The Rac GTPase effector p21-activated kinase is essential for hematopoietic stem/progenitor cell migration and engraftment

Hematopoietic stem and progenitor cell (HSPC) engraftment is a highly orchestrated series of biological processes that result in the ability of intravenously administered cells to lodge in the bone marrow (BM) medullary cavity. Subsequent retention and proliferation in the bone marrow space leads to reconstitution of the hematopoietic system. The capacity of the HSPC to transit out of the circulation into the bone marrow, commonly referred to as BM homing, somewhat resembles the migration of leukocytes to sites of inflammation. In the case of leukocytes, this process includes an initial phase of tethering and rolling along the vessel wall followed by firm adhesion to the endothelium via integrin activation, transendothelial migration, and finally chemotaxis through the extracellular matrix to the site of inflammation.^1-3  HSPCs migrate toward the medullary space in response to stimulation of the chemokine receptor CXCR4 by binding of stromal-derived factor 1α (SDF-1α). On arrival in the medullary cavity, cell surface molecules on HSPCs interact in specific microenvironment locations termed the stem cell “niche.”^4-6  Identifying the signaling pathways that regulate and integrate these HSPC processes is necessary to improve the engraftment efficiency of HSPCs, produce more effective mobilization of these cells, and develop better therapies for the treatment of hematological diseases.

Rac proteins are members of the Rho guanosine triphosphatase (GTPase) family of proteins that have been shown to integrate multiple extracellular signals and play important regulatory roles in HSPC functions such as cell migration, adhesion, proliferation, and survival.^7,8  Although, Rac1 and Rac2 proteins have been shown to regulate distinct biological processes, simultaneous deletion of both Rac1 and Rac2 reveal overlapping functions in HSPC survival, engraftment, retention and mobilization.⁹  Whether Rac proteins use distinct or shared downstream effector proteins to carry out these functions has yet to be determined.

The best characterized Rac effector proteins are the group A p21-activated kinases (Paks). Paks are a family of serine/threonine kinases consisting of 6 isoforms divided into 2 groups: group A (Pak1, Pak2, Pak3) and group B (Pak4, Pak5, Pak6) based on homology.^10,11  Among the group A Paks, Pak1 is highly expressed in brain, muscle, and spleen cells; Pak2 is ubiquitously expressed; and Pak3 is specifically expressed in the brain.¹¹  Group A Pak proteins are characterized by 2 canonical N-terminal proline-rich motifs that mediate interactions with Src-homology 3–containing proteins, a p21-binding domain that binds activated (guanosine triphosphate [GTP]-bound) Rac or Cdc42, and a C-terminal kinase domain. Pak A proteins exist as inactive homodimers. This inactive conformation results from the association of the autoinhibitory domain of 1 Pak molecule binding to and interfering with the C-terminal kinase cleft of the second Pak molecule. Activation of Pak is achieved though Rho GTPase binding to the p21-binding domain, relieving autoinhibition through a series of conformational changes allowing the protein to adopt a catalytically active state. The group B Pak proteins bind activated Cdc42, but not Rac, and use a distinct mechanism of activation in cells.^11-14  Paks have been implicated in tumor cell proliferation and survival and are being exploited as therapeutic targets in cancer several cancers.^15,16 

Previously, we have shown that Rac proteins have both unique, as well as overlapping, HSPC functions and are activated by stromal-derived factor 1/CXCR4, SCF/c-kit and fibronectin/β1 integrins, all ligands/receptors critical for HSPC engraftment. Paks have been shown to prime Mek1 for activation by Raf1,^17,18  thus integrating activation by cytokine receptors and mediating directional sensing in response to SDF-1α in migrating T lymphocytes¹⁹  and macrophages.²⁰  Now we present evidence that Pak A proteins have multiple roles in HSPC functions.

