2006). Thus, DSIF plays a pivotal role in transcription elongation in concert with several other transcription elongation factors.

Erythropoiesis involves a series of differentiation steps, and these steps are regulated by transcriptional mechanisms that are evolutionarily well conserved in vertebrates (Hsia & Zon 2005). In zebrafish embryonic hematopoiesis, erythrocytes originate from the lateral plate mesoderm, which gives rise to hemangioblasts, cardiac tissue, pronephric duct, and other tissues. Hemangioblasts differentiate into hematopoietic stem cells (HSCs) and vascular endothelial cells, and HSCs subsequently produce erythrocytes and myelocytes. Erythrocytes arise from posterior lateral plate mesoderm (PLPM), forming blood islands in the ventral tail, whereas myelocytes arise from anterior lateral plate mesoderm under the head. A key regulator of erythroid differentiation is gata1, which encodes a lineage-specific transcription factor that recognizes the WGATAR sequence and regulates erythroid and megakaryocytic gene expression (Ferreira et al. 2005). A zebrafish gata1 mutant, vlad tepes, exhibits a bloodless phenotype characterized by a severe reduction in erythroid progenitors in blood islands and few or no blood cells at the onset of circulation, despite normal expression of hematopoietic stem cell marker genes such as scl, lmo2, and gata2 (Lyons et al. 2002). Until recently, studies of hematopoietic gene expression focused on regulation of transcription initiation and chromatin structure by a variety of lineage-specific transcription factors. For example, the SCL complex, which consists of SCL/TAL1, LMO2, GATA2, GATA1, and other hematopoietic cofactors, regulates expression of a number of hematopoietic genes (Lecuyer & Hoang 2004). However, as suggested in some recent reports, transcription elongation also plays an important role in expression of hematopoietic genes, including gata1 (Sawado et al. 2003; Ito et al. 2006; Meier et al. 2006).

Although DSIF has biochemical features that suggest it is a general elongation factor, it functions in a gene- and tissue-specific manner. Immunostaining of Drosophila polytene chromosomes and a chromatin immunoprecipitation analysis indicated that DSIF is associated with numerous protein-coding genes, and this association has a pattern similar to that of Pol II (Andrulis et al. 2000; Kaplan et al. 2000 and our unpublished data). By contrast, comparative gene expression analysis showed that DSIF is required for only a small proportion of genes expressed in human HeLa cells (<1%) and zebrafish embryos (<5%) (Krishnan et al. 2008; Komori et al. 2009). Furthermore, in vivo analysis of Spt5 mutants showed the involvement of DSIF in a variety of developmental processes. For example, in zebrafish, embryos carrying the fogs30 and fogsk8 null alleles of foggy/spt5 (zspt5) show broad morphological defects, whereas a hypomorphic allele of zspt5, fogm806, causes neuron-specific defects (Guo et al. 2000; Keegan et al. 2002). In Drosophila, a hypomorphic allele of spt5, W049, causes locus-specific defects in segmental patterning (Jennings et al. 2004). Taken together, these reports suggest that DSIF has gene- and tissue-specific roles despite its general localization on chromosomes. How does this specificity arise? Identifying features common to target genes of DSIF may elucidate how DSIF functions at specific genes and tissues. Krishnan et al. (2008) have already identified some target genes of zSpt5 by microarray analysis of zebrafish fogsk8 mutant embryos. By computational analysis, they found that genes repressed by zSpt5 tend to have short introns and short gene lengths, but no distinctive characteristics of genes stimulated by zSpt5 were observed. Immediate-early genes such as hsp70 and c-fos, which are rapidly induced by extracellular stimuli, are well-known target genes of DSIF (Andrulis et al. 2000; Keegan et al. 2002; Yamada et al. 2006; Chen et al. 2009). DSIF may therefore play a specific role in the expression of rapidly or highly inducible genes, but this remains to be demonstrated. The role of CTR phosphorylation in the gene- and tissue-specific roles of DSIF also remains unclear.

