Content uploaded by Rebecca Ann Wingert
Author content
All content in this area was uploaded by Rebecca Ann Wingert on Apr 21, 2018
Content may be subject to copyright.
Developmental Cell, Vol. 8, 109–116, January, 2005, Copyright ©2005 by Elsevier Inc. DOI 10.1016/j.devcel.2004.12.001
Short ArticleLoss of Gata1 but Not Gata2
Converts Erythropoiesis to
Myelopoiesis in Zebrafish Embryos
inhibit each other’s function by physically interfering
with the transcriptional control of their target genes,
suggesting that the levels of each directly determines
cell fate (Graf, 2002). Despite overexpression studies
demonstrating the importance of GATA1 and PU.1 levels
Jenna L. Galloway,
1
Rebecca A. Wingert,
1
Christine Thisse,
2
Bernard Thisse,
2
and Leonard I. Zon
1,
*
1
Stem Cell Program and
Division of Hematology/Oncology
Children’s Hospital Boston in controlling erythroid or myeloid cell fates, no genetic
evidence has shown that loss of these factors altersDana-Farber Cancer Institute
Howard Hughes Medical Institute myeloid versus erythroid lineage decisions during devel-
opment.Harvard Medical School
Boston, Massachusetts 02115 In zebrafish, expression of hematopoietic genes such
as gata1 and gata2 identifies the first primitive erythroid
3
Institut de Ge
´ne
´tique et de Biologie
Mole
´culaire et Cellulaire precursors at the 5-somite stage in the posterior meso-
derm (reviewed in Davidson and Zon, 2004). After 10UMR 7104
CNRS/INSERM/ULP somites, blood precursors express mature erythroid
markers such as embryonic globin. By24 hr postfertiliza-1 rue Laurent Fries, BP10142
CU de Strasbourg tion (hpf), these cells are encircled by endothelial cells
of the trunk axial vein, and together form the intermedi-67404 Illkirch Cedex
France ate cell mass (ICM), the teleost equivalent of the mam-
malian yolk sac blood island. After 24 hpf, the erythro-
blasts enter circulation, where they later mature into
primitive erythrocytes. Zebrafish myelopoiesis is firstSummary
detected by pu.1 expression in the anterior mesoderm
at the 6-somite stage (Lieschke et al., 2002). pu.1 tran-The differentiation of hematopoietic progenitors into
erythroid or myeloid cell lineages is thought to depend scripts are also found in the ICM blood precursors be-
tween 10 and 18 somites, and at 20 somites maturingupon relative levels of the transcription factors gata1
and pu.1. While loss-of-function analysis shows that myeloid cells in the posterior ICM and anterior head
region express myeloperoxidase (mpo), a granulocytegata1 is necessary for terminal erythroid differentia-
tion, no study has demonstrated that loss of gata1 marker, and l-plastin, a macrophage marker (Bennett et
al., 2001).alters myeloid differentiation during ontogeny. Here
we provide in vivo evidence that loss of Gata1, but In this study, we examined the role of gata1 and gata2
in determining blood cell fate during zebrafish hemato-not Gata2, transforms primitive blood precursors into
myeloid cells, resulting in a massive expansion of poiesis. Despite an absence of erythroid cells in Gata1-
deficient embryos (Lyons et al., 2002), hematopoieticgranulocytic neutrophils and macrophages at the ex-
pense of red blood cells. In addition to this fate change, cells present in the ICM of these embryos were found
to differentiate into granulocytes and macrophages. Byexpression of many erythroid genes was found to be
differentially dependent on Gata1 alone, on both Gata1 expression analysis, confocal microscopy, and cell his-
tology, we show ICM blood precursors have convertedand Gata2, or independent of both Gata factors, sug-
gesting that multiple pathways regulate erythroid gene into myeloid cells in the absence of Gata1. In contrast,
loss of Gata2 did not shift differentiation of the ICM cellsexpression. Our studies establish a transcriptional
hierarchy of Gata factor dependence during hemato- to myeloid lineages. Red blood cell genes were found
to require either Gata1 or both Gata1 and Gata2 or werepoiesis and demonstrate that gata1 plays an integral
role in directing myelo-erythroid lineage fate decisions expressed in the absence of both Gata proteins, sug-
gesting that other transcription factors regulate theirduring embryogenesis.
