Regulation of Hematopoietic Stem Cell Migration and Function

A dissertation presented by Ellen Marie Durand

to The Division of Medical Sciences in partial fulfillment of the requirements for the degree of Doctor of Philosophy in the subject of Human Biology and Translational Medicine

Harvard University Cambridge, Massachusetts April 2014

! 2014 – Ellen Marie Durand All rights reserved.

Dissertation Advisor: Dr. Leonard Zon

Ellen M. Durand

Regulation of Hematopoietic Stem Cell Migration and Function Abstract

Hematopoietic stem cell transplantation (HSCT) is an effective treatment for blood disorders and autoimmune diseases. Following HSCT, these cells must successfully migrate to the marrow niche and replenish the blood system of the recipient. This process requires both non-cell and cell-autonomous regulation of hematopoietic stem and progenitor cells (HSPCs). A transgenic reporter line in zebrafish allowed the investigation of factors that regulate HSPC migration and function. To directly observe cells in their endogenous microenvironment, confocal live imaging was used to track runx1:GFP+ HSPCs as they arrive and lodge in the niche. A novel cellular interaction was observed that involves triggered remodeling of perivascular endothelial cells during niche formation. A chemical screen identified the TGF-beta pathway as a regulator of HSPC and niche interactions. Chemical manipulation of HSPCs was used to improve engraftment and repopulation capability following transplantation. Runx1:GFP fish treated with prostaglandin E2 (PGE2) during embryogenesis exhibit increased runx1+ cells in the AGM and CHT, consistent with previous in situ data. This increase in HSPCs is maintained into adulthood, even in the absence of prolonged PGE2 exposure. Kidney marrow from these treated fish can outcompete control marrow in transplantation assays. The ability of PGE2 to confer a long-term advantage on sorted mouse marrow populations in competitive transplantation assays was tested. I found that PGE2-treated

iii

short-term (ST)-HSCs, but not long-term (LT)-HSCs show enhanced transplantability in recipients compared to control animals. My studies demonstrate that the effects of PGE2 on HSC function persist over substantial time despite transient exposure. A population of short-term HSCs can engraft and give rise to long-term multilineage reconstitution following PGE2 treatment. Collectively, our studies have led to novel insights regarding the pathways involved in HSC migration, homing, and repopulation.

iv

Table of Contents

Abstract……………………………………………..………………………………….iii Acknowledgements……………………………………………...………………….....vi Chapter 1: Introduction……………….………………………………………………1 Chapter 2: Live imaging of hematopoietic stem cell lodgement reveals dynamic endothelial niche remodeling……………….………..22 Chapter 3: Prostaglandin E2 regulates long-term repopulation activity of ST-HSCs……………….…………………………..…………60 Chapter 4: Discussion and future directions..…….……………………………..….86 References……………………………………………………………………………107 Appendix……………………………………………………………………………..115 Supplementary figures………………………………………………………116 ProstaglandinE2 review article………………………………….………….144 Zebraflash bioluminescence imaging………………………………………149 The blood balance…………………………………………………………...159 Lineage Regulators………………………………………………………….161

v

Acknowledgements I thought my acknowledgements would be the easiest part of thesis writing. Instead I found myself avoiding them until the last possible second, literally. Not because I don’t have anyone to thank, but quite the opposite. This thesis would not have been possible without many people, and writing this section makes me realize how soon I will be moving on. Joining the Zon lab one of the first, and best, decisions I made in grad school. I started rotating in my 3rd week and by week 4 I knew I would be joining. Len is a great mentor- he is invested in his lab members’ scientific growth AND their happiness. The Zon lab is simply a fun place to be year-round: March madness season turns to fantasy baseball season turns to Halloween planning season turns to Halloween season turns to fantasy football season. Repeat. I will be tremendously sad to leave the lab, but most of all I will miss the friends I have made. My first bay on the 8th floor was like a mini family; Jared and I were the clumsy, newest lab babies and our post doc parents Katie and Owen had to make sure we didn’t break anything. I learned and continue to learn so much from all of them, and I am so thankful that I got to spend my first year in the lab as their baymate. After moving down to the coveted 7th floor, I was lucky enough to join the carebear bay…need I say more? Katie, Alison, and Colleen made every day at work more fun. Filmwise, sporcle, and “totally looks like” were routine ways to pass the time. I am so lucky to have remained friends with these girls. When Alison and Colleen graduated, Vera and Ellen (what are the odds?!) joined the bay. I am always thankful for their suggestions and support.

vi

I was fortunate to have fantastic peer mentors in the lab as well. I first rotated with Pulin, who continued to help me in many aspects of my project once I joined the lab. Teresa probably answered more of my questions than anyone. Besides being ridiculously knowledgeable about blood, she is one of the coolest people I know. I want to also thank the technicians I’ve worked with, Margot and Becca, for preventing me from messing up my own experiments from time to time. Finally, I need to thank the entire lab, especially my fellow grad students. You all have made this place feel like a second home for the past 5 years. To my family, I cannot give you all the thanks on paper that you deserve. You have shown me constant support and love throughout my entire life. I can’t remember a time in the past few months where I haven’t heard my parents ask: “What can we do for you? What do you need to make your life easier?” I don’t think I could have made it through grad school without you, I love you all so much and I dedicate this thesis to you.

vii

Chapter 1

Introduction

Attributions I wrote a review article entitled “Newly emerging roles for prostaglandin E2 regulation of hematopoiesis and hematopoietic stem cell engraftment”1 (Durand and Zon, Curr Op in Hematol. 2010 17:308–312). Excerpts from this review are updated and included in the introduction chapter. The review in its entirety is included in the appendix.

2

Review Hematopoietic stem cells (HSCs) are a rare cell type with the ability to self renew and differentiate. In addition to maintaining hematopoietic homeostasis, HSCs must respond to stress conditions and fluctuate between a quiescent and an actively cycling state. Following HSC transplantation, these cells must successfully migrate to the marrow niche and replenish the blood system of the recipient. Although much is known about the importance and clinical relevance of HSCs, very little is known regarding the mechanisms of HSC function. Utilizing the zebrafish as a model system has allowed us to study novel pathways, identified through chemical screening, involved in HSC migration, homing, and repopulation.

Hematopoietic Stem Cells HSCs are a rare cell population with the ability to both self-renew and differentiate into all mature blood cells via a life-long process called hematopoiesis1,2,3. Vertebrate hematopoiesis occurs in two waves, the primitive wave and the definitive wave. During the primitive wave, erythrocytes and macrophages are produced to help oxygenate and remodel the developing embryo. The definitive wave is marked by the emergence of HSCs, which will support the hematopoietic system for the life of the organism. In the mouse, the first long-term HSCs are detected in the aorta–gonad– mesonephros (AGM) region by embryonic day 10.5 and later colonize the fetal liver, thymus, spleen, and bone marrow4. Key transcription factors that play a role in the development of the hematopoietic system include runx1, scl, and c-myb. The bone marrow is composed of a diverse set of cell types, including osteoblasts,

3

endothelial cells, stromal cells, adipocytes, and mature blood cells, which form a regulatory environment called the niche5. HSCs must respond to regulatory signals in the marrow niche and either maintain the HSC pool through self-renewal, or produce mature blood cells by dividing into progenitor cells. Through this cellular hierarchy, HSCs are able to maintain hematopoietic homeostasis throughout the lifespan of an organism.

Zebrafish Hematopoiesis Despite the extensive evolutionary divergence between bony fish and mammals, the teleost Danio rerio, more commonly called the zebrafish, remains a robust model for studying hematopoiesis. Genetic pathways regulating hematopoiesis have been conserved from zebrafish to mammals, including two distinct waves of hematopoiesis, outlined in figure 16. Zebrafish undergo primitive hematopoiesis in an intraembryonic region called the intermediate cell mass (ICM), compared with the extraembryonic yolk sac in mammals7. The primitive wave produces lineage-committed erythroid and myeloid progenitor cells, which express specific transcription factors such as gata1 and pu.18. Zebrafish definitive hematopoiesis occurs in the ventral wall of the dorsal aorta, the aorta-gonad-mesonephros (AGM) equivalent region in the zebrafish. Definitive hematopoiesis produces all lineages of blood cells and functional HSCs, which express transcription factors such as runx1, cmyb, CD41, and lmo29. The zebrafish equivalent of mammalian bone marrow is the kidney, which supports definitive hematopoiesis from 4 days post fertilization (dpf) through adulthood. Whole kidney marrow (WKM) transplanted into lethally irradiated recipient fish has been shown to rescue recipient fish and repopulate all blood lineages.

4

Figure 1. Spatial and temporal representation of hematopoiesis in the zebrafish The primitive wave of hematopoiesis is depicted in yellow and the definitive wave is depicted in blue. The locations of different sites of hematopoiesis are depicted in a 24hpf embryo (left) and a 72hpf embryo (right). The timline for expression of blood-specific markers at different sites of hematopoiesis is shown in the bottom panel.

FACS analysis of whole kidney marrow can be used to analyze the hematopoietic system and differentiate cell populations10. A member of our lab has successfully developed a competitive transplant assay system in which the homing and engraftment of HSCs into irradiated recipient fish can be assessed using fluorescence microscopy and FACS analysis (Figure 2). Using a transparent fish called Casper, a double mutant in pigment genes nacre and roy, allows for easy visualization of the transplanted cell populations using fluorescence microscopy11. This competitive transplant assay allows

5

for analysis of different HSC conditions and their effect on engraftment efficiency during WKM transplant. Aside from genetic similarity, the zebrafish has other advantages for studying hematopoiesis. Unlike mammals, zebrafish red blood cells are nucleated, making it possible to utilize them in certain experiments such as ChIP and DNase hypersensitivity mapping. Zebrafish embryos are also transparent, which makes it possible to view blood circulation with a dissecting microscope12. In addition, a pair of zebrafish can produce 100-200 embryos per week, making them a useful system for large-scale chemical and genetic screens.

Figure 2. Competitive transplantation and readout schematic The diagram on the left is a schematic of competitive transplantation in adult zebrafish. Marrow cells are harvested from GFP+ donor fish and treated with chemical or vehicle. 20,000 treated WKM cells are mixed with 80,000 untreated mCherry+ cells and injected into irradiated recipient fish. Several months following transplantation, FACS analysis is performed on WKM of recipient fish. The graph on the right is a representative plot that shows the major blood lineages detectable by FACS.

6

Hematopoietic Stem Cell Niche The bone marrow niche for HSCs is a complex microenvironment that regulates hematopoietic stem and progenitor cell (HSPC) function and supports both self-renewal and differentiation processes13. Many interacting cell types and signaling networks exist in the niche environment, which have a mixture of direct and indirect effects on HSCs14. The niche protects HSCs from cellular stress, particularly by limiting the accumulation of reactive oxygen species. In addition, it is thought that osteoblastic cells help maintain HSCs in a quiescent state, which is necessary to prevent excessive proliferation and exhaustion, although this effect may not be direct15. Within the bone marrow, a perivascular niche comprised of a network of sinusoidal vessels provides a connection between circulating cells and the marrow niche. It is common for HSCs to localize to sinusoids, suggesting that the cell types that surround these vessels play a role in HSC maintenance16. For example, mesenchymal stem cells (MSCs) are found proximal to niche blood vessels and express fibroblast activation protein (FAP). Knock down of FAP+ cells results in anemia and hypocellularity in the bone marrow17–19. Endothelial cells in the perivascular niche have also been shown to regulate HSC maintenance both in vitro and in vivo, however it is unknown whether this effect is direct or indirect20,21. Although the bone marrow niche is a highly dynamic structure, comprised of numerous cell types and signaling networks and closely linked to circulation, our current methods of studying this environment are lacking. Experimental protocols looking at fixed sections and sorted cells lack information about the dynamic cellular interactions occurring the in the niche space. In addition, our knowledge about HSC migration to and subsequent colonization of the niche remains largely unknown. Understanding these

7

trafficking and engraftment mechanisms is critically important because they are mimicked during clinical stem cell mobilization and transplantation.

Hematopoietic Stem Cell Transplantation HSCs are arguably the most studied type of stem cell, and transplantation of HSCs has been used in the clinic for several decades as an effective therapy for leukemia, lymphoma, blood disorders, and autoimmune diseases1,2. Hematopoietic stem cell transplantation (HSCT) is achieved through transplantation of bone marrow, mobilized peripheral blood, or umbilical cord blood-derived HSPCs. Preconditioning for HSCT entails ablation of the recipients’ bone marrow, leaving patients at risk for infection, anemia, and thrombocytopenia. Transplanted HSCs must efficiently home to the niche, differentiate, and repopulate the host hematopoietic system. Recent studies have focused on the use of chemicals to enhance collection and transplantation efficiency of HSCs and HSPCs in order to shorten engraftment time and diminish risk for myeloablated patients22,23. This effort includes improving HSC expansion prior to transplantation, as well as improving the efficacy of HSPC homing and niche interaction following transplantation. One way to test for successful engraftment in patients is to measure neutrophil recovery. Improving this parameter by even a few days could have a dramatic effect on patient outcome.

Phenotypic Characterization of HSCs Blood cells are an extremely heterogeneous population comprised of stem cells with varying degrees of repopulation activity, multipotent progenitor cells, and fully

8

differentiated cells. For decades we have been able to study the functionality of these cells using elegant in vivo repopulation techniques24. The gold standard to identify a bona fide HSC is to perform bone marrow transplantation and observe multilineage reconstitution for at least 6-12 months following transplant. However, true HSCs are very rare amid all blood cells. In mice only about 1 in 105 bone marrow cells is a longterm reconstituting stem cell; in humans this number is closer to 1 in 106 cells25. In the past 25 years or so, a significant number of antibody combinations have been developed for both mouse and human to distinguish between stem and progenitor cells, and differentiated lineages. Murine HSCs were initially identified as lineage-, Sca1+, c-Kit+ (LSK) cells in the bone marrow26. The Nakauchi lab demonstrated through single cell transplantation experiments that LSKCD34- cells give rise to multilineage reconstitution in 40% of recipient mice, demonstrating a tremendous improvement in the purification of HSCs27. Similarly, Sean Morrison’s lab identified the signaling lymphocyte activation molecule (SLAM) family as a reliable and robust method for isolating HSCs. SLAM family receptors mark a variety of immune cells and are differentially expressed in HSPC populations28. They demonstrated 47% of LSKCD48-CD150+ cells can give rise to longterm multilineage reconstitution in recipient mice. Alternatively, progenitor populations have been observed that are LSKCD34+Flk2+, or LSKCD48+CD150-, which remain multipotent, but have only transient reconstitution capabilities29. In contrast to surface marker characterizations in mouse, almost all human HSCs reside in the CD34+ fraction. Human HSCs are most commonly defined by a CD34+Thy1+CD38-CD45RA- phenotype30–32. However, these markers alone are not enough to exclusively distinguish cells with long-term repopulation capacity from those

9

without. Sophisticated marker combinations allow us to study the functionality of more purified populations, as well as the molecular factors that underlie these differences. This is especially useful when studying the heterogeneity of human blood cells, and could have important clinical implications for patients receiving HSCT.

Regulation of HSC Homing and Engraftment The cytokine stromal-derived factor 1 (SDF-1) is a strong chemo-attractant for CD34+ cells, which express its receptor, CXCR4. The migration potential of HSPCs towards a gradient of SDF-1 has been demonstrated both in vitro and in vivo. In the bone marrow, SDF-1 –expressing stromal cells keep HSPCs in the niche. Mobilized HSPCs have decreased expression of CXCR4, making it an attractive target to manipulate both migration and homing33. During preconditioning for HSCT, SDF-1 levels in bone marrow are increased, allowing for enhanced trafficking of HSCs to the niche. It has recently been reported that complement is also activated upon myeloablation, and that a specific protein component fragment, C3a, increases the responsiveness of HSCs to SDF134. Mice treated with the small molecule SB290157, a C3a receptor (C3aR) antagonist, displayed enhanced mobilization following granulocyte colony signaling factor (G-CSF) treatment. In addition, these cells exhibit impaired engraftment35. Interestingly, this engraftment defect is blocked when C3aR-deficient cells are primed with C3a prior to transplantation. Following transplantation, cells must home to the bone marrow niche to effectively repopulate the recipient blood system. Improving homing and engraftment efficiency is crucial for the recovery of myeloablated patients. As mentioned previously,

10

HSCs expressing CXCR4 will migrate towards a gradient of SDF-1 present in bone marrow, whereas mobilized HSCs exhibit reduced CXCR4 expression. This suggests that maintaining a functional interaction between CXCR4 and SDF-1 is essential for proper homing. CD26/DPP4 is a membrane-bound extracellular peptidase found on a subpopulation of CD34+ cells. It is known to cleave SDF-1 at the N-terminus and prevent its interaction with CXCR4, resulting in decreased migration of these cells to the niche36,37. Treatment of CD34+ cells with Diprotin A, a small molecule inhibitor of CD26/DPP4, greatly enhances homing and engraftment following transplantation. Similar effects were observed in CD26/DPP4- deficient cells. This effect is dependent on CXCR4, as co-treatment with AMD3100 blocks the effect of Diprotin A treatment on homing38.

Umbilical Cord Blood Transplantation Trafficking of HSCs from the bone marrow to peripheral blood, termed mobilization, is an important element of maintaining hematopoietic homeostasis. In addition, mobilized HSCs are generally the preferred source of stem cells for autologous and allogeneic-donor transplantation. However, the frequency of steady-state HSCs present in peripheral blood at any time during homeostasis is not high enough for successful transplantation and engraftment39. Therefore, prior to transplantation, autologous patient and allogeneic donor HSCs are mobilized using chemotherapy or cytokines. Efficient mobilization of HSCs is essential in order to yield enough stem cells for HSCT. G-CSF is commonly used to induce mobilization of HSCs to peripheral blood prior to collection. There are several disadvantages to G-CSF, including a lengthy

11

administration protocol, an unpredictable variable response, and rare but serious side effects40. Umbilical cord blood (UCB) has emerged as a promising source of transplantable HSCs for patients without a related or immune-matched donor. Although the stringency for HLA-matching is reduced, there are insufficient numbers of HSCs in a single UCB unit to engraft an adult patient in the United States. To compensate, two UCB units are given as the standard of care during transplantation (Figure 3)41,42. The ability to expand HSCs in culture prior to transplantation has important clinical applications. Despite a growing body of research on the mechanisms governing HSC self-renewal, previous attempts at culturing HSCs in vitro while maintaining their stem cell properties have failed to generate large numbers in culture. Expansion of UCB HSCs prior to transplantation could potentially reduce the number of UCB units necessary for successful engraftment. Utilizing small molecules to expand HSCs in vitro is particularly appealing as chemicals can be removed before cells are introduced to patients.

Prostaglandin E2 A chemical screen for HSC formation in zebrafish A small molecule screen in zebrafish identified 16,16-dimethyl-Prostaglandin E2 (PGE2) as a mediator of HSC formation during embryonic hematopoiesis43. Zebrafish embryos treated with PGE2 during embryogenesis displayed increased expression of HSC markers runx1 and cmyb in the dorsal aorta. Expressions of these markers were used as a proxy for HSCs as both are expressed in all HSCs and are required for proper HSC generation in the AGM region44,45. Endogenous requirement for PGE2 in HSC formation

12

Figure 3. Review of hematopoietic stem cell transplantation HSCT is achieved through transplantation of cells derived from the patient, as in autologous transplant. Alternatively, a healthy donor marrow or a cord blood unit cells can be used in allogeneic transplant, During cord blood transplantation, two cords are required for an adult patient. HSCs are usually mobilized or eliminated prior to transplant. The transplanted cells must successfully home to and engraft in the niche following transplantation.

was confirmed using morpholino knock out of the PGE2 biosynthetic enzymes Cox-1 and Cox-2, resulting in decreased HSCs. Irradiation recovery experiments were used as an injury model to demonstrate HSC activation. Treatment with PGE2 enhanced the rate of kidney marrow recovery in irradiated recipients compared with control treated animals, suggesting PGE2 treatment as a potential therapy to improve the recovery time of myeloablated patients.

13

Conservation in mammals The effects of PGE2 on HSCs were tested using murine transplantation assays. Bone marrow cells were treated ex vivo with PGE2 and transplanted into irradiated recipients to assess either short-term progenitor potential or long-term-HSC function. Colony forming units (CFU-Spleen) assays demonstrated that PGE2 enhances HSPC activity. Competitive repopulation with limiting dilution analysis showed that ex vivo PGE2 treatment increased the number of repopulating HSCs without disrupting differentiation. A study by Hoggatt et al.46 confirmed enhanced murine HSC engraftment following exposure to PGE2, observing a four-fold increase in HSCs up to 20 weeks following transplantation. Whole bone marrow (WBM) harvested from primary recipients at 20 weeks post transplant maintained PGE2-induced enhancement of HSC frequency in secondary recipients without additional PGE2 treatment. These findings suggest that short-term PGE2 exposure results in a long-term repopulation advantage of transplanted WBM. The effects of PGE2 on HSCs could be the result of an increase in HSC number, homing capability, proliferation, survival, or a combination thereof. To evaluate the homing capabilities of PGE2-treated WBM, labeled cells were transplanted into irradiated recipients. A significant increase in homing of PGE2-treated LSK cells to recipient marrow was observed compared with control. Consistent with the murine LSK population, PGE2-treated UCB cells displayed enhanced homing to the marrow. This homing effect is partially attributed to an increase in CXCR4 expression following PGE2 treatment. PGE2 treatment of murine LSK and human CD34+ cells increased expression

14

of CXCR4. In addition to its affect on homing, PGE2 increased Survivin, an important regulator of HSC survival and proliferation, in LSK and UCB cells following pulse treatment. Further supporting a role in HSC proliferation, SLAM LSK cells treated with PGE2 displayed a significant increase in the percentage of cycling cells. Until recently there existed no data for in vivo treatment of PGE2 in the murine system. Frisch et al.47 utilized a continuous in vivo treatment regimen to study the effects of PGE2 in mice. Consistent with previous in vitro data, it was observed that in vivo dmPGE2 treatment significantly increased the LSK population without disrupting hematopoietic differentiation. To determine which subpopulation(s) in LSK cells are directly affected by PGE2, SLAM receptors were used to distinguish the long-term-HSC subset from short-term-HSCs and multipotent progenitors (MPPs). In vivo PGE2 treatment preferentially increased the short term (ST)-HSC/MPP subpopulation without changing the frequency of long term (LT)-HSCs. Consistent with ex vivo treatment transplantation assays, transplanted cells from mice treated with PGE2 in vivo demonstrated increased engraftment and hematopoietic reconstitution of host animals; however, this competitive advantage was eventually lost, most likely due to exhaustion.

PGE2 in clinical trials Human CD34+ cells treated with PGE2 and transplanted into immunodeficient mice demonstrate enhanced transplantation abilities. The number of human CD45+ cells detected in the peripheral blood and bone marrow of recipient mice was ~2-fold higher in PGE2-treated groups compared to control48. This data suggested that PGE2 could be used to expand UCB cells prior to transplantation. A phase 1 trial was recently completed to

15

test expansion and subsequent transplantation of UCB HSCs following ex vivo PGE2 exposure, which included 12 patients and optimized treatment conditions49. Each patient received two cord blood units (CBUs) infused in succession; the first CBU was treated with PGE2, the second CBU was untreated. Neutrophil engraftment was observed in 100% of patients in the trial, compared to 90% of historic control. The median time to engraftment was 17.5 days versus 21 days, respectively. PGE2-treated CBU displayed enhanced engraftment by outcompeting the untreated CBU in 10 out of 12 patients. Initial results demonstrate that this PGE2 treatment regimen is clinically well tolerated. It may be possible in the future to achieve engraftment in patients receiving only a single CBU that is expanded with PGE2. Collectively, these studies have clearly defined a role for PGE2 in the function of HSPCs, yet the mechanisms behind its effects remain largely unknown.