We show that activation of Pak A proteins is required for HSPC engraftment and inhibition of Pak activity is associated with aberrant regulation of diverse cellular functions such as chemokine-induced reorganization of the actin cytoskeleton, migration, proliferation, and survival. Using a genetic approach to further define this pathway, we demonstrate that Pak2-deficient, but not Pak1-deficient cells, show a complete lack of HSPC engraftment. Thus, Pak-dependent signaling pathways appear to coordinately regulate both cytoskeletal and proliferative pathways in HSPC. Dissection of these separate pathways using individual Pak KO mice may offer new targets for modulating normal and leukemic stem cell functions.

All procedures involving mice followed Children’s Hospital Boston Institutional Animal Care and Use Committee guidelines. C57Bl/6J, B6.SJL, and non-obese diabetic/severe combined immuno-deficiency (NOD/SCID) mice were obtained from Jackson Laboratories. C57Bl/6J mice were mated with B6.SJL mice to generate wild-type (WT) (CD45.2+/CD45.1+) heterozygous donor mice. Pak1^−/− mice have been previously described,²¹  and Pak2^flox/flox conditional knock-out mice were generated by homologous recombination using standard techniques, the derivation of which will be reported in detail in a subsequent publication.

Isolation and transduction of primary lin^-Sca1⁺c-kit⁺ (LSK) were performed according to our previously published methods,²²  and also see supplemental “Methods.” Retroviral vector constructs and antibodies can be found in supplemental “Methods.”

Bone marrow transplantation experiments on transduced LSK were performed according to our previously published methods,²²  and also see supplemental “Methods.”

In vitro chemotaxis in response to stromal derived factor-1 α (SDF1α) was analyzed by time-lapse microscopy as previously described⁷  and further described in the supplemental “Methods.” Cell tracks were visualized using Velocity software (PerkinElmer, Waltham, MA); total cell displacement (eucleadian distance) was calculated using ImageJ software (National Institutes of Health, Bethesda, MD). Transwell assays were performed as previously described^7,23  and described in detail in the supplemental “Methods.”

Green fluorescent protein (GFP)+ LSK cells were layered onto fibronectin-coated coverslips and stimulated with 100 ng/mL of SDF-1α. Cells were subsequently fixed and stained according to previously described methods and are further described in the supplemental “Methods.”

For the determination of proliferation/cell-cycle LSK cells were cultured in the presence of 5-bromo-2′-deoxy-uridine (BrdU) for 1 hour then fixed, permeabilized, and stained using the APC BrdU Flow Kit (BD Biosciences, San Jose, CA), following the manufacturer’s instructions. To determine differences in apoptosis, cells were stained with Annexin V-APC (BD Biosciences) and 7AAD (Invitrogen, Carlsbad, CA).

The 1.25 × 10⁵ cells were fixed and stained according to previously published methods²⁴  using Alexa647-conjugated p-ERK (p44/42) and pAKT (Cell Signaling Technology, Danvers, MA) and the analysis is further described in supplemental “Methods.”

Datasets were compared by two-tailed t tests and P values less than 0.05 were considered statistically significant.