erythroid expression. Furthermore, our results distin-
guish the functions of Gata1 and Gata2 and highlightIntroduction
the importance of Gata1 in promoting erythroid and sup-
pressing myeloid cell fate decisions in zebrafish.The fate of common myelo-erythroid (CMPs) progenitors
to become megakaryocyte-erythroid progenitors or my-
eloid progenitors is thought to be determined by GATA1 Results
and PU.1 (reviewed in Graf, 2002). Overexpression of
GATA1, a zinc finger transcription factor essential for Loss of Gata1 Results in Expanded Myelopoiesis
erythroid differentiation, reprograms myeloid cells to The mutual antagonism between pu.1 and gata1 and
undergo erythroid and megakaryocytic differentiation their coexpression in the ICM led us to ask if gata1
(Pevny et al., 1995; Iwasaki et al., 2003). Forced expres- regulates pu.1 expression during zebrafish blood devel-
sion of PU.1, an Ets family transcription factor, represses opment. We designed two gata1 morpholinos (MOs) and
erythropoiesis and promotes myeloid differentiation in found them to cause anemia at 36 hpf, phenocopying
erythroid cell lines (Yamada et al., 2001). PU.1 and GATA1 the zebrafish gata1 mutant, vlad tepes (vlt), while a
4-base pair mismatch MO did not (Lyons et al., 2002;
Supplemental Table S1 at http://www.developmentalcell.
*Correspondence: zon@enders.tch.harvard.edu
Developmental Cell
110
Figure 1. Loss of Gata1 Results in Expanded Myelopoiesis
(A) pu.1 expression in control and gata1 MO-injected embryos at 14 somites (a, b), 18 somites (c, d), and 24 hpf (e, f). Embryos (a–d) were
flat-mounted with anterior (left) and posterior (right). gata1 MO-injected embryos express pu.1 at 18 somites (d, brackets) and 24 hpf (f) while
wild-type embryos have downregulated pu.1 transcripts (c, brackets, e) in the ICM precursors. pu.1 is also expressed in anterior myeloid
precursors (a–f). gata1 MO-injected embryos have reduced expression of e1 globin (h) in the ICM at 24 hpf compared to wild-types (g).
(B) The number of cells expressing mpo and l-plastin are increased in the vessels of gata1 morphants (b, f, d, h) compared to wild-type
embryos (a, e) and gata2 morphants (c, g). g1/g2 morphants (d, h) have decreased numbers of mpo- and l-plastin-expressing cells compared
to the gata1 morphants alone (b, f).
(C) Expression of c-myb (a–d) and ikaros (e–h) is maintained in the ICM cells of gata1,gata2, and g1/g2 morphants at 20 somites. pu.1
expression persists in the ICM of gata1 (j) and g1/g2 (l) morphants, but not gata2 morphants (k) or wild-type embryos (i). Expression of runx1
(m–p) and cebp␣(q–t) persists in gata1 (n, r) and g1/g2 morphants (p, t) but not wild-type embryos (m, q) or gata2 morphants (o, s) at 22 hpf.
cebp␣is expressed at low levels in some ICM cells and in gut endoderm (q–t). ICM cells expressing mpo at 22 hpf are reduced in gata2
morphants (w) and lost in gata1 (v) and g1/g2 morphants (x) compared to wild-type embryos (u). Insets (u–x) are of cranial views of the same
embryos showing normal mpo-expressing cells in wild-type (u) and gata1 morphants (v) and reduced numbers of mpo-expressing cells in
gata2 (w) and g1/g2 morphants (x).
com/cgi/content/full/8/1/109/DC1/). pu.1 expression in gene expression in vitro (Yamada et al., 2001), expres-
sion of the granulocyte marker, mpo, and the macro-the ICM and anterior blood region of gata1 morphants
resembled that of wild-type embryos at 5, 12, and 14 phage marker, l-plastin, were analyzed in Gata1-defi-
cient embryos. Examination of vlt mutants and gata1somites (Figure 1A, a and b; data not shown). Persistent
expression of pu.1 was found in the ICM precursors of morphants at 32 hpf and 4 dpf revealed a significant
increase in the number of mpo (n ⫽36; 100%) andvlt mutants and gata1 morphants from 18 somites to 24
hpf (n ⫽47; 100%; Figure 1A, d and f), whereas in wild- l-plastin (n ⫽37; 100%) positive cells (Figure 1B, b, d,
f, and h; Supplemental Figure S1). These data suggesttype embryos, pu.1 expression was downregulated in
the ICM after 18 somites (Figure 1A, c and e). These data that ICM precursors have differentiated into macro-
phages and neutrophils in the absence of Gata1. Expres-suggest that Gata1 is required to limit pu.1 expression in
ICM precursors after 18 somites. sion of the T cell marker, rag1, was normal in Gata1-
deficient day 4 embryos, demonstrating that Gata1Since forced expression of PU.1 activates myeloid
Gata1 in Zebrafish Myelo-Erythroid Differentiation
111
exclusively regulates myelo-erythroid decisions in ze- Cell Morphology of Gata1-Deficient Cells
To examine Gata1-deficient ICM cell morphology, wild-brafish (n ⫽16; 100%; data not shown).