Synthesis and mechanism of action Prostaglandins are lipid compounds of the eicosanoid family that play a major role in inflammatory and immune response as well as various other tissue responses50,51. They are found in a majority of tissues and are produced by a number of cell types. When synthesized, prostaglandins function locally as autocrine or paracrine lipid mediators. Prostaglandins are derived from arachidonic acid, which is oxidized by Cox-1 and Cox-2 to form PGG2. PGG2 is subsequently reduced to PGH2, from which all three classes of prostaglandins originate. Prostaglandin E synthase acts on PGH2 to produce PGE252. Cox-1 is constitutively expressed in most tissues, whereas Cox-2 is silenced in normal physiological conditions but can rapidly activate downstream targets during times

16

of stress. PGE2 acts via four G-protein coupled E-prostanoid receptors PTGER1-4 (EP1-4), resulting in various and sometimes opposing downstream effects53. PGE2 has been shown to be both a vasodilator (in arterial and venous beds) and a vasoconstrictor (in the trachea and intestine), supporting evidence of multiple functional E-prostanoid receptors54. These receptors are loosely classified based on their function as relaxants or constrictors of smooth muscle cells. EP2 and EP4 are relaxant receptors and result in production of cAMP upon binding of PGE2. Despite this common pathway, additional unique functions have been shown for each receptor. PGE2 bound to EP2 but not EP4 can result in activation of the EGF receptor leading to increased invasion of colon cancer cells. EP1 receptors are constrictors, where binding of PGE2 increases intracellular calcium levels in smooth muscle cells. The diverse effects of PGE2 signaling are further demonstrated through EP3 receptor activity. Different splice variants of EP3 receptors have been implicated in cAMP induction and inhibition, as well as generation of IP3. In addition, PGE2 binds E-prostanoid receptors with different affinities, with a higher affinity for EP3 and EP4, and a lower affinity for EP1 and EP255. In zebrafish, the PGE2 receptors expressed on HSCs are Ptger2 and Ptger4. Knock down of ptger2 or ptger4 results in a decrease of runx/cmyb expression in the AGM that is not rescued with PGE2 treatment43. This suggests that PGE2 signals through Ptger2 and Ptger4 to regulate HSC formation in zebrafish embryos. Interestingly, binding of PGE2 to EP2 or EP4 can lead to activation of GSK-3/!-catenin signaling pathway via protein kinase A and phosphoinositide 3-kinase, respectively56. Furthermore, a recent study demonstrated an in vivo conserved genetic interaction

17

between PGE2 and Wnt at the level of the !-catenin stabilization in the zebrafish57. PGE2 treatment significantly increases expression of a !-catenin responsive GFP reporter in the AGM region, while treatment with indomethacin, a non-selective COX inhibitor, decreased expression. Co-localization of the GFP reporter and lmo2 indicated Wnt activity was present in HSCs and endothelial cells of the dorsal aorta. Induction of Wnt signaling during embryogenesis increased runx1+ cells in the AGM, however treatment with indomethacin blocked this increase, suggesting PGE2 is required for this effect. A zebrafish cmyb:GFP reporter line was used to demonstrate that PGE2 and Wnt signaling synergize to increase HSC number in the AGM. Combined treatment of PGE2 and Wnt activation enhanced the effect of Wnt activation alone. These results suggest that PGE2 is required for Wnt-mediated effects on HSC development and can enhance Wnt activity in vivo. PGE2 and Wnt interaction was tested in the adult zebrafish using a kidney marrow recovery assay. FACS analysis of recovering marrow indicates that Wnt activation significantly increased HSPC population and this was blocked by indomethacin. These results support interplay of PGE2 and Wnt signaling in the adult zebrafish. To determine whether this interaction is conserved in mammals, purified LSK cells were transplanted into irradiated mice. Recipients were treated with 6-bromoindirubin-3’-oxime (BIO), a GSK3! inhibitor, indomethacin, or both. CFU-S assays following treatment demonstrated an increase in progenitor cells after BIO treatment; this effect was reduced to baseline levels upon treatment of indomethacin, demonstrating that PGE2 mediates Wnt activity on mammalian hematopoietic progenitor cells.

18

All 4 EP receptors are expressed on murine progenitor populations (LSK) as well as more purified stem cell populations (LSKCD48-CD150+). EP1-EP4 receptors are also found on human CD34+CD38- cells46. This data, combined with the observed genetic interaction between PGE2 and Wnt, suggests the EP2 and EP4 receptors in particular might play an important role in the regulation of HSCs. Knockdown of EP4 in LSK cells results in decreased repopulation ability following transplantation, as well as a skewing towards T cell and myeloid lineages during differentiation58. In contrast, EP2 knockout mice exhibit normal hematopoiesis, reconstitution capabilities, and differentiation. In addition, treatment of LSK cells with an EP4 agonist, but not an EP2 agonist, lead to an increase in phosphorylation of GSK3! and !-catenin. Co-treatment of a PKA inhibitor with PGE2 blocked the activation of GSK3!. This data suggests that PGE2 acts through EP4 to activate the cAMP/PKA pathway in HSPCs.

19

Thesis Review In this thesis we have used zebrafish and mouse models to study the endogenous microenvironment and trafficking of HSPCs in the niche. We have also studied the longterm repopulating capabilities of PGE2-treated HSPCs following transplantation. To accomplish this, we developed a novel zebrafish transgenic reporter line that enabled us to visualize and purify zebrafish HSPCs. We demonstrate that the runx1:GFP labels a population of stem and progenitor cells in zebrafish close to that of LSK in mice. We used this line to observe a novel and essential cellular behavior that involves triggered remodeling of perivascular endothelial cells upon arrival of an HSPC in a new site of hematopoiesis. A chemical screen during zebrafish embryogenesis revealed novel compounds that regulate cellular behaviors, such as adhesion and division, in the CHT. Imaging of fetal liver explants revealed that our observed endothelial niche remodeling is conserved in mammals. Runx1:GFP fish treated with PGE2 during embryogenesis exhibit increased runx1+ cells in the AGM and CHT, consistent with previous in situ data. This increase in HSPCs is maintained into adulthood, even in the absence of prolonged PGE2 exposure. Kidney marrow from these treated fish can outcompete control marrow in transplantation assays. This advantage appears to be due to an effect on the HSPC pool as a whole, rather than an individual cell basis. The long-term effect of PGE2 on HSCT is conserved in mammals; WBM cells treated with PGE2 maintain increased chimerism levels in recipient mice over 1-year post transplantation. These cells display enhanced transplantability in competitive secondary transplantations without additional PGE2 treatment.

20

Further purification of murine HSPCs using cell surface marker antibody combinations demonstrated that PGE2 affects a specific population of HSCs that are phenotypically LSKCD48-CD34+CD150+. Historically thought to only have transient multilineage potential, we show that following PGE2 treatment these cells display long term multilineage reconstitution that persists at least 1-year post primary transplant. Gene expression data and ChIP-seq analysis in human CD34+ cells suggests that PGE2 may play a role in the quiescent state of these cells. Our studies provide a glimpse into the dynamic interactions between HSPCs and their endogenous niche. In addition, we show that a population of short-term HSCs can engraft and give rise to long-term multilineage reconstitution following PGE2 treatment. Collectively, we have gained novel insights in the pathways involved in HSC migration, homing, and repopulation.

21

Chapter 2

Live imaging of hematopoietic stem cell lodgement reveals dynamic endothelial niche remodeling

Attributions This chapter is to be submitted as a co-first author paper with Owen Tamplin, a post-doctoral fellow in the lab. Owen established the Runx1:GFP and Runx1:mCherry transgenic zebrafish lines. I characterized these lines by designing and performing all FACS experiments to detect Runx1+ cells in embryonic and adult zebrafish and compared the new Runx1 lines to previously used HSPC zebrafish lines. I also designed and performed the limit dilution transplantations in adult fish and performed FACS analysis on embryonic transplants. Owen performed all confocal microscopy and time-lapse movies using the Runx1 transgenic lines. EM microscopy was done by Sarah Childs at Alberta Children’s Hospital Research Institute, University of Calgary. Owen performed the chemical screen in zebrafish embryos. I established a FACS protocol to study the effect of Lycorine on Runx1+ cells and designed and performed the experiments for microarray on Lycorine-treated cells. We both designed experiments for studying murine fetal liver HSCs. Owen performed microscopy for fetal liver explants, and I characterized these cells and performed the FACS analysis/sorts. The addendum to this chapter includes work done in the mouse system to further study the interactions between the CXCR4 and S1P signaling pathways during mobilization and homing that is not included in the publication. Owen designed and performed the chemical screen in zebrafish that identified a role for the S1P pathway in CHT engraftment, as well as optimized dose response and interactions with CXCR4 in the zebrafish. I designed and performed all mouse experiments, including treating cells and FACS analysis, mobilization, and homing experiments.

23

Abstract Hematopoietic stem and progenitor cells (HSPCs) are able to reconstitute and sustain the entire blood system, making them critical for clinical transplantation as a treatment for leukemia and blood disease. To directly observe HSPCs in their endogenous microenvironment, we established a highly specific transgenic reporter in the zebrafish. Using confocal live imaging to track an HSPC as it arrives and lodges in the perivascular niche, we observed a novel cellular interaction—a small group of endothelial cells remodel around a single HSPC to form a surrounding pocket. To resolve the ultrastructure of this niche, we correlated live imaging of these rare cellular events with high-resolution 3D scanning electron microscopy data. Endothelial niche formation triggered by HSPC arrival is evolutionarily conserved in the mouse fetal liver. A chemical screen identified small molecule regulators of HSPC niche colonization. Our work establishes a new dynamic model of niche formation during stem cell engraftment.

Introduction Hematopoietic stem and progenitor cells (HSPCs) self-renew and give rise to all blood cell types. During development and throughout adulthood, HSPCs will undergo many migration events. Definitive HSPCs arise from the hemogenic endothelium of the dorsal aorta (DA)59–61 are released into circulation, and then seed an intermediate hematopoietic niche before colonizing the adult marrow. In mammals, this intermediate tissue is the fetal liver (FL), and in zebrafish it is the caudal hematopoietic tissue (CHT), a vascular plexus in the ventral tail of the embryo62,63. After rapid expansion in the intermediate niche, HSPCs will leave and go on to seed the adult marrow, which in

24

mammals is bone and in zebrafish is kidney64. During life the changing requirements for HSPCs in the marrow and other tissues of the body dictate that stem cells continue migrating between different locations65. HSPC trafficking during development and in the adult is highly regulated, enabling release of stem cells into circulation, followed by homing and engraftment into a niche. Understanding these trafficking mechanisms is critically important because they are mimicked during clinical stem cell mobilization and transplantation.

The HSPC niche is a complex microenvironment that maintains the stem cell pool throughout life, while balancing the production of multilineage progenitors. It has been thought that the endosteum of the bone marrow and its osteoblastic cells create a unique region that favors HSPC quiescence15. The bone marrow also contains a complex network of sinusoidal vessels that act as an interface between circulation and the niche. In fact, most HSPCs are proximal to these vessels and are therefore considered to be in a perivascular niche28,66,67. Studies have shown that endothelial cells (EC) have distinct properties that enable them to support and expand associated HSPCs68,69. However, there are many different cell types, such as perivascular mesenchymal cells, stromal cells, and arterioles that also play a role in the niche67,70,71. Together these studies build a model of the many interacting cell types and signals that make up the HSPC niche microenvironment. Yet the static images we have acquired from fixed samples do not give us a dynamic view of the interactions between endogenous cell types in the niche.

25

Live imaging of the adult HSPC niche is a challenging task. Homing of circulating HSPCs to the adult bone marrow has been observed in a number of elegant studies72–76. These studies have been technically limited to multiphoton intravital microscopy of surgically accessed bone, such as the skull, or bone explants such as the femur. There has been the caveat that labeled, sorted and transplanted HSPCs could differ physiologically from endogenous HSPCs. Also, in the case of myelosuppressed recipients there can be significant damage to the perivascular niche69. In the embryo, confocal time-lapse microscopy has been used to capture the emergence of HSPCs from the hemogenic endothelium of the DA in both mouse explants and live zebrafish embryos59–
61

, but the immigration of HSPCs into secondary sites of colonization has not been

extensively studied.

The zebrafish embryo is particularly amenable to live imaging studies. Here we sought to develop a novel transgenic zebrafish that we could use to follow HSPCs during migration. Definitive hematopoiesis progresses quickly in the zebrafish: HSPC birth in the DA, migration to the CHT, seeding of the thymus, and colonization of the kidney marrow takes only five days62. Conserved hematopoietic regulatory genes have led to the development of HSPC transgenic reporter lines, although none of these are entirely specific (e.g. cmyb:EGFP, cd41:EGFP77, and runx1P2:EGFP78). To establish a more specific HSPC line, we utilized a regulatory element from the first intron of the mouse Runx1 locus that is +23 kb downstream of the ATG in the first promoter to drive expression of a reporter79. In mouse, the Runx1+23 enhancer drives early expression in the primitive wave of hematopoiesis, then in the definitive wave from HSPC birth in the

26

DA to long-term repopulating hematopoietic stem cells (LTR-HSC) in adult bone marrow. We used the Runx1+23 enhancer from mouse to establish a novel HSPC-specific transgenic line in zebrafish. The ability to track endogenous HSPC in the live embryo allowed us to observe dynamic interactions with other cell types in the niche. We have discovered a novel and essential cellular behavior that involves triggered remodeling of perivascular EC upon arrival of an HSPC in a new site of hematopoiesis. Using 3D reconstruction of high resolution serial section electron microscopy scans, we have an accurate view of the stem cell and the adjacent cells in the niche. Furthermore, we have shown that a similar event occurs in the mouse FL, suggesting this is a highly conserved step during lodgement of an HSPC in its niche.

Materials and Methods Zebrafish maintenance and lines Zebrafish and mice were maintained in accordance with Animal Research Guidelines at Children's Hospital Boston. Transgenic zebrafish lines were crossed and selected by fluorescent microscopy. Time-lapse live imaging was performed using a spinning disk confocal with an incubated chamber and moving XY stage. Zebrafish embryos and mouse embryo explants were mounted and imaged as previously described59,80. We screened a chemical library of ~2400 known bioactive compounds at concentrations optimized for whole embryo treatments. After treatment, we checked for secondary defects (e.g. stopped circulation, toxicity, developmental delay). Next, we performed WISH with HSPC markers cmyb and runx1 as previously described81.

27

Imaging Staged transgenic zebrafish embryos were selected and mounted for imaging in 1% LMP agarose with E3 media and tricaine as described. Some zebrafish embryos had 0.003% PTU (1-Phenyl-2-thiourea) added to the media to block melanogenesis. Zebrafish embryos and FL explants were imaged in MatTek glass bottom dishes or multi-well plates (No. 1.5 cover slip). Zebrafish live imaging was performed in an incubated chamber at 28°C. Mouse embryo explant live imaging was performed in an incubated chamber at 37°C with humidified CO2 with culture media (DMEM, 20% FCS, glutamine, sodium pyruvate, 2-mercaptoethanol, 1% penicillin-streptomycin, recombinant mouse IL3 (R&D Systems; final concentration 50 ng/ml)). Confocal microscopy was performed using a Yokogawa spinning disk and Nikon inverted Ti microscope. Our microscope configuration allowed imaging of multiple embryos within a 2-5 minute interval using a moving XY stage, as well as acquisition of z-stacks through the entire CHT (1-2 µm optical slices) in multiple fluorescent channels. Objectives lenses (Nikon): 20x Plan-Apo DIC N.A. 0.75; 40x Plan-Apo phase N.A. 0.95 dry; 40x Apo LWD WI NA 1.15 lambda S. Image acquisition was done with a single Andor iXon DU-897 EM-CCD camera (512x512 pixels) or dual Andor iXon x3 EM-CCD cameras (512x512 pixels) and Andor iQ or NIS-elements computer software. Fixed transgenic zebrafish embryos were scanned using a Nikon C2si confocal system and NiE upright microscope. These embryos were briefly fixed for 10 minutes with PEM fixation buffer (dH20, EGTA 10 mM, MgSO4 1 mM, PIPES 100 mM, Triton X-100 0.1%, PFA 4%; Cold Spring Harbor Protocols, 2009, doi:10.1101/pdb.rec11730), then washed and mounted with DRAQ5 (1:500) for staining of nuclei.

28

Imaging analysis Image processing and rendering was done using Fluorender, Imaris (Bitplane), NIS-elements (Nikon), Volocity (PerkinElmer) and ImageJ/Fiji. The MTrackJ plugin was used for manual cell tracking. Lineage trees were created using Endro. Point-to-point measurements were made with Imaris. Manual tracing, segmentation, and surface rendering of objects was performed using Imaris. Confocal z-stack images are presented as single slices, maximum projections, or 3D rendered projections. In some cases, background subtraction was performed, and brightness and contrast was adjusted in one or more channels of a multi-channel image.

Immunohistochemistry of adult WKM Adult zebrafish were fixed with 4% paraformaldehyde, paraffin-embedded, sectioned, and stained with hematoxylin and eosin (H&E). Immunohistochemistry was performed using anti-GFP monoclonal antibody (clone JL-8). The Dako Mouse Envision kit with EDTA antigen retrieval was used for visualization. Incubation in the primary antibody was 60 minutes, and 30 minutes in the secondary, followed by DAB development for 5 minutes and counterstained with hematoxilyn.

Adult-to-Adult HSPC Transplantation WKM cells from 3-month Runx1:mCherry;ubi:GFP fish were isolated and sorted by FACS. Double-positive cells were transplanted into irradiated casper recipients (n=20-48 recipients per cell dose) along with untreated helper marrow at the following ratios: 1:20,000; 5:20,000; 10:20,000; 25:20,000; 50:20,000. Transplantation procedures

29

were performed as previously described82. At 3 months post-transplant, WKM from recipient fish was collected and analyzed by FACS to detect chimerism levels of mCherry and GFP-positive cells in the marrow. We confirmed multi-lineage reconstitution by observing differentiated GFP-positive cells in multiple cell populations, as determined by forward and side scatter profiles. Stem cell frequency was determined using ELDA software (confidence interval=0.95).

Embryo-to-Embryo HSPC Transplantation Adapted from previously published protocols10,83. Runx1:mCherry;ubi:EGFP or Runx1:GFP;ubi:mCherry 3 dpf embryos were collected and chopped finely with a razor blade. Embryos were dissociated in a 1:65 dilution of Liberase TM (Roche) in PBS, incubated at 37°C for 20 minutes before addition of PBS/5% FCS to stop the reaction. Dissociated cells were passed through a 45 µM filter, spun, and resuspended in PBS/5% FCS. mCherry+/GFP+ cells were collected using a FACSAria cell sorter (BD Biosciences). Collected cells were resuspended in PBS at an estimated concentration of 400 cells/microliter with 0.5% rhodamine-dextran(10k) as a marker for injection. A microinjection needle (without filament) was back-filled with the cell suspension. The drop volume was calibrated to 1 nanoliter and each embryo was injected with 1, 2, or 4 drops. This gave an estimated cell dose of 0.4, 0.8 or 1.6 cells per embryo. Drops were injected into the sinus venosus (i.e. duct of Cuvier) of 48 hpf wild-type AB embryo recipients. Embryos were held in place using agarose injection ramps. Approximately 30 embryos were injected per dose and 12-26 embryos per group survived to adulthood (3-5 months). WKM was then analyzed for percentage of engrafted Runx+ cells using a LSR

30

II flow cytometer (BD Biosciences). Any recipients with positive cells detected above background were scored as engrafted. Flow cytometry data was analyzed using FACSDiva and FlowJo software.

Serial block face scanning electron microscopy and 3D reconstructions Immediately at the end of live imaging time-lapse acquisition 60 hpf embryos were fixed in 2.5% glutaraldehyde and 4% paraformaldehyde in a 0.1 M sodium cacodylate buffer. Embryos were embedded in 5% low melting point agarose in 0.1M sodium cacodylate buffer for orientation. Samples were submitted to Renovo Neural Inc (Cleveland, USA) for further processing. Samples were stained with heavy metals following the protocol from the National Center for Microscopy and Imaging Research; http://ncmir.ucsd.edu/sbfsem-protocol.pdf. Samples were embedded in Epon resin, and mounted onto pins (detailed protocol available from Renovo Neural). Serial blockface images (analogous to serial sectioning) were obtained using a Zeiss Sigma VP scanning electron microscope equipped with a Gatan 3View in-chamber ultramicrotome. A series of 500-1000 images were acquired at 2 kV using at 15,000 magnification from the region of interest. Image and stack resolution was 10 nm/pixel with 100 nm slices. Images were registered and resized as necessary using ImageJ/Fiji software. Images were imported into the program IMOD 4.5 then aligned and reconstructed using dual-axis tomography. Cells were manually outlined and 3D reconstructions were generated. Movies were rendered using IMOD and Fiji.

31

Microarrays Runx1:GFP;kdrl:RFP embryos were collected and raised in E3 media at 28.5C. Embryos were treated from 2-3 dpf with E3/1% DMSO or E3/1%DMOS/75 uM Lycorine (~100-150 embryos per group). Embryos were dissociated and sorted as above. Three populations were collected: GFP+ HSPC (~1-2k cells/experiment), RFP+ EC (~10-20k cells/experiment), and negative cells (~100k cells/experiment; total embryo as a comparator population). In total, 18 samples were collected: 3 biological replicates x 2 treatment conditions x 3 cell populations. Cells were sorted directly in Trizol LS. Trizol extraction was performed as per the manufacturer's instructions, with the addition of GenElute LPA (Sigma, Cat.#56575). Total RNA was amplified and labeled using the NuGEN Ovation Pico WTA System V2 and Encore Biotin Module kits, respectively. Affymetrix ZebGene-1_0-st microarrays were hybridized, washed, and stained using Ambion kits. Microarrays were scanned using the Genechip Scanner 3000 7G.

Fetal liver preparations Live fetal liver explants: Pregnant wild-type C57/BL6 and Ly6a-GFP mice were dissected at E11-11.5 (vaginal plug observation was E0). Embryos were removed from the uterus in PBS with 10% FCS and penicillin-streptomycin. Embryos were staged as E11-E11.5 by counting >42 somite pairs. FLs were removed from the embryo and treated as described by Boisset and colleagues (“protocol b”). Conjugated antibodies used for detection were CD31-FITC (PECAM-1, BD Biosciences, Rat anti-mouse, clone MEC 13.3) and c-kit-APC (CD117, BD Biosciences, Rat anti-mouse, clone 2B8).

32

Fixed fetal livers: Embryos were dissected as above and fixed in 2% PFA on ice for 20 minutes. Embryos were then rinsed and dehydrated in methanol. FLs were removed, mounted, cleared and imaged using the method by Yokomizo et al. Livers from E11.5 Ly6a-GFP mice were stained with rat anti-CD144, rabbit anti-RUNX1, and chicken anti-GFP. Primary and secondary antibody details are below. Specimens were scanned using a Zeiss LSM 710 confocal microscope with a 25x oil objective. Images were acquired using multi-track sequential mode and Zeiss Zen software. Pinhole was set at 1 Airy unit, steps were 1.14 "m per z-section. 3D projections were made using Volocity software.

33

Results Establishment of a highly specific HSPC transgenic zebrafish line To observe and study endogenous HSPC, we established a transgenic zebrafish line with the mouse Runx1+23 enhancer driving either cytoplasmic EGFP (Runx:GFP) or nuclear localized mCherry (Runx:mCherry). These two lines were crossed to demonstrate that the green and red fluorescent proteins marked the same cell populations (Figure S1A). Time-lapse live imaging showed that Runx+ cells emerge from the hemogenic endothelium of the DA (Figure S1B and Movie S1). Intercrossing the Runx:mCherry line with cmyb:EGFP and cd41:EGFP lines showed that Runx+ cells marked an overlapping population of HSPC in all major hematopoietic sites of the embryo, including the DA, CHT, thymus and kidney (Figure S1C-L and data not shown). We detected Runx:GFP+ cells in the adult kidney marrow using anti-GFP immunohistochemistry and fluorescence-activated cell sorting (FACS; Figure S1M-P). Next, we followed HSPCs after their emergence, when they travel through circulation to arrive in and seed the CHT niche. At the early stages of FL colonization in mouse at embryonic day (E) 11 it is estimated there is only about one transplantable HSPC84, predicting that the number of Runx+ cells in the CHT would be rare. We quantified the number of Runx:GFP+ cells in the CHT at different developmental stages. We found there were only an average of 1, 2, or 5 Runx:GFP+ high cells (and 2, 6, or 10 Runx:GFP+ low cells) in the CHT at 48, 58, and 80 hpf, respectively (Figure 4A-F). Cell numbers in the Runx:mCherry line were comparable, except from 72 hpf there was a greater number of Runx:mCherry+ low cells that are likely progeny (Figure S1G-L). Looking at Runx:GFP+ cells in the CHT together with a vascular transgenic reporter

34

(kdrl:RFP85 or kdrl:Hsa.hras-mCherry) we found that HSPC were closely associated with the EC of the caudal vein plexus (Figure 4A-C). To further understand the association of Runx+ cells with the CHT niche, we crossed the Runx:GFP and cxcl12a:DsRed2 transgenic lines to observe HSPC together with stromal cells86. To quantify the proximity of these two cell types within 3D confocal z-stacks, we measured the distance between a Runx+ HSPC and its nearest cxcl12a+ stromal cell (Figure 4G-I). This analysis showed that 60% of HSPC are in contact with a cxcl12a+ cell and 85% are "3 microns distance away. We performed timelapse live imaging of CHT colonization in these double transgenic embryos and found that arrival and expansion of most HSPC occurred in close proximity to stromal cells (data not shown). This provided further support that cxcl12a+ cells are an important component of the CHT microenvironment. We confirmed that endothelial and stromal cells are closely associated in the CHT62, similar to the perivascular niche in fetal and adult bone marrow67,87, and showed that cxcl12a+ cells are distributed in abluminal spaces that underlie kdrl+ EC (Figure 4J). The HSPC-specific transgenic line, together with endothelial and stromal markers, has allowed us to demarcate HSPCs and their niche in the developing zebrafish embryo.

