The role of jak2a in zebrafish hematopoiesis

The Janus kinase (Jak)/signal transducer and activation of transcription (stat) cascade is a ubiquitous intracellular pathway that transduces signals from extracellular ligands. In mammals, the Jak family of protein tyrosine kinases consists of 4 members (Jak1-Jak3, Tyk2), of which Jak2 transduces signals of hematopoietic cytokines to enhance erythroid cell proliferation.¹  The recent identification of a gain-of-function mutation of Jak2 in human polycythemia vera (PV) has revised the diagnosis and perhaps future treatment of this disorder.²  In PV, substitution of valine by phenylalanine at amino acid 617 in the autoregulatory JH2 domain (V617F) results in constitutive activation of Jak2 (Jak2^V617F), conferring proliferative and survival advantage to erythroid progenitors.³  As a result, patients suffer from erythrocytosis and hence thrombosis and bleeding. A similar gain-of-function mutation in the Jak JH2 domain has also been described in the single Drosophila Jak, which is encoded by the hopscotch (hop) locus. A glutamic acid to lysine substitution at residue 695 (E695K) in the JH2 domain of the Hop protein resulted in hyperphosphorylation of the Drosophila Stat (D-Stat) and overproduction of primitive blood cells.⁴ 

The zebrafish has emerged as a model organism of the study of hematopoiesis during embryonic development and neoplastic transformation.⁵  In this organism, primitive and definitive hematopoiesis arise successively in the intermediate cell mass (ICM) and the ventral wall of the dorsal aorta in the developing embryos.⁶  In zebrafish, the gene encoding for Jak2 has undergone duplication and subsequent specialization: jak2a is expressed predominantly in the ICM, and jak2b is expressed predominantly in the developing eyes and pronephric ducts.⁷  However, functional studies of jak2a in zebrafish are lacking. We have previously demonstrated using microarray analysis that jak2a expression was significantly up-regulated in the ICM of the zebrafish chordin morphant that is characterized by expansion of primitive hematopoiesis.⁸  This suggested that jak2a may be involved in the hematopoietic expansion of this morphant.

In this study, we performed both qualitative and quantitative analyses to specifically examine the role of jak2a in zebrafish hematopoiesis using morpholino (MO) knock-down. Furthermore, we investigated to determine if jak2a activation is involved in the hematopoietic expansion of the chordin morphant and evaluated the effects of constitutive activation of jak2a on hematopoiesis.

Danio rerio (wild-type) were obtained from a local aquarium and were maintained and raised under standard conditions at 28°C. Wild-type embryos were obtained from natural spawning and were staged according to Kimmel et al.⁹  Transgenic zebrafish lines Tg(gata1:GFP)¹⁰  and Tg(fli1:GFP)¹¹  were obtained from Tsinghua University (gift from Dr Anming Meng) and Zebrafish Information Network, Eugene, Oregon). Antisense MOs (Gene-Tools, Philomath, OR) were designed to target the 5′ untranslated region (UTR) of the respective genes. MOs were also designed at the intron-exon junction (splicing site [SS]) to induce defective splicing (Table S1, available on the Blood website; see the Supplemental Materials link at the top of the online article). Where appropriate, a random sequence MO was used as a control,^12,13  which has no effect on the morphology and erythropoiesis of the developing embryos (unpublished data, A.C.H.M., January 2007). Solutions of these were prepared and injected into embryos at the 1- to 4-cell stage, which were maintained at 28°C until analyzed. In some experiments, jak2a activity was suppressed by incubating 1-cell–stage embryos with a soluble Jak2 inhibitor, AG490 (Calbiochem, San Diego, CA)¹⁴  until analysis. Protocols for whole-mount in situ hybridization and O-dianisidine staining have been described previously.^8,12,13 

