Functions of flt3 in zebrafish hematopoiesis and its relevance to human acute myeloid leukemia

The zebrafish has emerged as a model for the study of embryonic hematopoiesis.^1,2  In primitive hematopoiesis, pu.1 is first expressed at 12 hours postfertilization (hpf) in the anterior lateral plate mesoderm (ALPM) where it plays a role in driving cellular differentiation toward myeloid fate, independent of primitive erythropoiesis that occurs in the posterior lateral plate mesoderm.³  After formation of the rostral blood island at 16 hpf, the pu.1⁺ myeloid progenitor cells begin to spread on the yolk sac,⁴  and subsequently switch on expression of pan-leukocyte gene l-plastin,^5-7  as well as those genes associated with macrophage (csf1ra, mfap4, cxcr3.2, mpeg1, and ptpn6)⁸  and neutrophil lineages (lyc and mpo).⁵  The erythromyeloid progenitors that arise autonomously in the posterior blood island (PBI) also generate myeloid cells associated with l-plastin, pu.1, and mpo expression at 30 hpf.⁹ 

Definitive hematopoiesis arises from the ventral wall of the dorsal aorta (DA) as evidenced by the expression of c-myb and runx-1 at 36 hpf.^9,10  It subsequently migrates to the caudal hematopoietic tissue (CHT), thence the kidney marrow, where lifelong hematopoiesis occurs.¹¹ gcsf and cebpα are important regulators of granulopoiesis.¹² 

Recent studies also demonstrated that myeloid and erythroid cell fates from the myeloerythroid progenitor cells are precisely controlled by the interplay between pu.1 and gata1^12,13  and the sumoylation of cebpα promotes myeloid cell expansion at the expense of erythroid development.¹⁴ 

FMS-like tyrosine kinase 3 (FLT3) is a class III receptor tyrosine kinase normally expressed in human hematopoietic stem and progenitor cells (HSPCs).¹⁵  FLT3 consists of 5 immunoglobulin-like motifs in the extracellular domain, a transmembrane domain, a juxtamembrane domain (JMD), and 2 tyrosine kinase domains (TKDs) in the intracellular region.^16,17  Under normal circumstances, binding of FLT3 to its ligand results in FLT3 homodimerization and phosphorylation of TKD and hence activation of multiple effectors, providing the proliferative and survival signals to HSPCs.^16-19  Flt3 function on hematopoiesis has been evaluated by targeted deletion in mice and serial transplantations. Specifically, Flt3 knockout was compatible with postnatal survival with only defective B-lymphoid development at baseline.²⁰  In competitive and serial transplantation, although T-lymphoid and myeloid reconstitutions were reduced in primary recipients, they remained stable in secondary recipients. A more recent study only revealed defective B-lymphoid but no effect on hematopoietic stem cell (HSC), T-lymphoid, or myeloid reconstitution upon primary and secondary transplantations.²¹  These observations supported the proposition that Flt3 signaling is dispensable for HSC self-renewal. The role of Flt3 during embryonic hematopoiesis is currently unclear.

Internal tandem duplication (ITD) at the JMD and TKD of FLT3 occurs in 30% of human acute myeloid leukemia (AML) and is associated with inferior clinical prognosis. Single-base mutations at TKD occur in 5% of cases whose prognostic significance is less clear.²²  These gain-of-function changes resulted in constitutive activation of FLT3 and downstream effectors including signal transducer and activator of transcription 5 (STAT5),²³  AKT,²⁴  and extracellular signal-regulated kinase (ERK).²⁵  Newer tyrosine kinase inhibitors (TKIs) including sorafenib and AC220 were effective in inducing remission. However, the effects were transient and emergence of mutant clones, particularly at the residue 835 of the TKD, where aspartate was replaced by tyrosine (D835Y) hence conferring resistance to TKI, was common,^26,27  suggesting that the targets are likely to be evolving²⁸  and that a robust platform and animal model to identify more effective and personalized TKI in real time is needed.