To identify primary target genes of zSpt5 and to compare the induction patterns of target genes and nontarget genes, we carried out a time-course microarray analysis and morpholino oligonucleotide (MO)-mediated gene knockdown using early-stage zebrafish embryos. Our results suggest that zSpt5 plays an erythroid-specific role in early embryogenesis through the induction of gata1 gene expression and that the zSpt5 CTR is important to this regulation.


MO-mediated knockdown of zspt5 phenocopies fogsk8

A search for target gene candidates of zSpt5 using a microarray analysis has already been carried out in fogsk8 homozygous mutant embryos, which have severely truncated alleles of foggy/spt5 (Krishnan et al. 2008). This analysis used embryos at 24 h postfertilization (hpf), but phenotypic differences between fogsk8 homozygotes and heterozygotes or wild-type embryos are already evident around 19 hpf, suggesting that this analysis might have been unsuitable for identifying direct target genes of zSpt5.

To search for direct target genes of zSpt5, we carried out MO-mediated knockdown of zspt5 in zebrafish embryos. Because microinjection of an MO into fertilized embryos knocks down its target in all injected embryos, knockdown embryos can be directly subjected to microarray analysis even before abnormal phenotypes are evident. At 24 hpf, almost of all the embryos injected with an MO targeting zspt5 mRNA (zspt5KD) exhibited several abnormalities, including shorter tail length, impaired pigmentation, and cardiac edema (Fig. 1A–C and Table S1 in Supporting Information), which are phenotypes similar to those seen in the null mutants of zspt5 such as fogsk8 and fogs30 (Keegan et al. 2002). These phenotypes were rescued by coinjection of an mRNA expressing wild-type zSpt5 protein (see below and data not shown). These data indicate that the knockdown phenotype is specific to disruption of zspt5 functions and is not because of any unintended effects.

Figure 1.

 Analysis of zSpt5 protein level and phenotypes of the zspt5KD embryos. (A–C) Lateral views of live embryos at 24 h postfertilization (hpf), anterior to the left; wild-type embryos (WT; A), CtrlMO-injected embryos (Ctrl; B), and zspt5KD embryos (KD; C). Arrows represent, from the left, abnormal eyes, cardiac edema, and limited tail length in the zspt5KD embryos. (D) Embryo extracts prepared from Ctrl and KD embryos were immunoblotted with an anti-zSpt5 and an anti-actin antibody. Embryo extracts at 7, 9, 11, 13, 15, and 21 hpf were examined. Actin was detected as a loading control.

Decrease in the zSpt5 protein level in zspt5KD embryos

Because MO-mediated knockdown inhibits de novo protein synthesis, but does not affect the stability of maternal proteins, the total zSpt5 protein level may not be affected by its knockdown at early stages of development. To determine when the total protein level begins to be affected, we analyzed changes in the zSpt5 protein level over time in zspt5KD embryos by immunoblot analysis using anti-zSpt5 antibodies that were newly prepared in this study (Fig. S1 in Supporting Information). At 7 and 9 hpf, the zSpt5 protein level in zspt5KD embryos was not changed in comparison with control MO-injected (Ctrl) embryos, and a decrease was first observed at 11 hpf. The level of zSpt5 protein gradually decreased from 13 to 15 hpf, and at 21 hpf, more than 90% of the zSpt5 protein was depleted (Fig. 1D). Therefore, we reasoned that MO-mediated knockdown of zspt5 may cause primary effects as early as 11 hpf.