The functional stem cell characteristics of Runx+ HSPC in the zebrafish To assess the functional characteristics of cells in the Runx+ pool, we performed limiting dilution transplantation as a standard assay to determine hematopoietic stem cell content. We purified Runx:mCherry+ cells from the kidney marrow of adult transgenic donors. These donors also carried the ubiquitous ubi:GFP transgene88 so that

35

Figure 4. A novel transgenic zebrafish line specifically marks HSPC in the CHT niche (A-C) The CHT is colonized by Runx:GFP+ HSPC (green) that are closely associated with kdrl:mCherry+ EC (red). The caudal artery (CA) is dorsal to the CHT, and the caudal vein (CV) is ventral, with circulation running posterior (arrow right) and anterior (arrow left), respectively. A cluster of 3 Runx:GFP+ high cells are outlined with a dashed line and 4 Runx:GFP+ low cells are indicated with arrowheads. 58 hpf embryo. (D-F) The number of Runx:GFP+ high and low cells are quantified, together with the length of the CHT as an indicator of stage. (G) Percentage of HSPCs in the CHT scored by distance to the nearest stromal cell (n=168 total cells from 25 embryos). (H) Detail of Runx:GFP+ HSPCs (green) in proximity to cxcl12a:DsRed2+ stromal cells (red). Arrowheads mark HSPCs in contact with stromal cell. Circle marks HSPC with a 2 µm gap between it and the stromal cell. 60 hpf embryo. (I) cxcl12a:DsRed2+ stromal cells (red) underlie kdrl:GFP+ EC (green). 40 hpf embryo. Note: Confocal images are 3D rendered depth projections of 20-30 µm z-stacks. Scale bars: (A,I) 25 µm; (H) 10 µm. See also Figure S1.

36

multilineage contribution could be assessed if the Runx:mCherry HSPC enhancer was down-regulated upon differentiation. Sorted double positive cells were diluted in a range between 1 and 50 then transplanted into irradiated recipients. Survival rates improved in recipients that received a greater number of donor HSPCs, and at 3 months posttransplantation the kidney marrow of recipients was dissected for FACS analysis. Runx:mCherry+ cells were present in the kidney marrow and had contributed ubi:GFP+ progeny to all lineages (Figure S2A). Statistical analysis of engraftment over a range of cell doses estimated a stem cell frequency of approximately 1/35 (Figure 5). We consider this to be an underestimate because the donors and recipients are non-isogenic, and are therefore not immune matched, leading us to assume that some donor cells would be rejected. The Runx+ HSPC pool can be sorted with a single transgenic marker, and no additional labeling of cell surface markers, to a purity that is within the range of the wellcharacterized KSL (c-Kit+ Sca-1+ Lin-) population in mouse89. Based on the ability of a small number of adult Runx+ cells to engraft long-term, self-renew, and produce all lineages, we have demonstrated there is substantial stem cell content within this pool of cells. However, there are no antibodies available in zebrafish to further purify this population, or to distinguish between a stem cell and a progenitor. Therefore we will continue to use the term HSPC to describe the Runx+ cells in this study. We also wanted to evaluate the stem cell characteristics of the Runx+ population in the embryo. A number of studies have shown that hematopoietic stem cells isolated from different tissues of the mouse embryo have the capacity to reconstitute hematopoiesis in the adult84,90,91. However, technical limitations in mouse have made it difficult to assess the stem cell potential of an HSPC transplanted from one

37

Figure 5. Runx+ cells in the adult and embryo are functional HSPC Summary of results from (A) adult-to-adult and (B) embryo-to-embryo limiting dilution transplantation experiments. (C) Embryo-to-embryo transplantation recipients with engraftment of Runx:GFP+ cells in the kidney marrow at 3 months (above background; >0.001%). Representative kidney marrow FACS plots of (D) an adult-to-adult transplantation recipient with Runx:mCherry+ HSPC and ubi:GFP+ lineages, and (E) embryo-to-embryo transplantation recipient with Runx:GFP+ HSPC and ubi:mCherry+ lineages. See also Figure S2.

embryo to another. Based on a previous approach83 we have further developed an HSPC transplantation assay that is unique to zebrafish. HSPCs are sorted from a pool of double transgenic Runx+ and ubi+ donor embryos that were 3 dpf. Only 1-2 cells were injected into the circulation of a wild-type recipient embryo at 2 dpf. A dilution series established the number of cells injected for each experiment. At 2 dpf the CHT is being colonized by endogenous HSPCs but the thymus has not formed, allowing introduction of exogenous cells without the possibility of immune rejection. Recipient embryos are then raised to adulthood and their kidney marrow is 38

FACS analyzed for engraftment at 3-5 months. We scored engraftment as any detectable Runx+ cells above background (Figure 5 and Figure S2). This rationale was chosen because approximately one donor HSPC will be competing with endogenous stem cells in an unconditioned wild-type recipient embryo, and there is no precedent to predict chimerism in this scenario. The transplanted cells must seed the CHT, migrate to the kidney, and persist into adulthood where they will self-renew and contribute to all lineages. We identified recipients with Runx+ HSPCs and ubi+ progeny in the kidney marrow, as well as some that had ubi+ cells in the peripheral blood (Figure S2 and data not shown). Statistical analysis estimated the stem cell frequency of the Runx+ population in the 3 dpf embryo to be approximately 1/2.88 cells (Figure 5 and Figure S2). These results were representative of three independent experiments—two with Runx:GFP;ubi:mCherry double transgenic lines and a third with the opposite Runx:mCherry;ubi:GFP transgenic combination. Together, our limiting dilution transplantation in adults, as well as our novel embryo transplantation assay, demonstrates that Runx+ cells mark a highly purified HSPC population with functional stem cell characteristics.

Dynamic visualization of stem cell colonization of the CHT niche To directly observe the interaction of HSPCs with surrounding EC during CHT colonization, we performed time-lapse live imaging with spinning-disk confocal microscopy. We were able to routinely acquire image series with high temporal resolution (1-2 minutes per 2-channel confocal z-stack) for up to 16 continuous hours. A widefield image was acquired using a 20x objective because the rare lodgement of a

39

single HSPC could occur anywhere in the length of the CHT (Figure 4A). We observed HSPC arrival in the CHT via circulation, followed by adherence to endothelial walls (Figure S3A and Movie S2). Next, cells underwent rapid extravasation to the abluminal side of the endothelial wall (<5 minutes; Figure S3A). Once HSPCs lodged in the CHT, we made a striking and novel observation: a small group of EC actually remodeled around a single HSPC to form a stem cell pocket, which we call “endothelial cuddling” (Figure 6A and S3B). Within the 12-16 hour limit of time-lapse acquisition, an HSPC would make one of three cell division decisions: symmetric, asymmetric, or no division. In this example the division is asymmetrical, with one daughter cell crawling out of the endothelial niche and the other remaining (Figure S3C). After HSPC lodgement in the CHT, we used higher magnification (40-60x) to better detail the close association and contact of surrounding ECs (Figure 6B). To quantify the number of surrounding ECs, we briefly fixed and stained embryos with a nuclear dye, then imaged using scanning confocal microscopy (Figure 6C). The ECs in contact with a single HSPC were outlined with membrane-bound mCherry and their nuclei were counted. In each pocket we typically observed 5-6 ECs surrounding a single HSPC. Time-lapse live imaging of endogenous Runx+ HSPCs in the embryo using our novel transgene has revealed striking interactions with perivascular ECs in the niche microenvironment.

40

Figure 6. Remodeling of the perivascular niche after HSPC arrival in the CHT (A) 4 frames selected from time-lapse Movie S2 between 40-42 hpf (hours:minutes). Upper row is a merge of Runx:GFP+ HSPC (green, middle row) and kdrl:RFP EC (red, lower row). A group of surrounding EC (broken circle) remodel around a single HSPC soon after its arrival. (B) Higher magnification (60x) live image of a single Runx:GFP+ HSPC surrounded by kdrl:mCherry at 78 hpf (orthogonal views). CV: caudal vein. (C) 3D rendered projection of scanning confocal image from fixed 80 hpf embryo. (i) merge, (ii) kdrl:mCherry EC, (iii) Runx:GFP+ HSPC, (iv) DRAQ5 nuclei, (v) kdrl:mCherry projection with modeled green HSPC and 5 blue surrounding EC nuclei (arrowhead indicates HSPC in EC surround). All views: dorsal up, ventral down. Scale bars: 10 µm. See also Figure S3.

High-resolution ultrastructure of the perivascular niche surrounding an HSPC We sought to better reveal the cellular interactions between an HSPC and its perivascular niche that would not otherwise be visible with confocal light microscopy. To do this we used correlative light and electron microscopy92. First, to confirm that an HSPC was lodged in the CHT, we performed time-lapse live imaging of Runx:GFP;kdrl:mCherry embryos as above (Figure 6A). Imaging multiple stage-matched 41

embryos in parallel, we confirmed that 1-2 HSPCs per embryo had not only migrated to the CHT, but had also lodged and triggered an EC remodeling event that was stable for more than 6 hours (Figure 7A-B). Embryos were then fixed and embedded for electron microscopy (EM). Serial block face scanning EM captured large sections of the CHT at high resolution (Figure 7C; XY: 10 nm/pixel, Z: 100 nm/slice). Based on a number of cellular and anatomical markers (e.g. melanocytes, somites, vessels), the position of the lodged HSPC in time-lapse could be correlated with EM serial sections (Figure S4). 3D reconstruction of EM scans shows that the HSPC lodges in a region just below the ventral endothelial surface of the dorsal aorta. As predicted from our confocal microscopy analysis (Figure 6C), the HSPC is surrounded by at least 5 ECs. There are two stromal cells in proximity to the HSPC, but only one is in contact in a narrow region near the mid-point of the cell (Figure 7D). This is also consistent with our confocal microscopy of Runx+ HSPC and cxcl12a+ stromal transgenic lines (Figure 4G-I). An observation that could not have been made with confocal microscopy was that a fourth cell type (i.e. mesenchymal-like cells with melanophore inclusions), was found together with the HSPC, ECs, and stromal cells (Figure 7D,F and Movie S3). These two cells wrap the HSPC tightly and intertwine with each other. We found another example of an HSPC that is not associated with mesenchymal-like cells, or in proximity to stromal cells, but is surrounded by approximately 5 ECs, showing there are different configurations of this HSPC-EC niche (Figure S4). We were also able to identify Runx+ HSPC in timelapse movies that were transiently migrating through the CHT and had not lodged in the perivascular niche. These HSPC had only adhered to the vessel walls and in EM sections

42

Figure 7. Serial block face scanning EM of CHT after time-lapse (A) Last frame of CHT time-lapse (60 hpf). Arrowhead marks HSPC lodged >6 hours. Runx:GFP (green), kdrl:mCherry (red), brightfield (blue). Anterior left, posterior right, dorsal top, ventral bottom. (B) Detail of region in (A) marked by box. (C) Single section and orthogonal slice from serial block face EM scans. Lodged HSPC is purple, surrounding EC nuclei are green and numbered, stromal cells are blue. (D) High resolution EM of HSPC lodged in perivascular niche just ventral to the dorsal aorta (circulating RBC visible). The HSPC (outlined in pink) is in direct contact with one stromal cell (light blue), is wrapped in 2 fibroblastic cells with melanophore inclusions (yellow and orange), and is surrounded by EC (3 of 6+ shown in green shades). (E,F) 3D rendered models of HSPC and niche in the same view as (D). Fibroblastic cells are not shown in (E) but are shown in (F). See also Figure S4 and Movie S3.

43

did not have the characteristic EC surround structure (data not shown). This further emphasized that the HSPC-EC niche formation event is rare and distinct.

A chemical genetic screen to discover small molecule regulators of CHT niche colonization Having tracked HSPC lodgement in the CHT niche, we wanted to identify the signaling mechanisms that regulate these cellular events. We performed a chemical genetic screen by introducing individual small molecules to embryos during the peak window of HSPC migration to the CHT (48-72 hpf; Figure 8). Hits were scored if they increased or decreased hematopoietic progenitor markers cmyb and runx1 with similar activities to control compounds (±)11,12-epoxyeicosatrienoic acid (EET; P.L. and L.I.Z., unpublished) or the CXCR4 antagonist, AMD3100, respectively. We chose AMD3100 as a positive control in the embryo because: 1) the duplicated homologs for CXCR4 and CXCL12 are expressed in the DA and CHT (cxcr4a/b and cxcl12a/b; Figure S5A); 2) the cxcl12a transgenic reporter93 is expressed in CHT stromal cells (Figure 4G-J); 6) AMD3100 blocks CXCL12-dependent migration of kidney marrow cells93. After treating embryos with a range of AMD3100 doses, we observed dose-dependent reduction of HSPC markers in the CHT (Figures 8B,E and S5B). Following these results, we screened ~2400 known bioactive compounds, and identified 40 individual compounds that increased and 107 that decreased CHT hematopoiesis.

44

Figure 8. Chemical genetic screen to identify small molecule regulators of CHT niche colonization (A-D) WISH of cmyb/runx1 probes in the CHT of 72 hpf embryos. Treatments from 48 hpf as indicated. (A) DMSO control with normal expression. (B) Low expression after 25 µM AMD3100 treatment. High expression after treatment with 40 µM SB-431542 (C) or 25 µM lycorine (D). (E) Percentage of total embryos scored with cmyb/runx1 expression levels as shown in (A-D). (F) Overview of chemical screen. (G) Selected frames from time-lapse movie of a Runx:GFP (green) x kdrl:RFP (red) double transgenic embryo treated with TGF-! inhibitor, SB-431542. In <90 minutes, an HSPC (a) undergoes rounds of division and releases daughter cells (b,c,e) into circulation (caudal artery, CA; circulation to posterior, right). (H) Lineage tree analysis of HSPC tracked in parallel time-lapse movies, with and without SB431542 treatment, from 44-60 hpf. One line is one continuously tracked HSPC from the time it enters the CHT field of view until it exits. Branch points mark cell divisions. The red box marks the lineage of the cells shown in (G). See also Figure S5.

45

Live imaging with small molecule regulators reveals distinct cellular behaviors within the CHT We tested a subset of our chemical screen hits by adding individual compounds to the media of Runx:GFP+;kdrl:mCherry+ embryos during time-lapse imaging to directly observe the behavior of HSPC as they colonize the CHT. One of the hits in our screen that increased CHT hematopoiesis was SB-431542 (Figure 8C), a selective inhibitor of transforming growth factor (TGF)-! type I receptors, and most potently ALK5/TGFBR1. When added during time-lapse imaging, we observed normal HSPC arrival in the CHT and adherence to vessel walls. However, soon after arrival HSPCs began to undergo frequent rounds of asymmetric division (Figure 8G,H and Movie S4). One daughter cell would bud off into circulation, while the other remained adhered to the vessel wall. To quantify this difference, we performed lineage tree analysis of HSPCs during parallel time-lapse movies (i.e. control and treated stage-matched embryos imaged side-by-side; Figure 8H). Scoring cell divisions between 44 and 60 hpf, more HSPCs divided in SB431542 treated CHT movies (n=8-10 versus n=3-4 in DMSO treated controls). These data illustrate that inhibition of TGF-! receptor signaling using the chemical SB-431542 expands HSPC populations by increasing the frequency of cell divisions. This is consistent with previously published in vitro and in vivo data that showed TGF-! signaling negatively regulates HSPC proliferation94–99. Directly visualizing the effect of SB-431542 on HSPCs has demonstrated that we can rapidly identify cellular response to a specific signal in the endogenous niche.

46

A second compound we identified in our screen was lycorine, a natural alkaloid extracted from the Amaryllidaceae plant family that dose-dependently increased cmyb and runx1 in the CHT (Figure 8D and 9A). This drug does not have a defined target or mechanism of action, but is a candidate anti-inflammatory and anti-cancer drug (Kang et al., 2012; Lamoral-Theys et al., 2010; Yamazaki and Kawano, 2011). Running parallel time-lapse movies, we observed more HSPCs nestled in the CHT in lycorine-treated embryos compared to controls (Figure 9B,C). Over time, lycorine treatment dramatically increased HSPC number in the CHT (Figure 9D). We also scored the total amount of time each HSPC was resident in the CHT during the time-lapse, and found lycorine treatment produced a significant shift towards longer durations spent in the niche (Figure 9E; median difference 1.67 hours; Wilcoxon signed rank test, p=0.01). Lycorine treatment during the peak of CHT colonization also had a sustained effect on the total number of HSPC in the embryo. A pool of Runx:GFP+ embryos that were treated with lycorine from 2-3 dpf then washed off, had a significantly higher number of GFP+ HSPCs at 7 dpf compared to DMSO controls (Figure 9F). To obtain a better understanding of the molecular effects of lycorine on HSPC, we treated Runx:GFP;kdrl:RFP embryos from 2-3 dpf, then sorted Runx+ HSPCs and kdrl+ ECs for whole genome microarray analysis. (Figure S6). Lycorine treatment induces significant changes in both HSPC and EC that promotes the interaction of HSPC with the perivascular CHT niche.

47

Figure 9. Lycorine modulates HSPC-EC interactions in the CHT (A) Lycorine treatment dose-dependently increases the percentage of embryos with high cmyb/runx1 expression levels, as shown in Figure 5D. (B-C) Stage-matched frames at 42 hpf from parallel time-lapse movies of Runx:GFP;kdrl:RFP double transgenic embryos. There are more HSPCs (marked with arrowheads) in lycorine-treated (25 µM) embryos (C) compared to DMSO-treated controls (B). (D) HSPCs are counted in each frame of a time-lapse movie and then averaged each hour. 5 movies imaged in parallel: n=2 controls and n=3 lycorine-treated (25 µM). (E) Same movies as in (D) were scored for continuous hours an HSPC was tracked in the CHT. There was a significant difference in the median time HSPC spent in the CHT between controls and lycorine-treated embryos (1.67 hours; Wilcoxon signed-rank test, p=0.01; HSPC counts were normalized because the lycorine group had 3 embryos and the control group had only 2). (F) Pools of Runx:GFP+ embryos treated with lycorine from 2-3 dpf have significantly increased HSPC at 7 dpf (p=0.0004; FACS of 4 independent pools per dose). Error bars show mean ± s.e.m. See also Table S1.

Endothelial niche remodeling is conserved in mammalian fetal liver It was important to establish if the distinct endothelial niche structure we identified in the zebrafish CHT was also found in mammals. The equivalent tissue is the FL as it is the first tissue colonized by definitive HSPC from the dorsal aorta. The FL is an intermediate site of hematopoiesis where HSPC expand before leaving to colonize the

48

adult marrow and it produces the majority of blood during development90,100,101. To examine the earliest stages of FL colonization, we dissected E11.5 FL from Ly6a-GFP (Sca-1) mice, which have GFP+ HSPC102,103. Together with HSPC marker Runx1, and EC marker VE-cadherin (Cdh5), we found Ly6a-GFP+/Runx1+ HSPC in one of 3 compartments in the FL: 1) an abluminal space with no EC contact (Figure 10A); 2) adherent to EC on one side (Figure 10B); 3) inside an EC pocket (Figure 10C and Movie S5). The similarity of a single HSPC surrounded by a small group of EC in both the mouse FL and zebrafish CHT (compare Figure 6B and 10C), suggests that this cellular structure is important for stem cell lodgement in hematopoietic niches. The identification of a potentially conserved HSPC-endothelial niche structure raised the possibility that HSPC also trigger a dynamic remodeling of EC during colonization of the FL. The FL tissue in mouse is not directly accessible to confocal microscopy, so we instead applied a protocol for live imaging of embryo explants. We dissected wild-type and Ly6a-GFP E11.5 FL, soaked the explant FL in fluorescently conjugated antibodies to detect c-kit+ HSPC and CD31+ (Pecam1+) EC, and immediately performed live imaging for up to 4 hours. We observed c-kit+ hematopoietic cells adhered to the sinusoidal network of CD31+ EC (Figure 10D-E and S7). We followed a c-kit+ HSPC attached to the sinusoid as it migrated into a small group of EC (Figure 10D and Movie S6). Next, EC tightly surrounded the HSPC in what looked strikingly similar to endothelial niche remodeling in the zebrafish CHT (compare Figure 6A and 10D). Tracking a Ly6a-GFP+/c-kit+ HSPC in another explant, we observed one cell division (Figure 10E). Intriguingly, the daughter cell proximal to the sinusoid remained surrounded by EC, while the daughter cell distal to the sinusoid migrated away

49

Figure 10. Conserved remodeling of ECs around HSPCs in the mouse fetal liver (A-C) Fetal livers from E11.5 Ly6a-GFP mice were fixed and stained for immunofluorescence with antiVE-Cadherin (red), anti-Runx (blue), and anti-GFP (green) antibodies. We scored 59 Ly6a-GFP+/Runx1+ cells from 3 fetal livers and identified three different HSPC-EC configurations: (A) abluminal with no contact between HSPC and EC; (B) EC contact on one side of the HSPC; (C) HSPC surrounded on all sides with EC. See Movie S5. (D) An HSPC with c-kit+ accumulation (magenta) is shown adhered to CD31+ endothelial sinusoid (green) in one lobe of an E11.5 FL (arrowhead). The white box marks details below. Key time-lapse frames show in <90 minutes the HSPC migrates into a field of SEC. Soon after the SEC surround the HSPC to form a niche. See Movie S6. (E) Ly6a-GFP+ HSPC (green) is adhered to the abluminal side of a CD31+ sinusoid (red), with a c-kit+ contact point (blue). Following this same cell for >2 hours (1 frame/5 minutes), shows the cell divides, with one cell remaining in the endothelial niche and the other migrating away. See Movie S7. Confocal images are 3D rendered depth projections (A,B,C,E), orthogonal views (A,B,C below) or maximum projections (A) of z-stacks. Scale bars: 10 µm. See also Figure S6.

50

into the abluminal space (Figure 10E and Movie S7). In another time-lapse sequence, a ckit+ HSPC was observed undergoing discrete steps towards lodgement: adherence, extravasation, abluminal migration, and endothelial niche remodeling (Figure S7A-C). We also identified an accumulation of c-kit at the contact point between the HSPC and EC (Figure 10E and S7C). This observation is similar to data that showed c-kit condenses with clustering lipid rafts in cytokine-induced HSPC104,105. Even with the caveat that FL explants have been removed from circulation and are imaged ex vivo, our live imaging data strongly suggest an evolutionary conserved process of dynamic EC remodeling around a single HSPC in a hematopoietic niche.

Discussion The ability to follow endogenous HSPCs in the live zebrafish as they lodge in a niche has allowed us to capture never before seen cellular behaviors during this important event in the life of a stem cell. This high-resolution view of the zebrafish CHT has shown events that have not been visible in the mammalian bone marrow. We watched as HSPCs attached to the endothelial wall of a vessel, underwent extravasation, migrated into the abluminal space, then triggered a striking endothelial remodeling event to form a niche (summarized in Figure S3D). Live imaging of mouse FL explants revealed that these steps towards HSPC lodgement are highly conserved. This high resolution tracking of HSPC interaction with the perivascular niche opens many possibilities to further dissect the mechanisms of hematopoietic colonization. The combination of time-lapse live imaging and chemical genetics in the zebrafish has enabled us to rapidly screen compounds that have an effect on endogenous

51

HSPCs in their microenvironment. We have validated this approach by showing that signaling pathways with well-established roles in HSPC trafficking and proliferation, CXCR4-CXCL12 and TGF-!, respectively also regulate HSPCs in the zebrafish embryo. We found that treatment of the embryo during CHT colonization with the compound lycorine dose-dependently increased the number of hematopoietic progenitors. Analysis of live imaging data suggested that HSPCs accumulated in the CHT because of longer interaction times with the perivascular niche. The additional finding that the increase in HSPC number was sustained long after lycorine was removed raises the possibility that HSPC-niche interactions during development could have a role in determining the longterm size of the stem cell pool. We are currently testing this hypothesis at later timepoints and into adulthood. To reveal clues to the mechanism of lycorine action, we isolated HSPC and EC from lycorine-treated embryos. We found there was cell typespecific variations in gene expression that point to changes in the migratory and adhesive properties of HSPC and EC. Our results show there is a functional role for interactions between HSPC and EC during development, and this may be important for setting aside the long-term hematopoietic stem cells that will colonize the adult marrow. Our finding that HSPC arrival triggers cellular changes in the local perivascular niche raises interesting questions about what constitutes a hematopoietic stem cell niche. Rather than a static number of niches that can be either cleared or filled (Schofield, 1978), our results are suggestive of basic niche components that create a permissive environment for an arriving stem cell. Once attracted to these general locations in the marrow, the stem cell will move out of circulation and lodge in its new surroundings. The rapid remodeling of endothelial cells around a stem cell may provide a mechanism to

52

retain and protect these new arrivals. It will be important to determine how much plasticity there is in the number and locations that a stem cell can reside. We focused our study on the hematopoietic microenvironment in the zebrafish CHT and mouse fetal liver. These are intermediate niches are required to rapidly expand the stem cell pool. With continual advancements in imaging technology we will soon be able to resolve similar cellular dynamics in the adult marrow. Prior to HSPC transplantation, myelosuppression by chemotherapy or irradiation is used to condition a recipient and prepare their bone marrow to receive donor stem cells. Conceptually, this has been thought to “clear” or “open” the niche. However, studies have shown that myelosuppression causes significant damage to the bone marrow vascular epithelium that must be repaired for hematopoiesis to recover (Hooper et al., 2009). This has presented a dynamic model of the vascular niche that requires angiogenesis, together with engraftment of incoming HSPC, to rebuild a functional hematopoietic bone marrow. The ability to modulate these HSPC-EC interactions, as we and others have started to demonstrate, will likely have an important impact on clinical stem cell transplantation.