We made use of the zebrafish model to examine the role of flt3 in primitive and definitive hematopoiesis during embryonic development and validate a novel zebrafish model overexpressing human FLT3-ITD and FLT3-TKD with a view of developing a whole-organism platform for personalizing TKI treatment of AML.

Wild-type (WT) zebrafish were maintained under standard conditions and embryos were staged as described.^29,30  The study was approved by the Committee of the Use of Live Animals for Teaching and Research in The University of Hong Kong.

Total RNA was extracted (TRIzol; Invitrogen) from AB zebrafish embryos at 1 day postfertilization (dpf), and reverse transcribed into complementary DNA (cDNA) (SuperScript II Reverse Transcriptase; Invitrogen). A pair of primers (supplemental Table 1, see supplemental Data available on the Blood Web site) was designed to amplify the coding sequence of flt3 sequence by polymerase chain reaction (PCR) using Ex-Taq DNA polymerase (TaKaRa Biotechnology), cloned into pBluescript II SK(+) vector by EcoRI and BamHI and bidirectionally sequenced (Techdragon Limited, for all sequencing in this study) by primer walking (supplemental Table 1).

The morpholino (MO; Gene Tools, LLC) technology has been described.^31,32 flt3 MO (6 ng) targeting the 5′ untranslated regions (UTRs) (supplemental Table 1) was microinjected into 1-cell–stage embryos. The zebrafish flt3 UTR (containing the MO targeting sequence) and partial coding sequence (supplemental Table 1) was cloned into the pegfp-N3 vector (Clontech/BD Biosciences) to generate a flt3UTR-egfp fusion chimeric gene.

The partial cDNA (642 bp) sequence of flt3 was amplified by PCR (supplemental Table 1) and the PCR products were cloned into a pGEM-T-easy vector (Promega) and used as template to generate an antisense digoxigenin (DIG)–labeled RNA probe (DIG RNA Labeling Kit; Roche Applied Science) in vitro.^33,34  Double whole-mount in situ hybridization (WISH) was performed using the DIG- and fluorescein isothiocyanate–labeled probe as previously described.³⁵ 

Total RNA from 30 embryos was isolated using TRIzol reagent (Life Technologies) and reverse transcribed (SuperScript II Reverse Transcriptase; Invitrogen). Primers were shown in supplemental Table 1. Reverse transcription PCR has been described previously.^36,37 

Immunostaining was performed using anti-phospho-Histone H3 (Upstate) with slight modifications³⁸  (for details, refer to the supplemental Figure 7 legend).

Total RNA of AB zebrafish embryos at 24 hpf was extracted and reversed transcribed. Two primers (supplemental Table 1) were used to amplify the flt3 open reading frame. The PCR product was cloned into the pBluescript II SK(+) vector (pBSK-flt3) and confirmed by bidirectional sequencing. The pBSK-flt3 vector was linearized by SacII, purified (QIAquick PCR Purification Kit; QIAGEN), and used for in vitro transcription by T7 RNA polymerase (mMESSAGE mMACHINE T7 Kit; Ambion).

Total RNA from AML cell line KG-1 carrying WT FLT3 was extracted and reverse transcribed. pMSCV-neo-FLT3-ITD containing human FLT-ITD (21 bp duplication) was subcloned. WT FLT3 and FLT3-ITD were amplified using a reverse primer containing the T2a self-cleaving peptide sequence (supplemental Table 1) and inserted into the pegfp-N3 vector in frame with the egfp gene. The plasmids pegfp-N3, pHsFLT3-WT-T2a-egfp, and pHsFLT3-ITD-T2a-egfp were linearized by StuI and purified. Human FLT3-ITD was in-frame cloned into the pDsRed-Monomer-N1 vector to generate the FLT3-ITD-T2a-mRFP fusion gene (Clontech/BD Sciences). Human FLT3-ITD was also cloned into pBluescript II SK(+), and the human FLT3-ITD messenger RNA (mRNA) was in vitro transcribed for microinjection (300 pg per embryo).

