Pivotal role of Pten in the balance between proliferation and differentiation of hematopoietic stem cells in zebrafish

Understanding the development of blood stem cells has been a challenge for decades and still little is known about the ontogeny of hematopoietic stem/progenitor cells (HSPCs) that give rise to all blood lineages. All vertebrates possess 2 waves of hematopoiesis: the primitive wave and the definitive wave.^1-3  The primitive wave gives rise to primitive erythrocytes and macrophages, as well as megakaryocytes in mice⁴  and neutrophils in zebrafish,⁵  whereas the definitive wave or adult hematopoiesis generates all hematopoietic cell lineages. In mammals and zebrafish, HSPCs emerge from the ventral wall of the dorsal aorta in a conserved region known as the aorta-gonad-mesonephros (AGM).^6-8  After leaving the AGM, mammalian HSPCs transiently colonize the fetal liver before seeding the bone marrow. In zebrafish, HSPCs colonize the caudal hematopoietic tissue (CHT)⁹  before seeding the definitive hematopoietic organs, the thymus, and head kidney. In the hematopoietic organs, HSPCs commit to progenitors before maturing and giving rise to all blood cell populations.

HSPCs are tightly regulated in terms of dormancy and self-renewal, proliferation, and differentiation. Disrupting this balance leads to pathological consequences such as bone marrow failure or hematologic malignancy. For example, T-cell acute lymphoblastic leukemia (T-ALL) is an aggressive hematologic tumor arising from the malignant transformation of hematopoietic progenitors.¹⁰  To date, several genes and underlying pathways have been linked to T-ALL. One of these genes is the tumor suppressor, phosphatase and tensin homolog (PTEN). Deletion mutations in PTEN appear in 5% to 10% of T-ALL cases, and about 17% of patients lack PTEN expression.^10,11  Recent work using a conditional knockout mouse model demonstrated that loss of Pten in bone marrow HSPCs causes expansion of the short-term population and declining of the long-term population of HSPCs. Mice with Pten-deficient bone marrow HSPCs develop myeloproliferative disorder (MPD) eventually.^12-15 

PTEN is a phosphatase specific for the D3 position of the second messenger, phosphatidylinositol (3,4,5)-triphosphate, which is formed by phosphatidylinositol 3-kinase (PI3K). Thus, PTEN is one of the few known lipid phosphatases counteracting the PI3K-Akt (also known as protein kinase B [PKB]) pathway.¹⁵  Loss of Pten function is embryonic lethal in many organisms, including the mouse, Caenorhabditis elegans, and Drosophila.^16-18  The zebrafish genome encodes 2 pten genes with redundant function, designated ptena and ptenb.¹⁹  Single mutants display no morphologic phenotype and are viable and fertile, but mutants that retain only a single wild-type copy of pten develop hemangiosarcomas during adulthood.²⁰ Ptena^−/−ptenb^−/− mutants, hereafter referred to as Pten mutants, lack functional Ptena and Ptenb, are embryonic lethal at 5 to 6 days postfertilization (dpf), and display hyperplasia and dysplasia.¹⁹ 

We used zebrafish mutants lacking functional Pten to address hematopoiesis. Zebrafish embryos survive without circulating erythrocytes²¹  and hence provide an excellent system to investigate the early stages of hematopoiesis, which have not been addressed in other systems before. In particular, we investigated whether homing and proliferation of definitive HSPCs in the CHT, and their maturation into differentiated blood cells are Pten dependent. Here, we report that progenitors of all blood lineages are formed in the CHT in Pten mutants, but their differentiation into mature blood cells is arrested. Finally, we show that inhibition of PI3K signaling once the HSPCs have colonized the CHT allowed their differentiation into definitive blood cells.