53

Addendum A chemical screen for CHT colonization in zebrafish embryos demonstrated that signaling pathways with well-established roles in HSPC trafficking and proliferation also regulate HSPCs in zebrafish. AMD3100, a CXCR4 antagonist, dose-dependently decreased the numbers of HSPCs that localized to the CHT during embryogenesis. In addition, we identified a number of small molecules that regulate the sphingosine-1phosphate (S1P) pathway, which has previously been shown to play a role in chemotaxis and cell migration. Out of the 5 G-protein coupled receptors for S1P, S1PR1 is considered to be the one that is critical for blood cell migration. We looked at the expression pattern of S1PR1 receptor during the key stages of CHT colonization and found that it’s expressed in both the dorsal aorta and the CHT. To test if there was an interaction between CXCR4 and S1PR1 signaling, two important pathways that reduce CHT hematopoiesis, we used a chemical genetic interaction matrix, taking advantage of two clinically relevant small molecule regulators of these receptors: AMD3100 and FTY720 (Figure 11). This matrix demonstrated that there were combined doses of the two drugs that increased CHT hematopoiesis at concentrations where one drug alone had no effect. This result is highly suggestive of an interaction between these pathways in colonization of the CHT. This result suggested a potential for modulating both homing and engraftment in mammals. SDF-1, the ligand for CXCR4, and S1P, the ligand for S1PR1, are both powerful chemoattractants for murine HSCs, which express both receptor types. However, whereas SDF-1 levels are high in the bone marrow and cause HSCs to home to the niche via CXCR4, S1P levels are quite low in tissues, and are most highly expressed

54

Figure 11: Chemical interaction matrix of FTY720 and AMD3100 Zebrafish embryos were treated from 48hpf-72hpf with the doses of chemical indicated on the matrix. In situ hybridization was performed for runx1/cmyb+ cells. “0” score indicates no change in cell number from control, +2 indicates an increase of runx1/cmyb+ in the CHT, -2 indicates a decrease.

in the blood. The S1P agonist SEW2781 enhances HSC mobilization from the bone marrow following CXCR4 inhibition by AMD3100. This suggests an intricate connection between the CXCR4/SDF-1 and S1PR1/S1P signaling pathways to control HSC migration in and out of the bone marrow. Based on our chemical matrix and previous data, we hypothesized that during niche colonization HSCs respond and migrate towards high levels of S1P in the blood. Once in circulation, HSCs internalize S1PR1 and CXCR4 emerges on the surface of these cells. This allows HSCs to home to the

55

CHT, where CXCR4 internalizes and S1PR1 resurfaces once again, to enable cells to migrate to the next site. To examine the effect of these two drugs on CXCR4 and S1PR1 expression in mice, we treated whole bone marrow cells with varying doses of AMD3100 and FTY750 and performed FACS analysis to look at surface receptor expression. We found that treatment with either of the drugs alone, or in combination, resulted in dynamic changes in receptor expression (Figure 12). Treatment with higher doses of AMD3100 alone decreased CXCR4 expression while S1PR1 expression was increased. Conversely, treatment with FTY750 alone increased S1PR1 and decreased CXCR4 expression. Interestingly, combination treatments showed both increases and decreases in receptor expression dependent on dose.

Figure 12: S1PR1 and CXCR4 receptor expression data in WBM cells WBM was treated for 60 min at 37oC (n=8 total from 2 experiments), cells were sorted for lineage negative population and analyzed by FACS for receptor expression. The graph shows the change in number of WBM cells expressing receptor (log2, relative to control).

56

We next wanted to check if these changes in receptor expression level were correlated with changes in HSC homing and mobilization. We performed homing transplants using 4 X106 CD45.1 WBM cells treated with PBS, FTY750 alone, AMD 3100 alone, or a combination, for 1 hour. Cells were mixed with equal numbers of CD45.1.2 WBM cells and injected into CD45.2 recipient mice. 16 hours following transplantation, the marrow of recipient mice was harvested and stained for CD45 markers. As expected, we found a slight decrease in the homing capabilities of cells treated with AMD3100 alone (Figure 13). Interestingly, we found that the combination treatment resulted in a slight increase of homing potential, however these results must be repeated with optimized conditions.

Figure 13: Effect of AMD3100 and FTY720 treatment on homing of WBM cells WBM cells were treated with AMD3100, FTY720, or both for 1 hour. The graph indicates the ratio of treated cells: competitor cells in the marrow of recipient mice. These results were obtained 16 hours post transplantation. N=5 mice/group.

57

We also wanted to observe the combinatorial effects of these drugs on the mobilization of HSCs from the bone marrow to peripheral blood. In vivo treatment of mice with AMD3100 results in a dramatic increase of HSCs in circulation that peaks at 1-hour post injection and steadily decreases thereafter. We injected mice with PBS, AMD3100 alone, or AMD3100+FTY750 at multiple doses. At 1-hour post injection, we bled the treated mice and performed colony-forming assays to confirm the presence of HSCs in the peripheral blood. As expected, mice treated with PBS had essentially no circulating HSCs, evidenced by the absence of colonies at day 7, whereas mice treated with AMD3100 alone had high levels of circulating HSCs (Figure 14). The addition of FTY750 did not enhance mobilization over AMD3100 alone. However a low dose of FTY750 (0.01mg/kg) resulted in significantly more colonies than a high dose (1.0mg/kg).

Figure 14: Effect of AMD3100 and FTY720 treatment on mobilization of HSCs Graph shows the number of colonies on day 7 from peripheral blood that was isolated from mice 1 hourpost treatment with various drugs (as indicated in figure legend).

58

Our results suggest that we can modulate the processes of homing and mobilization using small molecule regulators of migration pathways and are consistent with recently published studies106,107. Further characterization of the CXCR4 and S1PR1 receptor dynamics in response to chemical modulation will be necessary to take advantage of this interaction. In the future, it should be possible to identify combinatorial doses of AMD3100 and FTY750 that favor either mobilization or homing. These treatment regimens could have useful clinical applications.
! ! ! ! ! ! ! ! ! ! ! ! ! ! ! !

59

! ! !

Chapter 3 Prostaglandin E2 regulates long-term repopulation activity of ST-HSCs

Attributions I designed and completed all zebrafish and mouse transplantation experiments. I analyzed transplant recipients with help from technicians Margot Brandt and Rebecca Maher. I performed all zebrafish and mouse FACS experiments, which included analysis of runx1:GFP fish, ex vivo treated cells, and sorted cells for transplantation. Owen Tamplin performed the confocal microscopy on dmPGE2-treated zebrafish. I performed cell sorts and treatments for the microarray and ChIP-seq experiments. Yi Zhou, Song Yang, and Eva Fast performed bioinformatics and data analysis for microarray and ChIPseq experiments.

61

Abstract 16,16- dimethyl prostaglandinE2 (dmPGE2) increases the number of embryonic HSCs in zebrafish and improves long-term engraftment of bone marrow and cord blood transplantation. The mechanism of signaling has been shown to involve CXCR4 upregulation and wnt activation, but the cellular activity of dmPGE2 remains to be determined. To investigate the long-term effects of dmPGE2, we transiently treated zebrafish embryos from gastrulation to 36 hours post fertilization (hpf), and grew up these animals to 3-4 months of age. Kidney marrow (KM) cells from adult zebrafish pulse-treated with dmPGE2 during embryogenesis were studied in a competitive transplantation experiment using GFP+ treated KM and untreated DsRed KM. The ratio of GFP+ cells present in KM 3 months post-transplant was higher in dmPGE2-treated groups compared to control (1.25 vs. 0.68, p<.001). We tested the ability of dmPGE2 to confer a long-term advantage on sorted marrow populations in competitive transplantation assays in the mouse. ST-HSCs, but not LT-HSCs or MPPs, showed enhanced reconstitution in recipient animals a full 12 months after dmPGE2 exposure, compared to control animals. The effects of dmPGE2 on HSC function persist over substantial time despite transient exposure, and suggest the long-term effects of dmPGE2 are due to a stimulatory effect on the ST-HSC population to acquire long-term activity. Gene expression data and ChIP-seq analysis in human CD34+ cells suggests that dmPGE2 may play a role in the quiescent state of these cells. Our studies are the first to demonstrate that a small molecule can prevent exhaustion of ST-HSCs, which facilitates long-term reconstitution following transplant.

62

Introduction Hematopoietic stem cell transplantation (HSCT) is an effective treatment for blood disorders and autoimmune diseases. Enhancing hematopoietic stem and progenitor cell (HSPC) frequency and function will shorten engraftment time and lower risk factors in myeloablated patients. 16,16- dimethyl prostaglandinE2 (dmPGE2) was previously identified in a chemical screen and found to regulate HSC development during zebrafish embryogenesis43. The effect of dmPGE2 on hematopoiesis is conserved in mammals. Several groups have demonstrated that ex vivo treatment of dmPGE2 confers an advantage on mouse marrow cells in competitive repopulation studies43,46. Treatment of human CD34+ cells with dmPGE2 prior to transplantation into immunodeficient mice results in increased chimerism in the peripheral blood and bone marrow of recipients48. Long-term transplantation studies have been carried out to analyze the repopulation kinetics of marrow cells following pulse exposure to dmPGE2. Recipient mice have normal blood counts and bone marrow histology through quinary transplants, with an absence of lineage bias; demonstrating that pulse treatment of dmPGE2 does not lead to a cancerous phenotype108. In addition, bone marrow cells pre-treated with dmPGE2 are protected from hematopoietic injury when exposed to chemotherapy reagents. Treated mice display accelerated recovery of hematopoiesis following total body irradiation109. These data suggest ex vivo treatment with dmPGE2 as a potential clinical therapy to enhance HSPC engraftment following transplantation. A recent phase 1 trial assessed the potential of dmPGE2 treatment to increase the effective dose of HSCs for umbilical cord blood (UCB) transplantation49. Although transplanting cord blood reduces the stringency for HLA matching, the concentration of

63

HSPCs per cord blood unit (CBU) is too low for successful engraftment in adult recipients. The current standard for UCB transplantation is to infuse a patient with 2 CBU. To determine whether dmPGE2 treatment improves engraftment of cells following UCB transplantation, patients were infused with one dmPGE2-treated CBU and one untreated CBU. Ex vivo pulse with dmPGE2 enhances the engraftment of CBU in patients as determined by the median time to neutrophil engraftment (17.5 days vs. 21 days). Chimerism analysis demonstrated that the dmPGE2-treated cord preferentially engrafted in >80% of patients. Historically, there is a 50% chance for either cord unit to reconstitute a patients’ hematopoietic system. These preliminary results have exciting therapeutic implications. It may be possible to eliminate the need for a second CBU, which would reduce strain on current resources. Shortening engraftment time will also reduce the risk of infection to myeloablated patients. Although UCB has emerged as a valuable source of HSPCs for transplantation, there are limitations for its use in adult patients. Recently, studies have aimed to chemically manipulate these cells ex vivo with the goal of expanding HSPCs prior to transplant, or to enhance homing and engraftment of these cells following transplantation. Improving either of these aspects would allow more patients to benefit from the advantages of cord blood transplantation. Using small molecules to treat entire CBUs can be somewhat complex; cord blood cells are a heterogeneous population that contains not only HSPCs but also mesenchymal stem cells, endothelial progenitors, and naïve immune cells110. In addition to its effects on HSPCs, dmPGE2 has immunomodulatory effects and promotes the survival of naïve UCB T cells through the Wnt/!-catenin pathway111. This protective effect could enhance recipient T-cell chimerism following

64

transplant. This has important clinical implications since delayed immunological reconstitution is a huge risk factor for transplant patients. The initial chemical screen in zebrafish that identified PGE2 as a regulator of HSPCs showed an increase in runx1 and cmyb positive cells, a heterogeneous population that includes HSCs, progenitors, and differentiated myeloid cells43. Follow-up studies in mice demonstrated that dmPGE2 enhances the homing and transplantation of murine LSK cells and human CD34+ cells, both heterogeneous populations46,48. Serial transplants in mice suggest that PGE2 is acting on a long-term repopulating cell, yet treated LT-HSCs do not demonstrate enhanced repopulation compared to control cells108. Consistent with these results, mice treated in vivo with dmPGE2 show no change in LT-HSC frequency, but display an increase of ST-HSCs/MPPs. These results suggest that a population of cells distinct from LT-HSCs mediate the effects of PGE2. ST-HSCs/MPPs are functionally defined by their multipotent but transient repopulation capacity27. Transplantation of ST-HSCs/MPPs from mice that are continuously treated with dmPGE2 into irradiated recipients results in enhanced shortterm, but not long-term reconstitution, most likely due to exhaustion. In contrast, we have found that ST-HSCs treated with an ex vivo pulse of dmPGE2 enhance long-term reconstitution for up to 1-year post transplant compared to control cells. Treated STHSCs maintain improved function in secondary transplants without additional exposure to dmPGE2. We have found through microarray and ChIP-seq analysis in human CD34+ cells that dmPGE2 induces gene expression that is associated with quiescence. Stimulation of this gene signature may prevent exhaustion of ST-HSCs. Our results show that dmPGE2 enhances the long-term potency of ST-HSCs, which were previously

65

thought to have transient reconstitution capabilities. Identifying the subpopulation of HSPCs that is responsive to dmPGE2 has important therapeutic implications and may help to improve pre-transplantation expansion protocols.

Materials and Methods Zebrafish maintenance and lines Zebrafish and mice were maintained in accordance with Animal Research Guidelines at Children's Hospital Boston. Embryo chemical treatments were performed in 1XE3 fish water.

HSPC population FACS sorting HSPC populations were sorted from freshly isolated WBM cells using a FACS ARIA (BD Biosystems) using the following antibodies: Lineage- Ter119, CD4, CD8, B220, Mac1, Gr-1 (PE-cy5) Sca1 (PE) cKit (APC) CD48 (Pacific Blue) CD34 (FITC) CD150 (PE-cy7). The following marker combinations were used to differentiate LTHSCs (LSKCD48-CD34-CD150+), ST-HSCs (LSKCD48-CD34+CD150+), and MPPs (LSKCD48-CD34+CD150-)

Adult zebrafish repopulation assays WKM cells from 3-month donor fish were isolated and sorted by FACS (Runx1+ cell transplants) or counted using a hemocytometer (WKM transplants). Donor cells were transplanted into irradiated casper recipients (30gy, split dose) along with untreated

66

helper marrow. Transplantation procedures were performed as previously described82. At 3 months post-transplant, WKM from recipient fish was collected and analyzed by FACS to detect chimerism levels of mCherry and GFP-positive cells in the marrow or by fluorescence microscopy. WBM repopulation assays Recipient mice (C57bl/6 CD45.2) are irradiated with 8Gy the day before transplantation. Bone marrow cells are freshly isolated from donor mice (SJL CD45.1). Spin down harvested marrow cells and resuspend cells in DMEM +2%FBS for treatment. Marrow cells are treated with 10uM dmPGE2 or DMSO for 2 hours on ice. Treated cells are competed against freshly isolated CD45.2 cells. For sorted transplants, HSPC populations are FACS-sorted as described above. Individual populations are then treated with 10uM dmPGE2 or DMSO for 2 hours on ice and transplanted along with helper marrow (200K or 400K). For 5-FU transplants, donor mice were treated with 5 FU or PBS as previously described112. Marrow was harvested two days later and treated with 10uM dmPGE2 or DMSO for 2 hours on ice and transplanted along with 200K helper marrow cells.

PB chimerism analysis Peripheral blood was isolated from recipient mice by RO bleed every 4 weeks following transplantation. Red blood cells were lysed and white blood cells were analyzed using the following antibody combinations for multilineage engraftment: Ter119 (PE-cy5) Mac1 (Alexa680) Gr-1 (PE-cy7) B220 (Pacific blue) CD3 (APC) CD45.1(PE) and CD45.2 (FITC).

67

pCREB ChIP-sequence Human CD34+ isolated cord blood cells were treated with 10uM dmPGE2 or DMSO for 2 hours at 37oC. ChIP was performed with a pCREB antibody from Santa Cruz (sc-7978) as previously described113.

Results PGE2 increases runx1+ HSPCs in embryo and adult zebrafish Treatment with dmPGE2 was previously shown to increase the number of runx1/cmyb+ cells in zebrafish embryos by in situ hybridization. We tested the effect of dmPGE2 on a runx1:GFP transgenic line that labels HSPCs. Embryos were treated from 3 somite stage (ss) to 36 hpf with 50µM dmPGE2 or DMSO as control. The chemical was washed off at 36 hpf and embryos were imaged using confocal microscopy at 60 hpf. Consistent with previous results, we observed an increase of runx1+ GFP cells in the caudal hematopoietic tissue (CHT) of dmPGE2-treated embryos compared to controls (Figure 15). Given that the increase of runx1+ cells was still present a full day after dmPGE2 was removed, we next looked at these cells in the whole kidney marrow (WKM) of adult fish. Runx1:GFP fish were treated with dmPGE2 or DMSO from 3ss-36hpf, at which point chemical was removed and embryos were grown to adulthood. FACS

68

analysis of WKM at 3 months revealed the number of runx1+ HSPCs was increased in fish previously treated with dmPGE2 (Figure 16). This indicates that the enhancement of HSPCs by dmPGE2 during embryogenesis is maintained through development to adulthood.

Figure 15: dmPGE2 increases Runx1+ cells in the CHT Zebrafish embryos were treated from 3ss-36hpf with 50uM dmPGE2 (bottom panel) or DMSO (top panel). The cartoon image demonstrates the region of the CHT depicted in the confocal image at 60hpf. HSPCs are labeled with GFP, endothelial cells are labeled with dsRed.

69

Figure 16: dmPGE2 increases Runx1+ cells in the adult kidney marrow Zebrafish embryos were treated from 3ss-36hpf with 50uM dmPGE2 or DMSO and then chemical was removed. Fish were grown up to adulthood and kidney marrow was dissected and FACS-analyzed for Runx1+:GFP cells.

Short-term exposure of PGE2 enhances long-term repopulation of the HSPC pool Our initial data from zebrafish and previous work by others in the murine system demonstrated that a short pulse of dmPGE2 leads to long-term effects on HSPCs43,46,108. We confirmed this in the mouse using competitive transplantation assays. Primary competitive transplants were performed using freshly isolated CD45.1 cells treated with dmPGE2 or DMSO. Treated cells were competed against fresh CD45.2 competitor marrow. FACS analysis on peripheral blood (PB) demonstrated robust chimerism in mice that received dmPGE2-treated cells compared to mice receiving cells treated with

70

DMSO. Recipients in the dmPGE2 group at 1-year post transplant maintain increased PB chimerism without lineage skewing (Figure 17). To test the robustness of treatment with dmPGE2 on repopulation capability, we performed competitive secondary transplants. WBM from primary recipient mice in each group was harvested at 1-year post transplant and FACS-sorted to isolate CD45.1 positive cells. These cells were transplanted in the presence of fresh CD45.2 competitor cells without additional dmPGE2 or DMSO treatment. PB chimerism analysis at 24 weeks post transplant indicates that dmPGE2treated cells were still able to out-compete control cells in secondary transplantation assays without disrupting differentiation status (Figure 18). Collectively, our results demonstrate that just a brief (2-hour) ex vivo exposure of WBM cells to dmPGE2 confers long-term repopulation advantage that is maintained through multiple rounds of transplantation. We next wanted to examine if dmPGE2 treatment months prior to transplantation would enhance the repopulation capacity of HSPCs. Since this type of experiment would be technically challenging in mice, we utilized the zebrafish to treat embryos during HSPC specification. We tested the ability of pulse treatment with dmPGE2 during zebrafish embryogenesis to confer a transplantation advantage on adult WKM cells. !-actin:GFP embryos were treated with dmPGE2 or DMSO. Following this treatment, the chemical was washed off and fish were left to grow to adulthood. After 3 months, WKM was harvested and prepared for transplantation into irradiated recipient fish. 20,000 WKM cells from each treated fish were competitively transplanted against 80,000 cells from untreated RedGlo fish into casper recipients (Figure 19a).

71

Figure 17: dmPGE2 increases long-term repopulation in primary recipients Analysis of PB chimerism was performed 1-year post transplantation in mice that received DMSO-treated cells (filled circles) or dmPGE2-treated cells (open circles). Multilineage reconstitution was determined by assessing B cells, myeloid cells, and T cells. Each circle represents an individual recipient.

Figure 18: dmPGE2 increases long-term repopulation in secondary recipients Analysis of PB chimerism was performed 20 weeks post secondary transplantation in mice that received DMSO-treated cells (filled circles) or dmPGE2-treated cells (open circles). Multilineage reconstitution was determined by assessing B cells, myeloid cells, and T cells. Each circle represents an individual recipient.

72

Figure 19a: Schematic of adult WKM transplant Zebrafish embryos were treated with DMSO or dmPGE2 from 3ss-36hpf. Chemical was removed and fish were grown to adulthood. Marrow from treated fish was transplanted into casper recipients.

Figure 19b: dmPGE2 treatment during embryogenesis enhances adult transplantation Each column represents recipients that received marrow from a single donor fish previously treated with DMSO (filled circles) or dmPGE2 (open circles). Each dot represents an individual recipient fish. The Y axis represents the ratio of green (treated) cells versus red (competitor) cells in recipient kidney marrow.

73

Four months following transplantation, recipient fish were imaged using a fluorescent microscope and the ratio of green to red cells (treated: untreated) in each kidney analyzed using ImageJ software. We observed that transient treatment with dmPGE2 during embryogenesis resulted in increased transplantation potential in adults (Figure 19b). We next tested this transplantation assay using runx1:GFP fish to examine at the effect of dmPGE2 on HSPCs specifically. Interestingly, when equal numbers of runx1+ cells were competitively transplanted into recipient fish, the advantage of dmPGE2 treatment was lost. This suggests that in vivo treatment of zebrafish with dmPGE2 results in a long-term repopulation advantage due to an increase in the runx1+ pool, rather than a change in inherent cell competitiveness. These results are consistent with murine transplantation experiments in which equal numbers of LSK SLAM cells from primary recipients mice did not demonstrate an advantage of dmPGE2 treatment in secondary transplants.

PGE2 preferentially affects ST-HSCs Previous transplantation studies using dmPGE2 have been performed on fairly heterogeneous cells populations, including runx1+ cells in zebrafish, WBM or LSK cells in mice, and CD34+ human cells. To differentiate the effects of dmPGE2 treatment on the repopulation potential of individual stem and progenitor cell populations, we performed competitive transplants using sorted mouse marrow. WBM was FACS-sorted into three populations that we will refer to as MPPs (LSKCD150-CD34+), ST-HSCs (LSKCD150+CD34+), and LT-HSCs (LSKCD150+CD34-). Cells were treated with dmPGE2 or DMSO and transplanted in various doses against competitor cells (outlined in Figure 20). Interestingly, sorted LT-HSCs or MPPs treated with dmPGE2 lost their

74

competitive advantage in transplantation assays. However, ST HSCs treated with dmPGE2 displayed increased PB chimerism in recipients that was stable 1-year post transplantation (Figure 21). In addition, when we further purify this population by selecting against CD48, the effect of dmPGE2 on repopulation is enhanced (Figure 22). Together, these data suggest that the long-term transplantation effects of dmPGE2 are mediated through the ST-HSC compartment.

Figure 20: Schematic of HSPC-sorted population transplants Whole bone marrow is harvested from donor mice and stained with FACS antibodies. Cells are sorted into LT-HSCs, ST-HSCs, or MPPs. Cells are treated ex vivo with DMSO or dmPGE2 before transplantation along with competitor marrow.

75

Figure 21: dmPGE2 treatment enhances long-term repopulation of ST-HSCs LSK cells were further purified for ST-HSCs using CD150 and CD34 (red box). Analysis of PB chimerism was performed at 4 weeks (left graph) and 24 weeks (right graph) post transplantation in mice that received DMSO-treated ST-HSCs (filled circles) or dmPGE2-treated ST-HSCs (open circles). Multilineage reconstitution was determined by assessing B cells, myeloid cells, and T cells. Each circle represents an individual recipient.

Figure 22: The effect of dmPGE2 is enhanced in a subset of ST-HSCs ST-HSCs were further purified by negative selection of CD48 (LSKCD48-CD34+CD150+). Analysis of PB chimerism was performed at 24 weeks post transplantation in mice that received DMSO-treated ST-HSCs (filled circles) or dmPGE2-treated ST-HSCs (open circles). Multilineage reconstitution was determined by assessing B cells, myeloid cells, and T cells. Each circle represents an individual recipient.