Site-directed mutagenesis was performed to generate the FLT3-TKD and FLT3-ITD-TKD (D835Y) mutation (supplemental Table 1). Plasmid (100 ng) and specific primers (1.25 pmol each) were used in PCR with 18 cycles using Platinum Pfx DNA polymerase platform. Cloning and Sanger sequencing followed standard protocols.

Sorafenib and sunitinib (LC Laboratories) were multikinase inhibitors used clinically for the treatment of hepatocellular and renal cell carcinomas.^39,40  AC220 (LC Laboratories) is a potent and relatively selective inhibitor of FLT3 currently evaluated in clinical trials for AML treatment.⁴¹  Embryos after gastrulation (6 hpf) were treated with drugs until collection for examination. Equivolume amounts of dimethylsulfoxide (DMSO) were used in parallel experiments using the same clutch of embryos as control.

The fluorescent images of embryos were taken under Olympus IX70 (Olympus Corporation) and a 10×/0.3 NA objective in 3% methylcellulose, with Olympus DP71 (Olympus Corporation) and Olympus DP-BSW basic software, processed with Adobe Photoshop (version 8.0.1) Bright-field images of embryos were taken under Nikon SMZ800 (Nikon Hong Kong Ltd.) and a P-Plan 1× objective in 3% methylcellulose with Nikon Digital Sight DS-Fil (Nikon Hong Kong Ltd.) and processed with Adobe Photoshop (version 8.0.1) and Image J (National Institutes of Health).

Data were expressed as means ± standard error of the mean. Comparisons between groups of numerical data were evaluated using the Student t test. Categorical data were evaluated using the χ² test. P values < .05 were considered statistically significant, and were represented with an asterisk (*P < .05 or **P < .01).

Zebrafish flt3 contained 2871 bp and encoded a 956-aa protein (supplemental Figure 1). FLT3 are found in different species (supplemental Figure 2) and full-length flt3 protein shared an overall 32%, 35%, and 34% sequence identity with those of human, mouse, and rat, respectively. The JMD and the activation loops of TKD are highly conserved (Figure 1A), implicating conserved functions of FLT3 signaling during evolution.

In adult fish, flt3 expression was found in gill, heart, spleen, and kidney (supplemental Figure 3A). At the embryonic stage, flt3 mRNA was detected as early as the 1-cell stage, indicating maternal transcripts (supplemental Figure 3B). Subsequently, flt3 expression increased from 6 hpf, peaked at 24 hpf, and maintained up to 48 hpf. WISH demonstrated ubiquitous flt3 expression at 3, 5, and 12 hpf (Figure 1B-D). The ubiquitous expression of flt3 was confirmed by histologic sectioning of embryos at 12 hpf (supplemental Figure 4A-B). At 18 hpf, expression at the anterior yolk sac could be demonstrated (Figure 1E). To further ascertain the hematopoietic origin of flt3-expressing cells, the cloche mutants in which primitive hematopoiesis was defective were examined. Normally, α-eHb1 was robustly expressed at the intermediate cell mass (ICM) of WT and heterozygous cloche embryos, and its absence served to ascertain the homozygous state of the cloche embryos. Based on this differential α-eHb1 expression, we demonstrated that flt3 was expressed in the WT and heterozygous cloche embryos (51 of 70; 72.86%) but not in the homozygous mutants (19 of 70; 27.14%) (Figure 1F-G; supplemental Figure 4C-J). At 24 hpf, flt3 was predominantly expressed in the cardiac mesoderm and pronephric ducts (supplemental Figure 4K). At 28 hpf, flt3 was expressed, albeit weakly, in the embryonic midline that included the region of the ventral wall of DA as shown in both WISH and paraffin sections (Figure 1H-K). At 32 hpf, flt3 was expressed in a small cluster of cells in the PBI (Figure 1L). At 48 hpf, flt3 expression was found near the developing heart and gut, which became more prominent at 72 hpf. On 4 dpf, its expression in the heart region disappeared and became restricted in the pharynx and intestine regions (supplemental Figure 4L-N).