Pak proteins regulate cell shape, adhesion, and migration, and these processes are critical for HSC homing and engraftment. To examine the role of Pak activation in HSPC, we measured the capacity of Pak inhibitory domain (PID)-expressing LSK cells to migrate in response to an SDF-1α gradient using a Dunn chamber and time-lapse video microscopy. We have previously demonstrated the expression and subsequent phosphorylation of Pak in HSPC cells after treatment with SCF and SDF-1α.^8,25  To inhibit Pak A proteins, we expressed the PID in WT HSPC cells. PID binds to the autoinhibitory domain of group A Pak proteins and inhibits subsequent activation.^26,27  Inhibition of Pak in HSPCs via expression of the PID results in simultaneous inhibition of all three Pak A proteins reducing the confounding variable of compensation by other highly homologous family members. PID was inserted into MSCV-IRES-enhanced green fluorescent protein (3) (MIEG3) retrovirus²⁸  to yield MIEG3-PID vector and NIH/3T3 cells were transduced with either MIEG3 expressing EGFP alone or MIEG3-PID co-expressing EGFP and PID. PDGF-stimulated transduced NIH/3T3 cells expressing PID led to significant inhibition of Pak activation as measured in an in vitro kinase assay (supplemental Figure 1). Freshly isolated LSK cells from WT (CD45.1+/CD45.2+) BM were transduced with MIEG3-PID or MIEG3 as a control. PID-expressing HSPC showed abnormal directed migration in response to an SDF-1α gradient (Figure 1A and supplemental Movies 1 and 2). The total cell displacement by PID-expressing cells in 1 hour was significantly reduced compared with MIEG3 controls (76.9 ± 62.1 vs 380.3 ± 142.5 units; mean ± SEM; PID vs MEIG; P < .01) (Figure 1B). PID-GFP+ LSK cells were also found to simultaneously display the presence of 2 or more lamellipodia in distinct regions around the cell (supplemental Movies 1 and 2) in contrast to MIEG3 controls.

To address whether inhibition of Pak proteins led to a loss of cytoskeletal machinery required for cell migration, we examined HSPC chemokinesis or random cell migration. We used in vitro transwell assays in which a uniform concentration of SDF-1α was added to both the top and bottom of transwells. We found a greater than 2-fold increase in chemokinesis in PID-expressing HSPC compared with MIEG3 controls (3.0 ± 0.2% vs 1.4 ± 0.2% of input cells, respectively; mean ± SEM; P < .01) (Figure 1C). Random migration in the absence of chemokine was also increased in PID-expressing HSPC compared with MIEG3 controls (0.40 ± 0.04% vs 0.11 ± 0.01% of input cells, respectively; mean ± SEM; P < .05) (Figure 1D). To determine if these defects in directed cell migration were the result of different levels of the chemokine receptor, CXCR4, on LSK transduced with PID compared with MIEG3 controls, we performed flow cytometry for expression of CXCR4 levels and found no significant differences (supplemental Figure 2). Taken together, the decrease in directed migration and increase in chemokinesis in PID-expressing HSPC suggests that appropriate activation of Pak A is required for directional migration in HSPC.

Regulation of cellular migration is dependent on the dynamics of actin reorganization and F-actin stabilization.²⁹  To determine the role Pak proteins play in coordinated organization of the cytoskeleton, we examined the change in cell morphology in response to chemokine stimulation in PID-expressing HSPC. In contrast to MIEG3 controls, PID-expressing HSPCs demonstrated exaggerated uropod-like structures after stimulation with SDF-1α (Figure 2A), similar to structures we have previously characterized in Rac-deficient neutrophils.³⁰  We quantified disruptions in the actin cytoskeleton by observing changes in cell shape and membrane protrusions after stimulation with SDF-1α using cell surface area and total cell perimeter as a measure of cell spreading. We found significant increases in both surface area and cell perimeter of PID-expressing LSK compared with controls (1949 ± 490.8 vs 1312 ± 202.6 uM²  and 194.5 ± 36.3 vs 136.8 ± 16.6 uM; mean ± SEM; PID vs MIEG3, respectively; P < .01) (Figure 2B-C). These data, together with live imaging microscopy results, demonstrate that PID-expressing HSPC display aberrant reorganization of the actin cytoskeleton in response to chemokine-stimulated migration and suggest a role for Pak A proteins in regulating actin dynamics during directed HSPC migration.