Down-regulation of erythropoietic genes including gata1 in zspt5KD embryos

To identify direct target gene candidates of zSpt5, we investigated gene expression changes caused by zspt5KD at 9, 12, 15, 18, and 21 hpf by microarray analysis. By comparing microarray expression data sets of zspt5KD embryos with those of Ctrl embryos at each stage, we found down-regulation of several erythropoietic genes, including gata1, alas2, hbae3, and hbae1 (Fig. 2A–C and data not shown). In Ctrl embryos, gata1 was expressed from 9 to 15 hpf and alas2, hbae3, and hbae1 were expressed from 15 to 21 hpf. By contrast, their expression was not substantially induced in zspt5KD embryos and was maintained at almost constant levels throughout these stages. Expression of other genes such as gata2, a hematopoietic stem cell marker gene, and flk-1, a vascular endothelial marker gene, was not affected (Fig. 2D,E). To confirm the microarray data, we carried out a whole-mount in situ hybridization (WISH) analysis of their expression. Compared with wild-type and Ctrl embryos, gata1 expression in zspt5KD embryos was modestly decreased at approximately 14 hpf (Fig. 3A–C) and was markedly reduced at approximately 16 and 21 hpf (Fig. 3D–I). A significant decrease in alas2 and hbae3 expression was observed at 21 and 18 hpf, respectively (Fig. 4A–F).

Figure 2.

 Time-course microarray analysis of zspt5KD embryos. Microarray expression profiles of erythropoietic genes gata1, alas2, and hbae3, an HSC gene gata2, and a vascular endothelial gene flk-1 are shown in A–E, respectively. Expression ratios of each gene in CtrlMO-injected embryos (Ctrl) and zspt5KD embryos (KD) at 9, 12, 15, 18, and 21 h postfertilization (hpf) relative to Ctrl at 9 hpf are plotted.

Figure 3.

 Reduced expression of gata1 in zspt5KD embryos. Expression of gata1 was analyzed by whole-mount in situ hybridization at 13–14 h postfertilization (hpf) (A–C), 15–16 hpf (D–F), and 19–21 hpf (G–I). Wild-type embryos (WT; A, D, G), CtrlMO-injected embryos (Ctrl; B, E, H), and zspt5KD embryos (KD; C, F, I) were compared. (A–F) Posterior views, dorsal to the top. (G–I) Lateral views, anterior to the right.

Figure 4.

 Zspt5KD embryos show severe defects in embryonic erythropoiesis. (A–F) Expression of alas2 at 19–21 h postfertilization (hpf) (A–C) and of hbae3 at 17–18 hpf (D–F) was analyzed by whole-mount in situ hybridization. Wild-type embryos (WT; A, D), CtrlMO-injected embryos (Ctrl; B, E), and zspt5KD embryos (KD; C, F) were compared. Lateral views, anterior to the right. (G,H) Hemoglobin production in Ctrl (G) and KD (H) at 36 hpf was examined by o-dianisidine staining. An arrow in (G) indicates hemoglobin stained by o-dianisidine. Lateral views, anterior to the left.

The stage at which gata1 gene induction occurred in Ctrl embryos (9–15 hpf) was almost coincident with the stage at which a decrease in the zSpt5 protein level was evident in zspt5KD embryos. Induction of alas2, hbae1, and hbae3, which are known to be transactivated by Gata1, was observed later than that of gata1. These results suggest that gata1 is one of primary targets of zSpt5 whereas alas2, hbae1, and hbae3 might be indirectly affected through gata1 repression.

Erythrocyte production is inhibited in zspt5KD embryos

As the above erythropoietic genes are responsible for erythropoiesis, the reduction of their expression is likely to cause defects in erythrocyte differentiation. Gata1 encodes an erythroid master regulator, and a zebrafish mutant carrying a non-sense mutation in gata1, vlad tepes, exhibits marked defects in erythropoiesis (Lyons et al. 2002). Alas2 encodes an enzyme that helps produce heme, the oxygen-carrying component of hemoglobin, and hbae3 encodes a major embryonic globin chain (Lyons et al. 2002). To examine the effects of zspt5KD on production of erythrocytes containing hemoglobin, we carried out whole-mount hemoglobin staining with o-dianisidine using embryos at 36 hpf. Circulating red blood cells were easily detected in Ctrl embryos (Fig. 4G), whereas red blood cells were markedly decreased in zspt5KD embryos (Fig. 4H). These data suggest that zspt5KD disrupts erythroid differentiation, probably through repressing the expression of gata1 and other erythropoietic genes.