type and Gata1-deficient ICM cells were isolated from
gata1-gfp transgenic embryos. The gata1-gfp line used
Loss of Gata2 Does Not Cause expresses gfp in the rostral myeloid cells, permitting
Expanded Myelopoiesis both erythroid and myeloid populations to be isolated
To determine if a myeloid expansion occurs in the ab- by flow cytometry (Supplemental Figure S2). GFP-sorted
sence of a related Gata family member also expressed in cells from 14 hpf wild-type and Gata1-deficient embryos
the ICM, a gata2 morpholino and a 4-base pair mismatch morphologically resembled hematopoietic precursors
control morpholino were designed (Supplemental Table (Figure 2A, a and b). GFP cells from wild-type 24 hpf
S1). An increase in mpo- and l-plastin-expressing cells embryos were round or polygonal in shape with coarse
in gata1 morphants at 32 hpf was not observed in gata2 condensed chromatin and a dark blue cytoplasm, char-
morphants (n ⫽30; 0%; Figure 1B, c and g), indicating acteristics that define proerythroblasts (Figure 2A, c). In
the myeloid expansion is specific to loss of gata1.In contrast, an increased number of cells from 24 hpf
mice, loss of GATA2 results in modestly decreased num- Gata1-deficient embryos resembled myeloblasts (wild-
bers of erythroid and myeloid precursors that differenti- type: n ⫽146; 25% myeloblast-like; gata1 MO: n ⫽56;
ate normally (Tsai and Orkin, 1997). Consistent with this, 41% myeloblast-like; Figure 2A, d). At 48 hpf, most wild-
the number of mpo- and l-plastin-expressing myeloid type GFP-sorted cells resembled erythroblasts (Figure
cells at 32 hpf appeared slightly decreased in the gata1/ 2A, e), while many Gata1-deficient cells possessed fea-
gata2 (g1/g2) double morphants (Figure 1B, d and h) tures of promonocytes or maturing granulocytes (wild-
compared to the gata1 morphants alone (Figure 1B, b type: n ⫽363; 13% myeloblast-like; gata1 MO: n ⫽
and f). These data suggest that gata1 and gata2 function 210; 43% myeloblast-like; Figure 2A, f, arrowheads). The
distinctly to regulate hematopoietic differentiation and increase in myeloid cells at 48 hpf provides further evi-
cell number, respectively. dence that ICM cells differentiate into myelomonocytes
in the absence of Gata1.
Loss of Gata1 Does Not Affect Expression
of Early Hematopoietic Genes Erythroid Precursors Are Converted into Myeloid
While the transcription factor genes scl,lmo2, and gata2 Precursors in the Absence of Gata1
are expressed in hematopoietic stem cells at 5 somites, As loss of GATA1 in mice causes erythroid progenitors
transcripts of c-myb,runx1, and ikaros are found in both to undergo cell death (Pevny et al., 1995), apoptosis was
primitive erythroid and myeloid progenitors (Davidson examined in zebrafish gata1 mutants by TUNEL staining.
and Zon, 2004). To determine if loss of Gata factors No appreciable increase in cell death was observed in
affects the expression of these genes, in situ hybridiza- vlt mutants (n ⫽6) between 12 somites and 24 hpf
tions were performed with the gata morphants. There (Figure 2B, a and c; data not shown; N. Paffett-Lugassy,
was no significant change in gata2,scl, and lmo2 ex- personal communication). Between 26 and 30 hpf, an
pression at 12 somites (data not shown) and c-myb (n ⫽approximate 2-fold increase in TUNEL-positive cells
41; 100%) and ikaros (n ⫽33; 100%; Figure 1C, a–h) was observed in the ICM region of vlt mutants (n ⫽17)
expression at 20 somites in gata1,gata2, and g1/g2 compared to wild-type and heterozygous siblings (n ⫽
morphants, indicating that loss of Gata factors does not 20; Figure 2B, b and d). This modest increase in cell
affect early blood formation. death detected in vlt mutants was not significant enough
In wild-type embryos, runx1 is expressed early in ante- to explain the loss of all ICM erythroid precursors.
rior myeloid precursors and ICM cells until 18 somites. To determine if ICM erythroid precursors are con-
In Gata1-deficient embryos, runx1 expression persisted verting into myeloid cells, double in situ hybridizations
in the ICM until 24 hpf (n ⫽37; 100%; Figure 1C, m–p), for e1 globin and pu.1 expression were performed on
suggesting that ICM cells have retained progenitor char- wild-type and Gata1-deficient 22 hpf embryos and ex-
acteristics. Increased expression of CCAAT/enhancer amined by confocal microscopy. In wild-type embryos,
binding protein ␣(cebp␣), a transcription factor found globin was expressed in ICM cells, while pu.1 was pre-
in myeloid progenitors essential for granulocytic matura- dominantly expressed in the anterior blood region (data
tion (Zhang et al., 1997), was also observed in Gata1- not shown; Figure 2C, a–c). Although pu.1 can be ex-
deficient embryos at 22 hpf (n ⫽24; 100%; Figure 1C, pressed in a few cells in the posterior ICM at 22 hpf,
q–t). Examination of mpo expression prior to circulation wild-type cells were never found to coexpress globin
revealed normal expression in the head region but a lack and pu.1 at this stage. In contrast, multiple ICM cells in
of mpo-expressing cells in the ICM of Gata1-deficient the gata1 morphants were found to coexpress pu.1 and
embryos at 24 hpf (n ⫽46; 100%; Figure 1C, u–x). While globin, suggesting that ICM cells were converting to the
blood cells are specified in Gata1-deficient embryos, an myeloid lineage (Figure 2C, d–f).