We have observed defects in HSPC directed cell migration and cytoskeletal re-organization as a result of Pak inhibition. However Paks are complex, multi-domain signaling molecules that mediate activation of several signaling pathways.^12,14  In other cell systems, Pak proteins have been demonstrated to regulate cell proliferation and survival.^12,31  To measure differences in proliferation in PID-expressing vs control HSPC, we examined transduced and cytokine-stimulated HSPC for BrdU incorporation. We found a significant decrease in proliferation of HSPC in PID-expressing LSK compared with control transduced cells (17.7 ± 3.4 vs 36.8% ± 2.4 of cells in S-phase; mean ± SEM; P < .01) (Figure 3A). Consistent with the decrease in the percentage of cells entering S-phase of the cell cycle, there was a corresponding increase of cells in G0-G1 and G2-M phases of the cell cycle (Figure 3A). PID-expressing HSPC also demonstrated a significant increase in cells undergoing apoptosis (Annexin V+) compared with controls (35.2 ± 10.6 vs 11.1 ± 1.6%; mean ± SEM; P < .05) (Figure 3B). Taken together, these data demonstrate that Pak A proteins regulate multiple HSPC functions important for HSPC engraftment. These functions not only involve directed cell migration and organization of the actin cytoskeleton but also HSPC proliferation and survival.

Next, to determine which signaling pathways are responsible for the observed Pak-dependent HSPC functions, we examined activation of kinase pathways in HSPC in response to cytokines. PID- and MIEG3-transduced LSK cells were isolated using fluorescence-activated cell sorter, starved, and then stimulated with growth cytokines. After stimulation, cells were fixed and stained for intracellular flow analysis using fluorescent tagged phospho-specific AKT and ERK antibodies. We found cytokine-induced increase in the levels of phospho-AKT and phospho-ERK in MIEG3-transduced LSK (Figure 3C). PID-expressing HSPC showed a significant impairment of activation of AKT and ERK (1.85 vs 1.38 and 1.61 vs 1.3 fold-increase in pAKT and pERK in MIEG3 vs PID-expressing LSK, respectively; 3 independent experiments; P < .05) (Figure 3D). These data suggest that Pak regulates both AKT and MAPK pathway activation in HSC in response to cytokines.

Next, to determine if Pak proteins are critical for HSPC engraftment, we tested whether Pak inhibition in HSPC altered in vivo reconstitution of hematopoiesis in a transplant model. 1.0 × 10⁵ GFP+-transduced LSK cells were isolated and co-transplanted with 5.0 × 10⁵ WT BoyJ (CD45.1+) whole BM into lethally irradiated WT C57Bl/6J (CD45.2+) recipients. Although animals transplanted with WT cells transduced with MIEG3 showed a sustained level of engraftment up to 22.2 ± 10.7% chimerism at 24-weeks posttransplant, animals transplanted with PID-transduced cells showed only 0.22 ± 0.19% donor chimerism at 3 weeks, decreasing to 0.07 ± 0.06% at 6 weeks with no detectable engraftment of these cells from 10-weeks postinfusion and onward (Figure 4A). To verify that PID-transduced LSK retained the capacity to differentiate into multiple lineages in progenitor-derived colonies in vitro, we performed colony forming unit assays. We found no significant differences in the number or type of colony forming units from PID-transduced LSK compared with MIEG3 controls (supplemental Figure 3). To determine if cells capable of full hematopoietic reconstitution were transduced in this protocol, we assessed the lineage contribution of MIEG3- and PID-transduced cells at 24-weeks post-BMT in the peripheral blood, spleen, and BM of recipient mice. Mice transplanted with MIEG3-transduced LSK cells demonstrated donor cell contribution to erythroid, myeloid, B-lymphoid, and T-lymphoid lineages, whereas PID-transduced cells failed to contribute to measurable engraftment in any of these lineages (supplemental Figure 4). The near complete absence of PID-transduced cells in recipient mice at early and later time points suggests Pak proteins play an important role in both short- and long-term hematopoietic reconstitution.