Ptena^−/−ptenb^−/−19  and Tg(cd41:eGFP)²²  were maintained, crossed, raised, and staged as described.^23,24 

Fluorescence images of transgenic embryos were acquired using a TCS-SPE confocal microscope (Leica) and processed with Image J (http://rsb.info.nih.gov/ij/). Embryos were anesthetized with tricaine²³  and mounted on a glass cover dish with 0.7% low-melting agarose and covered with standard E3 medium. Whole-mount images from in situ hybridizations and Sudan Black (SB)-stained embryos were taken using a Zeiss Axioplan microscope connected to a Leica DFC480 camera.

Whole-mount in situ hybridization was performed according to standard protocols.^25,26  Embryos stained with rag1 were used to measure rag1⁺ area with Image J software. Thirty-two hours postfertilization (hpf) embryos were genotyped after in situ hybridization according to Faucherre et al.¹⁹ 

Immunohistochemistry was performed using anti–phosphorylated Histone 3 (pHis3) (Abcam)¹⁹  with slight modifications. Briefly, 4 dpf embryos were fixed overnight in 2% paraformaldehyde and subsequently washed with phosphate-buffered saline (PBS) including 0.3% Triton X-100. Permeabilization was performed by incubating embryos for 5 minutes on ice in 2.5 mg/mL trypsin solution (Worthington Biochemical Corporation). Subsequently, embryos were rinsed 3 times for 5 minutes with PBS including 0.3% Triton X-100. Embryos were blocked for several hours with 0.1% bovine serum albumin (BSA), 2% lamb serum, 0.1% dimethylsulfoxide (DMSO) and 1% Triton-X100 in PBS and incubated with polyclonal anti-pHis3 antibody2 (1:1000 in blocking buffer) overnight. Embryos were washed with 0.3% Triton X-100 in PBS several times and incubated in blocking buffer for several hours. Secondary antibody (Alexa 568, 1:1000) was applied overnight and embryos were washed 10 × 10 minutes with 0.3% Triton X-100 in PBS and processed for imaging.

Ptena^+/−ptenb^−/− with Tg(cd41:eGFP) were crossed, offspring were mounted at 4 dpf, and the entire CHT was imaged by confocal microscopy. Green fluorescent protein (GFP)^low- and GFP^high-expressing cells were quantified using Volocity software.

Five-dpf-old embryos were fixed and stained using SB as described.⁵  Images of SB-stained embryos were taken and cells were counted using Image J software.

Embryos were incubated with 4μM LY 294002 from 82 hpf onwards, fixed at 5 dpf, and processed for in situ hybridization or SB staining. SB⁺ cells were counted in the head, yolk sac, trunk, tail, and CHT using Image J software.

To assess hematopoiesis systematically, we first analyzed the primitive wave of hematopoiesis. In situ hybridization was performed at 32 hpf using key markers that are associated with early erythrocytes (gata-1) and cells of the primitive myeloid lineage (pu.1, l-plastin, and c-fms).^2,5,9,27-29  Expression pattern analysis for gata1, pu.1, l-plastin, and c-fms revealed no differences between siblings and Pten mutant embryos (supplemental Figure 1, available on the Blood website), indicating that initiation of the primitive wave was not affected in Pten mutants.

To investigate definitive hematopoiesis in Pten mutants, we first determined expression of c-myb and scl that mark HSPCs^9,30  at 5 dpf. We detected expression of both genes in the CHT (Figure 1A-D brackets), indicating that homing of HSPCs was not dependent on Pten function. However, Pten mutants displayed more c-myb⁺ and scl⁺ cells, compared with wild-type embryos at 5 dpf (Figure 1A-D). We wondered whether the loss of function of Pten was associated with an elevated number of HSPCs in the CHT. To investigate this, we used the Tg(cd41:eGFP) line in which HSPCs express a low level of GFP (GFP^low) and thrombocytes a high level of GFP (GFP^high).³¹  The number of GFP^low+ HSPCs in the CHT was significantly increased in Pten mutants at 4 dpf, compared with siblings (Figure 1E-G arrowheads). Next, we investigated whether the elevated number of HSPCs was due to enhanced proliferation. Immunohistochemistry using pHis3-specific antibodies in the Tg(cd41:eGFP) line revealed an increased number of mitotic GFP^low+ HSPCs in Pten mutant embryos compared with siblings at 4 dpf (Figure 1H-M). Predominantly GFP^low+ HSPCs stained positive for pHis3 in Pten mutant embryos (Figure 1I,M). Our data demonstrate that Pten is not required for HSPCs to colonize the CHT, but loss of Pten led to enhanced proliferation of HSPCs.