A wild-type zebrafish jak2a clone in the mammalian expression vector pCS2⁺ has been described previously.⁷  Mutagenesis to generate a hyperactive E629K mutant equivalent to the Drosphila hop T42 mutation⁴  was performed using a QuikChange Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA), following the manufacturer's recommendations. Sequence analysis confirmed that the change had been correctly introduced in the absence of extraneous mutations. For rescue experiments, RNA encoding wild-type jak2a was generated by in vitro transcription and injected as described. Embryos were examined with a stereomicroscope (Nikon SMZ800; Nikon, Kawasaki, Japan) with a P-plan 1× objective. Images were captured with Nikon Coolpix E4500 (Nikon, Hong Kong, China) and processed with Adobe Photoshop Version 7.0 (Adobe, San Jose, CA). Fluorescence images were examined with Olympus IX70 (Olympus, Tokyo, Japan) with 10×/10.3 NA objective (P.T.O.). Images were captured with Olympus DP71 and acquired with Olympus DP-BSW basic software and processed with Adobe Photoshop version 7.0.

Total embryo protein extract was prepared by homogenizing 50 embryos at 18 hours after fertilization (hpf) in 50 μL protein extraction buffer (Tris.HCl [63 mM; pH 6.8], glycerol [10% vol/vol], β-mercaptoenthanol [5% vol/vol], and SDS [3.5% wt/vol]) and boiling for 1 minute. Protein extracts (100 μg) were fractionated on a 10% (wt/vol) SDS polyacrylamide gel and electrotransferred to nitrocellulose membrane (Protran; PerkinElmer Life Science, Boston, MA). The membrane was then blocked in blocking buffer (5% nonfat milk in Tris-buffered saline [TBS]) for 1 hour at room temperature, followed by overnight incubation at 4°C with primary antibody, either 1:250 rabbit anti-Stat5 antibody (Cell Signaling Technology, Beverly, MA) or 1:250 rabbit anti–phospho-Stat5 antibody (Zymed Laboratories, South San Francisco, CA). Afterward, the membrane was washed in TBS plus 0.05% (vol/vol) Tween-20 (TBST) followed by incubation with horseradish peroxidase (HRP)–linked electrochemiluminescence (ECL) donkey anti–rabbit IgG antibody (1:1000) at 4°C (Amersham Biosciences, Buckinghamshire, United Kingdom). The membrane was then washed with TBST followed by TBS before detection with SuperSignal West Pico Chemiluminescent Substrate (Pierce, Rockford, IL) and Hyperfilm ECL (Amersham Biosciences, Little Chalfont, United Kingdom). To reprobe the same membrane with another primary antibody, the membrane was incubated in stripping buffer (Tris.HCl [62.4 mM at pH 6.7], β-mercaptoenthanol [100 mM], SDS [2% wt/vol]) at 50°C for 15 minutes, followed by repeated washing in TBST.

Tg(gata1:GFP) embryos at 18 hpf were dechorionated and digested with 0.05% trypsin/EDTA solution (Invitrogen, Carlsbad, CA) for 15 minutes at 28°C, then completely dissociated to single-cell suspension by pipetting. Trypsin digestion was terminated by CaCl₂ (2 mM), and the whole suspension was filtered through a 40-μm cell strainer (BD Falcon, BD Biosciences Discovery Labware, Bedford, MA). Cells were washed and harvested in phosphate-buffered saline (PBS) with 2% (vol/vol) FBS, and the percentage of GFP⁺ cells were enumerated by flow cytometry.

Quantitative reverse transcriptase–polymerase chain reaction (Q-RT-PCR) was performed to examine the relative expression of genes specific for early progenitor (scl, lmo-2, gata2), erythroid (gata1, alpha and beta embryonic Hb 1, and αhe1 and βhe1), early myeloid (spi1), heterophilic granulocyte (myeloperoxidase, mpo), and macrophage (l-plastin) lineages using the ABI Prism 7700 Sequence Detector (PE Biosystems, Foster City, CA). Embryos at 12 and 18 hpf were dechorionated, and RNA was extracted using Trizol reagent as per the manufacturer's protocol (Invitrogen, Carlsbad, CA). A total of 30 embryos were included in each experiment. RNA was reverse-transcribed and PCR was performed using the SYBR Green Method. Primer sequences used for Q-RT-PCR are shown in Table S1.