Engraftment is a complex process involving the coordinated regulation of multiple physiological processes including cellular migration, localization to the appropriate niche in the medullary cavity, adhesion, and subsequent proliferation and differentiation.³²  To address the potential role(s) of Pak proteins in these processes, we first tested the capacity of the PID-expressing HSPC to home to the BM at 12 hours, and to be retained at 48 hours postinfusion compared with an equivalent number of MIEG3-transduced LSK cells (Figure 4B). The number of GFP+ cells that homed to the BM per hind limb was determined by flow analysis. At 12 hours postinfusion, there was a 78% reduction in PID-expressing cells that homed to the BM compared with control transduced cells. At 48 hours, the proportion of PID-GFP+ cells remaining in the BM was reduced by 94% compared with control cells. Thus, expression of PID was associated with reduced initial homing and reduced numbers of PID-transduced cells subsequently found in the BM compared with controls.

To determine if the defect in proliferation and engraftment of PID-expressing cells was related to defects in AKT or MAPK activation, we next transduced LSK with retroviral vectors expressing constitutively active (ca) forms of AKT or MEK1 (ERK) and co-expressing a truncated human-CD4 (hCD4) epitope, along with PID-GFP or MIEG3 constructs. GFP+ and hCD4+ cells were injected into lethally-irradiated C576BL/6J mice. At 3-weeks post-BMT we found that expression of caMEK1 but not caAKT was capable of partially rescuing the engraftment defect of PID-expressing HSPC (percentage of donor cells for PID + empty vector [hCD4], 3.1 ± 1.4%; PID + caMEK1, 15.8 ± 3.4%, and PID + caAKT, 0.5 ± 0.1%; mean ± SEM; P < .05 for empty vector vs caMEK1) (Figure 5A). However, by 6-weeks post-BMT, PID + caMEK1 cells were no longer significantly different than PID+ empty vector (hCD4) (supplemental Figure 5). When 250 000 sorted LSK were injected into lethally irradiated BoyJ recipients, there was also a trend toward rescue of the homing defect when PID-transduced HSPC expressing caMEK1 (GFP+/hCD4+) were compared with HSPC expressing PID alone at 12 hours postinjection (37.3 vs 52.6% of control; PID vs PID+caMEK1; N = 5-10). To determine if MEK1-induced rescue of short-term engraftment was also contributed to by increased proliferation and/or survival of PID-expressing, we performed proliferation and apoptosis assays on fluorescence-activated cell sorted cells 3 days posttransduction. Expression of caMEK1 was associated with a modest increase in proliferation in PID-expressing LSK compared with controls (PID + empty vector [hCD4] 6.1 ± 2.9% vs PID+ caMEK1; 9.5 ± 1.2% BrdU+ cells; mean ± SEM; N = 3 independent experiments; P < .05) (Figure 5B). This increase in proliferation occurred with a concomitant trend toward increased apoptosis that did not reach statistical significance (PID + empty vector [hCD4], 39.1 ± 3.7% and PID+ caMEK1 49.53 ± 18.8% Annexin V+ cells; mean ± SEM; N = 3 independent experiments; P = not significant) (Figure 5C). These results suggest that the PID-associated defect in MEK1-ERK signaling in response to cytokines was associated with reduced cell proliferation and loss of early engraftment of PID-expressing cells.