What are the consequences of lack of Pten function for the various definitive blood lineages generated from the HSPCs? We first investigated the lymphoid lineage. Pten mutants contained less rag1⁺ cells in the nascent thymus, compared with siblings (Figure 2A-D). Quantification of the rag1⁺ area in wild-type and Pten mutant embryos indicated a significant reduction in the rag1⁺ area in Pten mutants (Figure 2E). To assess whether reduced expression of rag1 was a consequence of reduced numbers of lymphoid progenitors, ikaros expression was evaluated. Surprisingly, ikaros expression was detected in the mutant thymus at the same level as in wild-type (Figure 2F-G inset). In addition, ectopic ikaros expression was detected in Pten mutants (Figure 2G) and ikaros expression was strongly enhanced in the CHT of Pten mutants (Figure 2H-I). The reduced number of rag1⁺ thymocytes and elevated number of ikaros⁺ lymphoid progenitors indicate that in Pten mutants, most lymphoid progenitors are arrested at a specific stage between ikaros and rag1 expression.

The Tg(cd41:eGFP) line revealed that the number of GFP^high thrombocytes in the CHT was significantly reduced by 4 dpf in the CHT (Figure 1E-F), suggesting that thrombocytic differentiation was impaired in Pten mutants.

To assess erythropoiesis, we used in situ hybridization and found elevated numbers of globin⁺ and gata1⁺ early erythroid progenitors in the CHT of Pten mutants (Figure 3A-D). Further differentiation of these early erythroid progenitors could not be addressed, for in fish, erythroid differentiation normally occurs very gradually over several days within the circulation.^21,32  The death of PTEN mutants by 5 to 6 dpf¹⁹  only allows us to detect the definitive committed (gata1⁺, globin⁺) early erythroid progenitors, that are mostly not yet circulating by that time and hence are located in the CHT.³³ 

Next, expression of myeloid cell markers was assessed. Clearly more l-plastin⁺, mpo⁺ (neutrophil lineage), and c-fms⁺ (macrophage lineage) cells dispersed in peripheral tissues were detected in Pten mutant than in wild-type embryos at 5 dpf (Figure 3E-J). SB stains the granules of neutrophils⁵  and was used to quantify mature neutrophils in Pten mutant embryos and siblings. SB⁺ cells in the head and yolk sac region are predominantly derived from the primitive wave, whereas SB⁺ cells in the trunk, tail, and CHT are mainly definitive wave-derived.⁵  The number of SB⁺ neutrophils was counted in the different areas of 6 wild-type and 7 Pten mutant embryos as illustrated in supplemental Figure 2. The number of SB⁺ cells from the primitive wave was enhanced more than twofold in Pten mutants (Figure 3K,L,O), whereas the number of SB⁺ cells from the definitive wave was reduced more than twofold (Figure 3M-O). The elevated number of mpo⁺ cells in the CHT, combined with reduced number of SB⁺ cells in the trunk, tail, and CHT, suggests an arrest in differentiation of neutrophils in the absence of Pten function. To verify whether cell death might cause the lack of differentiated cells, we performed acridine orange staining. Pten mutant embryos did not exhibit enhanced apoptosis in the CHT (supplemental Figure 3), suggesting that the absence of differentiated cells is rather due to perturbed maturation. These results demonstrate that in the absence of Pten all blood lineages are specified, yet differentiation is arrested, suggesting a requirement for Pten in terminal blood cell differentiation.