76

PGE2 may protect cycling cells from exhausting ST-HSCs are functionally distinguished from LT-HSCs by their transient ability to repopulate all lineages of the blood. This behavior reflects a key distinction in cycling status between LT-HSCs and ST-HSCs; it is thought that quiescence prevents depletion of LT-HSCs, thereby preserving long-term repopulation activity. dmPGE2 treatment has been shown to increase the percentage of cycling cells in vitro in all HSC populations, but the use of varying culture conditions and reporters of cell cycle status could be confounding factors. It is unclear how PGE2 affects cell cycle kinetics in vivo, or whether it preferentially acts on cells that are proliferating or quiescent. Given that dmPGE2treated ST-HSCs rapidly proliferate and give rise to long-term multilineage reconstitution in competitive repopulation assays, we wondered if PGE2 protects these cells from exhaustion. To test the effect of dmPGE2 on cycling cells, we performed transplants using mice treated with 5-fluorouracil (5-FU) as donors. Donor mice were injected with PBS or 5-FU such that HSCs will be forced into cycle112. WBM cells were harvested and treated with DMSO or dmPGE2 before being transplanted with competitor marrow into recipients (Figure 23). Consistent with previous results, mice that received cells from 5FU-treated donors displayed decreased chimerism compared to PBS controls at 4 weeks post transplant114 (Figure 24). Cells treated with dmPGE2 performed better in both experimental groups. Interestingly, by 12 weeks post transplant, recipient mice in the 5FU/dmPGE2 group had chimerism levels greater than PBS controls (Figure 24). These data suggest that cycling HSCs treated with dmPGE2 are able to contribute to hematopoiesis for at least 3 months following transplantation, long after control cells have depleted.

77

Figure 23: Schematic of 5-FU-treated donor transplants Whole bone marrow is harvested from donor mice that were treated with either 5-FU or PBS 2 days before transplant. Cells are treated ex vivo with DMSO or dmPGE2 before transplantation along with competitor marrow.

Figure 24: dmPGE2 enhances repopulation following 5-FU treatment Total donor marrow chimerism was analyzed in recipients of mice 4 weeks (left graph) and 12 weeks (right graph) post transplant. Donor cells were treated with DMSO (filled bars) or dmPGE2 (open bars). The x axis illustrates if donor mice were treated with PBS or 5-FU prior to transplant.

78

PGE2 induces a quiescent gene signature in human CD34+cells Microarray analysis of human UCB CD34+ cells treated with dmPGE2 revealed differentially regulated genes that play a role in cell surface receptor signal transduction, cell proliferation and migration49. This gene signature was confirmed using RNA sequencing on CD34+ cells (V. Binder, unpublished data). PGE2 robustly activates cAMP signaling in CD34+ cells through EP2 and EP4 receptors. We therefore performed ChIP-sequence analysis for activated cAMP response element-binding protein (pCREB) in human CD34+ cells. CREB is a transcription factor that binds cAMP response elements (CRE) on DNA. Using gene set enrichment analysis (GSEA), we compared genes that are up-regulated by PGE2 and bound by pCREB to published gene lists known to be involved in quiescence. Human CD34+ cells were mobilized and FACS-sorted for microarray analysis of quiescent and cycling cells115. We found a strong correlation between the genes that are up-regulated in quiescent cells and genes that are up-regulated by dmPGE2 treatment (Figure 25). In addition, close to half of the up-regulated genes in quiescent CD34+ cells are bound by pCREB after dmPGE2 treatment. A recent study in mice identified a set of genes that is down regulated in HSCs after just one cell division, which correlates with diminished repopulation potential116. GSEA shows genes that are specifically down following cell division are up-regulated when treated with PGE2 (Figure 25).

79

Figure 25: GSEA shows significant enrichment of quiescent gene expression by dmPGE2 Enrichment plots for mouse (left) and human (right) gene sets that have been previously associated with HSPC quiescence. FDR and p-value are all <0.05 and Normalized Enrichment Score (NES) = 1.4 (left) and 2.6 (right). ChIP-seq analysis shows that 33% (left) and 46% (right) of the genes in the leading edge are bound by pCREB in the presence of dmPGE2.

80

Discussion Our studies demonstrate that pulse exposure of dmPGE2 enhances long-term repopulation of HSPCs. Using a runx1:GFP zebrafish transgenic line, we have shown that treatment with dmPGE2 during development increases the number of embryonic HSPCs; an effect that is maintained through adulthood. This HSPC pool has improved engraftment and reconstitution in competitive transplantation assays, suggesting dmPGE2 increases functional HSPCs. However, this advantage is most likely attributed to an overall increase in cell number, rather than cell intrinsic memory from a signal. In contrast, mouse studies indicate that PGE2 preferentially enhances the function of STHSCs, without affecting cell number. PGE2 may induce a quiescent gene signature in this population that is normally found in more potent LT-HSCs. Variations in experimental systems could account for the different effects of dmPGE2 treatment on HSPCs. Zebrafish whole embryos are treated in vivo for a 24-hour developmental window, during which HSCs are being specified. In contrast, adult mouse marrow cells are treated ex vivo for 2 hours. Our methods of identifying HSPCs in zebrafish and mouse also differ greatly. Currently in the zebrafish, the runx1:GFP transgenic line is the most specific tool for isolating HSCs; however this population of cells most closely resembles that of LSK in the mouse. It is possible that PGE2 does preferentially affect a subset of runx1+ HSPCs that resembles the ST-HSC pool in mice, however we cannot distinguish this population with our current system. There is a distinction between ex vivo and in vivo treatment of dmPGE2 in mice. Prolonged in vivo exposure to dmPGE2 leads to increased proliferation and repopulation capacity of ST-HSCs/MPPs, followed by subsequent exhaustion and loss of

81

differentiation potential. Continuous exposure to dmPGE2 may induce multiple rounds of excessive proliferation from which cells cannot recover. Patients with mutations in the prostaglandin transporter gene, SLCO2A1, have elevated levels of circulating PGE2, which eventually leads to pancytopenia with bone marrow hypocellularity117. In addition, EP receptors have been shown to internalize upon constant stimulation, which would alter the response to PGE2. In contrast, ST-HSCs treated ex vivo with a brief pulse of dmPGE2 exhibit enhanced, long-term repopulation and do not deplete up to 1-year post transplant. These results have important clinical implications, as the latter treatment protocol is more amenable for human transplantation studies. Transplant recipients are at high risk for infection until they have a fully reconstituted hematopoietic system. PGE2 improves the homing and repopulation capacity of UCB donor cells, making it a valuable clinical resource. ST-HSCs/MPPs that are proliferative and actively cycling are the first cells to contribute to hematopoiesis following transplantation and are therefore critical for the initial recovery of myeloablated patients. Clonal analysis suggests that contribution from this progenitorlike pool is transient; once these cells exhaust, LT-HSCs provide long-term reconstitution (J. Sun, unpublished). PGE2 has a protective effect on cycling cells, specifically STHSCs, which may delay or prevent exhaustion and enable long-term contribution. Our studies show that PGE2 treatment improves transplantation by conferring the advantages of ST-HSCs and LT-HSCs—rapid contribution to hematopoiesis and long-term reconstitution—on ST-HSCs alone. GSEA demonstrates that genes which are up-regulated by PGE2 are enriched for genes associated with HSPC quiescence in human and mice. In addition, many of these

82

genes are targets of CREB, and are bound by pCREB following dmPGE2 treatment. Interestingly, about 60% of pCREB- bound sequences contain Runx1 binding motifs. We also observed that Runx1 target genes appear to be enriched in PGE2- up regulated gene lists. Runx1 plays a critical role in adult hematopoiesis44 and is required for HSC specification during embryogenesis43. Although not required for adult HSC self-renewal, Runx1-deficient HSCs display dysregulated cell cycle and cell survival pathways118. It will be interesting to compare the overlap of pCREB target genes with those of other transcription factors regulating HSC self-renewal and function. PGE2 may prevent exhaustion of cycling cells that are contributing to hematopoiesis by inducing a quiescent gene signature. Clonal tracking of HSCs in mice and non-human primates demonstrates that short-term clones repopulate the majority of mature blood immediately following transplantation119. After time, these short-term clones exhaust and are replaced by longterm clones that provide multi-lineage reconstitution for the duration of the organism’s life. We hypothesize that treatment with dmPGE2 extends the period of time posttransplantation that short-term clones can effectively contribute to the blood pool. A model to explain the enhanced transplantation potential of cells treated with dmPGE2 is summarized in figure 26. The improved repopulation effect of dmPGE2 on short-term HSCs may be enhanced by the additional support of long-term clones, however this has yet to be tested. Treatment of human cord blood cells with dmPGE2 induces a gene signature that is enriched in quiescent cells, presumably through activation of intracellular cAMP and associated transcription factors such as CREB and CREM. Induction of genes that are important for quiescence may support the maintenance of

83

short-term clones post-transplantation by improving engraftment potential, or preventing exhaustion of contributing cells. Additional experimentation should identify the specific genes required for this effect. Identifying a population of human cord blood cells that is specifically affected by dmPGE2 treatment will have important clinical implications. Improving expansion protocols prior to transplantation may decrease the number of required units and increase the availability of donor cords. In addition, further characterization of the specific genes induced by dmPGE2 treatment may give us additional insights into fundamental mechanisms regulating HSC self-renewal and function.
! ! ! ! ! !

84

!

!

!

!

!

! Figure 26: Summary model of dmPGE2 effect on HSC repopulation The top panel represents a normal transplantation setting after HSPCs are transplanted into irradiated recipients. Initially, short-term clones give rise to all mature blood cells in the peripheral blood. Once these clones exhaust, long-term clones take over and reconstitute the hematopoietic system. The bottom panel shows that when HSPCs are treated with dmPGE2, short-term clones give rise to multilineage reconstitution for an extended time post-transplantation. This effect is potentially due to an induction of genes associated with quiescence in these ST cells.

!

85

! ! ! !

Chapter 4

Concluding discussion and future directions

Concluding Discussion:
Generation of a specific HSPC transgenic zebrafish line We have developed a novel HSPC-specific transgenic zebrafish using a regulatory element that is +23 kb downstream of the ATG in the first promoter of mouse Runx1 to drive expression of a reporter79. Time-lapse live imaging showed that Runx1+ cells emerge from the hemogenic endothelium of the dorsal aorta, enter circulation, and seed the CHT, where definitive hematopoiesis first occurs. FACS analysis and cytospins confirmed that these cells are spherical with a high nucleus to cytoplasmic ratio, which is consistent with the morphology of murine HSCs. We developed an embryonic transplantation assay to calculate the frequency of HSCs in Runx1:GFP embryos. Embryonic limit dilution analysis estimated the stem cell frequency of the Runx1+ population in the 3 dpf embryo to be approximately 1/2.88 cells. A number of studies have shown that hematopoietic stem cells isolated from different tissues of the mouse embryo have the capacity to reconstitute hematopoiesis in the adult84,90,91. However, technical limitations in mouse have made it difficult to assess the stem cell potential of an HSPC transplanted from one embryo to another. Our studies are the first to document the functional capacity of embryonic HSPCs in an embryo transplant. FACS analysis of adult kidney marrow confirmed that Runx1+ cells go on to seed the adult niche. These cells give rise to long-term multilineage reconstitution when transplanted into irradiated recipients. This is consistent with studies in mice that demonstrate Runx1+ fetal liver cells successfully transplant into adult mice. Through adult limit dilution analysis of Runx1+ cells, we calculated a stem cell frequency of approximately 1/35. This is most likely an underestimate as zebrafish donors and

87

recipients are non-isogenic and are therefore not immune matched, which leads to immune rejection of some donor cells. Currently there are no antibodies available in zebrafish to further purify this population, or to distinguish between a stem cell and a progenitor. Our results demonstrate that the HSPC pool in zebrafish can be sorted with a single transgenic marker to a purity that is within the range of the well-characterized LSK (Lin-Sca1+cKit+) population in mouse89. Our characterization of the Runx1 transgenic line, including a novel embryo transplantation assay and limiting dilution transplantation in adults, demonstrates that Runx1+ cells mark a highly purified HSPC population with functional stem cell characteristics.

Live imaging reveals a novel endothelial cell behavior during stem cell colonization We performed time-lapse live imaging of Runx1:GFP embryos and directly observed distinct steps of HSPC engraftment, including a novel cellular behavior of remodeling endothelial cells. Upon entering circulation from the dorsal aorta, Runx1+ cells adhered to endothelial vessel walls in the CHT and underwent rapid extravasation to the abluminal side of the endothelial wall. Once an HSPC lodged in the CHT, a small group of endothelial cells remodeled around the stem cell to form a closely associated niche. This niche formation event was typically completed in 30-60 minutes and was marked by a single HSPC surrounded by a group of 5-6 endothelial cells. Through live imaging, we found that endothelial cell niche remodeling is conserved in fetal liver explants. Ly6a-GFP+/c-kit+ HSPCs underwent discrete steps towards lodgement: adherence, extravasation, abluminal migration, and endothelial niche remodeling. 3D reconstruction of high-resolution serial section electron microscopy scans

88

confirmed that HSPCs lodged in the CHT are surrounded by 5-6 endothelial cells. There are also stromal cells in proximity to the HSPC, which is consistent with our confocal microscopy of Runx1+ HSPC and cxcl12a+ stromal transgenic lines. Cell types not observed by confocal microscopy, including a fibroblastic mesenchymal cell with melanophore inclusions, were found together with the HSPC, endothelial cells and stromal cells. Melanophore inclusions are indicative of neural crest origin, and such cells have been found in the adult marrow71,120,121. HSPC-endothelial cell niches were heterogeneous, but the formation of niches was specific to more primitive HSCs. Our high-resolution reconstruction of a zebrafish HSPC niche recapitulates many features of the perivascular niche in mice, which contains a diverse set of cell types including mesenchymal stem and progenitor cells, endothelial cells, and nerves. An advantage of our system is the ability to observe dynamic interactions of these cell types in vivo. After remodeling of endothelial cells, HSPCs made one of three cell fate decisions: it remained quiescent, underwent symmetrical division, or it divided asymmetrically where one daughter cell exited the endothelial niche. Similarly, we observed in fetal liver explants that Ly6a-GFP+/c-kit+ HSPCs underwent cell divisions where the daughter cell proximal to the sinusoid remained surrounded by endothelial cells, while the daughter cell distal to the sinusoid migrated away into the abluminal space. Zebrafish CHT endothelial cells do not directly differentiate into budding HSPCs, as is seen in the hemogenic endothelium of the dorsal aorta. This is consistent with the model that the CHT is equivalent to fetal liver and is not a site of de novo stem cell production122 but instead must be seeded by HSPC arriving through circulation123. Our

89

studies provide dynamic observations of HSPC niche interactions during development that reveal a novel endothelial cell remodeling.

Chemical genetics reveals small molecule regulators of CHT niche colonization Treatment of zebrafish embryos with AMD3100 reduced the number of HSPCs in the CHT. This is consistent with the role of CXCR4-CXCL12 signaling during adult homing and engraftment and served as a proof-of-principle for the screen. A chemical screen identified 147 compounds that regulate CHT hematopoiesis. We combined selected chemical hits together with our live imaging assay to associate specific signals with the distinct engraftment steps we previously described. One screen hit, a selective inhibitor of TGF-! type I receptors, SB-431542, increased CHT hematopoiesis via increased frequency of HSPC cell divisions. HSPCs arrived normally in the CHT and adhered to the vessel wall, but before each cell engrafted, they underwent excessive rounds of division. These data illustrate that inhibition of TGF-! receptor signaling expands the HSPC population by causing extra cell divisions. This is consistent with previous studies for the role of TGF-! in quiescence that show knock down of TGF-! in mice results in excessive HSC proliferation95. Another hit from the screen DMOG, which stabilizes HIF-1# under normoxic conditions124, increased the number of runx1+ cells in the CHT by altering HSPC localization. HSPCs in DMOG-treated embryos extended protrusions into the abluminal space and preferentially migrated to those regions. The HIF-1# inhibitor YC-1 also increased HSPCs in the CHT, but these cells were retained in the vessel lumen. Our data suggest HIF-1# stability and a hypoxic state promotes HSPC migration into abluminal

90

spaces within the niche. HIF-1# is expressed in the zebrafish CHT (Rojas et al., 2007). Consistent with our results, HIF-1# is an important HSPC maintenance factor in mammals. HIF-1#-deficient HSCs exhibit excessive proliferation, and overstabilization of HIF-1# reduces the transplantation potential of HSCs125. We identified lycorine, a natural alkaloid extracted from the Amaryllidaceae plant family, as a novel regulator of hematopoiesis. Treatment with lycorine increased CHT hematopoiesis by retaining HSPCs in the niche. Time-lapse imaging confirmed that lycorine treatment increased the number of HSPCs that resided in endothelial cell niches, as well as the total duration each HSPC spent in the CHT. The increased number of HSPCs observed in the CHT is maintained after treatment. Together, these results demonstrate that lycorine treatment modulates the number of endothelial niches that are triggered upon HSPC arrival in the CHT, and also the time that HSPC stay within these niches. Several chemicals that regulate the sphingosine-1-phosphate (S1P) pathway induced alterations in the migration pattern of HSPCs. Consistent with our results, S1P is an important signal in adult HSPC and lymphocyte trafficking126. Embryos injected with s1pr1 morpholino had reduced numbers of HSPCs in the CHT. S1PR1deficient HSPCs were unable to extravasate into the CHT after adherence to the endothelial wall. These experiments support a cell autonomous role for s1pr1 in HSPC extravasation during colonization of the CHT. We and others106,107 have demonstrated that combination of S1P agonists and CXCR4 antagonists facilitates HSC mobilization from the niche in a G-CSF-independent manner. These results suggest that the current standard of G-CSF mobilization in patients could be improved by manipulation of the

91

S1P/CXCR4 signaling axis. Our chemical screen demonstrates the ability to modulate HSPC- endothelial cell interactions, at several key steps, and will likely have an important impact on clinical stem cell transplantation.

Short-term exposure of PGE2 enhances long-term repopulation of the HSPC pool A similar chemical screen identified PGE2 as a regulator of HSPCs in the AGM. Zebrafish embryos treated with dmPGE2 during embryogenesis displayed an increase of Runx1+ cells in the AGM, CHT, and adult kidney marrow, without additional treatment. Whole kidney marrow from treated fish demonstrated improved repopulation compared to control cells in competitive transplantation assays. However, Runx1+ cells did not demonstrate an advantage during transplantation on a cell-by-cell basis. These results are consistent with murine transplantation experiments in which equal numbers of LSK SLAM cells from primary recipients mice did not demonstrate an advantage of dmPGE2 treatment in secondary transplants108. Our data suggest that zebrafish treated in vivo with dmPGE2 results in a long-term repopulation advantage due to an increase in the Runx1+ pool, rather than a change in inherent cell competitiveness. Mouse marrow cells treated ex vivo with a short pulse of dmPGE2 displayed improved repopulation capabilities compared to control cells in primary transplants. Increased peripheral blood chimerism levels were maintained in recipients up to 1-year post transplant. Treated cells continued to demonstrate a transplantation advantage in secondary recipients without additional dmPGE2 treatment. This is consistent with previous transplantation studies that show a single pulse with dmPGE2 improves the repopulation capacity of HSPCs through quinary transplants46,108. Collectively, our results demonstrate that just a brief (2-hour) ex vivo

92

exposure of WBM cells to dmPGE2 confers long-term repopulation advantage that is maintained through multiple rounds of transplantation, however this is not due to improvement in inherent cell function of LT-HSCs.

PGE2 preferentially affects ST-HSCs and prevents exhaustion of cycling cells We found that treatment with dmPGE2 enhanced long-term repopulation by modulating a population of ST-HSCs that are phenotypically LSKCD34+CD150+. FACS-sorted ST-HSCs treated with dmPGE2 displayed improved reconstitution in recipients that was stable 1-year post transplantation, and was maintained through secondary transplantation. This advantage was specific to the ST-HSC compartment; there was no improvement in repopulation of dmPGE2-treated LT-HSCs or MPPs. The effect of dmPGE2 on the repopulation capacity of ST-HSCs was significantly enhanced following negative selection of CD48 expression. CD48+ cells account for close to 30% of LSKCD34+CD150+ cells. This illustrates the heterogeneity of purified HSPCs and suggests that PGE2 affects a small subset of the ST-HSC population. A study by the Calvi lab demonstrated that mice continuously treated in vivo with dmPGE2 displayed specific expansion of the ST-HSC/MPP pool47. Sorted STHSCs/MPPs from treated mice quickly lost repopulation potential following transplantation, suggesting these cells exhausted. It has been previously described that LSKCD34+ cells have only transient multilineage potential that is lost by ~12 weeks-post transplant27. In contrast, we demonstrate that ex vivo pulse treatment with dmPGE2 confers long-term repopulation capabilities on ST-HSCs that are maintained through serial transplantation.

93

We performed transplants with mice treated with 5-FU as donors to test the effect of dmPGE2 on cycling cells. Previous work showed that dmPGE2 treatment increased the percentage of cycling cells in vitro in all HSC populations46. It is not known how PGE2 affects cell cycle kinetics in vivo, or whether it preferentially acts on cells that are proliferating or quiescent. Cells treated with dmPGE2 performed better in both experimental groups (5-FU or PBS). Mice that received cells from 5-FU-treated donors displayed decreased chimerism levels, irrespective of treatment group, compared to PBS controls at 4 weeks-post transplant. This is consistent with previous studies that demonstrate the impaired repopulation potential of 5-FU-treated cells114. Recipient mice in the 5-FU/dmPGE2 group showed higher chimerism levels than PBS controls at 12 weeks-post transplant. These data suggest that cycling HSCs treated with dmPGE2 robustly contribute to hematopoiesis for at least 3 months following transplantation, long after control cells have depleted. Collectively, our data suggest that ex vivo pulse treatment with dmPGE2 prevents exhaustion of ST-HSCs and facilitates long-term reconstitution following transplant.

PGE2 induces a quiescent gene signature in human CD34+cells We found a strong correlation between genes that are up-regulated in quiescent HSCs and genes that are up-regulated by dmPGE2 in human CD34+ cells. RNA sequencing of human CD34+ cells treated with dmPGE2 revealed an induction of genes involved in cell surface receptor signal transduction (FOS, PTGER4, S1PR1), cell proliferation (TGFB2, PTGS2), and migration (CXCR4, S1PR1). We find a high degree of overlap with published microarray gene lists of dmPGE2-treated human CD34+ cells.

94

PGE2 robustly activates cAMP signaling in CD34+ cells through EP2 and EP4 receptors48. We identified a subset of genes that are bound by activated cAMP response element-binding protein (pCREB) following dmPGE2 treatment in human CD34+ cells. GSEA demonstrated a strong correlation between previously defined genes that are upregulated in quiescent cells, and genes that are differentially expressed and bound by pCREB following dmPGE2 treatment. A recent study developed a novel label retention Tet-off system in mice to monitor the cell division history of HSPCs116. In their system, HSPCs take up H2B:GFP until Dox is administered. Following Dox chase, label retention is assessed to determine the cycle history of HSPCs. HSCs that have previously undergone cell division have greatly impaired repopulation capabilities compared to cells that were previously quiescent. The authors identified a set of genes that are significantly down regulated following HSC division. GSEA demonstrated these genes are enriched in the gene signature induced by dmPGE2 treatment in human CD34+ cells. Our data suggest that PGE2 induces a gene signature associated with quiescent cells that have high repopulation potential. This is a potential mechanism by which PGE2 prevents exhaustion in cycling cells, such as the ST-HSC population.

95

Concluding Remarks
In this thesis, we have studied and characterized the endogenous microenvironment and trafficking of zebrafish HSPCs in the niche. We developed a novel zebrafish transgenic reporter line that enabled us to visualize and purify HSPCs. This line was used to observe a novel and essential cellular behavior that involves triggered remodeling of perivascular endothelial cells upon arrival of an HSPC in a new site of hematopoiesis. Live imaging of fetal liver explants confirmed that this endothelial remodeling effect is conserved in mammals. Using runx1:GFP zebrafish, we observed a long-term effect of PGE2 on hematopoiesis that is conserved in mammals; WBM cells treated with PGE2 maintain increased chimerism levels in recipient mice over 1-year post transplantation and display enhanced transplantability in competitive secondary transplantations without additional PGE2 treatment. We demonstrate that PGE2 affects a population of HSPCs that was not previously thought to have long-term repopulation potential. Gene expression data and ChIP-seq analysis in human CD34+ cells suggests that PGE2 may protect cycling cells from exhaustion by inducing a quiescent gene signature in these cells. Our studies provide a glimpse into the dynamic interactions between HSPCs and their endogenous niche. In addition, we show that a population of short-term HSCs can engraft and give rise to long-term multilineage reconstitution following dmPGE2 treatment. Collectively, we have gained novel insights in the pathways involved in HSC migration, homing, and repopulation.