Normal expression of HSC genes upstream of gata1 in zspt5KD embryos

Defective erythropoiesis in zspt5KD embryos might result from defects of HSC development. To investigate this possibility, we examined the expression of HSC genes such as scl, lmo2, and gata2 using WISH analysis. These genes are expressed in HSCs and are required for their differentiation into erythrocytes or myelocytes in embryonic hematopoiesis. These genes also regulate gata1 expression (Hsia & Zon 2005). In zebrafish embryos, their expression at the PLPM indicates the presence of HSCs that are destined to differentiate into erythrocytes (Hsia & Zon 2005). In zspt5KD embryos at approximately 14 hpf, expression of scl, lmo2, and gata2 at the PLPM was unaltered (Fig. 5A–I). Biklf encodes a Krüppel-like transcription factor that directly activates the gata1 gene promoter and is expressed in blood islands during zebrafish embryogenesis (Kawahara & Dawid 2001). Biklf expression was also not changed in zspt5KD embryos at approximately 15 hpf (Fig. 5J–L). These data suggest that zspt5KD does not affect the development of hematopoietic lineages in PLPM during zebrafish embryogenesis.

Figure 5.

 Normal expression patterns of HSC marker genes in zspt5KD embryos. Expression of scl (A–C), lmo2 (D–F), and gata2 (G–I) at 13–14 h postfertilization (hpf) and of biklf (J–L) at 14–15 hpf was analyzed by whole-mount in situ hybridization. Wild-type embryos (WT; A, D, G, J), CtrlMO-injected embryos (Ctrl; B, E, H, K), and zspt5KD embryos (KD; C, F, I, L) were compared. The arrow in (L) represents normal expression of biklf in the erythroid region. Posterior views, dorsal to the top.

Normal expression of vascular endothelial and pronephric marker genes in zspt5KD embryos

We next examined whether zspt5KD affects development of other mesodermal tissues such as blood vessels and pronephric ducts, which, like erythrocytes, are derived from PLPM. Vascular endothelial cells differentiate from hemangioblasts, mesodermal progenitors that also produce HSCs. Both hemangioblasts and the pronephros differentiate from the PLPM. Using WISH analysis, we examined the expression of vascular endothelial cell markers, fli-1 and flk-1, and a pronephric marker, pax2.1 (Drummond et al. 1998; Hsia & Zon 2005). Fli-1 and flk-1 showed normal expression in the dorsal midline of zspt5KD embryos (Fig. 6A–F). Similarly, pax2.1 expression in the PLPM was also normal in zspt5KD embryos (Fig. 6G,H). Altogether, these results suggest that zspt5KD has a specific effect on erythroid differentiation from HSCs among several lineages derived from the PLPM.

Figure 6.

 Normal expression patterns of vascular endothelial and pronephric markers in zspt5KD embryos. Expression of fli-1 (A–C) and flk-1 (D–F) at 15–16 hpf and of pax2.1 (G, H) at 13–14 hpf was analyzed by whole-mount in situ hybridization. Wild-type embryos (WT; A, D, G), CtrlMO-injected embryos (Ctrl; B, E), and zspt5KD embryos (KD; C, F, H) were compared. (A–F) Dorsal views, anterior to the top. (G, H) Posterior views, dorsal to the top.

CTR of zSpt5 is critical to gata1 expression

DSIF has both negative and positive effects on transcription. To determine which function is responsible for gata1 expression during zebrafish embryogenesis, we carried out complementation analysis using two zSpt5 point mutants deficient in a particular function, fogm806 and T4A. fogm806 is a hypomorphic allele that carries an amino acid substitution of aspartic acid for valine 1012 and is defective for the negative function of DSIF, as determined by in vitro transcription analysis (Guo et al. 2000). We previously characterized a human Spt5 mutant, T4A, bearing seven threonine-to-alanine substitutions in G-S-Q/R-T-P pentapeptide repeats of the CTR (Yamada et al. 2006). These threonine residues are phosphorylated by P-TEFb, and this phosphorylation is necessary for the positive function of DSIF. As the pentapeptide repeats of the CTR are well conserved in zSpt5 (Fig. 7A), we introduced corresponding mutations into zSpt5.