Embryos were injected with MO (flt3MO) targeting the 5′-UTR of the flt3 gene. The embryos could tolerate up to 6 ng without severe toxicity and this dose was used throughout the study. To ascertain the molecular targeting, we injected a flt3UTR-egfp chimeric gene containing the flt3MO targeting site in the 5′ terminus (supplemental Figure 5) and demonstrated mosaic egfp expression (Figure 1M). The latter could be quenched by flt3MO (Figure 1N-O; flt3UTR-egfp, 53.03% ± 7.93%; flt3UTR-egfp + MO, 5.10% ± 0.92%, P = .008). The majority of flt3 morphants had no significant developmental defects up to 2 dpf (Figure 1P-Q). Less than 10% of embryos showed deformity with curved tail, pericardial edema, and developmental delay and were excluded from subsequent analysis. Among embryos with normal morphology, l-plastin (Figure 2A-B; supplemental Figure 6M) but not pu.1 (supplemental Figure 6I-J) expression in the anterior yolk sac region was significantly downregulated at 18 hpf. At 22 hpf, both csf1r and mpeg1 expression, which denote macrophage, were significantly reduced (supplemental Figure 6O-P). mpo expression was only modestly reduced (supplemental Figure 6N). These observations suggested that flt3 knockdown more specifically perturbed macrophage derivation during primitive myelopoiesis. Importantly, a proportion of primitive macrophages in the anterior yolk sac in the control embryos (injected with scramble sequence MO) underwent proliferation as shown by overlapping signals of egfp (indicative of mpeg1 expression in macrophages) and Alexa Fluor 594 (indicative of cellular proliferation with positive phospho-histone H3 [Ser10] immunostaining) (Figure 2D-F and supplemental Figure 7D-F, white arrow). In contrast, the primitive macrophages in flt3 morphant showed significantly reduced proliferative signals (Figure 2C,G-I; supplemental Figure 7H-J). Proliferative cells (nonmacrophage) were also identified in the heads and tail regions (supplemental Figure 7G,K). There was no change in apoptosis as shown by TUNEL (terminal deoxynucleotidyltransferase-mediated dUTP nick end labeling) staining in these primitive macrophages (supplemental Figure 7L-S). Therefore, these observations provided unequivocal evidence that flt3 knockdown inhibited proliferation of primitive macrophages during embryonic development. The effects were specific to myeloid development but not a general one on ALPM as scl and lmo2 expression in ALPM were unaffected (supplemental Figure 6A-D). These observations supported the proposition that flt3 signaling is important for the maintenance of primitive myelopoiesis after their myeloid commitment driven by pu.1. Genes associated with erythropoiesis (gata1, α-eHb1) in ICM were also unaffected (supplemental Figure 6E-H). Therefore, an effect of flt3 on cellular specification and myeloid-erythroid skewing was less likely.

The effect of flt3 knockdown on definitive hematopoiesis was also examined. c-myb expression was consistently downregulated in the ventral wall of DA at 36 hpf (Figure 2J-K) and the CHT at 72 (Figure 2L-N) and 96 hpf (Figure 2O-P). cd41:egfp⁺ cells in CHT of Tg(cd41:egfp) flt3 morphants were also significantly reduced at 72 hpf (Figure 2Q-S). Moreover, the expression of recombination activating gene 1 (Figure 2T-U, rag1) and T-cell–specific light-chain kinase (Figure 2V-W, lck) in the thymus were also downregulated. To confirm the specificity of these phenotypes, downregulation of c-myb and rag1 by flt3MO (percentage of embryos with normal gene expression in morphants: c-myb: 15.0% ± 3.6%; rag1: 13.2% ± 0.7%) could be rescued by coinjecting the embryos with WT flt3 mRNA (Figure 3A-B) (c-myb: 75.4% ± 7.4%; P = .008; rag1, 78.7% ± 3.2%; P = .002, N = 3 experiments).