96

Future Directions
Improving HSC purification in zebrafish We described a novel transgenic zebrafish that we could use to follow HSPCs during migration. Adult limit dilution analysis estimated a stem cell frequency of approximately 1/35. This is a dramatic enhancement of purity over previously identified markers of zebrafish HSPCs such as cmyb and CD41. However, this level of purity could most likely be improved upon, especially considering that murine HSCs can be isolated at a frequency closer to 1/2. CD41 marks two populations of zebrafish blood cells; CD41hi expressing cells are thrombocytes and CD41lo expressing cells are transplantable HSPCs127. In addition to HSPCs, Runx1 also marks a population of thrombocytes. This suggests CD41 as a potential secondary marker for further purifying zebrafish HSCs. HSPCs from Runx1:mCherry/CD41:GFP double transgenic fish could be sorted out to compare the transplantation efficiency of Runx1+CD41lo and Runx1+CD41- cells to Runx1+ cells alone. Limit dilution analysis of each population would confirm if CD41 further purifies the stem cell frequency of the Runx1+ HSPC pool. I would expect that a population of Runx1+/CD41lo cells would have a higher stem cell frequency than Runx1+ cells alone. However, it is unlikely that these two markers alone will demonstrate purity close to that of LT-HSCs in mice. We have performed microarrays on sorted embryonic Runx1+ HSPCs and Flk1+ endothelial cells, which may implicate other genes that distinguish HSCs from other cell types in the embryo. We also looked at gene expression of Runx1+ cells at different stages of development. Since we know that adult Runx1+ cells retain HSPC function, genes that are common across all time-points will be of particular interest. HSPC-

97

specific genes can first be tested in co-expression studies with Runx1+ cells, and eventually transplantation assays. Currently there are no antibodies available in zebrafish to further purify HSPCs, or to distinguish between a stem cell and a progenitor. Improved purification of HSCs combined with the strengths of the zebrafish system would be beneficial to the field, and is especially important for our niche studies, as we may uncover unique niche dynamics that are specific to different HSPC subpopulations. Finally, it will be interesting to identify the developmental stage at which definitive HSPCs are mature enough to repopulate adult recipients. Runx1+ cells could be sorted at different developmental stages and transplanted into adult recipients. This will determine whether maturation in the CHT is sufficient for HSPCs to maintain adult hematopoiesis. If HSPCs in the CHT cannot repopulate adult recipients, it is likely that further maturation is required in the kidney marrow, the next site of hematopoiesis.

Testing downstream functional consequences of niche perturbations We identified several compounds that affect endogenous HSPCs in their interaction with the niche microenvironment. However, we do not know what subsequent functional consequences these perturbations may have on HSPCs. An important follow-up study would be to elucidate the downstream effects of altering HSPC-niche interactions. Zebrafish embryos can be treated with chemicals that we previously described to affect distinct stages of engraftment, such as: SB-431542, DMOG, lycorine, and S1P agonists. During treatment, embryos could be imaged using confocal microscopy to confirm the affected phenotype. Following CHT colonization, chemical would be washed off and fish would be grown to adulthood. HSPC migration

98

and colonization of the kidney marrow could be assayed using FACS analysis for Runx1+ cells. In this assay, it is possible to observe the presence or absence of Runx1+ cells, as well as determine the number of HSPCs. To test the functionality of these cells, Runx1+ cells could be sorted out and transplanted into irradiated recipients. I would expect that a TGF-! inhibitor, like SB-431542, might lead to increased numbers of Runx1+ cells in the kidney marrow due to excessive proliferation, but that these cells would not perform as well in a functional transplantation assay. If DMOG also affects the localization of Runx1+ cells in the kidney marrow, the total number of HSPCs will be increased. Similarly, I expect that these cells will have decreased repopulation capacity since stabilization of HIF-1# in the mouse reduces transplantation potential of HSCs. Lycorine demonstrates a similar effect to PGE2; a pulse treatment results in increased numbers of Runx1+ cells long after the chemical is washed off. Based on our PGE2 results, I would expect that lycorine-treated embryos would have higher numbers of HSPCs in kidney marrow, but would not have altered repopulation potential on a cell-by-cell basis. It is also possible that chemical treatment during CHT colonization will not affect later stages of adult hematopoiesis. Instead, we can treat adult fish with these chemicals to determine if they directly play a role in mediating adult HSPC-niche interactions. In this experiment, we can also look at peripheral blood of treated fish. If any chemicals induce mobilization of HSPCs, we would detect Runx1+ cells in circulation. The combination of time-lapse live imaging and chemical genetics in zebrafish embryos allowed us to identify novel chemicals that affect different and distinct aspects of HSPC

99

migration and colonization. We can now test these chemicals on endogenous adult HSPCs and their microenvironment.

Characterizing additional cell types that interact with HSPCs in the niche Confocal microscopy of Runx1:GFP; cxcl12a:DsRed double transgenic lines demonstrated that Runx1+ HSCPs are in close proximity to stromal cells in the CHT. Therefore, it would be interesting to repeat live imaging studies with our previously identified chemicals using Runx1:GFP; cxcl12a:DsRed fish. In particular, chemicals that mediate the localization of HSPCs may modify the architecture of HSPC-stromal interactions. Future studies should also focus on visualizing how Runx1+ HSPCs interact with multiple niche cells at once. Additional transgenic lines for Flk1, cxcl12a, and nestin should be generated driving CFP or far red expression so that more cell types can be visualized. Chemical treatments of these embryos would give a more complete picture of HSPC niche dynamics. We used high-resolution serial section electron microscopy (EM) scans to better reveal cellular interactions between HSPCs and the niche. As predicted from our confocal microscopy analysis, the HSPC is surrounded by at least 5 endothelial cells with stromal cells in close proximity to the HSPC. EM revealed an additional cell type that was not detected by confocal microscopy. A fibroblastic mesenchymal cell, with what appears to be melanophore inclusions, was found together with the HSPC, endothelial cells, and stromal cells. An important future experiment is to repeat EM scans on embryos that have been treated with chemical. The benefit of this experiment is twofold: treated embryos that display the expected CHT phenotype can be pre-selected

100

before EM scanning, which will give high resolution detail of the perturbed endothelial cell remodeling event. Secondly, EM scans will provide a more complete picture of additional cell types in the niche that cannot be visualized using confocal microscopy. Lastly, analysis of more EM scans will determine whether this remodeling event is specific to Runx1+ HSPCs in the CHT. There is dynamic cellular migration and trafficking in the CHT during this stage of development. I expect that our observed endothelial remodeling event is specific to HSPCs that are lodged in the CHT, as opposed to more mature hematopoietic cells. The combination of these two microscopy tools will be a powerful assay for piecing together the endogenous microenvironment of HSPCs.

Identifying genes that are targets of PGE2 in ST-HSCs GSEA on data from RNA sequencing, microarray, and ChIP sequencing analysis identified a correlation between genes that are up-regulated and bound by pCREB in the presence of dmPGE2, and genes that are up-regulated in quiescent cells. The majority of this data is from human CD34+ cells, and therefore future work should focus on discovering specific gene targets of PGE2 in the ST-HSC population. Previous microarray experiments on sorted LT-HSCs, ST-HSCs, and MPPs treated with or without dmPGE2 were inconclusive in identifying a PGE2-induced gene signature. Microarray experiments should be repeated, with minor changes to the previous protocol. Cells were harvested, stained and sorted for different HSPC populations (LT/ST/MPP), and treated with dmPGE2 or DMSO for 2 hours on ice before RNA collection. It is possible that a 2hour treatment on ice is not conducive to observe differential gene expression. In

101

addition, cells may need a “rest” period between sorting and chemical treatment, as they might already be in an activated stress state. It is possible that our population of ST-HSCs (LSKCD48-CD34+CD150+) remains heterogeneous. If PGE2 only affects a small subset of these cells, we might not detect expression changes on whole genome microarray. Single-cell PCR experiments could be performed using a successful protocol that was developed by the Orkin lab. This would enable us to test the heterogeneity of response of sorted cell populations to dmPGE2 treatment at the single cell level. PGE2-regulated genes identified from human CD34+ cells should be tested, as well as known mouse lineage genes. Although we did not identify a dmPGE2 signature specific to ST-HSCs in microarray experiments, we discovered genes that are differentially regulated between LT-HSCs, ST-HSCs, and MPPs. It will be important to test genes from this list, as we are using a unique antibody combination to most published expression analyses. Once genes have been identified that are regulated by PGE2 in ST-HSCs, they should be tested in functional assays. If the gene list is large, morpholino knock-downs or CRISPR in fish could serve as a mini screen to test target genes. Runx1:GFP zebrafish embryos would be injected with morpholino to the gene of interest, and treated with dmPGE2. Runx1+ cells would be assayed in the AGM of 36hpf zebrafish. If a particular gene is necessary for PGE2-mediated increase of Runx1+ cells, Runx1 expression will resemble control embryos, or might be decreased. Genes that are found to inhibit the effect of dmPGE2 treatment on Runx1+ cells in the fish (or the initial gene list if it is small) should be tested in mouse transplantation assays. Sorted ST-HSCs would be infected with a single shRNA designed to target the candidate gene list. Following

102

infection, cells would be treated with dmPGE2 or DMSO and transplanted into irradiated recipients. After 24 weeks, long-term reconstitution will be assessed by PB chimerism. If a gene is specifically required for enhanced repopulation by PGE2, those recipients will display chimerism levels similar to control cells. It is possible that certain genes may be required for normal reconstitution; knockdown of these genes would show an effect in transplant of DMSO-treated cells. These experiments will identify genes that are regulated by PGE2 in ST-HSCs, and are required for its effect on HSC function.

Investigating epigenetic regulation of HSCs by PGE2 The long-term transplantation effects of a brief pulse of dmPGE2 could, in part, be mediated by epigenetic changes. It is possible that PGE2 induces a stable change in epigenetic state that enables long-term repopulation by ST-HSCs. PGE2 has been shown to affect DNA methylation patterns in the intestine and increases global DNA methylation via induction of DNMT3a in fibroblasts128. To test the effect of dmPGE2 on DNA methylation in zebrafish, Runx1+ cells from treated embryos would be sorted out at different stages following treatment. Whole-genome bisulfite sequencing could be used to identify changes in DNA methylation. Alternatively, methylation analysis could be performed solely on the previously identified gene list. It would be interesting to follow methylation patterns at different times post-treatment. Runx1+ cells would most likely acquire changes in DNA methylation on some HSC maintenance or survival genes induced by dmPGE2 treatment. This methylation pattern would still be observed in adult Runx1+ cells. In addition, DNA methylation analysis can be performed on sorted mouse marrow populations as well as human CD34+ cells. It would be interesting to compare

103

methylation changes in zebrafish, induced by in vivo treatment during embryogenesis, and mouse or human cells, which are pulse-treated ex vivo. DNMT3a regulates HSC differentiation in mice, and is highly enriched in primitive HSC populations129. Based on data from other systems, it is likely that dmPGE2 will affect the methylation status of these cells. Epigenetic mechanisms are the most likely candidate to explain the long-term effects of PGE2 on HSPC function.

Investigating PGE2 and cycling cells dmPGE2 treatment increases the percentage of cycling cells in vitro in all HSC populations, but it is not known how PGE2 affects cell cycle kinetics in vivo, or whether it preferentially acts on cells that are proliferating or quiescent. A recent study by Qui et al demonstrates that current methods of analyzing cell cycle, such as Ki67, Hoescht, and Draq5 staining only provide a snapshot of the quiescent state of a cell that is not always indicative of a cell’s cycling history116. Using an inducible, GFP label-retention transgenic mouse, the authors tracked cell division history of HSPCs. This system enables cells to be analyzed based on current quiescent state, as well as cell cycle history. Our initial transplantation experiments showed a modest effect of dmPGE2 treatment on LSKCD34+CD150+ cells that was significantly enhanced when the population was further purified using CD48 expression. Interestingly, Qui et al found that negative selection of CD48 enriches for a more dormant HSC population (0.19% of LSKCD48+CD150- cells are highly quiescent, compared to 3.37% of LSKCD48-CD150- cells). Future experiments should test if PGE2 preferentially affects quiescent or cycling cells. Using this transgenic mouse system, LSKCD48-CD34+CD150+ ST-HSCs can be

104

split into highly dormant and more active HSCs based on their levels of GFP retention. Each population would be treated with dmPGE2 or DMSO and compared in repopulation assays. Chimerism analysis would reveal if PGE2 preferentially acts on one of these cell populations, as they are phenotypically identical. To determine if PGE2 acts preferentially on cells that are actively cycling, LSKCD48-CD34+CD150+ ST-HSCs stained with Draq5 can be sorted into different cycling populations based on DNA content. These experiments will provide insight into the cell population that is affected by PGE2. Identifying a population of HSPCs that is specifically affected by dmPGE2 treatment will have important clinical implications. A phase 1 trial was recently completed to test expansion and subsequent transplantation of cord blood HSCs following ex vivo dmPGE2 exposure. Improving expansion protocols prior to transplantation may decrease the number of required units and increase the availability of donor cords.

Exploring additional expansion protocols for UCB cells Although UCB has emerged as a valuable source of HSPCs for transplantation, there are limitations for its use in adult patients. Delayed neutrophil and platelet recovery in adult recipients is attributed to low HSPC numbers in a single cord unit130. Double cord transplantation improves the overall survival and recovery time in adult patients, but has an increased risk of acute GVHD131. Recent studies have focused on developing expansion protocols, including chemical manipulation of UCB ex vivo, to expand HSPCs prior to transplant or enhance homing and engraftment of these cells following transplantation- reviewed in22. Improving either of these aspects would allow more

105

patients to benefit from the advantages of cord blood transplantation. In the future it will be important to investigate ex vivo expansion of UCB through chemical manipulation of signaling pathways, in addition to prostaglandin, that regulate HSC function. The Notch signaling pathway has previously been implicated in hematopoiesis; Notch1 is expressed in human CD34+ cells132 and overexpression of Notch1 improves the repopulation potential of these cells133. Ex vivo treatment of UCB with the notch ligand Delta1 led to expansion of HSPCs and shortened time to neutrophil engraftment in patients, although this effect was only transient134. Future work should focus on optimizing expansion protocols that manipulate notch signaling in UCB. The Wnt pathway has also been identified as an important regulator of HSPC function. Treatment of UCB with 6bromoindirubin 3'-oxime (BIO), an inhibitor of GSK-3!, expands HSPCs ex vivo, and improves the engraftment and repopulation potential of these cells in immunodeficient mice135. Interestingly, treatment with BIO leads to activation of notch in UCB cells, and may promote notch signaling during ex vivo expansion135. In addition, prostaglandin and Wnt signaling pathways interact to regulate HSC function57, suggesting that targeting multiple pathways may enhance ex vivo expansion of UCB.

!

106

References
1. Shizuru, J. a, Negrin, R. S. & Weissman, I. L. Hematopoietic stem and progenitor cells: clinical and preclinical regeneration of the hematolymphoid system. Annu. Rev. Med. 56, 509–38 (2005). Kondo, M. et al. Biology of hematopoietic stem cells and progenitors: implications for clinical application. Annu. Rev. Immunol. 21, 759–806 (2003). Reya, T. Regulation of hematopoietic stem cell self-renewal. Recent Prog. Horm. Res. 58, 283–95 (2003). Orkin, S. H. & Zon, L. I. Hematopoiesis: an evolving paradigm for stem cell biology. Cell 132, 631–44 (2008). Zhang, J. et al. Identification of the haematopoietic stem cell niche and control of the niche size. Nature 425, 0–5 (2003). Chen, A. T. & Zon, L. I. Zebrafish blood stem cells. J. Cell. Biochem. 108, 35–42 (2009). Davidson, A. J. & Zon, L. I. The “definitive” (and “primitive”) guide to zebrafish hematopoiesis. Oncogene 23, 7233–46 (2004). Davidson, A. J. et al. Cdx4 Mutants Fail To Specify Blood Progenitors and Can Be Rescued By Multiple Hox Genes. Nature 425, 300–6 (2003). De Jong, J. L. O. & Zon, L. I. Use of the zebrafish system to study primitive and definitive hematopoiesis. Annu. Rev. Genet. 39, 481–501 (2005). Traver, D. et al. Transplantation and in vivo imaging of multilineage engraftment in zebrafish bloodless mutants. Nat. Immunol. 4, 1238–46 (2003). White, R., Sessa, A., Burke, C. & Bowman, T. Transparent adult zebrafish as a tool for in vivo transplantation analysis. Cell Stem Cell 2, 183–189 (2008). Berman, J. N., Kanki, J. P. & Look, a T. Zebrafish as a model for myelopoiesis during embryogenesis. Exp. Hematol. 33, 997–1006 (2005). Scadden, D. T. The stem-cell niche as an entity of action. Nature 441, 1075–9 (2006). Morrison, S. J. & Spradling, A. C. Stem cells and niches: mechanisms that promote stem cell maintenance throughout life. Cell 132, 598–611 (2008). Calvi, L. M. et al. Osteoblastic cells regulate the haematopoietic stem cell niche. 841–846 (2003). doi:10.1038/nature02041.1. Kiel, M. J. et al. SLAM family receptors distinguish hematopoietic stem and progenitor cells and reveal endothelial niches for stem cells. Cell 121, 1109–21 (2005). Morikawa, S. et al. Prospective identification, isolation, and systemic transplantation of multipotent mesenchymal stem cells in murine bone marrow. J. Exp. Med. 206, 2483–96 (2009). Tran, E. et al. Immune targeting of fibroblast activation protein triggers recognition of multipotent bone marrow stromal cells and cachexia. J. Exp. Med. 210, 1125–35 (2013).

2.

3.

4.

5.

6. 7.

8.

9.

10.

11.

12.

13. 14.

15.

16.

17.

18.

107

19.

Roberts, E. W. et al. Depletion of stromal cells expressing fibroblast activation protein-! from skeletal muscle and bone marrow results in cachexia and anemia. J. Exp. Med. 210, 1137–51 (2013). Yao, L., Yokota, T., Xia, L., Kincade, P. W. & McEver, R. P. Bone marrow dysfunction in mice lacking the cytokine receptor gp130 in endothelial cells. Blood 106, 4093–101 (2005). Li, W., Johnson, S. a, Shelley, W. C. & Yoder, M. C. Hematopoietic stem cell repopulating ability can be maintained in vitro by some primary endothelial cells. Exp. Hematol. 32, 1226–37 (2004). Cutler, C. & Ballen, K. K. Improving outcomes in umbilical cord blood transplantation: state of the art. Blood Rev. 26, 241–6 (2012). Horn, B. & Cowan, M. J. Unresolved issues in hematopoietic stem cell transplantation for severe combined immunodeficiency: need for safer conditioning and reduced late effects. J. Allergy Clin. Immunol. 131, 1306–11 (2013). Till, J. E. & McCulloch, E. a. A direct measurement of the radiation sensitivity of normal mouse bone marrow cells. 1961. Radiat. Res. 178, AV3–7 (2012). Repopulating, H. et al. Quantitative Analysis Reveals Expansion of Human. 186, 619–624 (1997). Ikuta, K. & Weissman, I. Evidence that hematopoietic stem cells express mouse c-kit but do not depend on steel factor for their generation. PNAS 89, 1502-1506 (1992). Osawa, M., Hanada, K., Hamada, H. & Nakauchi, H. Long-term lymphohematopoietic reconstitution by a single CD34-low/negative hematopoietic stem cell. Science 273, 242–5 (1996). Kiel, M. J. et al. SLAM family receptors distinguish hematopoietic stem and progenitor cells and reveal endothelial niches for stem cells. Cell 121, 1109–1121 (2005). Kiel, M. J., Yilmaz, O. H. & Morrison, S. J. CD150- cells are transiently reconstituting multipotent progenitors with little or no stem cell activity. Blood 111, 4413–4; author reply 4414–5 (2008). Cells, H., Marrow, B., Lansdorp, B. P. M., Sutherland, H. & Eaves, C. Selective Expression of CD45 Isoforms on Functional Subpopulations of CD34 + Hemopoietic Cells from Human Bone Marrow By P.M. Lansdorp, Hj. Sutherland, and Cj. Eaves. 172, (1990). Conneally, E., Cashman, J., Petzer, a & Eaves, C. Expansion in vitro of transplantable human cord blood stem cells demonstrated using a quantitative assay of their lympho-myeloid repopulating activity in nonobese diabetic-scid/scid mice. Proc. Natl. Acad. Sci. U. S. A. 94, 9836–41 (1997). Bhatia, M., Wang, J. C., Kapp, U., Bonnet, D. & Dick, J. E. Purification of primitive human hematopoietic cells capable of repopulating immune-deficient mice. Proc. Natl. Acad. Sci. U. S. A. 94, 5320–5 (1997). Aiuti, B. A., Webb, I. J., Bleul, C. & Springer, T. New Mechanism to Explain the Mobilization of CD34 # Progenitors to Peripheral Blood. 185, (1997). Ratajczak, J. et al. Mobilization studies in mice deficient in either C3 or C3a receptor (C3aR) reveal a novel role for complement in retention of hematopoietic stem/progenitor cells in bone marrow. Blood 103, 2071–8 (2004). Wysoczynski, M. et al. Defective engraftment of C3aR-/- hematopoietic stem progenitor cells shows a novel role of the C3a-C3aR axis in bone marrow homing. Leukemia 23, 1455–61 (2009).

20.

21.

22.

23.

24.

25. 26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

108

36.

Factor-, C. S. C., Ii, K. W. C., Hangoc, G. & Hal, E. Cell surface peptidase CD26/Dipeptidylpeptidase IV regulates CXCL12/Stromal cell-derived factor-1 alpha- mediated chemotaxis of human cord blood CD34+ progenitor cells. J. Immunol. 169, 7000–7008 (2014). Ii, K. W. C., Cooper, S. & Broxmeyer, H. E. Cell surface peptidase CD26 / DPPIV mediates GCSF mobilization of mouse progenitor cells. 101, 4680–4686 (2003). Christopherson, K. W., Hangoc, G., Mantel, C. R. & Broxmeyer, H. E. Modulation of hematopoietic stem cell homing and engraftment by CD26. Science 305, 1000–3 (2004). Abkowitz, J. L., Robinson, A. E., Kale, S., Long, M. W. & Chen, J. Mobilization of hematopoietic stem cells during homeostasis and after cytokine exposure. Blood 102, 1249–53 (2003). Thomas, J., Liu, F. & Link, D. C. Mechanisms of mobilization of hematopoietic progenitors with granulocyte colony-stimulating factor. Curr. Opin. Hematol. 9, 183–9 (2002). Schoemans, H. et al. Adult umbilical cord blood transplantation: a comprehensive review. Bone Marrow Transplant. 38, 83–93 (2006). Rocha, V. & Gluckman, E. Clinical use of umbilical cord blood hematopoietic stem cells. Biol. Blood Marrow Transplant. 12, 34–41 (2006). North, T. E. et al. Prostaglandin E2 regulates vertebrate haematopoietic stem cell homeostasis. Nature 447, 1007–11 (2007). North, T. E. et al. Runx1 expression marks long-term repopulating hematopoietic stem cells in the midgestation mouse embryo. Immunity 16, 661–72 (2002). Mukouyama, Y. S. et al. Hematopoietic cells in cultures of the murine embryonic aorta-gonadmesonephros region are induced by c-Myb. Curr. Biol. 9, 833–6 (2002). Hoggatt, J., Singh, P., Sampath, J. & Pelus, L. M. Prostaglandin E2 enhances hematopoietic stem cell homing, survival, and proliferation. Blood 113, 5444–55 (2009). Frisch, B. J. et al. In vivo prostaglandin E2 treatment alters the bone marrow microenvironment and preferentially expands short-term hematopoietic stem cells. Blood 114, 4054–63 (2009). Goessling, W. et al. Prostaglandin E2 enhances human cord blood stem cell xenotransplants and shows long-term safety in preclinical nonhuman primate transplant models. Cell Stem Cell 8, 445– 58 (2011). Cutler, C. et al. Prostaglandin-modulated umbilical cord blood hematopoietic stem cell transplantation. Blood 122, 3074–81 (2013). Smyth, E. M., Grosser, T., Wang, M., Yu, Y. & FitzGerald, G. a. Prostanoids in health and disease. J. Lipid Res. 50 Suppl, S423–8 (2009). Nathan, C. Points of control in inflammation. Nature 420, 846–52 (2002). Smith, W. L., DeWitt, D. L. & Garavito, R. M. Cyclooxygenases: structural, cellular, and molecular biology. Annu. Rev. Biochem. 69, 145–82 (2000). Regan, J. W. EP2 and EP4 prostanoid receptor signaling. Life Sci. 74, 143–153 (2003).

37.

38.

39.

40.

41.

42.

43.

44.

45.

46.

47.

48.

49.

50.

51. 52.

53.

109

54.