Figure 7.

 Forced expression of the T4A mutant does not efficiently rescue the deficiency of gata1 expression caused by zspt5KD, compared to wild-type or fogm806 mutant of zSpt5. (A) Protein sequence comparison of human Spt5 (hSpt5) with wild-type zSpt5, fogm806, and T4A. The positions of the acidic domain; the Spt4 binding domain; the KOW motifs, which have homology to the Escherichia coli NusG protein; and the C-terminal repeats (CTR) are represented. In fogm806, valine 1012 is substituted with aspartate. T4A has seven threonine-to-alanine substitutions in the GSQ/RTP pentapeptide repeats of the CTR. (B–H) Expression of gata1 was analyzed by whole-mount in situ hybridization at 14–16 h postfertilization (Posterior view). In each panel, the type of the embryos and the numbers of affected and total embryos analyzed are indicated. (I) zSpt5 protein level in each embryo was determined by immunoblot analysis. Actin was detected as a loading control.

First, we examined whether forced expression of wild-type zSpt5 restores gata1 expression in zspt5KD embryos. As demonstrated previously, gata1 expression was decreased in almost of all the zspt5KD embryos (65/66, 99%; Fig. 7B,C). Although coinjection of eGFP mRNA did not restore gata1 expression (66/66, 100%; Fig. 7D), coinjection of an MO-resistant mRNA coding for wild-type zSpt5 restored gata1 expression in approximately 64% of the coinjected embryos (43/67; Fig. 7E). These data show that the reduction of gata1 expression results from a specific effect of zspt5MO.

Next, fogm806 and T4A were tested using the same procedure. In approximately 61% (19/31) of the embryos coinjected with fogm806 mRNA, gata1 expression was restored (Fig. 7F). The percentage of rescue was almost equal to the value obtained using wild-type zSpt5. By contrast, T4A restored gata1 expression in only approximately 20% (16/81) of the coinjected embryos (Fig. 7G,H). To determine the expression levels of the exogenous zSpt5 proteins, we carried out an immunoblot analysis. T4A was found to be less abundantly expressed than exogenously expressed wild-type zSpt5 or fogm806 (Fig. 7I and Fig. S2 in Supporting Information), but more highly expressed than endogenous zSpt5 (Fig. 7I); this expression is therefore likely to be sufficient to compensate for the loss of endogenous zSpt5 protein. These results suggest that gata1 expression during zebrafish embryogenesis requires an intact CTR of zSpt5.


In this study, we discovered by microarray analysis that zSpt5 is involved in the expression of erythropoietic genes, including gata1, during embryonic erythropoiesis. The function of zSpt5 is specific to differentiation of erythrocytes among tissues arising from the PLPM and is dependent on an intact CTR. This study has several implications for the physiological and developmental roles of the transcription elongation factor DSIF.

How zSpt5 could directly regulate gata1 gene expression

The purpose of this study was to identify target genes that DSIF regulates during development and to understand how DSIF plays gene- or tissue-specific roles. Our results suggest that expression of the erythroid master regulator gata1 is positively regulated by zSpt5, a subunit of DSIF. Gata1 is a transcription factor that regulates erythropoietic gene expression and is essential for erythroid differentiation. Indeed, our results indicate that zSpt5 is required for erythrocyte production, although precisely how DSIF regulates gata1 expression is unclear. DSIF functions as a positive transcription elongation factor through the CTR of Spt5 (Wada et al. 1998a; Yamada et al. 2006; Chen et al. 2009). Our results suggest that the Spt5 CTR is important to gata1 expression. Gata1 induction is initiated at approximately the same time as zspt5KD begins to decrease the zSpt5 protein level (approximately 11 hpf), and gata1 expression is significantly inhibited by zspt5KD. By contrast, expression of genes upstream of gata1, such as scl, lmo2, gata2, and biklf, are not affected by zspt5KD. Hence, zSpt5 plays an important role in gata1 gene expression and erythropoiesis through its transcriptional activation domain.