absence of mpo-expressing cells at 24 hpf and persis-
tent runx1 expression suggests that the ICM cells are
temporarily delayed in their differentiation. In contrast, A Hierarchy of Gata-Dependent Erythroid
Gene Expressiongata2 morphants never displayed persistent expression
of pu.1 (n ⫽15; 0%), runx1 (n ⫽49; 0%), and cebp␣To understand the functional differences between gata1
and gata2 in hematopoietic regulation, expression of(n ⫽15; 0%) after 18 somites, indicating that the myeloid
expansion is specific to loss of Gata1. erythroid genes obtained from an in situ hybridization
Developmental Cell
112
Figure 2. Erythroid Precursors Are Converted into Myeloid Precursors in the Absence of Gata1
(A) gata1-gfp cells from wild-type (a) and gata1 splice MO-injected (b) embryos at 10 somites (14 hpf) resemble hematopoietic precursors.
At 24 hpf, most wild-type gata1-gfp cells resemble proerythroblasts (c, arrowheads), but myeloid cells with indented nuclei are observed.
Cells from gata1 splice MO-injected embryos have myeloid features such as vacuoles and indented nuclei (d, arrowheads). Many erythrocytes
and proerythroblasts are isolated from 48 hpf wild-type embryos (e, arrowheads) as well as a few myeloid cells. Cells from 48 hpf gata1 splice
MO-injected embryos have vacuoles and indented nuclei (f, arrowheads), suggesting that they are myelomonocytes (scale bars are 10 m).
(B) Wild-type (a) and vlt mutant (c) embryos at 22 hpf have similar numbers of TUNEL-positive cells. vlt mutant embryos have an approximate
2-fold increase in apoptosis in their ICM region (d, inset) compared to their wild-type siblings (b, inset) at 28 hpf.
(C) Confocal imaging of transverse sections of DAPI-stained (blue) embryos that have undergone double in situ hybridization for e1 globin
(green) and pu.1 (red). Wild-type embryos express e1 globin and not pu.1 in their ICM cells at 22 hpf (a–c). gata1 MO-injected embryos
express both e1 globin and pu.1 in their ICM cells at 22 hpf (d–f). Some cells coexpress (yellow) both genes (d–f, arrowheads; n, notochord;
nt, neural tube; scale bar equals 40 m).
screen was examined in gata1,gata2, and g1/g2 MO- six genes were unaffected by loss of Gata2, demonstra-
ting that their expression is specifically dependent oninjected embryos (screen described in Supplemental
Experimental Procedures). At 18 somites, epsin (in- gata1.
ICM expression of another group of erythroid genesvolved in clathrin-mediated endocytosis) and GTP bind-
ing-protein 1 (gtpbp1) were normally expressed in ICM was reduced in Gata1-deficient embryos. A decrease in
e1 globin (n ⫽33; 100%), alas2 (n ⫽27; 100%), car-cells; however, in Gata1-deficient embryos, their expres-
sion was absent (Figure 3A, a–d). Expression of biliverdin bonic anhydrase (n ⫽39; 100%), and gata1 (n ⫽33;
100%) expression in gata1 morphants and vlt mutantsreductase (heme degradation), SH3 domain binding pro-
tein 5 (SH3BP5),HIF1␣-like gene (hypoxic response), was observed (Figure 3B, b, f, j, and n; data not shown).
Previous work in vlt and mouse GATA1-deficient celland 5⬘nucleotidase type B (cause of human hemolytic
anemia; Bianchi et al., 2003) were absent in the ICM of lines also found multiple red cell-specific genes ex-
pressed in the absence of gata1 (Weiss et al., 1994;Gata1-deficient embryos at 20–24 hpf (Figure 3A, e–h;
data not shown). The absence of blood gene expression Lyons et al., 2002). Similar to the reduction in myeloid
cell number, gata2 morphants had a subtle decrease inin Gata1-deficient embryos is consistent with mamma-
lian studies indicating that GATA1 regulates virtually all erythroid cell number as detected by globin (n ⫽16;
100%), alas2 (n ⫽13; 100%), carbonic anhydrase (n ⫽red blood cell genes (Orkin, 1992). Furthermore, these
Gata1 in Zebrafish Myelo-Erythroid Differentiation
113
Figure 3. Erythroid Genes Are Differentially Regulated by Gata1 and Gata2
(A) Embryos (a–d) have been flat-mounted and photographed to show only the posterior tail region (anterior, left; posterior, right). Normal
expression of gdpbp1 and epsin (a, c) is lost in ICM erythroid precursors of vlt mutants at 18 hpf (b, d). Expression of biliverdin reductase
and SH3BP5 at 20 hpf (e) and 22 hpf (g), respectively, is absent in vlt mutant ICM cells (f, h).