Breyer, R. M., Bagdassarian, C. K., Scott, A. & Breyer, M. D. P ROSTANOID R ECEPTORS$: Subtypes. (2001). Hata, A. N. & Breyer, R. M. Pharmacology and signaling of prostaglandin receptors: multiple roles in inflammation and immune modulation. Pharmacol. Ther. 103, 147–66 (2004). Fujino, H., West, K. a & Regan, J. W. Phosphorylation of glycogen synthase kinase-3 and stimulation of T-cell factor signaling following activation of EP2 and EP4 prostanoid receptors by prostaglandin E2. J. Biol. Chem. 277, 2614–9 (2002). Goessling, W. et al. Genetic interaction of PGE2 and Wnt signaling regulates developmental specification of stem cells and regeneration. Cell 136, 1136–47 (2009). Ikushima, Y. M. et al. Prostaglandin E(2) regulates murine hematopoietic stem/progenitor cells directly via EP4 receptor and indirectly through mesenchymal progenitor cells. Blood 121, 1995– 2007 (2013). Bertrand, J. Y. et al. Haematopoietic stem cells derive directly from aortic endothelium during development. Nature 464, 108–11 (2010). Boisset, J.-C. et al. In vivo imaging of haematopoietic cells emerging from the mouse aortic endothelium. Nature 464, 116–120 (2010). Kissa, K. & Herbomel, P. Blood stem cells emerge from aortic endothelium by a novel type of cell transition. Nature 464, 112–5 (2010). Murayama, E. et al. Tracing Hematopoietic Precursor Migration to Successive Hematopoietic Organs during Zebrafish Development. Immunity 25, 963–975 (2006). Orkin, S. H. & Zon, L. I. Hematopoiesis: an evolving paradigm for stem cell biology. Cell 132, 631–644 (2008). Traver, D. et al. Transplantation and in vivo imaging of multilineage engraftment in zebrafish bloodless mutants. Nat. Immunol. 4, 1238–1246 (2003).

55.

56.

57.

58.

59.

60.

61.

62.

63.

64.

65.

Laird, D. J., von Andrian, U. H. & Wagers, A. J. Stem cell trafficking in tissue development, growth, and disease. Cell 132, 612–630 (2008). Nombela-Arrieta, C. et al. Quantitative imaging of haematopoietic stem and progenitor cell localization and hypoxic status in the bone marrow microenvironment. Nat. Cell Biol. (2013). doi:10.1038/ncb2730 Sugiyama, T., Kohara, H., Noda, M. & Nagasawa, T. Maintenance of the Hematopoietic Stem Cell Pool by CXCL12-CXCR4 Chemokine Signaling in Bone Marrow Stromal Cell Niches. Immunity 25, 977–988 (2006). Butler, J. M. et al. Endothelial cells are essential for the self-renewal and repopulation of Notchdependent hematopoietic stem cells. Cell Stem Cell 6, 251–264 (2010). Hooper, A. T. et al. Engraftment and reconstitution of hematopoiesis is dependent on VEGFR2mediated regeneration of sinusoidal endothelial cells. Cell Stem Cell 4, 263–274 (2009).

66.

67.

68.

69.

110

70.

Oguro, H., Ding, L. & Morrison, S. J. SLAM family markers resolve functionally distinct subpopulations of hematopoietic stem cells and multipotent progenitors. Cell Stem Cell 13, 102–16 (2013). Méndez-Ferrer, S. et al. Mesenchymal and haematopoietic stem cells form a unique bone marrow niche. Nature 466, 829–834 (2010). Köhler, A. et al. Altered cellular dynamics and endosteal location of aged early hematopoietic progenitor cells revealed by time-lapse intravital imaging in long bones. Blood 114, 290–298 (2009). Lewandowski, D. et al. In vivo cellular imaging pinpoints the role of reactive oxygen species in the early steps of adult hematopoietic reconstitution. Blood 115, 443–452 (2010). Lo Celso, C. et al. Live-animal tracking of individual haematopoietic stem/progenitor cells in their niche. Nature 457, 92–6 (2009). Takaku, T. et al. Hematopoiesis in three dimensions: human and murine bone marrow architecture visualized by confocal microscopy. Blood (2010). doi:10.1182/blood-2010-02-268466 Xie, Y. et al. Detection of functional haematopoietic stem cell niche using real-time imaging. Nature 457, 97–101 (2009). Lin, H.-F. et al. Analysis of thrombocyte development in CD41-GFP transgenic zebrafish. Blood 106, 3803–3810 (2005). Lam, E. Y. N. et al. Zebrafish runx1 promoter-EGFP transgenics mark discrete sites of definitive blood progenitors. Blood 113, 1241–1249 (2009). Nottingham, W. T. et al. Runx1-mediated hematopoietic stem-cell emergence is controlled by a Gata/Ets/SCL-regulated enhancer. Blood 110, 4188–4197 (2007). Boisset, J.-C. et al. In vivo imaging of haematopoietic cells emerging from the mouse aortic endothelium. Nature 464, 116–20 (2010). Thisse, C. & Thisse, B. High-resolution in situ hybridization to whole-mount zebrafish embryos. Nat. Protoc. 3, 59–69 (2008). Pugach, E. K., Li, P., White, R. & Zon, L. Retro-orbital injection in adult zebrafish. J. Vis. Exp. 4–6 (2009). doi:10.3791/1645 Stachura, D. L. & Traver, D. Cellular dissection of zebrafish hematopoiesis. Methods Cell Biol 101, 75–110 (2011). Kumaravelu, P. et al. Quantitative developmental anatomy of definitive haematopoietic stem cells/long-term repopulating units (HSC/RUs): role of the aorta-gonad-mesonephros (AGM) region and the yolk sac in colonisation of the mouse embryonic liver. Development 129, 4891–4899 (2002). Huang, H., Zhang, B., Hartenstein, P. A., Chen, J. N. & Lin, S. NXT2 is required for embryonic heart development in zebrafish. BMC Dev Biol 5, 7 (2005). Glass, T. J. et al. Stromal cell-derived factor-1 and hematopoietic cell homing in an adult zebrafish model of hematopoietic cell transplantation. Blood 118, 766–774 (2011).

71.

72.

73.

74.

75.

76.

77.

78.

79.

80.

81.

82.

83.

84.

85.

86.

111

87.

Ara, T. et al. Long-term hematopoietic stem cells require stromal cell-derived factor-1 for colonizing bone marrow during ontogeny. Immunity 19, 257–267 (2003). Mosimann, C. et al. Ubiquitous transgene expression and Cre-based recombination driven by the ubiquitin promoter in zebrafish. Development 138, 169–177 (2011). Osawa, M. et al. In vivo self-renewal of c-Kit+ Sca-1+ Lin(low/-) hemopoietic stem cells. J. Immunol. (Baltimore, Md 1950) 156, 3207–3214 (1996). Medvinsky, A. & Dzierzak, E. Definitive hematopoiesis is autonomously initiated by the AGM region. Cell 86, 897–906 (1996). Müller, A. M., Medvinsky, A., Strouboulis, J., Grosveld, F. & Dzierzak, E. Development of hematopoietic stem cell activity in the mouse embryo. Immunity 1, 291–301 (1994). Mironov, A. A. & Beznoussenko, G. V. Correlative microscopy: a potent tool for the study of rare or unique cellular and tissue events. J Microsc 235, 308–321 (2009). Glass, T. J. et al. Stromal cell-derived factor-1 and hematopoietic cell homing in an adult zebrafish model of hematopoietic cell transplantation. Blood 118, 766–74 (2011). Batard, P. et al. TGF-(beta)1 maintains hematopoietic immaturity by a reversible negative control of cell cycle and induces CD34 antigen up-modulation. J. Cell Sci. 113 ( Pt 3, 383–390 (2000). Larsson, J. et al. TGF-beta signaling-deficient hematopoietic stem cells have normal self-renewal and regenerative ability in vivo despite increased proliferative capacity in vitro. Blood 102, 3129– 35 (2003). Sitnicka, E., Ruscetti, F. W., Priestley, G. V, Wolf, N. S. & Bartelmez, S. H. Transforming growth factor beta 1 directly and reversibly inhibits the initial cell divisions of long-term repopulating hematopoietic stem cells. Blood 88, 82–88 (1996). Soma, T., Yu, J. M. & Dunbar, C. E. Maintenance of murine long-term repopulating stem cells in ex vivo culture is affected by modulation of transforming growth factor-beta but not macrophage inflammatory protein-1 alpha activities. Blood 87, 4561–4567 (1996). Yamazaki, Y. & Kawano, Y. Inhibitory effects of herbal alkaloids on the tumor necrosis factoralpha and nitric oxide production in lipopolysaccharide-stimulated RAW264 macrophages. Chem Pharm Bull 59, 388–391 (2011). Yamazaki, S. et al. TGF-beta as a candidate bone marrow niche signal to induce hematopoietic stem cell hibernation. Blood 113, 1250–1256 (2009). Morrison, S. J., Hemmati, H. D., Wandycz, A. M. & Weissman, I. L. The purification and characterization of fetal liver hematopoietic stem cells. Proc. Natl. Acad. Sci. U. S. A. 92, 10302– 10306 (1995). Sánchez, M. J., Holmes, A., Miles, C. & Dzierzak, E. Characterization of the first definitive hematopoietic stem cells in the AGM and liver of the mouse embryo. Immunity 5, 513–525 (1996). Ma, X. et al. Expression of the Ly-6A (Sca-1) lacZ transgene in mouse haematopoietic stem cells and embryos. Br J Haematol 116, 401–408 (2002). Ma, X., Robin, C., Ottersbach, K. & Dzierzak, E. The Ly-6A (Sca-1) GFP transgene is expressed in all adult mouse hematopoietic stem cells. Stem Cells 20, 514–521 (2002).

88.

89.

90.

91.

92.

93.

94.

95.

96.

97.

98.

99.

100.

101.

102.

103.

112

104.

Jahn, T., Leifheit, E., Gooch, S., Sindhu, S. & Weinberg, K. Lipid rafts are required for Kit survival and proliferation signals. Blood 110, 1739–1747 (2007). Yamazaki, S. et al. Cytokine signals modulated via lipid rafts mimic niche signals and induce hibernation in hematopoietic stem cells. EMBO J. 25, 3515–3523 (2006). Golan, K., Kollet, O. & Lapidot, T. Dynamic Cross Talk between S1P and CXCL12 Regulates Hematopoietic Stem Cells Migration, Development and Bone Remodeling. Pharmaceuticals (Basel). 6, 1145–69 (2013). Juarez, J. G. et al. Sphingosine-1-phosphate facilitates trafficking of hematopoietic stem cells and their mobilization by CXCR4 antagonists in mice. Blood 119, 707–16 (2012). Hoggatt, J., Mohammad, K. S., Singh, P. & Pelus, L. M. Prostaglandin E2 enhances long-term repopulation but does not permanently alter inherent stem cell competitiveness. Blood 122, 2997– 3000 (2013). Porter, R. L. et al. Prostaglandin E2 increases hematopoietic stem cell survival and accelerates hematopoietic recovery after radiation injury. Stem Cells 31, 372–83 (2013). Yu, X., Gu, Z., Wang, Y. & Wang, H. New strategies in cord blood cells transplantation. Cell Biol. Int. 37, 865–74 (2013). Li, L. et al. Prostaglandin E2 promotes survival of naive UCB T cells via the Wnt/"-catenin pathway and alters immune reconstitution after UCBT. Blood Cancer J. 4, e178 (2014). Wilson, A. et al. Hematopoietic stem cells reversibly switch from dormancy to self-renewal during homeostasis and repair. Cell 135, 1118–29 (2008). Lee, T., Johnstone, S. & Young, R. Chromatin immunprecipitation and microarray-based analysis of protein location. Nat. Protoc. 1, 729–748 (2006). Harrison, D. E. & Lerner, C. P. Most primitive hematopoietic stem cells are stimulated to cycle rapidly after treatment with 5-fluorouracil. Blood 78, 1237–40 (1991). Graham, S. M., Vass, J. K., Holyoake, T. L. & Graham, G. J. Transcriptional analysis of quiescent and proliferating CD34+ human hemopoietic cells from normal and chronic myeloid leukemia sources. Stem Cells 25, 3111–20 (2007). Qiu, J., Papatsenko, D., Niu, X., Schaniel, C. & Moore, K. Divisional History and Hematopoietic Stem Cell Function during Homeostasis. Stem Cell Reports 2, 1–18 (2014). Diggle, C. P. et al. Prostaglandin transporter mutations cause pachydermoperiostosis with myelofibrosis. Hum. Mutat. 33, 1175–81 (2012). Cai, X. et al. Runx1 loss minimally impacts long-term hematopoietic stem cells. PLoS One 6, e28430 (2011). Kim, S. et al. Dynamics of HSPC Repopulation in Nonhuman Primates Revealed by a DecadeLong Clonal-Tracking Study. Cell Stem Cell 14, 473–485 (2014). Nagoshi, N. et al. Ontogeny and multipotency of neural crest-derived stem cells in mouse bone marrow, dorsal root ganglia, and whisker pad. Cell Stem Cell 2, 392–403 (2008).

105.

106.

107.

108.

109.

110.

111.

112.

113.

114.

115.

116.

117.

118.

119.

120.

113

121.

Yamazaki, S. et al. Nonmyelinating Schwann cells maintain hematopoietic stem cell hibernation in the bone marrow niche. Cell 147, 1146–1158 (2011). Murayama, E. et al. Tracing hematopoietic precursor migration to successive hematopoietic organs during zebrafish development. Immunity 25, 963–75 (2006). JOHNSON, G. R. & MOORE, M. A. S. Role of stem cell migration in initiation of mouse foetal liver haemopoiesis. Nature 258, 726–728 (1975). Jaakkola, P. et al. Targeting of HIF-alpha to the von Hippel-Lindau ubiquitylation complex by O2regulated prolyl hydroxylation. Science 292, 468–72 (2001). Takubo, K. et al. Regulation of the HIF-1alpha level is essential for hematopoietic stem cells. Cell Stem Cell 7, 391–402 (2010). Rivera, J., Proia, R. L. & Olivera, A. The alliance of sphingosine-1-phosphate and its receptors in immunity. Nat. Rev. Immunol. 8, 753–63 (2008). Ma, D., Zhang, J., Lin, H., Italiano, J. & Handin, R. I. The identification and characterization of zebrafish hematopoietic stem cells. Blood 118, 289–97 (2011). Huang, S. K. et al. Prostaglandin E! increases fibroblast gene-specific and global DNA methylation via increased DNA methyltransferase expression. FASEB J. 26, 3703–14 (2012). Challen, G. A. et al. Dnmt3a is essential for hematopoietic stem cell differentiation. Nat. Genet. 44, 23–31 (2012). Laughlin, M. J. et al. Outcomes after transplantation of cord blood or bone marrow from unrelated donors in adults with leukemia. N. Engl. J. Med. 351, 2265–75 (2004). Ballen, K. K. et al. Double unrelated reduced-intensity umbilical cord blood transplantation in adults. Biol. Blood Marrow Transplant. 13, 82–9 (2007). Milner, B. L. A., Kopan, R., Martin, D. I. K. & Bernstein, I. D. A Human Homologue of the Drosophila Developmental Gene,. 2057–2062 (1994). Varnum-Finney, B. et al. Pluripotent, cytokine-dependent, hematopoietic stem cells are immortalized by constitutive Notch1 signaling. Nat. Med. 6, 1278–1281 (2000). Delaney, C. et al. Notch-mediated expansion of human cord blood progenitor cells capable of rapid myeloid reconstitution. Nat Med. 16, 232-6 (2010). Ko, K.H. et al. GSK-3b Inhibition Promotes Engraftment of Ex Vivo-Expanded Hematopoietic Stem Cells and Modulates Gene Expression. Stem Cells. 29, 108-118 (2010).

122.

123.

124.

125.

126.

127.

128.

129.

130.

131.

132.

133.

134.

135.

114

Appendix

Supplemental Information (Tamplin, Durand et al., 2014)

Supplemental Figure, Table and Movie Legends

Figure S1. Characterization of stable Runx:GFP and Runx:mCherry transgenic zebrafish lines. (A) Intercross of Runx:GFP and Runx:mCherry lines to show overlapping expression in the CHT (arrowheads). dpf: days post fertilization (B) Emergence of Runx:mCherry+ HSPC (red nuclei; arrowheads) in the kdrl:GFP+ endothelial cells (ECs) of the ventral wall of the dorsal aorta (DA; green). See Movie S1. hpf: hours post fertilization (C,E,G,I,K) Cross of Runx:mCherry and cd41:EGFP lines. (D,F,H,J,L) Cross of Runx:mCherry and cmyb:EGFP lines. Co-expression at stages and in embryonic sites of hematopoiesis as marked (i.e. thymus and CHT). We predict the larger number of Runx:mCherry+ cells in 3 and 5 dpf embryos are HSPC progeny. (M) Section of Runx:GFP zebrafish kidney stained with hematoxylin and eosin (H&E). (N) Immunohistochemistry of adjacent section to (M). Anti-GFP shows rare GFP+ cells in the kidney marrow (brown; arrowheads; counterstained with hematoxilyn). (O) Fluorescence-activated cell sorting (FACS) of whole kidney marrow (WKM) from adult Runx:GFP transgenic zebrafish. Runx:GFP+ cells represent on average 0.133% of the total WKM.

116

(P) Forward scatter (FSC) and side scatter (SSC) were used to distinguish characteristic WKM populations (Traver et al., 2003). Runx:GFP+ cells were found predominately in the lymphoid/HSPC gate.

Figure S2. Runx:mCherry positive cells in the embryo are functional HSPC. (A) Summary of results from embryo-to-embryo limiting dilution transplantation experiments. (B) Embryo-to-embryo transplantation recipients with engraftment of Runx:mCherry+ cells in the kidney marrow at 3 months (scored if above background; >0.001%). (C) Representative WKM FACS plots of an embryo-to-embryo transplantation recipient with Runx:mCherry+ HSPC and ubi:GFP+ lineages.

Figure S3. Time-lapse live imaging sequence of CHT colonization. See Movie S2. Selected frames of time-lapse are shown (1 frame/2 minutes). First row frames are a merge of Runx:GFP+ HSPC (green; second row) and kdrl:RFP+ ECs (red; third row in B). Times are hours:minutes post fertilization. (A) HSPC (arrowhead) extravasates by squeezing through endothelial wall. (B) ECs remodel around HSPC to form niche (broken circle). (C) After HSPC division the daughter cells undergo a 90º rotation (arrowheads). The upper cell becomes migratory and crawls out of niche. (D) A diagram summarizing the steps of HSPC lodgement in the perivascular niche. The times shown correspond to the time-lapse frames shown above. Other possible cell division decisions are shown.

117

Figure S4. Serial block face scanning EM of CHT after time-lapse. (A,B) Embryo 1 (same embryo as shown in Figure 4). Comparison of 3D projections from (A) confocal z-stack data and (B) serial EM scans to confirm the position of the lodged HSPC. (A) 3D rendered projection of z-stack (Imaris software) showing one lodged Runx:GFP+ HSPC (green; arrowhead) and kdrl:mCherry+ ECs (red). Scale bar: 20 microns. (B) Vessel lumen manually outlined, surface rendered, and shown in red (Imaris software). 3D projection overlaid on a single EM section. Rotation performed to match orientation of image in (A). Orientation: D (dorsal), V (ventral), A (anterior), P (posterior). CA (caudal artery). Circulation is towards the posterior (left; arrow). Major vessels outlined. (C-G) Embryo 2. A second independent example of serial block face EM sections on a lodged HSPC after time-lapse. (C) Last frame of CHT time-lapse (60 hpf). Arrowhead marks HSPC lodged >6 hours. Runx:GFP green; kdrl:mCherry red; brightfield blue. Anterior left, posterior right, dorsal top, ventral bottom. (D) Detail of region in (C) marked by box. The brightness of the GFP channel was adjusted independently of the mCherry channel to clearly show the position of the Runx:GFP+ HSPC. Scale bar 50 microns. (E) Single section and orthogonal slice from serial block face EM scans. Lodged HSPC is purple, surrounding EC nuclei are green and numbered. Part of the HSPC is missing because it was not captured in serial EM sections. (F,G) High resolution EM sections with cells of interest outlined as follows: EC (dark green, bright green, blue, light blue, yellow, orange); HSPC (purple); FB (fibroblastic mesenchymal cell; white); RBC (red blood cell; red). (F) Direct contact between the midsection of the pink HSPC

118

and dark green EC is visible. (G) Gaps are visible between the HSPC and surrounding cells. There are some contacts with the fibroblastic mesenchymal cell (FB; white). This FB cell does not have the melanophore inclusions observed in Embryo 1 (Figure S4G and data not shown; compare to Figure 4D). There are no stromal cells visible in this example. For 3D reconstruction see Movie S4. Scale bars: 5 microns.

Figure S5. Embryonic expression patterns of CXCR4-CXCL12 homologs in zebrafish and dose-dependency of CXCR4 antagonist AMD3100. (A) Whole mount in situ expression patterns of cxcr4a, cxcr4b, cxcl12a, and cxcl12b between 34-38 hpf. For each example a detail of the trunk and tail is shown, with the DA and caudal hematopoietic tissue (CHT), respectively, marked with dashed white boxes. All genes are expressed in the head and lateral line. cxcr4b but not cxcr4a is expressed in the CHT. cxcl12a and cxcl12b are expressed in the DA and CHT. (B) Dose-dependent decrease in cmyb/runx1 expression levels with increasing concentrations of CXCR4 antagonist AMD3100. Wild-type AB embryos were treated from 48-72 hpf and scored as those in Figure 5A,B. The percentage of total is shown.

Figure S6. Time-lapse live imaging of E11.5 fetal liver (FL) explant with labeled HSPC and ECs. (A) A time-lapse sequence from a FL lobe showing a c-kit+ HSPC (magenta) adhered to a CD31+ sinusoid (green). In ~30 minutes, the anterior of the cell protrudes (arrowhead) and extravasates to the abluminal side of the sinusoid (division marked by dotted line). Luminal versus abluminal sides of the sinusoid were confirmed in single z slices of

119

confocal stacks (data not shown). Scale bar: 10 microns. Times are hours:minutes after start of explant imaging. (B) After extravasation, the HSPC migrates on the abluminal side of the sinusoid. (C) ECs surround the HSPC as a rosette (dotted line). There is an accumulation of c-kit at the contact point between the HSPC and the ECs (compare to similar c-kit accumulation in Figure 7D,E). (D) A diagram summarizing the steps of HSPC lodgement in the FL. The times shown correspond to the time-lapse frames shown above. A morphologically symmetric division is observed in Figure 7D. Other possible cell division decisions are shown.

Table S1. Ingenuity Pathway Analysis (IPA) of microarray data from Lycorinetreated embryos. Double transgenic Runx:GFP;kdrl:RFP embryos were treated with DMSO (1%) or Lycorine (75 µM) from 48-72 hpf. Embryos from each pool were dissociated and sorted into three populations: 1) GFP+ HSPC; 2) RFP+ EC; 3) negative whole embryo. Three biological replicates were collected and analyzed. Lycorine-treated versus DMSO were compared for both HSPC and EC populations. Availabe Entrez Gene IDs were used for IPA analysis. (A) The Lycorine-treated EC population was enriched for genes associated with adhesion and extravasation. (B) The Lycorine-treated HSPC population was enriched for genes associated with adhesion and activation.

120

Movie S1. Emergence of Runx:mCherry+ HSPC from the hemogenic endothelium of the DA. Corresponds to Figure S1B. Runx:mCherry+ nuclei (red) are seen in the kdrl:GFP+ DA (green) at ~28 hpf. The cells round up and protrude on the ventral side of the DA. Time-lapse live imaging captured at 2 minutes/frame and rendered at 6 frames/second.

Movie S2. HSPC niche colonization in the zebrafish CHT reveals novel steps to engraftment. Corresponds to Figure 3A and S3. A double transgenic embryo with kdrl:RFP+ ECs (red) and Runx:GFP+ HSPC (green) imaged during colonization of the CHT (~38-43 hpf). An arriving HSPC attaches to the endothelial wall, undergoes extravasation and migration to the abluminal side, followed by endothelial remodeling into a pocket, cell division, a 90º turn, and migration of one daughter cell out of the niche. Time-lapse live imaging captured at 2 minutes/frame and rendered at 6 frames/second.

Movie S3. Reconstructed cells from TEM tomography are shown in relationship to each other in a 3D model. Corresponds to Figure 4D-F. A lodged hematopoietic progenitor stem cell (HSPC; purple), is closely associated with two putative stromal cells (light and dark blue), one of which directly contacts the HSPC. These three cells are nestled close to the endothelial cell wall of the dorsal aorta (individual cells outlined in different shades of green). Erythrocytes (red) are seen both within the dorsal aorta (as in Figure 4E,F upper), and in the sinusoids of the CHT in proximity to one of the stromal

121

cells (as in Figure 4E,F lower). Two additional fibroblastic mesenchymal cells with melanophore inclusions wrap the HSPC tightly (white and yellow).

Movie S4. Reconstructed cells from TEM tomography from a second embryo are shown in relationship to each other in a 3D model. Corresponds to Figure S4F,G (Embryo 2). A lodged HSPC (purple) is closely associated with EC (dark green, bright green, dark blue, light blue). One EC (dark green) directly contacts the HSPC. Part of an RBC (red) is seen on the luminal side of an EC. Gaps are visible between the HSPC and surrounding cells. There are some contacts with the fibroblastic mesenchymal cell (FB; white) that tightly wraps the HSPC. There are no stromal cells visible in this example.