To ensure that the amplification efficiency for all marker genes in the Q-RT-PCR was equal, RNA was extracted before the project from a group of 40 embryos at 24 hpf followed by reverse transcription. Serial dilutions of samples ranging from 0.0025 to 125 ng input cDNA were frozen at −20°C in small aliquots (standards). Once thawed, each aliquot was used only once. Relative quantification of each gene expression was performed by the comparative C_T methods.

Data are expressed as means (± SEM; calculated by dividing the SD by the square root of the number of replicate experiments). Comparisons between numerical data were evaluated by paired Student t tests. A P value less than .05 was considered statistically significant.

This study was approved by the Committee of the Use of Laboratory and Research Animals (CULATR) at the University of Hong Kong.

We first examined the expression of jak2a during zebrafish embryonic development using whole-mount in situ hybridization. In agreement with another study,⁷  jak2a was expressed predominantly in the ICM of zebrafish embryos at 18 and 24 hpf (Figure 1A,B,E). Expression of jak2a was significantly up-regulated within the expanded ICM of the chordin morphant at both time points (Figure 1C,D,F), confirming our previous observations based on microarray analysis.⁸ 

To investigate the role of jak2a in primitive hematopoiesis, we injected the jak2a^UTR-MOs into zebrafish embryos at the 1- to 4-cell stage. Initial experiments showed that most embryos tolerated 7.5 ng and developed normally at 24 and 48 hpf (Figure 1G-J). However, at higher doses, this MO caused excessive toxicity and lethality. Therefore, in all subsequent experiments, the embryos were injected with a dose of 7.5 ng (referred to hereafter as jak2a^UTR embryos). Erythropoiesis was reduced in these embryos at 48 hpf as shown by O-dianisidine staining (Figure 1K,L). To confirm the specificity of this result, a jak2a splice-site MO targeting an exon-intron junction (jak2a^SS-MO; 2 ng) was injected. These jak2a^SS embryos showed a similarly reduced erythropoiesis at 48 hpf (Figure 1M). Moreover, the efficacy of the jak2a^SS-MOs could be demonstrated by RT-PCR showing significantly reduced levels of jak2a transcript (Figure 1O-P). To further correlate tyrosine kinase activity with primitive hematopoiesis, we incubated freshly spawned embryos with AG490 (50 μM), a known soluble inhibitor of Jak2 (jak2a^AG embryos). At 48 hpf, O-dianisidine staining was also markedly reduced in these embryos, corroborating the proposition that jak2a regulates primitive hematopoiesis during zebrafish embryonic development (Figure 1N).

We performed quantitative analysis on the effects of jak2a knock-down on the gata1⁺ population using transgenic Tg(gata1:GFP) embryos at 18 hpf, before the onset of functional circulation. In these embryos, GFP⁺ cells represent cells committed to the erythroid lineage. In uninjected embryos, the GFP⁺ population constitutes 4.47% (± 0.18%) of the dissociated embryos. A significant reduction in the GFP⁺ population was observed in jak2a^UTR (3.56% ± 0.58%; P=.017), jak2a^SS (2.68% ± 0.40%; P=.01), and jak2a^AG (3.31% ± 0.18%; P=.003) embryos (Figure 2A). Furthermore, coinjection of embryos with both jak2a^UTR-MOs (7.5 ng) and jak2a^SS-MOs (2 ng) resulted in a further reduction of erythropoiesis (2.15% ± 0.10%; P < .001) compared with either of these MOs alone, demonstrating synergism and hence specificity of these MOs on jak2a gene function. Specificity of jak2a^UTR-MO was further demonstrated by a reversal of its inhibitory effect on erythropoiesis after rescue by wild-type jak2a mRNA (75 pg) coinjection (4.25% ± 0.03%; P=.86). Collectively, these observations confirmed the findings with O-dianisidine staining that jak2a knock-down resulted in a decrease in erythropoiesis.

To confirm that the reduced erythropoiesis in zebrafish embryos after down-modulation of jak2a was related to reduction in Jak/Stat signaling, we examined the phosphorylation status of endogenous stat5 protein in these embryos. Although the amount of stat5 protein was unchanged in jak2a^UTR, jak2a^SS, and jak2a^AG embryos, stat5 phosphorylation was significantly reduced compared with that of uninjected embryos (Figure 2B). The results supported the notion that stat5 mediates the downstream effects of jak2a.