To ascertain whether the effects of flt3 knockdown on definitive HSPCs could result from perturbation of angiogenesis and vasculogenesis,⁴²  these processes were examined in the flt3 morphants. In Tg(flk1:egfp) embryos, the intersegmental vessels (ISVs), the DA, and dorsal vein (DV) were intact at 24 hpf after flt3 knockdown (Figure 3C). The expression of flt4 (vein) and efnb2a (artery) was also unaffected (Figure 3D).

AC220 is a potent and relatively selective inhibitor of FLT3 and is currently being evaluated in clinical trials for the treatment of AML.^41,43  We tested whether AC220 could recapitulate the hematopoietic phenotypes of the flt3 morphants. AC220 (2.5 μmol/L) had no effect on embryo morphology but significantly reduced the expression of l-plastin in the anterior yolk sac at 18 hpf (Figure 4A-C) and c-myb in the ventral wall of DA at 32 (Figure 4D-E) and 96 hpf (Figure 4J-K) as well as rag1 in the thymus at 96 hpf (Figure 4L-M). Vasculogenesis, as shown by flt4 (Figure 4F-G, DV) and efnb2α (Figure 4H-I, DA), expression was intact.

We first cloned the human FLT3-ITD into an egfp plasmid, and FLT3-ITD and egfp were separated by a T2a self-cleaving peptide. The FLT3-ITD sequence was confirmed by DNA sequencing (supplemental Figure 8). The FLT3ITD-T2a-egfp plasmid was linearized and microinjected into 1-cell stage embryos (100 pg per embryo). FLT3ITD-T2a-egfp expressed in a mosaic pattern (supplemental Figure 9A-D) with evidence of egfp⁺ cells could be identified in the anterior yolk sac (supplemental Figure 9E-F) and the CHT (supplemental Figure 9G-H).

Notably, FLT3-ITD expression was associated with a significant expansion of the primitive myeloid cells that were characterized by the expression of pu.1, mpo, cebpα, and l-plastin at 18 (supplemental Figure 10A-H) and 36 hpf (Figure 5A-D). The responses based on WISH were defined arbitrarily as intermediate, in which increased gene expression was identified but no clustering was found, and severe, in which significant clustering of gene expression could be identified on the surface of yolk (supplemental Figures 11-12). FLT3-ITD expression based on the FLT3-ITD-T2a-mRFP plasmid also significantly increased egfp⁺ population in dissociated Tg(mpo:egfp) embryos at 36 hpf (Figure 5E-F) from a baseline of 0.40% ± 0.27% to 0.57% ± 0.29% (P = .038). The myeloid expansion appeared to be less than that in WISH, as the expression of FLT3-ITD-mRFP was less efficient than that of FLT-ITD-EGFP. Quantification of double mpo:egfp⁺ and mRFP populations was not performed due to the small populations of cells detected. To further examine the morphology of the myeloid cell expansion, egfp⁺ cells were isolated by fluorescence-activated cell sorter (FACS). In the latter experiments, mRNA (supplemental Figure 13A-B) was used instead of plasmid DNA as it gave rise to a slightly stronger myeloid expansion in embryos that would facilitate isolation of egfp⁺ cells (supplemental Figure 13C-D). The sorted egfp⁺ cells had a monocytic morphology, consistent with its myeloid nature (Figure 5G-J).