Movie S5. 3D reconstruction of an HSPC surrounded by EC in an E11.5 FL. Corresponds to Figure 7C. A 3D rendered projection of a confocal z-stack using Volocity software. This example shows a Ly6a-GFP+/Runx1+ HSPC surrounded by VECadherin+ EC. Contacts with EC are visible on all sides of the HSPC.

Movie S6. Time-lapse live imaging detail of an E11.5 FL explant showing EC remodeling around an HSPC. Corresponds to Figure 7D. A c-kit+ HSPC (magenta) migrates into a field of CD31+ sinusoidal ECs (green). The ECs then remodel around the HSPC. Maximum projection of confocal z-stack. ImageJ used for bleach correction over time. Time-lapse live imaging captured at 2 minutes/frame and rendered at 6 frames/second. Times are hours:minutes after start of explant imaging.

122

Movie S7. Time-lapse live imaging detail of an E11.5 FL explant showing HSPC division. Corresponds to Figure 7E. A white arrowhead tracks the movement of a Ly6aGFP+ (green)/c-kit+(blue) HSPC that is adhered to CD31+ sinusoidal ECs (red). The HSPC undergoes division, with one daughter remaining surrounded by EC and the other migrating away. Note: Ly6a-GFP also marks some ECs and macrophages in the FL. Times are hours:minutes after start of explant imaging. Scale bar is 10 microns.

123

Supplemental Experimental Procedures

All animals were handled according to approved Institutional Animal Care and Use Committee (IACUC) of Boston Children's Hospital protocols.

Vectors and transgenesis

All PCR was performed using the High Fidelity Advantage 2 PCR Kit (Clontech). The Runx1 +23 enhancer (Nottingham et al., 2007) was PCR amplified from C57/BL6 mouse genomic DNA using the following primers: Forward (underlined XhoI and BamHI sites added) 5’-GGCTCGAGGGATCCGGGGTGGGAGGTGTAAGTTC-3’ 5’GGGGTGGGAGGTGTAAGTTC-3’ and Reverse (underlined BglII and NotI sites added) 5’- GGGCGGCCGCAGATCTCAGGTGTCAGCAACCCATC -3’. The PCR fragment was gel purified, XhoI/BglII digested, ligated into XhoI/BamHI digested Tol2kit (Kwan et al., 2007) #228 p5E-MCS vector and sequence verified. The mouse !globin minimal promoter was PCR amplified from C57/BL6 mouse genomic DNA using the following primers: Forward (underlined SpeI site added) 5’GGACTAGTCCAATCTGCTCAGAGAGGACA-3’ and Reverse (underlined SacII site added) 5’-GGCCGCGGGATGTCTGTTTCTGAGGTTGC-3’. The !-globin minimal promoter and Runx1+23 5’ entry vector were SpeI/SacII digested and ligated together. Multisite Gateway reactions were performed according to the Invitrogen protocol. The Runx1+23 5’ entry enhancer/minimal promoter construct was assembled with middle

124

entry vectors Tol2kit #383 pME-EGFP or Tol2kit #233 pME-NLS-mCherry, 3’ entry vector Tol2kit #302 p3E-polyA, and destination vector Tol2kit #394 pDestTol2pA2. Transgenic lines were established as previously described (Mosimann et al., 2011). At least two independent lines with 50% transmission from the F2 generation were established for each construct (Runx:GFP and Runx:mCherry).

Transgenic zebrafish lines

We are grateful to those who generously shared the following lines for this study: kdrl(flk1):GFP (specifically kdrl(flk1):GRCFP (Cross et al., 2003)); kdrl(flk1):RFP (Huang et al., 2005); cxcl12a(sdf-1a):DsRed2 (Glass et al., 2011); cmyb:EGFP (North et al., 2007); cd41:EGFP (Lin et al., 2005); ubi:EGFP and ubi:mCherry (Mosimann et al., 2011).

Imaging

Staged transgenic zebrafish embryos were selected and mounted for imaging in 1% LMP agarose with E3 media and tricaine as described (Bertrand et al., 2010). Some zebrafish embryos had 0.003% PTU (1-Phenyl-2-thiourea) added to the media to block melanogenesis. Zebrafish embryos and FL explants were imaged in MatTek glass bottom dishes or multi-well plates (No. 1.5 cover slip). Zebrafish live imaging was performed in an incubated chamber at 28°C. Mouse embryo explant live imaging was performed in an incubated chamber at 37°C with humidified CO2 with culture media (DMEM, 20% FCS,

125

glutamine, sodium pyruvate, 2-mercaptoethanol, 1% penicillin-streptomycin, recombinant mouse IL-3 (R&D Systems; final concentration 50 ng/ml)). Confocal microscopy was performed using a Yokogawa spinning disk and Nikon inverted Ti microscope. Our microscope configuration allowed imaging of multiple embryos within a 2-5 minute interval using a moving XY stage, as well as acquisition of z-stacks through the entire CHT (1-2 µm optical slices) in multiple fluorescent channels. Objectives lenses (Nikon): 20x Plan-Apo DIC N.A. 0.75; 40x Plan-Apo phase N.A. 0.95 dry; 40x Apo LWD WI NA 1.15 lambda S. Image acquisition was done with a single Andor iXon DU897 EM-CCD camera (512x512 pixels) or dual Andor iXon x3 EM-CCD cameras (512x512 pixels) and Andor iQ or NIS-elements computer software. Fixed transgenic zebrafish embryos were scanned using a Nikon C2si confocal system and NiE upright microscope. These embryos were briefly fixed for 10 minutes with PEM fixation buffer (dH20, EGTA 10 mM, MgSO4 1 mM, PIPES 100 mM, Triton X-100 0.1%, PFA 4%; Cold Spring Harbor Protocols, 2009, doi:10.1101/pdb.rec11730), then washed and mounted with DRAQ5 (1:500) for staining of nuclei.

Imaging analysis

Image processing and rendering was done using Fluorender (Wan et al., 2009), Imaris (Bitplane), NIS-elements (Nikon), Volocity (PerkinElmer) and ImageJ/Fiji (Schindelin et al., 2012). The MTrackJ plugin was used for manual cell tracking (Meijering et al., 2012). Lineage trees were created using Endrov (Henriksson et al., 2012). Point-to-point measurements were made with Imaris. Manual tracing, segmentation, and surface

126

rendering of objects was performed using Imaris. Confocal z-stack images are presented as single slices, maximum projections, or 3D rendered projections. In some cases, background subtraction was performed, and brightness and contrast was adjusted in one or more channels of a multi-channel image.

Immunohistochemistry of adult WKM

Adult zebrafish were fixed with 4% paraformaldehyde, paraffin-embedded, sectioned, and stained with hematoxylin and eosin (H&E). Immunohistochemistry was performed using anti-GFP monoclonal antibody (clone JL-8). The Dako Mouse Envision kit with EDTA antigen retrieval was used for visualization. Incubation in the primary antibody was 60 minutes, and 30 minutes in the secondary, followed by DAB development for 5 minutes and counterstained with hematoxilyn.

Adult-to-Adult HSPC Transplantation

WKM cells from 3-month Runx:mCherry;ubi:GFP fish were isolated and sorted by FACS. Double-positive cells were transplanted into irradiated casper recipients (n=20-48 recipients per cell dose) along with untreated helper marrow at the following ratios: 1:20,000; 5:20,000; 10:20,000; 25:20,000; 50:20,000. Transplantation procedures were performed as previously described (Pugach et al., 2009). At 3 months post-transplant, WKM from recipient fish was collected and analyzed by FACS to detect chimerism levels of mCherry and GFP-positive cells in the marrow. We confirmed multi-lineage

127

reconstitution by observing differentiated GFP-positive cells in multiple cell populations, as determined by forward and side scatter profiles (Traver et al., 2003). Stem cell frequency was determined using ELDA software (confidence interval=0.95; (Hu and Smyth, 2009)).

Embryo-to-Embryo HSPC Transplantation

Adapted from previously published protocols (Stachura and Traver, 2011; Traver et al., 2003). Runx:mCherry;ubi:EGFP or Runx:GFP;ubi:mCherry 3 dpf embryos were collected and chopped finely with a razor blade. Embryos were dissociated in a 1:65 dilution of Liberase TM (Roche) in PBS, incubated at 37°C for 20 minutes before addition of PBS/5% FCS to stop the reaction. Dissociated cells were passed through a 45 µM filter, spun, and resuspended in PBS/5% FCS. mCherry+/GFP+ cells were collected using a FACSAria cell sorter (BD Biosciences). Collected cells were resuspended in PBS at an estimated concentration of 400 cells/microliter with 0.5% rhodamine-dextran(10k) as a marker for injection. A microinjection needle (without filament) was back-filled with the cell suspension. The drop volume was calibrated to 1 nanoliter and each embryo was injected with 1, 2, or 4 drops. This gave an estimated cell dose of 0.4, 0.8 or 1.6 cells per embryo. Drops were injected into the sinus venosus (i.e. duct of Cuvier) of 48 hpf wildtype AB embryo recipients. Embryos were held in place using agarose injection ramps. Approximately 30 embryos were injected per dose and 12-26 embryos per group survived to adulthood (3-5 months). WKM was then analyzed for percentage of engrafted Runx+ cells using a LSR II flow cytometer (BD Biosciences). Any recipients with positive cells

128

detected above background were scored as engrafted. Flow cytometry data was analyzed using FACSDiva and FlowJo software.

Serial block face scanning electron microscopy and 3D reconstructions

Immediately at the end of live imaging time-lapse acquisition 60 hpf embryos were fixed in 2.5% glutaraldehyde and 4% paraformaldehyde in a 0.1 M sodium cacodylate buffer. Embryos were embedded in 5% low melting point agarose in 0.1M sodium cacodylate buffer for orientation. Samples were submitted to Renovo Neural Inc (Cleveland, USA) for further processing. Samples were stained with heavy metals following the protocol by Deerinck et al., 2010 (available from the National Center for Microscopy and Imaging Research; http://ncmir.ucsd.edu/sbfsem-protocol.pdf). Samples were embedded in Epon resin, and mounted onto pins (detailed protocol available from Renovo Neural). Serial blockface images (analogous to serial sectioning) were obtained using a Zeiss Sigma VP scanning electron microscope equipped with a Gatan 3View in-chamber ultramicrotome. A series of 500-1000 images were acquired at 2 kV using at 15,000 magnification from the region of interest. Image and stack resolution was 10 nm/pixel with 100 nm slices. Images were registered and resized as necessary using ImageJ/Fiji software (Schindelin et al., 2012). Images were imported into the program IMOD 4.5 (Kremer et al., 1996) then aligned and reconstructed using dual-axis tomography (Mastronarde, 1997). Cells were manually outlined and 3D reconstructions were generated. Movies were rendered using IMOD and Fiji.

129

Chemicals

Chemical screening was adapted from previously published methods in our lab (Kaufman et al., 2009; North et al., 2007; White et al., 2011). For chemical screening, stagematched AB embryos were dechorionated and arrayed 8-10/well in 96-well meshbottomed plates (Millipore). Embryos were treated with individual library chemicals (~2400 bioactives) from 48-72 hpf by placement directly in 96-well receiver plates containing small molecules diluted in E3 media + 1% DMSO (typical concentration 30 µM). After treatment and before fixation, we checked for secondary defects (e.g. stopped circulation, toxicity, developmental delay). WISH was performed as previously described (Thisse and Thisse, 2008), except 0.2% glutaraldehyde was added to 4% formaldehyde at the post-fixation step. Chemicals used in the study were as follows: AMD3100 (Sigma #A5602, optimal concentration 25 µM in dH2O, (Schols et al., 1997)); SB-431542 (Sigma #S4317, optimal concentration 40 µM in DMSO, (Inman et al., 2002)); Lycorine (Sigma #L5139, concentration between 25-75 µM in DMSO).

Microarrays

Runx:GFP;kdrl:RFP embryos were collected and raised in E3 media at 28.5C. Embryos were treated from 2-3 dpf with E3/1% DMSO or E3/1%DMOS/75 uM Lycorine (~100150 embryos per group). Embryos were dissociated and sorted as above. Three populations were collected: GFP+ HSPC (~1-2k cells/experiment), RFP+ EC (~10-20k cells/experiment), and negative cells (~100k cells/experiment; total embryo as a

130

comparator population). In total, 18 samples were collected: 3 biological replicates x 2 treatment conditions x 3 cell populations. Cells were sorted directly in Trizol LS. Trizol extraction was performed as per the manufacturer's instructions, with the addition of GenElute LPA (Sigma, Cat.#56575). Total RNA was amplified and labeled using the NuGEN Ovation Pico WTA System V2 and Encore Biotin Module kits, respectively. Affymetrix ZebGene-1_0-st microarrays were hybridized, washed, and stained using Ambion kits. Microarrays were scanned using the Genechip Scanner 3000 7G.

All ZebGene-1_0-st array data was processed using the BioConductor (Gentleman et al., 2004) package oligo (Carvalho and Irizarry, 2010). Arrays were assessed for quality with arrayQualityMetrics (Kauffmann et al., 2009) and normalized using the Robust Multichip Average (RMA, (Bolstad et al., 2003)) method at the probe level, collapsing probes into "core" transcripts based on the pd.zebgene.1.0.db annotation package (Benilton Carvalho. pd.zebgene.1.0.st: Platform Design Info for Affymetrix ZebGene-1_0-st. R package version 3.8.0.). Batch correction was performed with the sva (Leek et al., 2012) package and the ComBat method (Johnson et al., 2007). Control probes and those with either mean log transformed intensity values of less than 2.5 or standard deviations of less than 0.1 among all samples were removed. Probes were assigned to genes using Netaffx (Liu et al., 2003) annotations (ZebGene-1_0-st-v1.na33.3.zv9.transcript.csv); probes were not combined at the gene level, but were treated as independent assays. Differential expression statistics for pairwise and three-way (interaction terms) comparisons were generated by linear model for microarray data analysis (limma) (Smyth, 2004) using empirical Bayes shrinkage methods. Differentially expressed genes were assessed as

131

those with least a log fold expression change of 1 and an FDR (Benjamini and Hochberg, 1995) based adjusted p-value of less than 0.25. Genes annotated with a zebrafish Entrez Gene ID were used for “Ingenuity Pathway Analysis” (IPA; QIAGEN, www.qiagen.com/ingenuity).

Fetal liver preparations

Live fetal liver explants: Pregnant wild-type C57/BL6 and Ly6a-GFP mice were dissected at E11-11.5 (vaginal plug observation was E0). Embryos were removed from the uterus in PBS with 10% FCS and penicillin-streptomycin. Embryos were staged as E11-E11.5 by counting >42 somite pairs. FLs were removed from the embryo and treated as described by Boisset and colleagues (“protocol b”) (Boisset et al., 2010). Conjugated antibodies used for detection were CD31-FITC (PECAM-1, BD Biosciences, Rat antimouse, clone MEC 13.3) and c-kit-APC (CD117, BD Biosciences, Rat anti-mouse, clone 2B8).

Fixed fetal livers: Embryos were dissected as above and fixed in 2% PFA on ice for 20 minutes. Embryos were then rinsed and dehydrated in methanol. FLs were removed, mounted, cleared and imaged using the method by Yokomizo et al. (Yokomizo et al., 2012). Livers from E11.5 Ly6a-GFP mice were stained with rat anti-CD144, rabbit antiRUNX1, and chicken anti-GFP. Primary and secondary antibody details are below. Specimens were scanned using a Zeiss LSM 710 confocal microscope with a 25x oil objective. Images were acquired using multi-track sequential mode and Zeiss Zen

132

software. Pinhole was set at 1 Airy unit, steps were 1.14 !m per z-section. 3D projections were made using Volocity software.

Primary antibodies: anti-CD144 (VE-cadherin) (BD Pharmingen, Catalog# 550548, Clone: 11D4.1, Isotype: Rat IgG2a, "); anti-RUNX1 rabbit (Abcam, Catalog# ab92336, Clone # EPR3099); anti-GFP Chicken IgY (Invitrogen/Life technologies, Catalog# A10262).

Secondary antibodies: Goat Anti-Rabbit IgG (H+L) Alexa Fluor® 488 (Invitrogen/Life technologies, Catalog# A-11034); Goat Anti-Rabbit IgG (H+L) Alexa Fluor® 647 (Abcam, Catalog# 150079); Goat Anti-Chicken IgY (IgG) (H+L) Alexa Fluor® 647 (Jackson, Catalog# 103-605-155); Goat Anti-Rat IgG (H+L) Alexa Fluor® 555 (Abcam, Catalog# ab150158); Goat anti-chicken IgG (H+L) Alexa Fluor® 488 (Invitrogen/ Life Technologies, Catalog# A11039).

133

Supplemental References

Benjamini, Y., and Hochberg, Y. (1995). Controlling the false discovery rate: a practical and powerful approach to multiple testing. Journal of the Royal Statistical Society Series B (Methodological), 289-300. Bertrand, J.Y., Chi, N.C., Santoso, B., Teng, S., Stainier, D.Y.R., and Traver, D. (2010). Haematopoietic stem cells derive directly from aortic endothelium during development. Nature 464, 108-111. Boisset, J.-C., Van Cappellen, W., Andrieu-Soler, C., Galjart, N., Dzierzak, E., and Robin, C. (2010). In vivo imaging of haematopoietic cells emerging from the mouse aortic endothelium. Nature 464, 116-120. Bolstad, B.M., Irizarry, R.A., Astrand, M., and Speed, T.P. (2003). A comparison of normalization methods for high density oligonucleotide array data based on variance and bias. Bioinformatics 19, 185-193. Carvalho, B.S., and Irizarry, R.A. (2010). A framework for oligonucleotide microarray preprocessing. Bioinformatics 26, 2363-2367. Cross, L.M., Cook, M.A., Lin, S., Chen, J.N., and Rubinstein, A.L. (2003). Rapid analysis of angiogenesis drugs in a live fluorescent zebrafish assay. Arterioscler Thromb Vasc Biol 23, 911-912. Gentleman, R.C., Carey, V.J., Bates, D.M., Bolstad, B., Dettling, M., Dudoit, S., Ellis, B., Gautier, L., Ge, Y., Gentry, J., et al. (2004). Bioconductor: open software development for computational biology and bioinformatics. Genome Biol 5, R80. Glass, T.J., Lund, T.C., Patrinostro, X., Tolar, J., Bowman, T.V., Zon, L.I., and Blazar, B.R. (2011). Stromal cell-derived factor-1 and hematopoietic cell homing in an adult zebrafish model of hematopoietic cell transplantation. Blood 118, 766-774. Henriksson, J., Hench, J., Abou-Zied, A., Lüppert, M., Dethlefsen, J., Tong, Y.-G., Tang, L., Baillie, D., and Bürglin, T. (2012). Endrov – An integrated platform for image analysis and systems biology. Manuscript in preparation. Hu, Y., and Smyth, G.K. (2009). ELDA: extreme limiting dilution analysis for comparing depleted and enriched populations in stem cell and other assays. Journal of Immunological Methods 347, 70-78. Huang, H., Zhang, B., Hartenstein, P.A., Chen, J.N., and Lin, S. (2005). NXT2 is required for embryonic heart development in zebrafish. BMC Dev Biol 5, 7. Inman, G.J., Nicolas, F.J., Callahan, J.F., Harling, J.D., Gaster, L.M., Reith, A.D., Laping, N.J., and Hill, C.S. (2002). SB-431542 is a potent and specific inhibitor of transforming growth factor-beta superfamily type I activin receptor-like kinase (ALK) receptors ALK4, ALK5, and ALK7. Mol Pharmacol 62, 65-74. Johnson, W.E., Li, C., and Rabinovic, A. (2007). Adjusting batch effects in microarray expression data using empirical Bayes methods. Biostatistics 8, 118-127. Kauffmann, A., Gentleman, R., and Huber, W. (2009). arrayQualityMetrics--a bioconductor package for quality assessment of microarray data. Bioinformatics 25, 415416.

134

Kaufman, C.K., White, R.M., and Zon, L. (2009). Chemical genetic screening in the zebrafish embryo. Nature Protocols 4, 1422-1432. Kremer, J.R., Mastronarde, D.N., and McIntosh, J.R. (1996). Computer visualization of three-dimensional image data using IMOD. Journal of structural biology 116, 71-76. Kwan, K.M., Fujimoto, E., Grabher, C., Mangum, B.D., Hardy, M.E., Campbell, D.S., Parant, J.M., Yost, H.J., Kanki, J.P., and Chien, C.-B. (2007). The Tol2kit: a multisite gateway-based construction kit for Tol2 transposon transgenesis constructs. Developmental dynamics : an official publication of the American Association of Anatomists 236, 3088-3099. Leek, J.T., Johnson, W.E., Parker, H.S., Jaffe, A.E., and Storey, J.D. (2012). The sva package for removing batch effects and other unwanted variation in high-throughput experiments. Bioinformatics 28, 882-883. Lin, H.-F., Traver, D., Zhu, H., Dooley, K., Paw, B.H., Zon, L.I., and Handin, R.I. (2005). Analysis of thrombocyte development in CD41-GFP transgenic zebrafish. Blood 106, 3803-3810. Liu, G., Loraine, A.E., Shigeta, R., Cline, M., Cheng, J., Valmeekam, V., Sun, S., Kulp, D., and Siani-Rose, M.A. (2003). NetAffx: Affymetrix probesets and annotations. Nucleic Acids Res 31, 82-86. Mastronarde, D.N. (1997). Dual-axis tomography: an approach with alignment methods that preserve resolution. Journal of structural biology 120, 343-352. Meijering, E., Dzyubachyk, O., and Smal, I. (2012). Methods for cell and particle tracking. Methods Enzymol 504, 183-200. Mosimann, C., Kaufman, C.K., Li, P., Pugach, E.K., Tamplin, O.J., and Zon, L.I. (2011). Ubiquitous transgene expression and Cre-based recombination driven by the ubiquitin promoter in zebrafish. Development (Cambridge, England) 138, 169-177. North, T.E., Goessling, W., Walkley, C.R., Lengerke, C., Kopani, K.R., Lord, A.M., Weber, G.J., Bowman, T.V., Jang, I.-H., Grosser, T., et al. (2007). Prostaglandin E2 regulates vertebrate haematopoietic stem cell homeostasis. Nature 447, 1007-1011. Nottingham, W.T., Jarratt, A., Burgess, M., Speck, C.L., Cheng, J.-F., Prabhakar, S., Rubin, E.M., Li, P.-S., Sloane-Stanley, J., Kong-A-San, J., et al. (2007). Runx1-mediated hematopoietic stem-cell emergence is controlled by a Gata/Ets/SCL-regulated enhancer. Blood 110, 4188-4197. Pugach, E.K., Li, P., White, R., and Zon, L. (2009). Retro-orbital injection in adult zebrafish. Journal of visualized experiments : JoVE. Schindelin, J., Arganda-Carreras, I., Frise, E., Kaynig, V., Longair, M., Pietzsch, T., Preibisch, S., Rueden, C., Saalfeld, S., Schmid, B., et al. (2012). Fiji: an open-source platform for biological-image analysis. Nat Methods 9, 676-682. Schols, D., Struyf, S., Van Damme, J., Este, J.A., Henson, G., and De Clercq, E. (1997). Inhibition of T-tropic HIV strains by selective antagonization of the chemokine receptor CXCR4. J Exp Med 186, 1383-1388. Smyth, G.K. (2004). Linear models and empirical bayes methods for assessing differential expression in microarray experiments. Statistical applications in genetics and molecular biology 3, Article3. Stachura, D.L., and Traver, D. (2011). Cellular dissection of zebrafish hematopoiesis. Methods in cell biology 101, 75-110.

135

Thisse, C., and Thisse, B. (2008). High-resolution in situ hybridization to whole-mount zebrafish embryos. Nature Protocols 3, 59-69. 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 engraftment in zebrafish bloodless mutants. Nature Immunology 4, 1238-1246. Wan, Y., Otsuna, H., Chien, C.B., and Hansen, C. (2009). An interactive visualization tool for multi-channel confocal microscopy data in neurobiology research. IEEE Trans Vis Comput Graph 15, 1489-1496. White, R.M., Cech, J., Ratanasirintrawoot, S., Lin, C.Y., Rahl, P.B., Burke, C.J., Langdon, E., Tomlinson, M.L., Mosher, J., Kaufman, C., et al. (2011). DHODH modulates transcriptional elongation in the neural crest and melanoma. Nature 471, 518522. Yokomizo, T., Yamada-Inagawa, T., Yzaguirre, A.D., Chen, M.J., Speck, N.A., and Dzierzak, E. (2012). Whole-mount three-dimensional imaging of internally localized immunostained cells within mouse embryos. Nat Protoc 7, 421-431.

136

137

138

139

140

141

142

143

144

145

146

147

148

149

150

151

152

153

154

155

156

157

158

159

160

161

162

163

164

165

166

167

168

169

170

171

172

173