The effects of jak2a knock-down on hematopoiesis led us to examine its effects on angiogenesis as both processes arise from a common precursor, known as the hemangioblasts.¹⁵  Moreover, in the zebrafish cloche mutant, whose blood and vascular formation are defective, jak2a expression was significantly down-regulated.⁷  We injected jak2a^UTR-MOs into transgenic Tg(fli1:GFP) embryos at the 1- to 4-cell stage. At 48 hpf, both vasculogenesis and angiogenesis was not affected as shown by the intact axial and intersegmental vessels (Figure 2C,D). A patent circulation in these vessels was further confirmed by microangiography (data not shown).

In order to understand the regulatory role of jak2a in the hematopoietic hierarchy, we examined the expression of a panel of hematopoietic genes in the jak2a^UTR embryos by Q-RT-PCR (Table 1). At 12 hpf, expression of scl, lmo2, gata1, αhe1, and βhe1 were significantly reduced in jak2a^UTR embryos, while expression of mpo was unaffected. At 18 hpf, expression of genes encoding for all these hematopoietic genes were reduced. At both time points, spi1 appeared to be down-regulated, although the difference from uninjected embryos did not reach statistical significance. Expression of fli1 was not affected in the jak2a^UTR embryos. Intriguingly, expression of l-plastin was significantly up-regulated in jak2a^UTR embryos at both time points.

The up-regulation of jak2a expression in the chordin morphant in which there is expansion of primitive hematopoiesis led us to investigate whether increased jak2a activity may mediate the hematopoietic expansion in these embryos. Wild-type embryos injected with chordin MO (0.75 ng) induced expansion of ICM in nearly 90% of embryos (Figure 3A-B). However, coinjection of chordin and jak2a^UTR-MOs resulted in significant reduction in ICM expansion both in the average size of the expansion as well as the number of embryos showing significant expansion (defined arbitrarily by ICM of 5 somites or more; Figure 3C). Similar changes were recapitulated by coinjection of chordin and jak2a^SS-MOs or by incubating chordin MO–injected embryos with AG490 (Figure 3D). Using flow cytometry on Tg(gata1:GFP) embryos at 18 hpf, jak2a^UTR-MOs and jak2a^SS-MOs as well as AG490 were able to reduce the increase in the GFP⁺ population in the chordin morphant (Figure 3E). Q-RT-PCR showed that the expression of scl, lmo2, gata1, αhe1 and βhe1, spi1, mpo, and l-plastin were up-regulated in the chordin morphant at 12 and 18 hpf (Table 1). With the exception of l-plastin, the increases in gene expression were ameliorated by coinjection with Jak2^UTR-MOs.

The results from the chordin morphant suggested that activation of jak2a in zebrafish embryos may increase hematopoiesis. To test this hypothesis, we injected the plasmid containing the constitutive-active form of jak2a (jak2a^ca, 150 pg; Figure 4A) into 1-cell–stage wild-type or Tg(gata1:GFP) embryos and examined its effects on hematopoiesis, stat5 phosphorylation, and gene expression. The jak2a^ca embryos had normal morphology (Figure 4B-E). When jak2a^ca was injected into Tg(gata1:GFP), there was a significant increase in gata1⁺ cells (uninjected embryos, 4.26% ± 0.06% vs jak2a^ca embryos, 5.26% ± 0.14%; P=.019, n=4 separate paired experiments including a total of 120 embryos; Figure 4F,G). Furthermore, stat5 phosphorylation was significantly increased in the jak2a^ca embryos. Similar increase was observed in the chordin morphant embryos (Figure 4H). In addition, expression of gata1, αhe1, βhe1, spi1, mpo, and l-plastin was significantly increased. On the other hand, those encoding for scl, lmo2, and fli1 did not show any changes in the jak2a^ca embryos (Table 2).