Experiments were also performed to examine whether the myeloid expansion resulted from an increased flt3 signaling in the embryos. Overexpression of human FLT3-ITD in embryos was also associated with an increase in p-flt3, p-stat5, p-erk1/2, and p-akt at 36 hpf (Figure 6A). In addition, colocalization of mpo and egfp (surrogate for FLT3-ITD) was evident (Figure 6B-E). Furthermore, the effects of AC220 treatment on myeloid responses to FLT3-ITD were evaluated. AC220 had no effect on baseline pu.1⁺, mpo⁺, and cebpα⁺ populations in uninjected embryos (Figure 7A-F). However, it significantly ameliorated the myeloid expansion induced by FLT3-ITD expression based on pu.1⁺, mpo⁺, and cebpα⁺ expression (Figure 7G-L,S-U, P < .01 in all cases, χ² test, N = 3 experiments). Collectively, these observations supported the proposition that human FLT3-ITD expression stimulated endogenous flt3 signaling in zebrafish and led to expansion of myeloid populations. The possibility of aberrant sequestration of these myeloid populations in the Duct of Cuvier was unlikely as the circulatory system was intact based on direct visualization (supplemental Video 1) and myeloid expansion was also evident at 23 hpf, before the onset of circulation (supplemental Figure 10I-P).

Single-base mutation of the FLT3-TKD occurs in ∼5% AML and its prognostic significance is unclear.²²  In zebrafish embryos, overexpression of FLT3-TKD (D835Y) likewise induced a significant expansion of pu.1 (Figure 7M), mpo (Figure 7O), and cebpα (Figure 7Q) despite a smaller phenotypic penetrance for mpo and cebpα. Importantly, AC220 had no significant ameliorating effects on myeloid expansion in the FLT3-TKD embryos (Figure 7N,P,R-U, P > .05 in all gene expression studies, χ² test).

Both clinical and laboratory data demonstrated that the leukemia clone carrying the additional TKD mutation might emerge in FLT3-ITD⁺ AML treated with TKI (the “double mutation”), leading to drug resistance.^26,27  We addressed this issue in zebrafish embryos by injecting them with plasmid encoding the FLT3-ITD-TKD (D835Y) double mutation. Overexpression of the double mutants induced expansion of myeloid compartment as shown by the increase in pu.1, mpo, cebpα, and l-plastin expression (Figure 8E-H). However, AC220 did not significantly reduce the expansion of the myeloid compartment (Figure 8I-L and M-P, P > .05 in all gene expression studies, χ² test), suggesting that the double mutation has conferred resistance to AC220 in zebrafish embryos.

In the present study, we demonstrated that flt3 was highly conserved at the JMD and TKD and its targeting by MO and AC220 specifically reduced primitive myelopoiesis and definitive HSPC specification. The effects of knockdown on primitive myelopoiesis were seen as early as 18 hpf, with a preferential downregulation of l-plastin but not pu.1 expression. csf1r and mpeg1 expression at 22 hpf, before the onset of circulating, were significantly reduced, suggesting flt3 may play an important role in macrophage development during primitive myelopoiesis. Proliferation of primitive macrophages was reduced in the flt3 morphants as shown by phospho-Histone H3 staining in mpeg1⁺ cells. Apoptosis was not affected. Primitive erythroid and precursor cells were not affected, supporting the proposition that flt3 knockdown perturbed specifically macrophage development rather than erythromyeloid lineage specification. The effects on c-myb and cd41 populations in the ventral wall of DA as well as the lck and rag-1 populations in the thymus might reflect perturbation of the hematopoietic progenitor population. The results were consistent with the expression of flt3 in the ventral wall of DA and PBI. As the specification of hematopoietic stem and progenitors depends on blood flow through the axial circulation,⁴²  we also evaluated the process of vasculogenesis in the morphant embryos. Vasculogenesis and arteriovenous specification as well as the blood flow in the vasculature of the morphants (supplemental Videos 2-5) were intact, suggesting that flt3 might specify definitive HSPC after its derivation from the putative hemogenic endothelium.