(B) In situ hybridization for e1 globin (a–d), alas2 (e–h), carbonic anhydrase (i–l), and gata1 (m–p) at 20 hpf. Compared to wild-type embryos
(a, e, i, m), ICM expression of e1 globin,alas2,carbonic anhydrase, and gata1 was decreased in gata1 morphants (b, f, j, n) and absent in
g1/g2 morphants (d, h, l, p). Decreased number of cells expressing e1 globin,alas2,carbonic anhydrase, and gata1 are observed in gata2
morphants (c, g, k, o) compared to wild-types (a, e, i, m).
29; 100%), and gata1 (n ⫽10; 100%) expression at 20 KIAA0650 (n ⫽45; 100%), testhymin (n ⫽45; 100%),
and kelch repeat-containing protein (n ⫽54; 100%) weresomites (Figure 3B, c, g, k, and o). In contrast, globin
(n ⫽118; 91%), alas2 (n ⫽16; 100%), carbonic anhy- expressed in the ICM precursors of the gata1,gata2,
and g1/g2 double morphants at 12 somites (Figure 4A,drase (n ⫽44; 100%), and gata1 (n ⫽18; 100%) expres-
sion was lost in the g1/g2 double morphants compared e–o). We also found that biklf, a Kruppel-like transcrip-
tion factor expressed in early erythroid precursorsto wild-type embryos and gata1 or gata2 single mor-
phants (Figure 3B, d, h, l, and p). This absence of expres- (Oates et al., 2001), was expressed normally in gata1
(n ⫽21; 100%), gata2 (n ⫽16; 100%), and g1/g2 (n ⫽sion may result from the additive decrease in erythroid
gene expression, resulting from loss of Gata1 and Gata2. 16; 100%) morphants at 12 somites (Figure 4A, a–d).
The Gata-independent expression of biklf,testhymin,Alternatively, a decline in cell number caused by loss
of Gata2 combined with reduced erythroid gene expres- KIAA0650, and kelch repeat-containing protein sug-
gests that other hematopoietic transcription factors in-sion due to loss of Gata1 may lead to the absence of
expression. Congruent with the former conclusion, sug- duce their expression and that at least part of the ery-
throid program was initiated in gata1 morphants. Aftergesting Gata2 may regulate erythroid gene expression
in the absence of Gata1, mouse studies have found that 20 hpf, gata1 and g1/g2 morphants weakly expressed
biklf (n ⫽87; 100%), KIAA0650 (n ⫽20; 100%), andother GATA factors placed under GATA1 transcriptional
control can substitute for loss of GATA1 during embry- testhymin (n ⫽50; 100%) in the ICM compared to wild-
types or gata2 morphants (Figure 4B), suggesting thatonic hematopoiesis (Takahashi et al., 2000). Neverthe-
less, our results suggest that erythroid genes differ in Gata1 is required to maintainwild-type expression levels
of these genes. Additionally, their downregulation istheir requirements for Gata factors.
concomitant with the conversion of the ICM cells to the
myeloid lineage.Gata-Independent Erythroid Gene Expression
To determine whether loss of both Gata factors ablated
erythroid precursors, we utilized erythroid-specific Discussion
genes identified from the in situ hybridization screen.
The genes KIAA0650,testhymin, and kelch repeat-con- Hierarchy of Regulation of Erythroid
Gene Expressiontaining protein were never detected in the anterior
myeloid cells and were found to be lost in the ICM of 18- As gata1 is a major regulator of erythroid gene expres-
sion, we found that many erythroid genes were depen-somite stage moonshine (mon) embryos, which harbor a
mutation in TIF1-␥and lack primitive erythrocytes (Ran- dent on gata1 for expression. A subset of erythroid
genes had reduced, but notabsent, expression in Gata1-som et al., 2004; data not shown). This suggests that
these genes are specific to early ICM erythroid cells. deficient embryos. Similarly, the GATA1 null mouse and
Developmental Cell
114
Figure 4. Expression of Gata-Independent Erythroid Genes
(A) Embryos were flat-mounted (anterior, left; posterior, right). Expression of biklf (a–d), KIAA0650 (e–h), testhymin (i–l), and kelch repeat-
containing protein (m–p) is maintained in ICM precursors of 12 somite gata1,gata2, and g1/g2 morphants. Anterior expression of biklf stains
the hatching gland (a–d).