In this study, we examined the role of jak2a during embryonic hematopoiesis in zebrafish. Injection of jak2a^UTR-MOs significantly reduced primitive hematopoiesis, as shown qualitatively by O-dianisidine staining and quantitatively by flow cytometry of dissociated Tg(gata1:GFP) embryos and by real-time quantitative PCR. Enumeration of gata1⁺ cells at 18 hpf is specific for erythropoiesis, as thrombopoiesis, which is also under gata1 regulation, occurs at a later time point during embryonic development.¹⁶  Specificity of the MO was shown by the synergism between this and the jak2a^SS-MO, which induced defective splicing of jak2a pre-mRNA, and by use of a pharmacologic inhibitor to reproduce the observed phenotypes. Moreover, the hematopoietic defects of the jak2a^UTR embryos could be rescued by wild-type jak2a mRNA. The 2 MOs were different in their efficacies of perturbing hematopoiesis under basal and stimulated conditions. Whereas the efficacy of jak2a^SS-MO could be tested by RT-PCR, that of jak2a^UTR-MO was not testable due to the lack of specific antibody against zebrafish jak2a. Notwithstanding this limitation, the overall results were consistent with earlier observations showing reduced jak2a expression in the cloche mutant and corroborated with the proposition that jak2a mediates primitive hematopoiesis in zebrafish.^7,17  In mice, the function of Jak2 in embryonic hematopoiesis could not be examined mechanistically because homozygous Jak2^null mice are embryonic lethal at days 12.0 to 13.0 after coitus.^18,19  Therefore, the zebrafish system has provided us with a unique model for the examination of this gene with hitherto uncertain function during embryonic development, and we have made a number of observations that may shed light to our understanding of jak2a during normal and deregulated hematopoiesis.

First, we demonstrated that jak2a plays a nonredundant role in the initiation of primitive hematopoiesis. In particular, genes associated with early hematopoietic specification (scl and lmo2) as well as erythroid (gata1, αhe1, and βhe1) and myeloid (spi1 [early] and mpo [late]) differentiation were down-regulated in the jak2a^UTR embryos as early as 12 hpf. Therefore, jak2a might be required at the level of early hematopoietic progenitor/stem cells before they differentiate into distinct erythroid and myeloid lineages. We have recently shown that zebrafish stat5.1 is crucial for multilineage hematopoietic development (R. S. Lewis, C. Liongue, and A.C.W., manuscript submitted). Given the tight correlation between jak2a activation and stat5 phosphorylation observed in this study, it is likely that the jak2a-stat5.1 signaling module is critically involved in the cell-fate decision of early hematopoietic progenitor/stem cells. Intriguingly, l-plastin, which is associated with primitive macrophage development,²⁰  was significantly up-regulated in jak2a^UTR embryos at both 12 and 18 hpf. This lineage appears to be derived from an anterior hematopoietic precursor population distinct from the posterior ICM population.^21,22  Whether the paradoxical up-regulation of l-plastin represented skewing of hematopoietic progenitor cell-fate from erythromyeloid toward a macrophage lineage, or from a posterior to an anterior precursor population, warrants further examination.

Second, we demonstrated that erythrocytosis in zebrafish embryos is mediated by jak2a activation, consistent with enforced expression of a tel-jak2a fusion.²³  The proposition was first supported by the study of the chordin morphant, whose hematopoietic expansion was significantly ameliorated by jak2a knock-down. Furthermore, constitutive activation of jak2a also resulted in a significant increase (24%) in hematopoiesis, as defined quantitatively by flow cytometry in Tg(gata1:GFP) embryos. Such an increase in erythropoiesis might not manifest morphologically in the developing embryos, and nonquantitative whole-mount in situ hybridization did not show any consistent increase in gata1 staining (unpublished data, A.C.H.M., February 2007). However, consistent with the data from flow cytometry, we were able to demonstrate both an increase in erythroid (gata1, αhe1, and βhe1) and myelomonocytic gene expression (spi1, mpo and l-plastin) using quantitative real-time PCR as well as an increase in stat5 phosphorylation with Western blotting. Notably, constitutive activation of stat5.1 has also been shown to increase gata1⁺ and spi1⁺ populations in zebrafish embryos.²¹  The jak2a^ca used in the present study was generated based on the Drosophila Hop^T42 mutant carrying a gain-of-function mutation (E695K) at the JH2 domain.⁴  In this mutant, proliferation of primitive hematopoietic cells was increased, and the Drosophila Jak/Stat pathway was activated.²⁴  Our observations therefore highlight a conserved phenomenon in which gain-of-function mutations of the JH2 domain in Jak2 induce hematopoietic proliferation in Drosophila, zebrafish, and human via activation of Stat signaling. Intriguingly, constitutive jak2a activation had no effect on scl or lmo2 expression. Therefore, the effects of jak2a stimulation (presumably via stat5.1) appear to be restricted to the lineage-committed erythroid/myeloid precursors rather than the early hematopoietic progenitor cells.