Next, because PID expression indiscriminately inhibits all Pak A proteins, we used mouse knockout lines to determine the relative role of Pak 1 and Pak 2 in the engraftment process. Pak1^−/− mice are viable and healthy,²¹  and we found no significant differences in engraftment between Pak1^−/− vs Pak1^wt/wt donor chimerism at any time point from as early as 3 weeks (35.8 ± 5.3% vs 37.4 ± 3.6% donor chimerism; mean ± SEM; N = 6; P = not significant) up to 24 weeks (50.7 ± 8.3 vs 52.1 ± 12.4; mean ± SEM; N = 10; P = not significant). These animals were not studied further. Pak2 germline knock-out mice are embryonic lethal. Therefore, we used mice homozygous for floxed Pak2 alleles (Pak2^flox/flox) for subsequent studies. The derivation and genotyping of these animals will be described in detail in a subsequent publication. Freshly isolated Pak2^fl//fl and Pak2^WT/WT LSK cells were transduced with Cre-GFP or MIEG3 control retrovirus. LSK-transduced and GFP-sorted Pak2^fl/fl cells showed near complete deletion of Pak2 sequences (supplemental Figure 6). Transduced LSK cells were assayed for migration and analyzed for cell shape changes. Similar to LSK-expressing PID, analysis of Pak2^Δ/Δ cells demonstrated excessive cell spreading and protrusions (Figure 6A–B). Sorted GFP+ cells were injected into lethally irradiated C57BL/6J recipient mice. Pak2^Δ/Δ cells showed ∼50% reduction in homing to the bone marrow compared with Pak2^WT/WT cells transduced with Cre when analyzed 12 hours after injection (100+/−13 vs 50.4+/−15%; WT vs Pak2^Δ/Δ; P < .05; N = 9) (Figure 6C). Finally, GFP+ cells were then co-transplanted with 5.0 × 10⁵ WT BoyJ (CD45.1+) WBM into sublethally irradiated (2.8 Gy) NOD/SCID recipients. The use of immunodeficient mice as recipients was necessary because the Pak2^flox/flox mice were on a mixed genetic background. Compared with Pak2^WT/WT cells transduced with Cre recombinase, Pak2^flox/flox + Cre recombinase cells failed to contribute to recipient hematopoiesis, as demonstrated by near complete absence of donor chimerism in the peripheral blood in all time points examined posttransplant (Figure 6D). To verify Pak2^Δ/Δ cells retain the ability to differentiate into multiple lineages, as well as the capacity to form progenitor-derived colonies, we performed colony forming assays and found no significant differences in the number of colonies or in the frequency of the colony type after expression of Cre-recombinase compared with Cre-transduced WT bone marrow and noted by the presence of erythroid, myeloid, and mixed colony types (supplemental Figure 7). Previously, we noted, along with others, that expression of Cre-recombinase in LSK leads to some toxicity, even in WT cells, however, Pak2-deficient LSK display a significant decrease in engraftment beyond what was seen with WT-LSK transduced with Cre-recombinase and no decrease in colony formation compared with these cells. These data strongly suggest that Pak2 is required for HSPC engraftment.

HSPC engraftment is a complex process that requires the coordination of multiple pathways regulating distinct cellular processes, such as chemotaxis, extravasation from the circulation to the BM microenvironment, lodgment to the stem cell niche and subsequent survival, proliferation, and differentiation.³²  This process requires HSPC to integrate extracellular cues from chemokines, cytokines, and integrin receptors. Alterations in signaling pathways downstream of these receptors can result in multiple HSPC abnormalities, and in some cases malignant transformation. Here we studied the potential role of group A Paks, major downstream targets of Rac, and Cdc42 Rho GTPases,¹²  in HSPC functions. Although Paks have been studied in a variety of cell lines, a direct role for Paks in HSPC biology has yet to be identified. In other model systems, Pak proteins regulate multiple cellular functions such as transcription, translation, migration, focal adhesion turnover, organization of the actin cytoskeleton, proliferation, and survival.^{10,12,19,33,34}  In the present work, we used 2 approaches – transgenic expression of the PID to inhibit all group A Paks, as well as individual gene knockouts of Pak1 and Pak2 – to establish that group A Paks are required for HSPC functions critical for engraftment including directed migration, homing and survival/proliferation and that Pak2, but not Pak1, is the critical isoform in engraftment of these cells post-transplant. In particular, we demonstrate that Pak2-deficient HSPC fail to reconstitute hematopoiesis in irradiated murine recipients. Although the analyses of the differential roles of Pak1 and Pak2 were done using different experimental systems, these data imply that Paks, and particularly Pak2, coordinate both cytoskeletal and kinase pathways involved in the complex processes of HSC engraftment. Additional analysis of the relative contributions of other group A Paks, including Pak1, to engraftment are underway.