Our data suggest that proliferation of definitive blood cell progenitors was enhanced, while concomitantly differentiation was arrested in Pten mutant embryos. To address directly whether the lack of Pten function was causative, we treated Pten mutant embryos with the PI3K inhibitor, LY294002, from 82 hpf onwards, that is, after the HSPCs had colonized the CHT. Pten mutant embryos displayed enhanced numbers of HSPCs at 5 dpf, compared with siblings (Figure 4A), consistent with our data at 4 dpf (Figure 1E-G). Treatment with 4µM LY294002 from 82 hpf onwards significantly decreased the number of CD41-GFP^low cells in Pten mutants and siblings as well (Figure 4A). Expression of the progenitor marker, c-myb, which was enhanced in the Pten mutants at 5 dpf, was also diminished by treatment with LY294002 from 82 hpf (Figure 4B-E) in both mutant and wild-type embryos, consistent with the CD41-GFP^low data. These data suggest that inhibition of PI3K signaling after colonization of the CHT suppressed proliferation of HSPCs.

Next, we addressed whether the blood cell lineage commitment was affected by treatment with LY294002 from 82 hpf onward. Expression of gata-1, l-plastin, and ikaros were determined for the erythroid, myeloid, and lymphoid lineages, respectively. While 5 dpf Pten mutant embryos exhibited elevated expression of gata-1, l-plastin, and ikaros in the CHT compared with wild-type embryos (Figure 4F-Q), treatment with LY294002 from 82 hpf onward suppressed this upregulation of gata-1, l-plastin, and ikaros in Pten mutant embryos to levels that were comparable to wild type (Figure 4F-Q). These results indicate that LY294002-induced inhibition of elevated PI3K activity in Pten mutants leads to suppression of the enhanced numbers of blood cell progenitors committed to the erythroid, myeloid, and lymphoid lineages.

Finally, we addressed whether arrested differentiation in Pten mutant embryos is caused by elevated PI3K signaling. Rag1 expression that is greatly reduced in Pten mutant embryos at 5 dpf (Figure 2A-D) was fully restored after treatment with LY294002 (Figure 4R-U). Furthermore, quantification of SB⁺ mature neutrophils at 5 dpf following late LY294002 treatment demonstrated a dramatic increase in the number of definitive wave-derived mature neutrophils in the CHT, trunk, and tail of Pten mutants, but not siblings (Figure 4V). In contrast, LY294002 treatment did not significantly affect the number of primitive wave-derived SB⁺ cells in the head and yolk sac of Pten mutants and siblings (Figure 4V). Taken together, these data indicate that inhibition of PI3K signaling in Pten mutant embryos suppressed the enhanced proliferation of HSPCs in the CHT and at the same time released the blood cell differentiation arrest.

PTEN is a lipid phosphatase that counteracts PI3K activity and is one of the most frequently mutated tumor suppressor genes in a wide range of cancer types, including leukemia.^12-15,34,35  PI3K signaling regulates hematopoietic stem cell (HSC) function. For instance, deletion of the downstream factors, Akt1 and Akt2, increases HSC quiescence,³⁶  whereas activation of PI3K signaling promotes HSC proliferation and depletion.^12,13  In the mouse embryo, HSPCs emerge from the dorsal aorta at embryonic day 10.5 (E10.5) and transiently colonize the fetal liver before seeding the organs of adult hematopoiesis.^8,37-39  Similarly, zebrafish HSPCs emerge from the ventral wall of the dorsal aorta and transiently colonize the CHT, which is homologous to the fetal liver in mouse, before colonization of the adult organs of hematopoiesis.⁹  We used zebrafish mutant embryos lacking functional Pten to investigate whether loss of Pten affects homing of HSPCs to the CHT and further we addressed whether Pten is required for blood lineage specification and differentiation.