The present data are insufficient to conclude that the positive function of DSIF directly regulates gata1 gene expression. Previously, we demonstrated that DSIF directly promotes expression of the immediate-early gene c-fos at the transcription elongation level; ChIP analysis showed that translocation of Pol II from promoter-proximal to downstream regions is inhibited by knockdown of Spt5 (Yamada et al. 2006; Chen et al. 2009). Krishnan et al. (2008) showed, by in vivo ChIP analysis using zebrafish embryos, that Pol II and Spt5 occupy both the 5′ and 3′ ends of several genes that are down-regulated in fogsk8. An in vivo knockdown ChIP analysis would also be useful to demonstrate that zSpt5 directly regulates gata1 expression. Because, however, only a small number of gata1-expressing cells exist in zebrafish embryos (approximately 300 cells at 14 hpf), it is difficult to collect the number of gata1-expressing cells, usually 106, required for a single ChIP reaction (Long et al. 1997). Therefore, in this study, we could not determine whether or not zSpt5 directly regulates gata1 expression at the transcription elongation level.

Factors that specify the role of DSIF in erythropoiesis

This study suggests that zSpt5 plays a positive, erythroid-specific role through its CTR. A similar positive, tissue-specific role in differentiation of cardiac myocytes has been suggested: in a null allele of zspt5, fogs30, expression of the myocardial marker cmlc2 is markedly decreased, whereas expression of gata4 and nkx2.5 in precardiac mesoderm is normal (Keegan et al. 2002). An analysis of fogm806, a hypomorphic allele of zspt5 lacking the negative function of DSIF, suggests that the negative function plays a neuron-specific role (Guo et al. 2000). Furthermore, a previous microarray analysis using fogsk8 showed that only a small number of genes are either positively or negatively regulated by zSpt5 (Krishnan et al. 2008). These observations suggest that DSIF has gene- and tissue-specific roles. By contrast, DSIF has biochemical properties as a general transcription elongation factor. For example, Spt5 is ubiquitously expressed and colocalizes with Pol II on most genes (Yamaguchi et al. 1999b; Andrulis et al. 2000; Kaplan et al. 2000; Rahl et al. 2010). Such general properties of DSIF do not account for its gene- and tissue-specific roles, suggesting that other unknown factors might be involved.

Although a variety of DNA-binding transcription factors play important roles in tissue- and lineage-specific gene expression, physical interaction between DSIF and such transcription factors has not, to our knowledge, been reported. By contrast, P-TEFb is known to be recruited to target genes through interaction with tissue- and lineage-specific transcription factors and plays specific roles in development and differentiation (Peterlin & Price 2006). For example, the MEF2 family of transcription factors, which is critical to differentiation of cardiac and skeletal muscle cells, activates transcription by recruitment of P-TEFb (Nojima et al. 2008). P-TEFb also interacts with hematopoietic transcription factors such as Tif1γ, LDB1, and GATA1, which are components of the SCL complex (Meier et al. 2006; Elagib et al. 2008; Bai et al. 2010). Such specific action of P-TEFb might confer specificity to DSIF.

A recent report by Bai et al. (2010) demonstrated that fogs30 and fogm806 do not cause, but rather suppress, erythropoietic defects in zebrafish Tif1γ mutant embryos, indicating that the negative function of DSIF has a role in erythropoiesis. The authors propose that on erythropoietic genes, DSIF and the Paf1 complex cause promoter-proximal pausing and that pause release occurs when Tif1γ interacting with the SCL complex recruits P-TEFb and FACT to the genes. By contrast, our results suggest that DSIF positively regulates erythropoiesis through the Spt5 CTR, which does not fit with the aforementioned model. Perhaps, after the pause release by Tif1γ and the SCL complex, DSIF may also function as a positive regulator in downstream regions of erythropoietic genes.