Here we demonstrate in HSPC that Pak-inhibition leads to defective ERK and AKT activation, resulting in reduced proliferation and increased apoptosis in response to growth factors. Alterations in ERK activation likely contributes directly to the engraftment defect in Pak-deficient cells, because expression of caMEK increases proliferation of these cells in vitro and partially restores homing and engraftment of Pak-deficient cells in vivo. It is not clear whether the lack of full rescue of caMEK expression in Pak-deficient cells results from the insufficient levels of ERK/MEK1 activation to rescue the concomitant increased apoptosis that we noted in these cells, or because multiple domains of Pak, and therefore multiple downstream pathways, are required to restore full engraftment activity. Studies to resolve these issues are underway.

Transducing LSK with retrovirus containing the PID domain, we demonstrate a marked defect in directed migration of HSPC in response to SDF-1α and preserved or even enhanced actin assembly leading to excessive cell spreading and adhesion. Some functions of Pak, especially certain cytoskeleton effects are independent of Pak kinase activity.^33,35  Thus, although not well characterized in hematopoietic cells, in other cells the actin and cytoskeletal effects of Pak activation appear to be mediated by a scaffold function that is dependent on proline-rich domains and interaction with Src-homology 3-domain containing proteins such as Nck and PIX.^10,12  PID inhibition of Pak activity may leave this scaffold function intact, possibly explaining the excessive cell spreading and adhesion. In addition, phosphorylation of Pak and its regulation of actin remodeling in some cells is dependent on the presence of Nck-binding sites on Pak.³⁶  We are currently investigating the potential scaffolding role of Pak2 in HSPC functions.

The Rho GTPases are regulated by guanine exchange factors and GTPase-activating proteins (GAPs) in a cell and agonist-specific fashion and result in the activation of 1 or more effector proteins. Dissecting the complex regulation of these pathways and defining the functional interactions between GTPases and their effector proteins in specific HSPC functions remains a major challenge. We have recently demonstrated that Vav1, a Rac guanine exchange factor, is critical for HSPC perivascular homing and retention in the bone marrow.²⁵  Along with the data presented here, we have thus defined a Rac-signaling axis that includes the chemokine receptor CXCR4, c-kit, and β1 integrins, the guanine exchange factors Vav1, Rac/Cdc42, and Pak2,^7-9  which is essential for HSPC engraftment and retention in the hematopoietic microenvironment. A more precise understanding of the roles of the individual components and their regulation of this pathway that are critical for engraftment and sustained hematopoiesis, will likely lead to the identification of new targets for therapeutic intervention and manipulation.

The authors thank members of our laboratory and the laboratories of Scott Armstrong and Ben Ebert for helpful discussions and advice. We thank Harvey Lodish and Jing Zhang for the constitutively active MEK1, AKT, and MICD4 constructs.

This work was supported by grants from the National Institutes of Health 5R01DK062757-11 (D.A.W.), R01CA142928-3, and R01CA117884-2 (J.C.), 5 T32 HL 007574-29, and 5 T32 HL 066987-08 (A.M.D).

Contribution: A.M.D., M.D.M., J.C., and D.A.W. designed research. A.M.D., R.K., S.D.V., P.N.G.R., M.K.M., R.M., C.E.H., M.R., and S.W.L. performed research. A.M.D., M.D.M., and D.A.W. analyzed and interpreted data. R.M. and J.C. contributed new reagents and provided advice. A.M.D., J.C., and D.A.W. wrote the manuscript. All the authors reviewed versions of the manuscript and approved the final version.

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

Correspondence: David A. Williams, Division of Hematology/Oncology, Children’s Hospital Boston, 300 Longwood Ave, Karp Family Research Laboratories 08125.3, Boston, MA 02115; e-mail: dawilliams@childrens.harvard.edu.