(B) ICM expression of biklf (a–d) at 20 hpf and KIAA0650 (e–h) and testhymin (i–l) at 22 hpf was decreased but present in gata1 morphants
(b, f, j) and g1/g2 MO-injected embryos (d, h, l) compared to wild-type embryos (a, e, i). gata2 morphants (c, g, k) had decreased numbers
of ICM cells expressing biklf,KIAA0650, and testhymin.
GATA1-deficient cell lines exhibited residual expression KIAA0650 (erythroid progenitors). Although attempts to
rescue erythropoiesis in Gata1-deficient embryos byof erythroid genes (Weiss et al., 1994; Pevny et al., 1995).
Our data showing loss of expression of these genes in gata2 overexpression were technically confounded by
gata2 mRNA toxicity (data not shown), studies in mouseG1/G2-deficient embryos suggests that in the absence
of Gata1, Gata2 regulates expression of a subset of have demonstrated that GATA2, when placed under the
control of the GATA1 promoter, can substitute forerythroid genes in erythroid precursors. In support of
this, zebrafish g1/g2 morphants, unlike mouse G1/G2 GATA1 during embryonic hematopoiesis (Takahashi et
al., 2000). This is consistent with the hypothesis thatnulls in which yolk sac blood cells undergo apoptosis
(Fujiwara et al., 2003), have primitive blood cells ex- gata2 regulates some erythroid target genes in the ab-
sence of Gata1.pressing c-myb and ikaros (common progenitors) and
biklf,kelch repeat-containing protein,testhymin, and We discovered that the novel genes kelch repeat-
Gata1 in Zebrafish Myelo-Erythroid Differentiation
115
In Situ Hybridizations and TUNEL
containing protein,testhymin, and KIAA0650 were ex-
Antisense mRNA probes were made for c-myb,biklf,l-plastin,mpo,
pressed independently of Gata factors. In contrast to
pu.1,e1 globin, and gata1 (Thompson et al., 1998; Bennett et al.,
genes like c-myb and ikaros that are found in G1/G2-
2001; Oates et al., 2001; Lieschke et al., 2002). Riboprobes of genes
deficient CMPs, these genes had erythroid-specific ex-
isolated from the in situ hybridization screen were described in
pression, suggesting that they exclusively mark the ery-
Supplemental Experimental Procedures. Confocal imaging was
throid progenitor cell compartment. The early blood
done on paraffin-embedded 10 m sections that underwent double
in situ hybridization (Brent et al., 2003) with Tyramide Signal Amplifi-
genes biklf,scl, and lmo2 may regulate expression of
cation (Perkin Elmer, Molecular Probes). TUNEL was performed us-
these Gata-independent genes, and future studies will
ing the ApopTag Peroxidase In Situ Apoptosis detection kit
further elucidate erythroid gene regulation during hema-
(Chemicon).
topoiesis.
Gene Knockdown by Morpholinos
Gata1 but Not Gata2 Regulates Myelo-Erythroid
Morpholino oligos were designed against the ATG of gata1 (5⬘-CTG
Fate Decisions
CAAGTGTAGTATTGAAGATGTC-3⬘), the first exon/intron boundary of
Our findings demonstrate that gata1 is necessary to
gata1 (5⬘-GTTTGGACTCACCTGGACTGTGTCT-3⬘), and the third
exon/intron boundary of gata2 (5⬘-CATCTACTCACCAGTCTGCGC
promote erythroid cell fate and repress myeloid differen-
TTTG-3⬘). Control morpholinos containing four base pair mismatches
tiation. Similar to vlt, GATA1-deficient mice lack mature
were also designed against the gata1 splice MO (5⬘-GTTCGGACT
red blood cells, and in vitro differentiation of GATA1⫺
CGCCTGTACTGTGTAT-3⬘) and the gata2 splice MO (5⬘-CATCCAC
murine ES cells reveals an increase (⬎5-fold) in myeloid
TCACTAGTCTACGCTGTG-3⬘). Morpholinos were resuspended in
colonies (S. Orkin, personal communication). Despite
nuclease-free water, and one nanoliter was injected at the 1- to
these similarities, an increase in myeloid cells was not
4-cell stage at the following concentrations: 1 mM gata1 MO, 0.2
mM gata2 splice MO, 0.4 mM gata1 splice MO, 0.2 mM gata2 splice
observed in GATA1
⫺
and GATA1/GATA2⫺
/
⫺mice and
mismatch MO, and 0.4 mM gata1 splice mismatch MO.
GATA1⫺cultured yolk sac cells (Fujiwara et al., 1996,
2003; Pevny et al., 1995), suggesting that the mouse yolk
Flow Cytometry and Histological Analysis
sac environment may not support this cell fate alteration.