The regulatory sequence of gata1 contains consensus E-box and WGATAR motifs. The SCL complex containing LDB1 and GATA1 binds to this motif and regulates the expression of a number of erythropoietic genes, including gata1 (Lecuyer & Hoang 2004). Recently, genome-wide occupancy profiles for several components of the SCL complex have been reported (Soler et al. 2010); these show that binding sites of the SCL complex are often found between 1 and 3 kb downstream from transcription start sites on erythropoietic genes containing a TATA-less promoter. Although this report did not clearly present the occupancy of downstream regions of gata1, gata1 also has a TATA-less promoter. Because the SCL complex interacts with P-TEFb and Tif1γ as described previously, the SCL complex and Tif1γ are likely to recruit P-TEFb to the downstream regions of gata1. Such regulator complexes might also be necessary for DSIF to exert its positive function in downstream regions of genes, as biochemical and cell culture studies show that P-TEFb-mediated phosphorylation of the Spt5 CTR is required for the positive function of DSIF.

Therefore, the action of DSIF on gene expression is likely to depend on gene-specific behaviors of P-TEFb. Its activity and localization are controlled by core promoter type (TATA box or TATA-less), cis elements (such as E-boxes), or interactions with DNA-binding transcription factors (e.g., the SCL complex and Tif1γ). We speculate that for a subset of erythropoietic genes, including gata1, recruitment of P-TEFb to the downstream regions by the SCL complex and Tif1γ plays a crucial role in the positive function of DSIF. Further investigations are needed to clarify the genome-wide occupancy of transcription elongation factors including DSIF and P-TEFb using erythropoietic cells or tissues at developmental stages. Such research would lead to further elucidation of the gene- and tissue-specific roles of DSIF.

Experimental procedures

Fish maintenance and injection

Zebrafish and embryos were maintained at 28 °C and staged as previously described (Westerfield 1995). Fertilized embryos were injected with approximately 5 nL of MOs or mRNAs at the one-cell or two-cell stage and observed or collected at the indicated stages.

Microinjection of morpholinos and mRNA

Morpholinos modified with a 3′-end carboxyfluorescein were obtained from GeneTools. An MO targeting zspt5 mRNA was designed to target 25 bases including the initiation codon of zspt5 mRNA (zspt5MO; 5′-GTCCTCGCTGTCTGACATCTCTTTG-3′). CtrlMO is the GeneTools standard control morpholino. Approximately 2.5 pmol of each morpholino was injected into embryos. Dose effects of zspt5MO on embryogenesis are shown in Table S1 in Supporting Information.

For in vitro mRNA synthesis, expression constructs of the wild-type and each mutant zSpt5 were generated by subcloning the open reading frame with an N-terminal Flag tag into pCS2+. To make zspt5 mRNA resistant to zspt5MO, we introduced silent mutations into the sequence around the MO target site. The wild-type sequence 5′-ATGTCAGACAGCGAGGAC-3′ was substituted with the MO-resistant sequence 5′-ATGAGTGATTCGGAAGAT-3′. As a negative control, an expression construct of eGFP was also generated. The capped synthetic mRNAs were generated from the expression constructs using the SP6 mMessage mMachine system (Ambion), and approximately 750 pg of each mRNA was coinjected with zspt5MO where indicated.

Construction of zspt5 mutants

The fogm806 construct was described previously (Guo et al. 2000). For construction of T4A, a series of PCR was carried out using the following five overlapping reverse primers, which introduce substitutions of alanine to threonine in GSQ/RTP repeats of the zSpt5 CTR, as shown in Fig. 7A: 5′-atacatgggagcacgtgagccagtcccataaatgggtgcctgggaaccatacatcggggctcgtagatgggtggaac-3′, 5′-agcctgtgagccatagtgaggagcacgacttccatcatggaggggtgcctggctaccatacatgggagcacgtga-3′, 5′-ggattgttggggtcccacgcaccactttggccaggtgccctgctcccatcatgaagaggagcctgtgagccatagtg-3′, 5′-ctcgtcatcgtaggcaaactcatactcgtcatcaggccgtgatggtgtgtttggattgttggggtcccac-3′, 5′-atccaggcgtctgagggttgggtgtcccaccatagccttgaggagacggcgagggctcgtcatcgtaggcaaa-3′.