We have also evaluated the effects of FLT3 activation during embryonic development. Intriguingly, expression of human FLT3-ITD was able to activate downstream signaling in zebrafish embryos, highlighting the conserved signaling machinery in response to FLT3 activation in this model. Primitive myelopoiesis was accentuated, as shown by the increased populations of mpo, cebpα, l-plastin, and pu.1 positive cells. Microinjection of mRNA encoding for FLT3-ITD, which enabled earlier and more ubiquitous FLT3-ITD expression, only affected myeloid development but not primitive precursors and erythropoiesis (supplemental Figure 14). Therefore, FLT3-ITD overexpression might induce proliferation of primitive myeloid cells rather than misspecification or skewing of lineage differentiation. Colocalization of FLT3-ITD expression and mpo could be demonstrated, supporting the proposition that constitutive activation of FLT3 might be driving the expansion of the myeloid compartment. The ectopic myeloid expansion corroborated with murine studies in which Flt3 knockin induced myeloid expansion reminiscent of human myeloproliferative neoplasms⁴⁴  and AML with short latency in the presence of cooperative oncogenic signaling.^45,46  Importantly, coexisting TKD in the same allele conferred the myeloid compartment with the resistance to AC220, consistent with the clonal emergence of AC220-resistant AML patients.

The results of this study provided an example whereby human AML can be modeled in this organism. To demonstrate the clinical relevance of this model, we made use of AC220 that has been tested in clinical trials for the treatment of AML. Interestingly, the myeloid expansion induced by TKD mutant expression was resistant to AC220, consistent with the clinical emergence of resistant clones carrying additional TKD at D835 position. Off-target effects of multikinase inhibitors could be readily demonstrated, as exemplified by the perturbation on arterial specification upon sorafenib and sunitinib treatment secondary to disruption of vascular endothelial growth factor (VEGF) signaling (supplemental Figure 15).

We envisioned that the model might be modified to suit specific clinical needs so that novel and patient-specific therapeutic agents can be identified at a whole-organism level in real time. For instance, multiple FLT3-ITD clones that often coexist in AML samples can be isolated at different stages of treatment and expressed in zebrafish embryos and the inhibitory effects of TKI on different ITD clones can be evaluated. The efficacies of novel TKI and inhibitors of downstream signals including pSTAT5,²³  pAKT,²⁴  and PIM1⁴⁷  toward resistant FLT3-ITD clones could also be evaluated. In addition to FLT3-ITD, other mutant leukemia genes have been evaluated. For instance, the mutant human nucleophosmin 1 (NPM1c) gene, when expressed in zebrafish embryos, has been shown to induce myeloid expansion with the mutant protein resided exclusively in the cytoplasm, reminiscent of human NPM1c⁺ AML.⁴⁸  With the advent of next-generation sequencing and the emergence of new mutations in AML and related diseases,⁴⁹  the zebrafish may provide a timely model whereby the pathogenetic significance and therapeutic potential of novel gene mutations can be evaluated.

The authors thank Dr Leonard Zon for the helpful comments on the data and Dr Zilong Wen for the procurement of embryos.

This work was supported by the General Research Fund (HKU771110 and HKU729809M), the Innovative Collaborative Research Programme and Zebrafish Core Facility at the Li Ka Shing Faculty of Medicine, as well as National Institutes of Health (General Medical Sciences) grant GM63904 (S.C.E.).

A.Y.H.L. is the Li Shu Fan Medical Foundation Professor in Haematology and received funding from the endowment.

Contribution: B.-L.H. designed and performed the experiments, analyzed the data, and wrote and approved the manuscript; C.H.M., X.S., and H.C.H.C. performed the cloning, western blot, and microinjection experiments and approved the manuscript; A.C.H.M. and S.C.E. designed the experiments, analyzed the data, and approved the manuscript; C.W.E.S. cloned the human FLT3-ITD and approved the manuscript; W.W.L.C. performed the cell staining and evaluated cellular morphology and approved the manuscript; W.Z. and Y.Z. prepared the cloche mutant embryos and approved the manuscript; and A.Y.H.L. designed the experiments, analyzed the data, and wrote and approved the manuscript.

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

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