Our data revealed that Pten is not required for colonization of the CHT by HSPCs. Surprisingly, quantification of CD41^low-positive HSPCs and analysis of c-myb and scl expression patterns revealed an increased number of HSPCs in the CHT during the course of development. PTEN is a tumor suppressor and loss of PTEN function is associated with enhanced cell proliferation. Reminiscent of studies in the mouse,^12,13  we found increased numbers of mitotic cells in the CHT and conclude that in the absence of Pten, HSPCs colonize the niche and hyperproliferate.

PTEN is frequently mutated in leukemia resulting in replacement of normal bone marrow cells with leukemic cells. Acute myeloid leukemia patients suffer from a drop in erythrocytes, platelets, and normal white blood cells. Our investigation revealed that HSPCs lacking functional Pten engage in all blood lineages, indicating that lineage specification is a Pten-independent process. However, definitive differentiation of the thromboid, myeloid, lymphoid, and possibly other blood lineages is arrested in Pten mutants. We observed elevated numbers of mitotic HSPCs in the CHT of Pten mutant embryos, but no obvious differences in apoptosis between wild-type and Pten mutant embryos, indicating that enhanced proliferation caused the observed elevated number of HSPCs in mutant embryos. It is well established that proliferation and differentiation of stem cells, including HSCs, are inversely correlated.⁴⁰  Our results are consistent with this notion. Yet, the potential of HSPCs to commit to distinct blood lineages appears not to be affected in Pten mutants, in that we observed progenitors of all the main lineages. Moreover, inhibition of PI3K after the HSPCs had colonized the CHT, released the arrest of differentiation, and resulted in the production of definitively differentiated blood cells, including SB⁺ neutrophils and rag1⁺ thymocytes, indicating that these cells can produce mature blood cells.

The role of Pten in mouse HSCs has been addressed by conditional knockout of Pten using the Mx-1 promoter driving Cre recombinase in adult bone marrow cells, respectively, the VE-cadherin promoter in fetal liver.^12,13,41  HSCs are driven into the cell cycle in the absence of Pten, resulting in mobilization of HSCs from the bone marrow or fetal liver and transient expansion of the spleen, leading to depletion of HSCs. These conditional Pten-deficient mice die of a MPD that resembles acute myeloid/lymphoid leukemia.^12,13,41  Our observations that HSPCs hyperproliferate in the CHT of Pten mutant zebrafish embryos are consistent with the expansion of bone marrow HSCs in conditional mouse models.

Here, we report that zebrafish mutant embryos lacking functional Pten develop multiple hematopoietic abnormalities during development. In particular, we show that whereas Pten is not required for HSPCs to colonize the CHT/niche, loss of Pten results in enhanced proliferation of HSPCs. Additionally, enhanced PI3K signaling in response to loss of Pten does not favor a certain blood lineage in that all progenitors are defined. However, differentiation into mature blood cells is impaired in Pten mutants, which is reversed by antagonizing PI3K signaling by treatment with LY294002. We conclude that the zebrafish Pten mutants are a powerful tool to further study hematopoietic abnormalities/hematologic malignancies and demonstrate that zebrafish mutants are a suitable model for drug screening.

The authors thank Mark Reijnen and the animal caretakers for excellent management of the fish facility. Microscopy was done at the Hubrecht Imaging Centre. The authors also thank Jeroen Paardekooper Overman for technical support.

This work was supported in part by a European Union (EU) (FP7) grant, ZF-CANCER (HEALTH-F2-2008-201439).

Contribution: S.C. and J.d.H. designed experiments with input from K.K. and P.H.; S.C. and R.K. performed experiments; and S.C. and J.d.H. wrote the manuscript with input from P.H. and K.K.

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

The current affiliation for S.C. and K.K. is Institut National de la Santé et de la Recherche Médicale, Unités Mixtes de Recherche 5235, Dynamique des Interactions Membranaires Normales et Pathologiques, Université Montpellier 2, Montpellier, France.

Correspondence: Jeroen den Hertog, Hubrecht Institute, Uppsalalaan 8, 3584 CT, Utrecht, The Netherlands; e-mail: j.denhertog@hubrecht.eu.