The role and regulation of friend of GATA-1 (FOG-1) during blood development in the zebrafish

Abstract

The nuclear protein FOG-1 binds transcription factor GATA-1 to facilitate erythroid and megakaryocytic maturation. However, little is known about the function of FOG-1 during myeloid and lymphoid development or how FOG-1 expression is regulated in any tissue. We used in situ hybridization, gain- and loss-of-function studies in zebrafish to address these problems. Zebrafish FOG-1 is expressed in early hematopoietic cells, as well as heart, viscera, and paraspinal neurons, suggesting that it has multifaceted functions in organogenesis. We found that FOG-1 is dispensable for endoderm specification but is required for endoderm patterning affecting the expression of late-stage T-cell markers, independent of GATA-1. The suppression of FOG-1, in the presence of normal GATA-1 levels, induces severe anemia and thrombocytopenia and expands myeloid-progenitor cells, indicating that FOG-1 is required during erythroid/myeloid commitment. To functionally interrogate whether GATA-1 regulates FOG-1 in vivo, we used bioinformatics combined with transgenic assays. Thus, we identified 2 cis-regulatory elements that control the tissue-specific gene expression of FOG-1. One of these enhancers contains functional GATA-binding sites, indicating the potential for a regulatory loop in which GATA factors control the expression of their partner protein FOG-1.

Figure 1

FOG-1 expression pattern during early hematopoiesis. The expression pattern of FOG-1 and GATA factors was visualized by WISH during zebrafish embryonic development (A-O). The maternal expression of FOG-1 (A-B) and GATA-2 (K-L) transcripts is initially noted at the 2-cell and sphere stages. The zygotic expression pattern of FOG-1 in the blood island is coincident with the expression of GATA-1 and GATA-2 in the LPM (C,H,M) and the ICM (D-E,I-J,N-O). FOG-1 is highly expressed in the ICM and neural tissues of a wild-type embryo at 24 hpf (P). Mutants involved in defective formation of hematopoietic (cloche, clo; Q) or erythroid progenitors (vlad tepes, vlt; R) are deficient in the expression of FOG-1 in the ICM (black arrow), whereas expression in the heart is preserved (white arrow, inset magnification). The expression of FOG-1 in the ICM is normal in a mutant with a defect in late erythroid maturation (frascati, frs; S).

Figure 2

FOG-1 is necessary for erythroid/thrombocytic development in zebrafish. Uninjected control embryos (A,D,G), FOG-1 MO-injected embryos (B,E,H), and FOG-1 MO coinjected with FOG-1 cRNA (C,F,I). Lateral views of 24-hpf embryos processed by WISH to reveal band3 expression (A-C). Lateral (D-Di,E-Ei,F-Fi) and ventral (Dii,Eii,Fii) views of 48 hpf embryos processed by o-dianisidine staining (D-F) to reveal hemoglobinized cells. Lateral view at the trunk level of 4-dpf transgenic zebrafish [Tg(cd41:eGFP)] embryos to reveal eGFP⁺ thrombocytes (G-I). FOG-1 morphant embryos display reduction or complete absence of band3 expression in the ICM (A-B black brackets, inset magnification). The morphant embryos show severe anemia when stained with o-dianisidine (D-E black arrows). Injection of FOG-1 MO in the transgenic zebrafish [Tg(cd41:eGFP)] produces complete absence of eGFP⁺ thrombocytes at 4 dpf (G-H white brackets). The anemic phenotype is fully rescued by coinjection of FOG-1 cRNA in the morphant embryos (C,F). In contrast, the thrombocytic phenotype is only partially rescued (I, white arrow). The bars represent quantification of the normalized phenotypes (mean ± SD) with the numbers of embryos analyzed in each category indicated above the bar (J-L). White, black, and gray bars represent embryos with normal, reduced, and absence of erythroid/thrombocytic markers, respectively. Data were derived from 3 independent experiments. The statistical significance, marked by lines for each paired condition, was analyzed using analysis of variance; *P < .001; **P < .02.

Figure 3

Loss of zebrafish FOG-1 results in defective lymphopoiesis in the thymic organs. Lateral views of 4-dpf control (wild-type [wt], A,E,I,M,O) and FOG-1 morphant embryos (B-C,F-G,J-K,N,P) labeled for the following markers: lck (A-C), rag-1 (E-G), ikaros (I-K,M-N), and pax9 (O-P). In control embryos, the thymic cells express normal levels of lck (A), rag-1 (E), and ikaros (I). In contrast, the morphant embryos show strong reduction or absence of lck (B-C), rag-1 (F-G), and ikaros (J-K) expression in the thymi. The expression of ikaros is maintained in the CHT, indicating normal formation of lymphoid progenitors (M,N, brackets and insets). In the lateral views of 4 dpf embryos, the expression of pax9 is normal in the FOG-1 MO-injected embryos, indicating that the initial phase of pharyngeal endoderm formation is not affected (O-Q). In addition, the embryos were processed with Alcian blue staining to delineate the morphologic architecture of the pharyngeal arch cartilages in control (wt; R) and morphant (S) embryos. The morphant embryos display dysplastic development of the pharyngeal arches, indicating a disruption of late endodermal derivatives: pharyngeal arch cartilage and thymic anlage (S). The bars represent quantification of the normalized phenotypes (mean ± SD) with the numbers of embryos analyzed in each category indicated above the bars. White, black, and gray bars indicate embryos with normal, reduced, and absent expression for lymphoid markers, respectively (D,H,L,Q). Results were derived from 3 independent experiments. The statistical significance, marked by lines for each paired conditions, was analyzed using analysis of variance; *P < .05.