Third, we demonstrated that neither knock-down or constitutive activation of jak2a had any effect on endothelial cell specification, as shown by Q-RT-PCR for fli-1 and the intact vasculature in Tg(fli1:GFP) embryos injected with jak2a^UTR-MOs. This issue has not been addressed in murine model, as Jak2 knock-out results in early embryonic lethality.^18,19  Remarkably, the jak2a^UTR embryos showed specific reduction of gene expression associated with early hematopoiesis but not vasculogenesis, and may be useful as a tool whereby these hitherto intertwined processes can be dissected. Our data also suggested that jak2a functions at a higher level in the hematopoietic hierarchy than phospholipase C gamma 1 (PLCγ1) during development, as specific knock-down of PLCγ1 reduces gata1, αhe1, and βhe1, but not scl or lmo2 expression.¹³ 

In the present study, the use of flow cytometry in dissociated transgenic embryos and Q-RT-PCR has enabled us to examine zebrafish hematopoiesis objectively and quantitatively. In fact, the differential gene expression induced by jak2a^UTR-MOs could not be differentiated reliably by whole-mount in situ hybridization (unpublished data, A.C.H.M., February 2007), which is nonquantitative and is subject to biases due to variations in embryo staining. Therefore, quantitative analyses such as those described herein represent a useful supplement to nonquantitative tests in future study of zebrafish embryonic hematopoiesis.

The results of this study are of clinical relevance. Recent identification of Jak2^V617F in PV²³  suggested that specific targeting of Jak2 signaling might lead to better treatment for this disorder. The jak2a^ca embryos may therefore shed light on our understanding of human PV and provide us with a robust model for the screening of potential therapeutic agents used for this and other Jak2-related blood diseases. A zebrafish equivalent of the jak2^V617F mutation has also been generated (unpublished data, A.C.W., January 2007), and this model will enable us to dissect the pathogenesis of jak2^V617F-induced neoplastic transformation. On the other hand, the up-regulation of jak2a in chordin morphants and the amelioration of expanded hematopoiesis by jak2a knock-down have provided us with 2 possible leads for further studies. First, a possible link between bone morphogenetic protein (BMP) and jak2a signaling should be examined (Figure 5). In particular, whether the increase in jak2a signaling in the embryos was due to a cross-talk between Smad and stat pathways as reported in neural stem cells²⁵  or a simple increase in hematopoietic cell population in the embryos should be further studied. Second, previous studies have demonstrated the pivotal role of BMP4 in mesoderm induction and hematopoietic differentiation during embryonic development,^26,27  and whether and how deregulation of BMP/Smad signaling might link to human leukemia would have to be further evaluated.

We thank Dr Anming Meng (Tsinghua University, China) for the generous gift of the Tg(gata1:GFP) fish line. We also thank Ms Jessie Fu, Babs Kwok, and Rachel Lin for performing some of the microinjection experiments; Dr Andrew Oates for the jak2a clone; and Mr Howard Chow and Ms Rowena Lewis for part of the molecular studies.

This work was supported by Research Grants Council grant (University of Hong Kong 7488/04M and 7520/06M) and small project funding from The University of Hong Kong.

Contribution: A.C.H.M. conducted the study and wrote the manuscript. A.C.W. conducted part of the experiments. R.L. analyzed the data and wrote the manuscript. A.Y.H.L. designed the research, analyzed the data, and wrote the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Anskar Y. H. Leung, Rm K418, K Block, Department of Medicine, Queen Mary Hospital, The University of Hong Kong, Pok Fu Lam Rd, Hong Kong, China; e-mail: ayhleung@hku.hk.