Approximately 200 to 400 uninjected or gata1 splice MO-injected
Loss of zebrafish gata2 does not alter red blood cell
gata1-gfp transgenic embryos were collected (14 hpf, 24 hpf, and
maturation or myeloid differentiation, but does decrease
48 hpf). Embryos were processed, isolated by GFP fluorescence
blood cell number, consistent with decreased numbers
using a FACSVantage flow cytometer (Beckton Dickinson), centri-
of progenitors in GATA2-deficient murine embryonic
fuged, and stained with May-Gru
¨nwald and Giemsa solutions as
described (Traver et al., 2003).
stem cells (Tsai and Orkin, 1997).
Two proposed models may explain the dramatic shift
Acknowledgments
toward myelopoiesis in Gata1-deficient embryos (Graf,
2002). In one model, the ICM contains both myeloid
We would especially like to thank A. Davidson and J. Rivera-Felici-
and erythroid progenitors; in the absence of Gata1, the
ano for critical reading and helpful advice, and acknowledge H.
erythroid progenitors undergo cell death while the my-
Stern, D. Traver, and C. Burns for critical reading, K. Dooley and D.
eloid progenitors expand and differentiate. In the sec-
Ransom for reagents, N. Paffett-Lugassy and S. Orkin for use of
ond model, ICM CMPs differentiate into erythroid or
unpublished data, S. Cho for assistance with confocal imaging, and
A. Flint for assistance with flow cytometry. L.I.Z. is supported by
myeloid cells, based on relative levels of gata1 and pu.1.
HHMI and the NIH (U01DK063328-01). B.T. and C.T. are supported
Our results support the latter model wherein the pres-
by INSERM, CNRS, the Ho
ˆpital Universitaire de Strasbourg, the
ence or absence of Gata1 determines the fate of ICM
Association pour la Recherche sur le Cancer, the Ligue Nationale
progenitors. In Gata1-deficient embryos, blood progeni-
Contre le Cancer, and the NIH (R01RR15402).
tors form normally as demonstrated by expression of
biklf,scl, and lmo2. Without Gata1 antagonism, pu.1
Received: April 19, 2004
expression is not downregulated after 18 somites. This
Revised: September 8, 2004
persistent pu.1 expression is likely the instructive signal
Accepted: December 1, 2004
that guides ICM cells to activate a myeloid differentiation
Published: January 3, 2005
program (Rhodes et al., 2004 [this issue of Develop-
mental Cell]). The decrease in cells expressing biklf and
References
the novel erythroid-specific genes and the presence of
Bennett, C.M., Kanki, J.P., Rhodes, J., Liu, T.X., Paw, B.H., Kieran,
ICM cells coexpressing globin and pu.1 provides strong
M.W., Langenau, D.M., Delahaye-Brown, A., Zon, L.I., Fleming, M.D.,
evidence for an ICM cell fate conversion when Gata1
and Look, A.T. (2001). Myelopoiesis in the zebrafish, Danio rerio.
is absent. Our study demonstrates that gata1 has an
Blood 98, 643–651.
important role in suppressing myeloid differentiation,
Bianchi, P., Fermo, E., Alfinito, F., Vercellati, C., Baserga, M., Ferraro,
presumably by antagonizing the effect on pu.1-initiated
F., Guzzo, I., Rotoli, B., and Zanella, A. (2003). Molecular character-
myelopoiesis. This work also provides genetic evidence
ization of six unrelated Italian patients affected by pyrimidine 5⬘-
that a hierarchy exists in the regulation of erythroid
nucleotidase deficiency. Br. J. Haematol. 122, 847–851.
genes and that gata1, but not gata2, is critical for de-
Brent, A.E., Schweitzer, R., and Tabin, C.J. (2003). A somatic com-
termining erythroid versus myeloid cell fates during em-
partment of tendon progenitors. Cell 113, 235–248.
bryonic development.
Davidson, A., and Zon, L.I. (2004). The ‘definitive’ (and ‘primitive’)
guide to zebrafish hematopoiesis. Oncogene 23, 7233–7246.
Experimental Procedures Fujiwara, Y., Browne, C.P., Cunniff, K., Goff, S.C., and Orkin, S.H.
(1996). Arrested development of embryonic red cell precursors in
Zebrafish Strains mouse embryos lacking transcription factor GATA-1. Proc. Natl.
Zebrafish breeding and staging were done as described (West- Acad. Sci. USA 93, 12355–12358.
erfield, 1994). vlt
m651
were genotyped (Lyons et al., 2002), and the
gata1-gfp line (Long et al., 1997) and wild-type Tu
¨were used. Fujiwara, Y., Chang, A.N., Williams, A.M., and Orkin, S.H. (2003).
Developmental Cell
116
Functional overlap of GATA-1 and GATA-2 in primitive hematopoi-
etic development. Blood 103, 583–585.
Graf, T. (2002). Differentiation plasticity of hematopoietic cells.
Blood 99, 3089–3101.