Figure 4

FOG-1 acts independently of GATA-1 during thymic development. Lateral views of uninjected/Tg(lck:eGFP)/control embryos (A,D,G,J) compared with embryos injected with GATA-1MO (B,E,H,K) or FOG-1MO (C,F,I,L). The pericardial effusion is noted in the FOG-1 morphant (C black arrow). Knocking down of either GATA-1 or FOG-1 shows drastic anemia when stained with o-dianisidine for hemoglobinized cells (D-F). The transgenic zebrafish [Tg(lck:eGFP)] injected with FOG-1 MO shows complete absence of eGFP⁺-T lymphocytes in the thymi (comparing G,J with I,L). In contrast, GATA-1 morphant shows normal levels of eGFP expression in the thymi (H,K).

Figure 5

FOG-1 is required for myelopoiesis in developing zebrafish. Uninjected control embryos (A,D,G), FOG-1 MO-injected embryos (B,E,H), vlt mutants (J), and vlt mutants injected with FOG-1 MO (K). Lateral views of embryos fixed at 22 and 24 hpf were processed by single-labeled WISH to reveal the expression of the myeloid markers, PU.1 (A-B) and mpo (D-E), respectively. Lateral views of embryos fixed at 48 hpf were processed by double-labeled WISH to reveal the expression of the myeloid-specific marker mpo (purple) and the erythroid-specific marker band3 (red; G-H,J-K). Black arrow indicates myeloid cells; white arrow, erythroid cells. Higher magnification views of the ICM (square bracket, D-E) and the ducts of Cuvier on the yolk (Ji-Ki). Loss of FOG-1 function results in an increased expression of myeloid-specific markers, PU.1/mpo (A-F) and decreased expression of the erythroid-specific marker, band3 (G-H). Mutation in the GATA-1 gene causes an increase in the number of mpo⁺ cells in the vlt mutants compared with wild-type embryos. Injection of FOG-1MO in vlt further expands mpo⁺ cells compared with vlt control embryos (compare J with K). Genotyping of injected embryos (FOG-1 MO) with increased myeloid markers was performed to verify their vlt (−/−) genotype (L). Bars represent quantification of the normalized phenotypes (mean ± SD) with the numbers of embryos analyzed in each category indicated above the bars (C,F,I,M). White, blue, and black bars represent embryos with normal, increased, and reduced expression for myeloid markers, respectively. Results were derived from 3 independent experiments. The statistical significance, marked by lines for each paired conditions, was analyzed using analysis of variance: *P < .001; **P < .05.

Figure 6

Loss of FOG-1 expands myelopoiesis at the expense of erythropoiesis. Lateral views at 24 hpf of Tg(GATA-1:eGFP) embryos uninjected (A,C) or injected with FOG-1MO (B,D). White arrows indicate eGFP expression in the myeloid precursor cells (m); square bracket, ICM expression in insets (C-D). Gated erythroid (red circle) and myeloid cells (blue circle) are shown using forward (FSC) and side scatter (SSC) flow cytometry (E-F). The eGFP⁺ cells were purified by fluorescence-activated flow cytometry (G-H). Histologic analysis of sorted eGFP⁺ cells from FOG-1 morphants shows myelomonocytic (H asterisks) and dyserythropoietic morphology (H arrows); in comparison, the sorted cells from control embryos are predominantly erythroblasts (G). A representative from 3 independent experiments is shown. Populations of cells within the gate are enumerated as mean percentages of total cells ± SD, showing statistically significant differences (*P < .05; E-F).

Figure 7

Conserved cis-enhancer fragment regulates the expression of FOG-1 in zebrafish blood and heart. Cis-regulatory modules in the mouse FOG-1 locus on chromosome 8 are compared with other vertebrates (A). The high regulatory potential (hi-RP) and GATA-binding sites (GATA-1_BS) are represented as black boxes (hiRP&GATA-1_BS), the exons as blue boxes, the FOG-1 enhancer candidates (FE1-FE4) as red boxes, and the conservation score as “CS.” These 4 FE fragments were interrogated for in vivo GATA-1–binding activity using ChIP assay and real-time PCR (B). The relative occupancy levels of GATA-1 are indicated by the fold enrichment at each of the sites shown, normalized to levels at the negative control region (2 kb 5′ to the GATA-1 gene enhancer HS1). The bar graphs represent quantification (mean ± SD) for GATA-1 binding (3 independent experiments) using nuclear extracts from differentiated MEL cells. The mouse GATA-2 (GATA-2, −2.8 kb) and the c-kit (c-kit, + 5 kb) promoters were used as positive controls for GATA-1 occupancy. *The only significant GATA-1 occupancy in the FOG-1 locus at the FE2 site (P < .001). PI indicates the preimmune sera control. WISH shows endogenous expression of FOG-1 mRNA in the intraembryonic blood island (ICM, red arrow) and Rohon-Beard neurons (yellow arrow) (C). The FE1 from mouse (D,F-G) and zebrafish (H-I) robustly drives the expression of eGFP in the ICM (red arrow), but only FE2 from either zebrafish or mouse is expressed in Rohon-Beard paraspinal neurons (E,J-K and yellow arrows in E,J). The developmental stages are properly indicated.