Preparation of embryo extracts

Collected embryos were frozen in liquid nitrogen and thawed in RIPA buffer [150 mm NaCl, 1% NP-40, 0.5% deoxycholate, 0.1% sodium dodecyl sulfate, 50 mm Tris–HCl (pH 8.0)]. After thorough sonication, the extracts were centrifuged for 15 min at 12 000 × g, and the supernatants were collected as embryo extracts.

Antibodies and immunoblot analysis

For the generation of an anti-zSpt5 antibody, rabbits were immunized with hexadecapeptides (CQSGAWDPNNPNTPSR; Fig. S1A in Supporting Information). The resulting polyclonal antibody was affinity-purified from the serum with the peptide antigen as described previously (Harlow & Lane 1988). An anti-actin antibody was purchased from Chemicon (MAB1501). Immunoblot assays were carried out as described previously (Wada et al. 1998a). Blots were developed with an ECL system (Amersham Biosciences).

o-Dianisidine staining

For histochemical staining of hemoglobin, embryos at 36 hpf were dechorionated with forceps and stained in 0.6 mg/mL o-dianisidine staining solution (40% ethanol, 0.01 m sodium acetate, 0.65% hydrogen peroxide; Sigma) for 15 min at room temperature in the dark. Stained embryos were washed in water and observed with a stereoscopic microscope.

Whole-mount in situ hybridization

Whole-mount in situ hybridization was carried out as described previously (Jowett 2001). The examined genes were cloned into pBlueScript SK+ by PCR from a zebrafish embryonic cDNA library, and digoxigenin-labeled antisense probes were transcribed in vitro with T7 or T3 RNA polymerase. The primers used for their cloning are shown in Table S2 in Supporting Information.

Microarray analysis

Total RNAs were extracted from embryos using Sepasol-RNA I super (Nacalai Tesque). Three independent preparations of RNAs were further purified using an RNeasy mini protocol for RNA cleanup (Qiagen). In vitro synthesis of cDNA and target cRNA preparations were carried out with 5 μg of total RNA in accordance with a One-Cycle Target Labeling protocol (Affymetrix). Aliquots of cRNA (20 μg) from each sample were fragmented, hybridized to the GeneChip Zebrafish Genome Array (Affymetrix) and processed in accordance with the manufacturer’s instructions. Scanned data were analyzed with the GC-RMA algorithm to estimate gene expression levels (James et al. 2007). The estimated gene expression levels were transferred to the GeneSpring software program (Silicon Genetics) and analyzed.


We are indebted to Dr. T. Kokubo for many helpful discussions and to Drs. K. Hoshijima, K. Kawakami, A. Kudo, and Y. Imai for providing zebrafish and technical support. We are grateful to Dr. T. Yung for many helpful suggestions in preparing our manuscript. We also thank S. Okabe for technical support and members of the Handa lab for helpful suggestions. This study was supported by a Grant-in Aid for Scientific Research on Priority Areas from the Ministry of Education, Culture, Sports, Science, and Technology (to T.W.) and a Tokyo Tech Award for Challenging Research (also to T.W.). In addition, this study was supported in part by a Grant from the 21st Century COE Program and the Global COE program from the Ministry of Education, Culture, Sports, Science, and Technology and a grant for Special Coordination Funds for Promoting Science and Technology from Japan Science and Technology Agency (JST). This work was also supported by a grant from the New Energy and Industrial Technology Development Organization, Japan (to H. H). T.T. is a research fellow of the Japan Society for the Promotion of Science.