Iwasaki, H., Mizuno, S., Wells, R.A., Cantor, A.B., Watanabe, S., and
Akashi, K. (2003). GATA-1 converts lymphoid and myelomonocytic
progenitors into the megakaryocyte/erythrocyte lineages. Immunity
19, 451–462.
Lieschke, G.J., Oates, A.C., Paw, B.H., Thompson, M.A., Hall, N.E.,
Ward, A.C., Ho, R.K., Zon, L.I., and Layton, J.E. (2002). Zebrafish
SPI-1 (PU.1) marks a site of myeloid development independent of
primitive erythropoiesis: implications for axial patterning. Dev. Biol.
246, 274–295.
Long, Q., Meng, A., Wang, H., Jessen, J.R., Farrell, M.J., and Lin,
S. (1997). GATA-1 expression pattern can be recapitulated in living
transgenic zebrafish using GFP reporter gene. Development 124,
4105–4111.
Lyons, S.E., Lawson, N.D., Lei, L., Bennett, P.E., Weinstein, B.M.,
and Liu, P.P. (2002). A nonsense mutation in zebrafish gata1 causes
the bloodless phenotype in vlad tepes. Proc. Natl. Acad. Sci. USA
99, 5454–5459.
Oates, A.C., Pratt, S.J., Vail, B., Yan, Y.l., Ho, R.K., Johnson, S.L.,
Postlethwait, J.H., and Zon, L.I. (2001). The zebrafish klf gene family.
Blood 98, 1792–1801.
Orkin, S.H. (1992). GATA-binding transcription factors in hematopoi-
etic cells. Blood 80, 575–581.
Pevny, L., Lin, C., D’Agati, V., Simon, M.C., Orkin, S.H., and Costan-
tini, F. (1995). Development of hematopoietic cells lacking transcrip-
tion factor GATA-1. Development 121, 163–172.
Ransom, D.G., Bahary, N., Niss, K., Traver, D., Burns, C., Paffett-
Lugassy, N., Saganic, W.J., Lim, C.A., Hersey, C., Zhou, Y., et al.
(2004). The zebrafish moonshine gene encodes transcriptional inter-
mediary factor 1␥, an essential regulator of hematopoiesis. PLoS
2, 1188–1196.
Rhodes, J., Hagen, A., Hsu, K., Deng, M., Xi Liu, T., Look, A.T., and
Kanki, J.P. (2004). Interplay of Pu.1 and Gata1 determines myelo-
erythroid progenitor cell fate in zebrafish. Dev. Cell 8, this issue,
97–108.
Takahashi, S., Shimizu, R., Suwabe, N., Kuroha, T., Yoh, K., Ohta,
J., Nishimura, S., Lim, K.C., Engel, J.D., and Yamamoto, M. (2000).
GATA factor transgenes under GATA-1 locus control rescue germ-
line GATA-1 mutant deficiencies. Blood 96, 910–916.
Thompson, M.A., Ransom, D.G., Pratt, S.J., MacLennan, H., Kieran,
M.W., Detrich, H.W., Vail, B., Huber, T.L., Paw, B., Brownlie, B.,
et al. (1998). The cloche and spadetail genes differentially affect
hematopoiesis and vasculogenesis. Dev. Biol. 197, 248–269.
Traver, D., Paw, B.H., Poss, K.D., Penberthy, W.T., Lin, S., and Zon,
L.I. (2003). Transplantation and in vivo imaging of multilineage en-
graftment in zebrafish bloodless mutants. Nat. Immunol. 4, 1238–
1244.
Tsai, F.Y., and Orkin, S.H. (1997). Transcription factor GATA-2 is
required for proliferation/survival of early hematopoietic cells and
mast cell formation, but not for erythroid and myeloid terminal differ-
entiation. Blood 89, 3636–3643.
Weiss, M.J., Keller, G., and Orkin, S.H. (1994). Novel insights into
erythroid development revealed through in vitro differentiation of
GATA-1 embryonic stem cells. Genes Dev. 8, 1184–1197.
Westerfield, M. (1994). The Zebrafish Book: A Guide for the Labora-
tory Use of Zebrafish (Brachydanio rerio), 2.1 Edition (Eugene, OR:
University of Oregon Press).
Yamada, T., Abe, M., Higashi, T., Yamamoto, H., Kihara-Negishi, F.,
Sakurai, T., Shirai, T., and Oikawa, T. (2001). Lineage switch induced
by overexpression of Ets family transcription factor PU.1 in murine
erythroleukemia cells. Blood 97, 2300–2307.
Zhang, D.E., Zhang, P., Wang, N.D., Hetherington, C.J., Darlington,
G.J., and Tenen, D.G. (1997). Absence of granulocyte colony-stimu-
lating factor signaling and neutrophil development in CCAAT en-
hancer binding protein alpha-deficient mice. Proc. Natl. Acad. Sci.
USA 94, 569–574.